1
0
mirror of https://passt.top/passt synced 2024-11-05 20:31:11 +00:00
passt/tap.c

1006 lines
23 KiB
C
Raw Normal View History

passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
// SPDX-License-Identifier: AGPL-3.0-or-later
/* PASST - Plug A Simple Socket Transport
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
* for qemu/UNIX domain socket mode
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
*
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
* PASTA - Pack A Subtle Tap Abstraction
* for network namespace/tap device mode
*
* tap.c - Functions to communicate with guest- or namespace-facing interface
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
*
* Copyright (c) 2020-2021 Red Hat GmbH
* Author: Stefano Brivio <sbrivio@redhat.com>
*
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
*/
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
#include <sched.h>
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
#include <stdio.h>
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
#include <errno.h>
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
#include <limits.h>
#include <string.h>
#include <net/ethernet.h>
#include <net/if.h>
#include <net/if_arp.h>
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
#include <netinet/in.h>
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
#include <arpa/inet.h>
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
#include <stdint.h>
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
#include <sys/epoll.h>
#include <sys/ioctl.h>
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <sys/uio.h>
#include <stdlib.h>
#include <unistd.h>
#include <netinet/ip.h>
#include <netinet/tcp.h>
#include <netinet/udp.h>
#include <netinet/ip_icmp.h>
#include <netinet/if_ether.h>
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
#include <linux/if_tun.h>
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
#include <linux/icmpv6.h>
#include "checksum.h"
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
#include "util.h"
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
#include "passt.h"
#include "arp.h"
#include "dhcp.h"
#include "ndp.h"
#include "dhcpv6.h"
#include "pcap.h"
#include "netlink.h"
#include "pasta.h"
#include "packet.h"
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
/* IPv4 (plus ARP) and IPv6 message batches from tap/guest to IP handlers */
static PACKET_POOL_NOINIT(pool_tap4, TAP_MSGS, pkt_buf);
static PACKET_POOL_NOINIT(pool_tap6, TAP_MSGS, pkt_buf);
#define TAP_SEQS 128 /* Different L4 tuples in one batch */
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
/**
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
* tap_send() - Send frame, with qemu socket header if needed
* @c: Execution context
* @data: Packet buffer
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
* @len: Total L2 packet length
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
* @vnet_pre: Buffer has four-byte headroom
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
*
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
* Return: return code from send() or write()
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
*/
int tap_send(const struct ctx *c, const void *data, size_t len, int vnet_pre)
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
{
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (vnet_pre)
pcap((char *)data + 4, len);
else
pcap(data, len);
if (c->mode == MODE_PASST) {
int flags = MSG_NOSIGNAL | MSG_DONTWAIT;
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (vnet_pre) {
*((uint32_t *)data) = htonl(len);
len += 4;
} else {
uint32_t vnet_len = htonl(len);
if (send(c->fd_tap, &vnet_len, 4, flags) < 0)
return -1;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
}
return send(c->fd_tap, data, len, flags);
}
return write(c->fd_tap, (char *)data + (vnet_pre ? 4 : 0), len);
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
}
/**
* tap_ip_send() - Send IP packet, with L2 headers, calculating L3/L4 checksums
* @c: Execution context
* @src: IPv6 source address, IPv4-mapped for IPv4 sources
* @proto: L4 protocol number
* @in: Payload
* @len: L4 payload length
* @flow: Flow label for TCP over IPv6
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
*/
void tap_ip_send(const struct ctx *c, const struct in6_addr *src, uint8_t proto,
const char *in, size_t len, uint32_t flow)
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
{
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
char buf[USHRT_MAX];
char *pkt = buf + 4;
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
struct ethhdr *eh;
eh = (struct ethhdr *)pkt;
/* TODO: ARP table lookup */
memcpy(eh->h_dest, c->mac_guest, ETH_ALEN);
memcpy(eh->h_source, c->mac, ETH_ALEN);
if (IN6_IS_ADDR_V4MAPPED(src)) {
struct iphdr *iph = (struct iphdr *)(eh + 1);
char *data = (char *)(iph + 1);
eh->h_proto = ntohs(ETH_P_IP);
iph->version = 4;
iph->ihl = 5;
iph->tos = 0;
iph->tot_len = htons(len + 20);
iph->id = 0;
iph->frag_off = 0;
iph->ttl = 255;
iph->protocol = proto;
udp: Connection tracking for ephemeral, local ports, and related fixes As we support UDP forwarding for packets that are sent to local ports, we actually need some kind of connection tracking for UDP. While at it, this commit introduces a number of vaguely related fixes for issues observed while trying this out. In detail: - implement an explicit, albeit minimalistic, connection tracking for UDP, to allow usage of ephemeral ports by the guest and by the host at the same time, by binding them dynamically as needed, and to allow mapping address changes for packets with a loopback address as destination - set the guest MAC address whenever we receive a packet from tap instead of waiting for an ARP request, and set it to broadcast on start, otherwise DHCPv6 might not work if all DHCPv6 requests time out before the guest starts talking IPv4 - split context IPv6 address into address we assign, global or site address seen on tap, and link-local address seen on tap, and make sure we use the addresses we've seen as destination (link-local choice depends on source address). Similarly, for IPv4, split into address we assign and address we observe, and use the address we observe as destination - introduce a clock_gettime() syscall right after epoll_wait() wakes up, so that we can remove all the other ones and pass the current timestamp to tap and socket handlers -- this is additionally needed by UDP to time out bindings to ephemeral ports and mappings between loopback address and a local address - rename sock_l4_add() to sock_l4(), no semantic changes intended - include <arpa/inet.h> in passt.c before kernel headers so that we can use <netinet/in.h> macros to check IPv6 address types, and remove a duplicate <linux/ip.h> inclusion Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-04-29 14:59:20 +00:00
iph->daddr = c->addr4_seen;
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
memcpy(&iph->saddr, &src->s6_addr[12], 4);
iph->check = 0;
iph->check = csum_unaligned(iph, (size_t)iph->ihl * 4, 0);
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
memcpy(data, in, len);
if (iph->protocol == IPPROTO_TCP) {
csum_tcp4(iph);
} else if (iph->protocol == IPPROTO_UDP) {
struct udphdr *uh = (struct udphdr *)(iph + 1);
uh->check = 0;
} else if (iph->protocol == IPPROTO_ICMP) {
struct icmphdr *ih = (struct icmphdr *)(iph + 1);
ih->checksum = 0;
ih->checksum = csum_unaligned(ih, len, 0);
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
}
if (tap_send(c, buf, len + sizeof(*iph) + sizeof(*eh), 1) < 0)
debug("tap: failed to send %lu bytes (IPv4)", len);
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
} else {
struct ipv6hdr *ip6h = (struct ipv6hdr *)(eh + 1);
char *data = (char *)(ip6h + 1);
eh->h_proto = ntohs(ETH_P_IPV6);
memset(ip6h->flow_lbl, 0, 3);
ip6h->payload_len = htons(len);
ip6h->priority = 0;
ip6h->saddr = *src;
udp: Connection tracking for ephemeral, local ports, and related fixes As we support UDP forwarding for packets that are sent to local ports, we actually need some kind of connection tracking for UDP. While at it, this commit introduces a number of vaguely related fixes for issues observed while trying this out. In detail: - implement an explicit, albeit minimalistic, connection tracking for UDP, to allow usage of ephemeral ports by the guest and by the host at the same time, by binding them dynamically as needed, and to allow mapping address changes for packets with a loopback address as destination - set the guest MAC address whenever we receive a packet from tap instead of waiting for an ARP request, and set it to broadcast on start, otherwise DHCPv6 might not work if all DHCPv6 requests time out before the guest starts talking IPv4 - split context IPv6 address into address we assign, global or site address seen on tap, and link-local address seen on tap, and make sure we use the addresses we've seen as destination (link-local choice depends on source address). Similarly, for IPv4, split into address we assign and address we observe, and use the address we observe as destination - introduce a clock_gettime() syscall right after epoll_wait() wakes up, so that we can remove all the other ones and pass the current timestamp to tap and socket handlers -- this is additionally needed by UDP to time out bindings to ephemeral ports and mappings between loopback address and a local address - rename sock_l4_add() to sock_l4(), no semantic changes intended - include <arpa/inet.h> in passt.c before kernel headers so that we can use <netinet/in.h> macros to check IPv6 address types, and remove a duplicate <linux/ip.h> inclusion Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-04-29 14:59:20 +00:00
if (IN6_IS_ADDR_LINKLOCAL(src))
ip6h->daddr = c->addr6_ll_seen;
else
ip6h->daddr = c->addr6_seen;
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
memcpy(data, in, len);
ip6h->hop_limit = proto;
ip6h->version = 0;
ip6h->nexthdr = 0;
if (proto == IPPROTO_TCP) {
struct tcphdr *th = (struct tcphdr *)(ip6h + 1);
th->check = 0;
th->check = csum_unaligned(ip6h, len + sizeof(*ip6h),
0);
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
} else if (proto == IPPROTO_UDP) {
struct udphdr *uh = (struct udphdr *)(ip6h + 1);
uh->check = 0;
uh->check = csum_unaligned(ip6h, len + sizeof(*ip6h),
0);
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
} else if (proto == IPPROTO_ICMPV6) {
struct icmp6hdr *ih = (struct icmp6hdr *)(ip6h + 1);
ih->icmp6_cksum = 0;
ih->icmp6_cksum = csum_unaligned(ip6h,
len + sizeof(*ip6h),
0);
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
}
ip6h->version = 6;
ip6h->nexthdr = proto;
ip6h->hop_limit = 255;
if (flow) {
ip6h->flow_lbl[0] = (flow >> 16) & 0xf;
ip6h->flow_lbl[1] = (flow >> 8) & 0xff;
ip6h->flow_lbl[2] = (flow >> 0) & 0xff;
}
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
if (tap_send(c, buf, len + sizeof(*ip6h) + sizeof(*eh), 1) < 1)
debug("tap: failed to send %lu bytes (IPv6)", len);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
}
}
PACKET_POOL_DECL(pool_l4, UIO_MAXIOV, pkt_buf);
/**
* struct l4_seq4_t - Message sequence for one protocol handler call, IPv4
* @msgs: Count of messages in sequence
* @protocol: Protocol number
* @source: Source port
* @dest: Destination port
* @saddr: Source address
* @daddr: Destination address
* @msg: Array of messages that can be handled in a single call
*/
static struct tap4_l4_t {
uint8_t protocol;
uint16_t source;
uint16_t dest;
uint32_t saddr;
uint32_t daddr;
struct pool_l4_t p;
} tap4_l4[TAP_SEQS /* Arbitrary: TAP_MSGS in theory, so limit in users */];
/**
* struct l4_seq6_t - Message sequence for one protocol handler call, IPv6
* @msgs: Count of messages in sequence
* @protocol: Protocol number
* @source: Source port
* @dest: Destination port
* @saddr: Source address
* @daddr: Destination address
* @msg: Array of messages that can be handled in a single call
*/
static struct tap6_l4_t {
uint8_t protocol;
uint16_t source;
uint16_t dest;
struct in6_addr saddr;
struct in6_addr daddr;
struct pool_l4_t p;
} tap6_l4[TAP_SEQS /* Arbitrary: TAP_MSGS in theory, so limit in users */];
/**
* tap_packet_debug() - Print debug message for packet(s) from guest/tap
* @iph: IPv4 header, can be NULL
* @ip6h: IPv6 header, can be NULL
* @seq4: Pointer to @struct tap_l4_seq4, can be NULL
* @proto6: IPv6 protocol, for IPv6
* @seq6: Pointer to @struct tap_l4_seq6, can be NULL
* @count: Count of packets in this sequence
*/
static void tap_packet_debug(const struct iphdr *iph,
const struct ipv6hdr *ip6h,
const struct tap4_l4_t *seq4, uint8_t proto6,
const struct tap6_l4_t *seq6, int count)
{
char buf6s[INET6_ADDRSTRLEN], buf6d[INET6_ADDRSTRLEN];
char buf4s[INET_ADDRSTRLEN], buf4d[INET_ADDRSTRLEN];
uint8_t proto = 0;
if (iph || seq4) {
inet_ntop(AF_INET, iph ? &iph->saddr : &seq4->saddr,
buf4s, sizeof(buf4s));
inet_ntop(AF_INET, iph ? &iph->daddr : &seq4->daddr,
buf4d, sizeof(buf4d));
if (iph)
proto = iph->protocol;
else if (seq4)
proto = seq4->protocol;
} else {
inet_ntop(AF_INET6, ip6h ? &ip6h->saddr : &seq6->saddr,
buf6s, sizeof(buf6s));
inet_ntop(AF_INET6, ip6h ? &ip6h->daddr : &seq6->daddr,
buf6d, sizeof(buf6d));
proto = proto6;
}
if (proto == IPPROTO_TCP || proto == IPPROTO_UDP) {
trace("tap: protocol %i, %s%s%s:%i -> %s%s%s:%i (%i packet%s)",
proto,
seq4 ? "" : "[", seq4 ? buf4s : buf6s, seq4 ? "" : "]",
ntohs(seq4 ? seq4->source : seq6->source),
seq4 ? "" : "[", seq4 ? buf4d : buf6d, seq4 ? "" : "]",
ntohs(seq4 ? seq4->dest : seq6->dest),
count, count == 1 ? "" : "s");
} else {
trace("tap: protocol %i, %s -> %s (%i packet%s)",
proto, iph ? buf4s : buf6s, iph ? buf4d : buf6d,
count, count == 1 ? "" : "s");
}
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
/**
* tap4_handler() - IPv4 and ARP packet handler for tap file descriptor
* @c: Execution context
* @in: Ingress packet pool, packets with Ethernet headers
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
* @now: Current timestamp
*
* Return: count of packets consumed by handlers
*/
static int tap4_handler(struct ctx *c, const struct pool *in,
const struct timespec *now)
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
{
unsigned int i, j, seq_count;
struct tap4_l4_t *seq;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (!c->v4 || !in->count)
return in->count;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
i = 0;
resume:
for (seq_count = 0, seq = NULL; i < in->count; i++) {
size_t l2_len, l3_len, hlen, l4_len;
struct ethhdr *eh;
struct iphdr *iph;
struct udphdr *uh;
char *l4h;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
packet_get(in, i, 0, 0, &l2_len);
eh = packet_get(in, i, 0, sizeof(*eh), &l3_len);
if (!eh)
continue;
if (ntohs(eh->h_proto) == ETH_P_ARP) {
PACKET_POOL_P(pkt, 1, in->buf, sizeof(pkt_buf));
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
packet_add(pkt, l2_len, (char *)eh);
arp(c, pkt);
continue;
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
iph = packet_get(in, i, sizeof(*eh), sizeof(*iph), NULL);
if (!iph)
continue;
hlen = iph->ihl * 4UL;
if (hlen < sizeof(*iph) || htons(iph->tot_len) != l3_len ||
hlen > l3_len)
continue;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
l4_len = l3_len - hlen;
if (iph->saddr && c->addr4_seen != iph->saddr) {
c->addr4_seen = iph->saddr;
proto_update_l2_buf(NULL, NULL, &c->addr4_seen);
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
l4h = packet_get(in, i, sizeof(*eh) + hlen, l4_len, NULL);
if (!l4h)
continue;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (iph->protocol == IPPROTO_ICMP) {
PACKET_POOL_P(pkt, 1, in->buf, sizeof(pkt_buf));
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (c->no_icmp)
continue;
packet_add(pkt, l4_len, l4h);
icmp_tap_handler(c, AF_INET, &iph->daddr, pkt, now);
continue;
}
uh = packet_get(in, i, sizeof(*eh) + hlen, sizeof(*uh), NULL);
if (!uh)
continue;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (iph->protocol == IPPROTO_UDP) {
PACKET_POOL_P(pkt, 1, in->buf, sizeof(pkt_buf));
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
packet_add(pkt, l2_len, (char *)eh);
if (dhcp(c, pkt))
continue;
}
if (iph->protocol != IPPROTO_TCP &&
iph->protocol != IPPROTO_UDP) {
tap_packet_debug(iph, NULL, NULL, 0, NULL, 1);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
continue;
}
#define L4_MATCH(iph, uh, seq) \
(seq->protocol == iph->protocol && \
seq->source == uh->source && seq->dest == uh->dest && \
seq->saddr == iph->saddr && seq->daddr == iph->daddr)
#define L4_SET(iph, uh, seq) \
do { \
seq->protocol = iph->protocol; \
seq->source = uh->source; \
seq->dest = uh->dest; \
seq->saddr = iph->saddr; \
seq->daddr = iph->daddr; \
} while (0)
if (seq && L4_MATCH(iph, uh, seq) && seq->p.count < TAP_SEQS)
goto append;
for (seq = tap4_l4 + seq_count - 1; seq >= tap4_l4; seq--) {
if (L4_MATCH(iph, uh, seq)) {
if (seq->p.count >= TAP_SEQS)
seq = NULL;
break;
}
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (!seq || seq < tap4_l4) {
seq = tap4_l4 + seq_count++;
L4_SET(iph, uh, seq);
pool_flush((struct pool *)&seq->p);
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
#undef L4_MATCH
#undef L4_SET
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
append:
packet_add((struct pool *)&seq->p, l4_len, l4h);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (seq_count == TAP_SEQS)
break; /* Resume after flushing if i < count */
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
}
for (j = 0, seq = tap4_l4; j < seq_count; j++, seq++) {
struct pool *p = (struct pool *)&seq->p;
uint32_t *da = &seq->daddr;
size_t n = p->count;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
tap_packet_debug(NULL, NULL, seq, 0, NULL, n);
if (seq->protocol == IPPROTO_TCP) {
if (c->no_tcp)
continue;
while ((n -= tcp_tap_handler(c, AF_INET, da, p, now)));
} else if (seq->protocol == IPPROTO_UDP) {
if (c->no_udp)
continue;
while ((n -= udp_tap_handler(c, AF_INET, da, p, now)));
}
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (i < in->count)
goto resume;
return in->count;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
}
/**
* tap6_handler() - IPv6 packet handler for tap file descriptor
* @c: Execution context
* @in: Ingress packet pool, packets with Ethernet headers
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
* @now: Current timestamp
*
* Return: count of packets consumed by handlers
*/
static int tap6_handler(struct ctx *c, const struct pool *in,
const struct timespec *now)
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
{
unsigned int i, j, seq_count = 0;
struct tap6_l4_t *seq;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (!c->v6 || !in->count)
return in->count;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
i = 0;
resume:
for (seq_count = 0, seq = NULL; i < in->count; i++) {
size_t l4_len, plen, check;
struct in6_addr *saddr, *daddr;
struct ipv6hdr *ip6h;
struct ethhdr *eh;
struct udphdr *uh;
uint8_t proto;
char *l4h;
eh = packet_get(in, i, 0, sizeof(*eh), NULL);
if (!eh)
continue;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
ip6h = packet_get(in, i, sizeof(*eh), sizeof(*ip6h), &check);
if (!ip6h)
continue;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
saddr = &ip6h->saddr;
daddr = &ip6h->daddr;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
plen = ntohs(ip6h->payload_len);
if (plen != check)
continue;
if (!(l4h = ipv6_l4hdr(in, i, sizeof(*eh), &proto, &l4_len)))
continue;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (IN6_IS_ADDR_LINKLOCAL(saddr)) {
c->addr6_ll_seen = *saddr;
if (IN6_IS_ADDR_UNSPECIFIED(&c->addr6_seen)) {
c->addr6_seen = *saddr;
}
} else {
c->addr6_seen = *saddr;
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (proto == IPPROTO_ICMPV6) {
PACKET_POOL_P(pkt, 1, in->buf, sizeof(pkt_buf));
if (c->no_icmp)
continue;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (l4_len < sizeof(struct icmp6hdr))
continue;
if (ndp(c, (struct icmp6hdr *)l4h, eh->h_source, saddr))
continue;
tap_packet_debug(NULL, ip6h, NULL, proto, NULL, 1);
packet_add(pkt, l4_len, l4h);
icmp_tap_handler(c, AF_INET6, daddr, pkt, now);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
continue;
}
if (l4_len < sizeof(*uh))
continue;
uh = (struct udphdr *)l4h;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (proto == IPPROTO_UDP) {
PACKET_POOL_P(pkt, 1, in->buf, sizeof(pkt_buf));
packet_add(pkt, l4_len, l4h);
if (dhcpv6(c, pkt, saddr, daddr))
continue;
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
*saddr = c->addr6;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (proto != IPPROTO_TCP && proto != IPPROTO_UDP) {
tap_packet_debug(NULL, ip6h, NULL, proto, NULL, 1);
continue;
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
#define L4_MATCH(ip6h, proto, uh, seq) \
(seq->protocol == proto && \
seq->source == uh->source && seq->dest == uh->dest && \
IN6_ARE_ADDR_EQUAL(&seq->saddr, saddr) && \
IN6_ARE_ADDR_EQUAL(&seq->daddr, daddr))
#define L4_SET(ip6h, proto, uh, seq) \
do { \
seq->protocol = proto; \
seq->source = uh->source; \
seq->dest = uh->dest; \
seq->saddr = *saddr; \
seq->daddr = *daddr; \
} while (0)
if (seq && L4_MATCH(ip6h, proto, uh, seq) &&
seq->p.count < TAP_SEQS)
goto append;
for (seq = tap6_l4 + seq_count - 1; seq >= tap6_l4; seq--) {
if (L4_MATCH(ip6h, proto, uh, seq)) {
if (seq->p.count >= TAP_SEQS)
seq = NULL;
break;
}
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (!seq || seq < tap6_l4) {
seq = tap6_l4 + seq_count++;
L4_SET(ip6h, proto, uh, seq);
pool_flush((struct pool *)&seq->p);
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
#undef L4_MATCH
#undef L4_SET
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
append:
packet_add((struct pool *)&seq->p, l4_len, l4h);
if (seq_count == TAP_SEQS)
break; /* Resume after flushing if i < count */
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
for (j = 0, seq = tap6_l4; j < seq_count; j++, seq++) {
struct pool *p = (struct pool *)&seq->p;
struct in6_addr *da = &seq->daddr;
size_t n = p->count;
tap_packet_debug(NULL, NULL, NULL, seq->protocol, seq, n);
if (seq->protocol == IPPROTO_TCP) {
if (c->no_tcp)
continue;
while ((n -= tcp_tap_handler(c, AF_INET6, da, p, now)));
} else if (seq->protocol == IPPROTO_UDP) {
if (c->no_udp)
continue;
while ((n -= udp_tap_handler(c, AF_INET6, da, p, now)));
}
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (i < in->count)
goto resume;
return in->count;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
}
/**
* tap_handler_passt() - Packet handler for AF_UNIX file descriptor
* @c: Execution context
* @now: Current timestamp
*
* Return: -ECONNRESET on receive error, 0 otherwise
*/
static int tap_handler_passt(struct ctx *c, const struct timespec *now)
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
{
struct ethhdr *eh;
ssize_t n, rem;
char *p;
redo:
p = pkt_buf;
rem = 0;
pool_flush(pool_tap4);
pool_flush(pool_tap6);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
n = recv(c->fd_tap, p, TAP_BUF_FILL, MSG_DONTWAIT);
if (n < 0) {
if (errno == EINTR || errno == EAGAIN || errno == EWOULDBLOCK)
return 0;
epoll_ctl(c->epollfd, EPOLL_CTL_DEL, c->fd_tap, NULL);
close(c->fd_tap);
return -ECONNRESET;
}
while (n > (ssize_t)sizeof(uint32_t)) {
ssize_t len = ntohl(*(uint32_t *)p);
p += sizeof(uint32_t);
n -= sizeof(uint32_t);
/* At most one packet might not fit in a single read, and this
* needs to be blocking.
*/
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (len > n) {
rem = recv(c->fd_tap, p + n, len - n, 0);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if ((n += rem) != len)
return 0;
}
/* Complete the partial read above before discarding a malformed
* frame, otherwise the stream will be inconsistent.
*/
if (len < (ssize_t)sizeof(*eh) || len > (ssize_t)ETH_MAX_MTU)
goto next;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
pcap(p, len);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
eh = (struct ethhdr *)p;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (memcmp(c->mac_guest, eh->h_source, ETH_ALEN)) {
memcpy(c->mac_guest, eh->h_source, ETH_ALEN);
proto_update_l2_buf(c->mac_guest, NULL, NULL);
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
switch (ntohs(eh->h_proto)) {
case ETH_P_ARP:
case ETH_P_IP:
packet_add(pool_tap4, len, p);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
break;
case ETH_P_IPV6:
packet_add(pool_tap6, len, p);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
break;
default:
break;
}
next:
p += len;
n -= len;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
}
tap4_handler(c, pool_tap4, now);
tap6_handler(c, pool_tap6, now);
/* We can't use EPOLLET otherwise. */
if (rem)
goto redo;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
return 0;
}
/**
* tap_handler_pasta() - Packet handler for tuntap file descriptor
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
* @c: Execution context
* @now: Current timestamp
*
* Return: -ECONNRESET on receive error, 0 otherwise
*/
static int tap_handler_pasta(struct ctx *c, const struct timespec *now)
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
{
ssize_t n, len;
int ret;
redo:
n = 0;
pool_flush(pool_tap4);
pool_flush(pool_tap6);
restart:
while ((len = read(c->fd_tap, pkt_buf + n, TAP_BUF_BYTES - n)) > 0) {
struct ethhdr *eh = (struct ethhdr *)(pkt_buf + n);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (len < (ssize_t)sizeof(*eh) || len > (ssize_t)ETH_MAX_MTU) {
n += len;
continue;
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
pcap(pkt_buf + n, len);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (memcmp(c->mac_guest, eh->h_source, ETH_ALEN)) {
memcpy(c->mac_guest, eh->h_source, ETH_ALEN);
proto_update_l2_buf(c->mac_guest, NULL, NULL);
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
switch (ntohs(eh->h_proto)) {
case ETH_P_ARP:
case ETH_P_IP:
packet_add(pool_tap4, len, pkt_buf + n);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
break;
case ETH_P_IPV6:
packet_add(pool_tap6, len, pkt_buf + n);
break;
default:
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
break;
}
if ((n += len) == TAP_BUF_BYTES)
break;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
}
if (len < 0 && errno == EINTR)
goto restart;
ret = errno;
tap4_handler(c, pool_tap4, now);
tap6_handler(c, pool_tap6, now);
if (len > 0 || ret == EAGAIN)
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
return 0;
if (n == TAP_BUF_BYTES)
goto redo;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
epoll_ctl(c->epollfd, EPOLL_CTL_DEL, c->fd_tap, NULL);
close(c->fd_tap);
return -ECONNRESET;
}
/**
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
* tap_sock_unix_init() - Create and bind AF_UNIX socket, listen for connection
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
* @c: Execution context
*/
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
static void tap_sock_unix_init(struct ctx *c)
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
{
int fd = socket(AF_UNIX, SOCK_STREAM, 0), ex;
struct epoll_event ev = { 0 };
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
struct sockaddr_un addr = {
.sun_family = AF_UNIX,
};
int i, ret;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (fd < 0) {
perror("UNIX socket");
exit(EXIT_FAILURE);
}
for (i = 1; i < UNIX_SOCK_MAX; i++) {
char *path = addr.sun_path;
if (*c->sock_path)
memcpy(path, c->sock_path, UNIX_PATH_MAX);
else
snprintf(path, UNIX_PATH_MAX - 1, UNIX_SOCK_PATH, i);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
ex = socket(AF_UNIX, SOCK_STREAM | SOCK_NONBLOCK, 0);
if (ex < 0) {
perror("UNIX domain socket check");
exit(EXIT_FAILURE);
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
ret = connect(ex, (const struct sockaddr *)&addr, sizeof(addr));
if (!ret || (errno != ENOENT && errno != ECONNREFUSED)) {
if (*c->sock_path) {
err("Socket path %s already in use", path);
exit(EXIT_FAILURE);
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
close(ex);
continue;
}
close(ex);
unlink(path);
if (!bind(fd, (const struct sockaddr *)&addr, sizeof(addr)) ||
*c->sock_path)
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
break;
}
if (i == UNIX_SOCK_MAX) {
perror("UNIX socket bind");
exit(EXIT_FAILURE);
}
info("UNIX domain socket bound at %s\n", addr.sun_path);
listen(fd, 0);
ev.data.fd = c->fd_tap_listen = fd;
ev.events = EPOLLIN | EPOLLET | EPOLLRDHUP;
epoll_ctl(c->epollfd, EPOLL_CTL_ADD, c->fd_tap_listen, &ev);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
info("You can now start qrap:");
info(" ./qrap 5 kvm ... -net socket,fd=5 -net nic,model=virtio");
info("or directly qemu, patched with:");
info(" qemu/0001-net-Allow-also-UNIX-domain-sockets-to-be-used-as-net.patch");
info("as follows:");
info(" kvm ... -net socket,connect=%s -net nic,model=virtio",
addr.sun_path);
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
/**
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
* tap_sock_unix_new() - Handle new connection on listening socket
* @c: Execution context
*/
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
static void tap_sock_unix_new(struct ctx *c)
{
struct epoll_event ev = { 0 };
int v = INT_MAX / 2;
struct ucred ucred;
socklen_t len;
len = sizeof(ucred);
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
/* Another client is already connected: accept and close right away. */
if (c->fd_tap != -1) {
int discard = accept4(c->fd_tap_listen, NULL, NULL,
SOCK_NONBLOCK);
if (!getsockopt(discard, SOL_SOCKET, SO_PEERCRED, &ucred, &len))
info("discarding connection from PID %i", ucred.pid);
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
if (discard != -1)
close(discard);
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
return;
}
c->fd_tap = accept4(c->fd_tap_listen, NULL, NULL, 0);
if (!getsockopt(c->fd_tap, SOL_SOCKET, SO_PEERCRED, &ucred, &len))
info("accepted connection from PID %i", ucred.pid);
if (!c->low_rmem &&
setsockopt(c->fd_tap, SOL_SOCKET, SO_RCVBUF, &v, sizeof(v)))
trace("tap: failed to set SO_RCVBUF to %i", v);
if (!c->low_wmem &&
setsockopt(c->fd_tap, SOL_SOCKET, SO_SNDBUF, &v, sizeof(v)))
trace("tap: failed to set SO_SNDBUF to %i", v);
ev.data.fd = c->fd_tap;
ev.events = EPOLLIN | EPOLLET | EPOLLRDHUP;
epoll_ctl(c->epollfd, EPOLL_CTL_ADD, c->fd_tap, &ev);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
}
static int tun_ns_fd = -1;
/**
* tap_ns_tun() - Get tuntap fd in namespace
* @c: Execution context
*
* Return: 0
*
* #syscalls:pasta ioctl openat
*/
static int tap_ns_tun(void *arg)
{
struct ifreq ifr = { .ifr_flags = IFF_TAP | IFF_NO_PI };
int flags = O_RDWR | O_NONBLOCK | O_CLOEXEC;
struct ctx *c = (struct ctx *)arg;
memcpy(ifr.ifr_name, c->pasta_ifn, IFNAMSIZ);
if (ns_enter(c) ||
(tun_ns_fd = open("/dev/net/tun", flags)) < 0 ||
ioctl(tun_ns_fd, TUNSETIFF, &ifr) ||
!(c->pasta_ifi = if_nametoindex(c->pasta_ifn))) {
if (tun_ns_fd != -1)
close(tun_ns_fd);
tun_ns_fd = -1;
}
return 0;
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
/**
* tap_sock_init_tun() - Set up tuntap file descriptor
* @c: Execution context
*/
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
static void tap_sock_tun_init(struct ctx *c)
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
{
struct epoll_event ev = { 0 };
NS_CALL(tap_ns_tun, c);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (tun_ns_fd == -1) {
err("Failed to open tun socket in namespace");
exit(EXIT_FAILURE);
}
pasta_ns_conf(c);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
c->fd_tap = tun_ns_fd;
ev.data.fd = c->fd_tap;
ev.events = EPOLLIN | EPOLLRDHUP;
epoll_ctl(c->epollfd, EPOLL_CTL_ADD, c->fd_tap, &ev);
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
}
/**
* tap_sock_init() - Create and set up AF_UNIX socket or tuntap file descriptor
* @c: Execution context
*/
void tap_sock_init(struct ctx *c)
{
size_t sz = sizeof(pkt_buf);
int i;
pool_tap4_storage = PACKET_INIT(pool_tap4, TAP_MSGS, pkt_buf, sz);
pool_tap6_storage = PACKET_INIT(pool_tap6, TAP_MSGS, pkt_buf, sz);
for (i = 0; i < TAP_SEQS; i++) {
tap4_l4[i].p = PACKET_INIT(pool_l4, TAP_SEQS, pkt_buf, sz);
tap6_l4[i].p = PACKET_INIT(pool_l4, TAP_SEQS, pkt_buf, sz);
}
if (c->fd_tap != -1) {
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
epoll_ctl(c->epollfd, EPOLL_CTL_DEL, c->fd_tap, NULL);
close(c->fd_tap);
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
c->fd_tap = -1;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
}
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
if (c->mode == MODE_PASST) {
if (c->fd_tap_listen == -1)
tap_sock_unix_init(c);
} else {
tap_sock_tun_init(c);
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
}
/**
* tap_handler() - Packet handler for AF_UNIX or tuntap file descriptor
* @c: Execution context
* @fd: File descriptor where event occurred
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
* @events: epoll events
* @now: Current timestamp, can be NULL on EPOLLERR
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
*/
void tap_handler(struct ctx *c, int fd, uint32_t events,
const struct timespec *now)
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
{
if (fd == c->fd_tap_listen && events == EPOLLIN) {
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
tap_sock_unix_new(c);
return;
}
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if (events & (EPOLLRDHUP | EPOLLHUP | EPOLLERR))
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
goto reinit;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
if ((c->mode == MODE_PASST && tap_handler_passt(c, now)) ||
(c->mode == MODE_PASTA && tap_handler_pasta(c, now)))
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
goto reinit;
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
return;
passt, pasta: Namespace-based sandboxing, defer seccomp policy application To reach (at least) a conceptually equivalent security level as implemented by --enable-sandbox in slirp4netns, we need to create a new mount namespace and pivot_root() into a new (empty) mountpoint, so that passt and pasta can't access any filesystem resource after initialisation. While at it, also detach IPC, PID (only for passt, to prevent vulnerabilities based on the knowledge of a target PID), and UTS namespaces. With this approach, if we apply the seccomp filters right after the configuration step, the number of allowed syscalls grows further. To prevent this, defer the application of seccomp policies after the initialisation phase, before the main loop, that's where we expect bad things to happen, potentially. This way, we get back to 22 allowed syscalls for passt and 34 for pasta, on x86_64. While at it, move #syscalls notes to specific code paths wherever it conceptually makes sense. We have to open all the file handles we'll ever need before sandboxing: - the packet capture file can only be opened once, drop instance numbers from the default path and use the (pre-sandbox) PID instead - /proc/net/tcp{,v6} and /proc/net/udp{,v6}, for automatic detection of bound ports in pasta mode, are now opened only once, before sandboxing, and their handles are stored in the execution context - the UNIX domain socket for passt is also bound only once, before sandboxing: to reject clients after the first one, instead of closing the listening socket, keep it open, accept and immediately discard new connection if we already have a valid one Clarify the (unchanged) behaviour for --netns-only in the man page. To actually make passt and pasta processes run in a separate PID namespace, we need to unshare(CLONE_NEWPID) before forking to background (if configured to do so). Introduce a small daemon() implementation, __daemon(), that additionally saves the PID file before forking. While running in foreground, the process itself can't move to a new PID namespace (a process can't change the notion of its own PID): mention that in the man page. For some reason, fork() in a detached PID namespace causes SIGTERM and SIGQUIT to be ignored, even if the handler is still reported as SIG_DFL: add a signal handler that just exits. We can now drop most of the pasta_child_handler() implementation, that took care of terminating all processes running in the same namespace, if pasta started a shell: the shell itself is now the init process in that namespace, and all children will terminate once the init process exits. Issuing 'echo $$' in a detached PID namespace won't return the actual namespace PID as seen from the init namespace: adapt demo and test setup scripts to reflect that. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2022-02-07 20:11:37 +00:00
reinit:
passt: Add PASTA mode, major rework PASTA (Pack A Subtle Tap Abstraction) provides quasi-native host connectivity to an otherwise disconnected, unprivileged network and user namespace, similarly to slirp4netns. Given that the implementation is largely overlapping with PASST, no separate binary is built: 'pasta' (and 'passt4netns' for clarity) both link to 'passt', and the mode of operation is selected depending on how the binary is invoked. Usage example: $ unshare -rUn # echo $$ 1871759 $ ./pasta 1871759 # From another terminal # udhcpc -i pasta0 2>/dev/null # ping -c1 pasta.pizza PING pasta.pizza (64.190.62.111) 56(84) bytes of data. 64 bytes from 64.190.62.111 (64.190.62.111): icmp_seq=1 ttl=255 time=34.6 ms --- pasta.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 34.575/34.575/34.575/0.000 ms # ping -c1 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 (2606:4700:3034::6815:147a): icmp_seq=1 ttl=255 time=29.0 ms --- spaghetti.pizza ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 28.967/28.967/28.967/0.000 ms This entails a major rework, especially with regard to the storage of tracked connections and to the semantics of epoll(7) references. Indexing TCP and UDP bindings merely by socket proved to be inflexible and unsuitable to handle different connection flows: pasta also provides Layer-2 to Layer-2 socket mapping between init and a separate namespace for local connections, using a pair of splice() system calls for TCP, and a recvmmsg()/sendmmsg() pair for UDP local bindings. For instance, building on the previous example: # ip link set dev lo up # iperf3 -s $ iperf3 -c ::1 -Z -w 32M -l 1024k -P2 | tail -n4 [SUM] 0.00-10.00 sec 52.3 GBytes 44.9 Gbits/sec 283 sender [SUM] 0.00-10.43 sec 52.3 GBytes 43.1 Gbits/sec receiver iperf Done. epoll(7) references now include a generic part in order to demultiplex data to the relevant protocol handler, using 24 bits for the socket number, and an opaque portion reserved for usage by the single protocol handlers, in order to track sockets back to corresponding connections and bindings. A number of fixes pertaining to TCP state machine and congestion window handling are also included here. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-07-17 06:34:53 +00:00
tap_sock_init(c);
passt: New design and implementation with native Layer 4 sockets This is a reimplementation, partially building on the earlier draft, that uses L4 sockets (SOCK_DGRAM, SOCK_STREAM) instead of SOCK_RAW, providing L4-L2 translation functionality without requiring any security capability. Conceptually, this follows the design presented at: https://gitlab.com/abologna/kubevirt-and-kvm/-/blob/master/Networking.md The most significant novelty here comes from TCP and UDP translation layers. In particular, the TCP state and translation logic follows the intent of being minimalistic, without reimplementing a full TCP stack in either direction, and synchronising as much as possible the TCP dynamic and flows between guest and host kernel. Another important introduction concerns addressing, port translation and forwarding. The Layer 4 implementations now attempt to bind on all unbound ports, in order to forward connections in a transparent way. While at it: - the qemu 'tap' back-end can't be used as-is by qrap anymore, because of explicit checks now introduced in qemu to ensure that the corresponding file descriptor is actually a tap device. For this reason, qrap now operates on a 'socket' back-end type, accounting for and building the additional header reporting frame length - provide a demo script that sets up namespaces, addresses and routes, and starts the daemon. A virtual machine started in the network namespace, wrapped by qrap, will now directly interface with passt and communicate using Layer 4 sockets provided by the host kernel. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-16 06:25:09 +00:00
}