passt/util.c

263 lines
5.6 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 07:25:09 +01:00
// SPDX-License-Identifier: AGPL-3.0-or-later
/* PASST - Plug A Simple Socket Transport
*
* util.c - Convenience helpers
*
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 07:25:09 +01:00
* Copyright (c) 2020-2021 Red Hat GmbH
* Author: Stefano Brivio <sbrivio@redhat.com>
*
*/
#include <stdio.h>
#include <stdint.h>
#include <stddef.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 07:25:09 +01:00
#include <unistd.h>
#include <linux/ipv6.h>
#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 07:25:09 +01:00
#include <net/ethernet.h>
#include <net/if.h>
#include <netinet/ip.h>
#include <netinet/tcp.h>
#include <netinet/udp.h>
#include <sys/epoll.h>
#include <syslog.h>
#include <stdarg.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 07:25:09 +01:00
#include "passt.h"
#define logfn(name, level) \
void name(const char *format, ...) { \
va_list args; \
\
va_start(args, format); \
vsyslog(level, format, args); \
va_end(args); \
}
logfn(err, LOG_ERR)
logfn(warn, LOG_WARNING)
logfn(info, LOG_INFO)
#ifdef DEBUG
logfn(debug, LOG_DEBUG)
#endif
/**
* csum_fold() - Fold long sum for IP and TCP checksum
* @sum: Original long sum
*
* Return: 16-bit folded sum
*/
uint16_t csum_fold(uint32_t sum)
{
while (sum >> 16)
sum = (sum & 0xffff) + (sum >> 16);
return sum;
}
/**
* csum_ipv4() - Calculate IPv4 checksum
* @buf: Packet buffer, L3 headers
* @len: Total L3 packet length
*
* Return: 16-bit IPv4-style checksum
*/
uint16_t csum_ip4(void *buf, size_t len)
{
uint32_t sum = 0;
uint16_t *p = buf;
size_t len1 = len / 2;
size_t off;
for (off = 0; off < len1; off++, p++)
sum += *p;
if (len % 2)
sum += *p & 0xff;
return ~csum_fold(sum);
}
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 07:25:09 +01:00
/**
* csum_ipv4() - Calculate TCP checksum for IPv4 and set in place
* @iph: Packet buffer, IP header
*/
void csum_tcp4(struct iphdr *iph)
{
struct tcphdr *th = (struct tcphdr *)((char *)iph + iph->ihl * 4);
uint16_t tlen = ntohs(iph->tot_len) - iph->ihl * 4, *p = (uint16_t *)th;
uint32_t sum = 0;
sum += (iph->saddr >> 16) & 0xffff;
sum += iph->saddr & 0xffff;
sum += (iph->daddr >> 16) & 0xffff;
sum += iph->daddr & 0xffff;
sum += htons(IPPROTO_TCP);
sum += htons(tlen);
th->check = 0;
while (tlen > 1) {
sum += *p++;
tlen -= 2;
}
if (tlen > 0) {
sum += *p & htons(0xff00);
}
th->check = (uint16_t)~csum_fold(sum);
}
/**
* ipv6_l4hdr() - Find pointer to L4 header in IPv6 packet and extract protocol
* @ip6h: IPv6 header
* @proto: Filled with L4 protocol number
*
* Return: pointer to L4 header, NULL if not found
*/
char *ipv6_l4hdr(struct ipv6hdr *ip6h, uint8_t *proto)
{
int offset, len, hdrlen;
struct ipv6_opt_hdr *o;
uint8_t nh;
len = ntohs(ip6h->payload_len);
offset = 0;
while (offset < len) {
if (!offset) {
nh = ip6h->nexthdr;
hdrlen = sizeof(struct ipv6hdr);
} else {
nh = o->nexthdr;
hdrlen = (o->hdrlen + 1) * 8;
}
if (nh == 59)
return NULL;
if (nh == 0 || nh == 43 || nh == 44 || nh == 50 ||
nh == 51 || nh == 60 || nh == 135 || nh == 139 ||
nh == 140 || nh == 253 || nh == 254) {
offset += hdrlen;
o = (struct ipv6_opt_hdr *)(unsigned char *)ip6h +
offset;
} else {
*proto = nh;
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 07:25:09 +01:00
return (char *)(ip6h + 1) + offset;
}
}
return NULL;
}
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 07:25:09 +01:00
/**
* sock_l4_add() - Create and bind socket for given L4, add to epoll list
* @c: Execution context
* @v: IP protocol, 4 or 6
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 11:33:38 +01:00
* @proto: Protocol number, host order
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 07:25:09 +01:00
* @port: Port, network order
*
* Return: newly created socket, -1 on error
*/
int sock_l4_add(struct ctx *c, int v, uint16_t proto, uint16_t port)
{
struct sockaddr_in addr4 = {
.sin_family = AF_INET,
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 11:33:38 +01:00
.sin_port = htons(port),
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 07:25:09 +01:00
.sin_addr = { .s_addr = INADDR_ANY },
};
struct sockaddr_in6 addr6 = {
.sin6_family = AF_INET6,
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 11:33:38 +01:00
.sin6_port = htons(port),
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 07:25:09 +01:00
.sin6_addr = IN6ADDR_ANY_INIT,
};
struct epoll_event ev = { 0 };
const struct sockaddr *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 11:33:38 +01:00
int fd, sl, one = 1;
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 07:25:09 +01:00
if (proto != IPPROTO_TCP && proto != IPPROTO_UDP &&
proto != IPPROTO_ICMP && proto != IPPROTO_ICMPV6)
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 07:25:09 +01:00
return -1; /* Not implemented. */
fd = socket(v == 4 ? AF_INET : AF_INET6,
proto == IPPROTO_TCP ? SOCK_STREAM : SOCK_DGRAM, proto);
if (fd < 0) {
perror("L4 socket");
return -1;
}
#define CHECK_SET_MIN_MAX(ipproto, proto_ctx, fd) \
if (proto == (ipproto)) { \
if (fd < c->proto_ctx.fd_min) \
c->proto_ctx.fd_min = (fd); \
if (fd > c->proto_ctx.fd_max) \
c->proto_ctx.fd_max = (fd); \
}
CHECK_SET_MIN_MAX(IPPROTO_ICMP, icmp, fd);
CHECK_SET_MIN_MAX(IPPROTO_ICMPV6, icmp, fd);
CHECK_SET_MIN_MAX(IPPROTO_TCP, tcp, fd);
CHECK_SET_MIN_MAX(IPPROTO_UDP, udp, fd);
#undef CHECK_SET_MIN_MAX
if (proto == IPPROTO_ICMP || proto == IPPROTO_ICMPV6)
goto epoll_add;
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 07:25:09 +01:00
if (v == 4) {
sa = (const struct sockaddr *)&addr4;
sl = sizeof(addr4);
} else {
sa = (const struct sockaddr *)&addr6;
sl = sizeof(addr6);
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 11:33:38 +01:00
setsockopt(fd, IPPROTO_IPV6, IPV6_V6ONLY, &one, sizeof(one));
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 07:25:09 +01:00
}
if (bind(fd, sa, sl) < 0) {
/* We'll fail to bind to low ports if we don't have enough
* capabilities, and we'll fail to bind on already bound ports,
* this is fine.
*/
close(fd);
return 0;
}
if (proto == IPPROTO_TCP && listen(fd, 128) < 0) {
perror("TCP socket listen");
close(fd);
return -1;
}
epoll_add:
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 07:25:09 +01:00
ev.events = EPOLLIN;
ev.data.fd = fd;
if (epoll_ctl(c->epollfd, EPOLL_CTL_ADD, fd, &ev) == -1) {
perror("L4 epoll_ctl");
return -1;
}
return fd;
}
/**
* timespec_diff_ms() - Report difference in milliseconds between two timestamps
* @a: Minuend timestamp
* @b: Subtrahend timestamp
*
* Return: difference in milliseconds
*/
int timespec_diff_ms(struct timespec *a, struct timespec *b)
{
if (a->tv_nsec < b->tv_nsec) {
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 11:33:38 +01:00
return (b->tv_nsec - a->tv_nsec) / 1000000 +
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 07:25:09 +01:00
(a->tv_sec - b->tv_sec - 1) * 1000;
}
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 11:33:38 +01:00
return (a->tv_nsec - b->tv_nsec) / 1000000 +
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 07:25:09 +01:00
(a->tv_sec - b->tv_sec) * 1000;
}