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

176 lines
4.0 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
*
* udp.c - UDP L2-L4 translation routines
*
* Copyright (c) 2020-2021 Red Hat GmbH
* Author: Stefano Brivio <sbrivio@redhat.com>
*
*/
/**
* DOC: Theory of Operation
*
*
* For UDP, no state machine or any particular tracking is required. Try to
* create and bind sets of 2^16 sockets, one for IPv4 and one for IPv6. Binding
* will fail on ports that are already bound, or low ports depending on
* capabilities.
*
* Packets are forwarded back and forth, by prepending and stripping UDP headers
* in the obvious way, with no port translation.
*
*/
#include <stdio.h>
#include <errno.h>
#include <limits.h>
#include <net/ethernet.h>
#include <net/if.h>
#include <netinet/in.h>
#include <stdint.h>
#include <stddef.h>
#include <string.h>
#include <sys/epoll.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <unistd.h>
#include <linux/ip.h>
#include <linux/ipv6.h>
#include <linux/udp.h>
#include <time.h>
#include "passt.h"
#include "tap.h"
#include "util.h"
static int udp4_sock_port[USHRT_MAX];
static int udp6_sock_port[USHRT_MAX];
/**
* udp_sock_handler() - Handle new data from socket
* @c: Execution context
* @s: File descriptor number for socket
* @events: epoll events bitmap
*/
void udp_sock_handler(struct ctx *c, int s, uint32_t events)
{
struct in6_addr a6 = { .s6_addr = { 0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0xff, 0xff,
0, 0, 0, 0 } };
struct sockaddr_storage sr, sl;
socklen_t slen = sizeof(sr);
char buf[USHRT_MAX];
struct udphdr *uh;
ssize_t n;
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)events;
n = recvfrom(s, buf + sizeof(*uh), sizeof(buf) - sizeof(*uh),
MSG_DONTWAIT, (struct sockaddr *)&sr, &slen);
if (n < 0)
return;
uh = (struct udphdr *)buf;
if (getsockname(s, (struct sockaddr *)&sl, &slen))
return;
if (sl.ss_family == AF_INET) {
struct sockaddr_in *sr4 = (struct sockaddr_in *)&sr;
struct sockaddr_in *sl4 = (struct sockaddr_in *)&sl;
memcpy(&a6.s6_addr[12], &sr4->sin_addr, sizeof(sr4->sin_addr));
uh->source = sr4->sin_port;
uh->dest = sl4->sin_port;
uh->len = htons(n + sizeof(*uh));
tap_ip_send(c, &a6, IPPROTO_UDP, buf, n + sizeof(*uh));
} else if (sl.ss_family == AF_INET6) {
struct sockaddr_in6 *sr6 = (struct sockaddr_in6 *)&sr;
struct sockaddr_in6 *sl6 = (struct sockaddr_in6 *)&sl;
uh->source = sr6->sin6_port;
uh->dest = sl6->sin6_port;
uh->len = htons(n + sizeof(*uh));
tap_ip_send(c, &sr6->sin6_addr, IPPROTO_UDP,
buf, n + sizeof(*uh));
}
}
/**
* udp_tap_handler() - Handle packets from tap
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
* @c: Execution context
* @af: Address family, AF_INET or AF_INET6
* @in: Input buffer
* @len: Length, including UDP header
*/
void udp_tap_handler(struct ctx *c, int af, void *addr, char *in, size_t len)
{
struct udphdr *uh = (struct udphdr *)in;
int s;
(void)c;
if (af == AF_INET) {
struct sockaddr_in sa = {
.sin_family = AF_INET,
.sin_port = uh->dest,
};
if (!(s = udp4_sock_port[ntohs(uh->source)]))
return;
sa.sin_addr = *(struct in_addr *)addr;
sendto(s, in + sizeof(*uh), len - sizeof(*uh), MSG_DONTWAIT,
(struct sockaddr *)&sa, sizeof(sa));
} else if (af == AF_INET6) {
struct sockaddr_in6 sa = {
.sin6_family = AF_INET6,
.sin6_port = uh->dest,
.sin6_addr = *(struct in6_addr *)addr,
};
if (!(s = udp6_sock_port[ntohs(uh->source)]))
return;
passt: Assorted fixes from "fresh eyes" review A bunch of fixes not worth single commits at this stage, notably: - make buffer, length parameter ordering consistent in ARP, DHCP, NDP handlers - strict checking of buffer, message and option length in DHCP handler (a malicious client could have easily crashed it) - set up forwarding for IPv4 and IPv6, and masquerading with nft for IPv4, from demo script - get rid of separate slow and fast timers, we don't save any overhead that way - stricter checking of buffer lengths as passed to tap handlers - proper dequeuing from qemu socket back-end: I accidentally trashed messages that were bundled up together in a single tap read operation -- the length header tells us what's the size of the next frame, but there's no apparent limit to the number of messages we get with one single receive - rework some bits of the TCP state machine, now passive and active connection closes appear to be robust -- introduce a new FIN_WAIT_1_SOCK_FIN state indicating a FIN_WAIT_1 with a FIN flag from socket - streamline TCP option parsing routine - track TCP state changes to stderr (this is temporary, proper debugging and syslogging support pending) - observe that multiplying a number by four might very well change its value, and this happens to be the case for the data offset from the TCP header as we check if it's the same as the total length to find out if it's a duplicated ACK segment - recent estimates suggest that the duration of a millisecond is closer to a million nanoseconds than a thousand of them, this trend is now reflected into the timespec_diff_ms() convenience routine Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-21 10:33:38 +00:00
sendto(s, in + sizeof(*uh), len - sizeof(*uh),
MSG_DONTWAIT | MSG_NOSIGNAL,
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 sockaddr *)&sa, sizeof(sa));
}
}
/**
passt: Assorted fixes from "fresh eyes" review A bunch of fixes not worth single commits at this stage, notably: - make buffer, length parameter ordering consistent in ARP, DHCP, NDP handlers - strict checking of buffer, message and option length in DHCP handler (a malicious client could have easily crashed it) - set up forwarding for IPv4 and IPv6, and masquerading with nft for IPv4, from demo script - get rid of separate slow and fast timers, we don't save any overhead that way - stricter checking of buffer lengths as passed to tap handlers - proper dequeuing from qemu socket back-end: I accidentally trashed messages that were bundled up together in a single tap read operation -- the length header tells us what's the size of the next frame, but there's no apparent limit to the number of messages we get with one single receive - rework some bits of the TCP state machine, now passive and active connection closes appear to be robust -- introduce a new FIN_WAIT_1_SOCK_FIN state indicating a FIN_WAIT_1 with a FIN flag from socket - streamline TCP option parsing routine - track TCP state changes to stderr (this is temporary, proper debugging and syslogging support pending) - observe that multiplying a number by four might very well change its value, and this happens to be the case for the data offset from the TCP header as we check if it's the same as the total length to find out if it's a duplicated ACK segment - recent estimates suggest that the duration of a millisecond is closer to a million nanoseconds than a thousand of them, this trend is now reflected into the timespec_diff_ms() convenience routine Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-21 10:33:38 +00:00
* udp_sock_init() - Create and bind listening sockets for inbound packets
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
* @c: Execution context
*
* Return: 0 on success, -1 on failure
*/
int udp_sock_init(struct ctx *c)
{
in_port_t port;
int s;
for (port = 0; port < USHRT_MAX; port++) {
passt: Assorted fixes from "fresh eyes" review A bunch of fixes not worth single commits at this stage, notably: - make buffer, length parameter ordering consistent in ARP, DHCP, NDP handlers - strict checking of buffer, message and option length in DHCP handler (a malicious client could have easily crashed it) - set up forwarding for IPv4 and IPv6, and masquerading with nft for IPv4, from demo script - get rid of separate slow and fast timers, we don't save any overhead that way - stricter checking of buffer lengths as passed to tap handlers - proper dequeuing from qemu socket back-end: I accidentally trashed messages that were bundled up together in a single tap read operation -- the length header tells us what's the size of the next frame, but there's no apparent limit to the number of messages we get with one single receive - rework some bits of the TCP state machine, now passive and active connection closes appear to be robust -- introduce a new FIN_WAIT_1_SOCK_FIN state indicating a FIN_WAIT_1 with a FIN flag from socket - streamline TCP option parsing routine - track TCP state changes to stderr (this is temporary, proper debugging and syslogging support pending) - observe that multiplying a number by four might very well change its value, and this happens to be the case for the data offset from the TCP header as we check if it's the same as the total length to find out if it's a duplicated ACK segment - recent estimates suggest that the duration of a millisecond is closer to a million nanoseconds than a thousand of them, this trend is now reflected into the timespec_diff_ms() convenience routine Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
2021-02-21 10:33:38 +00:00
if (c->v4) {
if ((s = sock_l4_add(c, 4, IPPROTO_UDP, port)) < 0)
return -1;
udp4_sock_port[port] = s;
}
if (c->v6) {
if ((s = sock_l4_add(c, 6, IPPROTO_UDP, port)) < 0)
return -1;
udp6_sock_port[port] = s;
}
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
}
return 0;
}