From: "Jim C. Brown" <jma5@umd.edu>
To: qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] Connecting vde and LAN
Date: Mon, 11 Jul 2005 11:02:04 -0400 [thread overview]
Message-ID: <20050711150204.GA7933@jbrown.mylinuxbox.org> (raw)
In-Reply-To: <Pine.LNX.4.61.0507110646340.12555@filer.marasystems.com>
[-- Attachment #1: Type: text/plain, Size: 1545 bytes --]
On Mon, Jul 11, 2005 at 09:50:11AM +0200, Henrik Nordstrom wrote:
> On Sun, 10 Jul 2005, Jim C. Brown wrote:
>
> I didn't mean to implement this in qemu directly, but rather in a helper
> similar to vdeq, but the primary method should be via vde_switch. Having
> this shortcut helper is a future extension.
>
Ooooh! Like slirpvde.
> Inspired by this discussion I now have a prototype vde_packet glue for
> vde_switch at http://www.henriknordstrom.net/examples/vde_packet.c,
> connecting the vde_switch to the LAN via a packet socket.
>
Excellent work! I've attached a modified version called vde_pcap.c, which
uses libpcap to do the dirty work instead. I haven't gotten all the linuxisms
out of it yet, but it should be easy to fix. One day, this might run on Windows
hosts.
>
> b) can't talk to the host itself. This is due to the packets going
> directly to the wire and never really "seen" by the host stack. Not sure
> yet if there is an easy way out, but I suppose it may be possible to set
> up a dummy tap with the same MAC and IP address as the base Ethernet
> device and duplicate broadcasts and packet directed to the host there,
> obviously assuming the administrator does not block this in firewalling..
>
> Regards
> Henrik
>
Alas, the accepted solution to allow pcap programs to talk to the host is to
use tuntap to create a tap device and connect the program to the tap device
instead of the real ethernet device.
--
Infinite complexity begets infinite beauty.
Infinite precision begets infinite perfection.
[-- Attachment #2: vde_pcap.c --]
[-- Type: text/plain, Size: 4958 bytes --]
/* Copyright (C) 2005 Jim Brown
* Copyright (C) 2005 Henrik Nordstrom
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111, USA.
*
* The VDE connection setup (open_vde) is loosely based on a similar
* functions in vdeq.c by Renzo Davoli
*/
#include <sys/socket.h>
#include <netpacket/packet.h>
#include <net/ethernet.h> /* the L2 protocols */
#include <sys/ioctl.h>
#include <fcntl.h>
#include <net/if.h>
#include <net/if_arp.h>
#include <sys/un.h>
#include <errno.h>
#include <linux/if_tun.h>
#include <stdio.h>
#include <string.h>
#include <stdint.h>
#include <stdlib.h>
#include <pcap.h>
#define SWITCH_MAGIC 0xfeedface
#define BUFSIZE 2048
#define ETH_ALEN 6
#define PCAP_READ_TIMEOUT 1
enum request_type {
REQ_NEW_CONTROL
};
struct request_v3 {
uint32_t magic;
uint32_t version;
enum request_type type;
struct sockaddr_un sock;
};
static int
open_vde(char *name, int intno, int group)
{
int pid = getpid();
struct request_v3 req;
int fdctl;
int fddata;
struct sockaddr_un sock;
if ((fdctl = socket(AF_UNIX, SOCK_STREAM, 0)) < 0) {
perror("socket");
exit(1);
}
sock.sun_family = AF_UNIX;
snprintf(sock.sun_path, sizeof(sock.sun_path), "%s", name);
if (connect(fdctl, (struct sockaddr *) &sock, sizeof(sock))) {
perror("connect");
exit(1);
}
memset(&req, 0, sizeof(req));
req.magic = SWITCH_MAGIC;
req.version = 3;
req.type = REQ_NEW_CONTROL + ((group > 0) ? ((geteuid() << 8) + group) << 8 : 0);
req.sock.sun_family = AF_UNIX;
sprintf(&req.sock.sun_path[1], "%5d-%2d", pid, intno);
if ((fddata = socket(AF_UNIX, SOCK_DGRAM, 0)) < 0) {
perror("socket");
exit(1);
}
if (bind(fddata, (struct sockaddr *) &req.sock, sizeof(req.sock)) < 0) {
perror("bind");
exit(1);
}
if (send(fdctl, &req, sizeof(req), 0) < 0) {
perror("send");
exit(1);
}
if (recv(fdctl, &sock, sizeof(struct sockaddr_un), 0) < 0) {
perror("recv");
exit(1);
}
if (connect(fddata, (struct sockaddr *) &sock, sizeof(struct sockaddr_un)) < 0) {
perror("connect");
exit(1);
}
/* note: fdctl is intentionally leaked. Closed on exit by the OS. */
return fddata;
}
void
set_nonblocking(int fd)
{
int fl = fcntl(fd, F_GETFL);
fcntl(fd, F_SETFL, (fl | O_NONBLOCK));
}
void
pcap_relay(u_char *args, const struct pcap_pkthdr *header, const u_char * packet)
{
int * vde_socket = (int *)args;
send(*vde_socket, packet, header->len, 0);
}
static void
wrap_dispatch(fd_set * rfd, pcap_t * handle, int vde_socket)
{
int n;
if (FD_ISSET(pcap_get_selectable_fd(handle), rfd)) {
n = pcap_dispatch(handle, -1, pcap_relay, (u_char *)&vde_socket);
}
}
int pcap_sendpacket(pcap_t * handle, const u_char * packet, int len)
{
if (send(pcap_fileno(handle), packet, len, 0) == len)
{
return 0;
}
else
{
return -1;
}
}
static void
relay(fd_set * rfd, int vde_socket, pcap_t * handle)
{
int n;
char packet[65536];
union {
struct sockaddr_ll ll;
struct sockaddr_un un;
} sa;
socklen_t sa_len = sizeof(sa);
if (FD_ISSET(vde_socket, rfd)) {
n = recvfrom(vde_socket, packet, sizeof(packet), 0, (struct sockaddr *)&sa, &sa_len);
if (n > 0)
pcap_sendpacket(handle, packet, n);
else if (n == 0)
exit(1);
else if (errno != EINTR) {
perror("recv:");
exit(1);
}
}
}
int
main(int argc, char **argv)
{
int vde_socket;
pcap_t * handle;
char errbuf[PCAP_ERRBUF_SIZE];
const u_char * packet;
fd_set rfd;
int fds;
if (argc != 3) {
printf("Usage: %s interface vde_socket\n", argv[0]);
exit(1);
}
handle = pcap_open_live(argv[1], BUFSIZ, 1, PCAP_READ_TIMEOUT, errbuf);
if (handle == NULL)
{
printf("%s\n", errbuf);
exit(1);
}
vde_socket = open_vde(argv[2], 0, 0);
set_nonblocking(vde_socket);
fds = pcap_get_selectable_fd(handle);
if (vde_socket > fds)
fds = vde_socket;
fds += 1;
FD_ZERO(&rfd);
int i;
for (;;) {
int n;
FD_SET(pcap_get_selectable_fd(handle), &rfd);
FD_SET(vde_socket, &rfd);
n = select(fds, &rfd, NULL, NULL, NULL);
if (n > 0) {
wrap_dispatch(&rfd, handle, vde_socket);
relay(&rfd, vde_socket, handle);
} else if (n < 0 && errno != EINTR) {
perror("select:");
}
}
}
next prev parent reply other threads:[~2005-07-11 15:07 UTC|newest]
Thread overview: 29+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-07-10 12:27 [Qemu-devel] Connecting vde and LAN Oliver Gerlich
2005-07-10 16:42 ` Henrik Nordstrom
2005-07-10 17:37 ` Jim C. Brown
2005-07-10 18:23 ` Oliver Gerlich
2005-07-10 18:58 ` Jim C. Brown
2005-07-11 2:21 ` Henrik Nordstrom
2005-07-11 2:33 ` Jim C. Brown
2005-07-11 7:50 ` Henrik Nordstrom
2005-07-11 15:02 ` Jim C. Brown [this message]
2005-07-11 23:01 ` Jim C. Brown
2005-07-12 2:49 ` Henrik Nordstrom
2005-07-12 22:25 ` Jim C. Brown
2005-08-04 10:14 ` Henrik Nordstrom
2005-08-05 16:54 ` Jim C. Brown
2005-08-10 19:07 ` Jim C. Brown
2005-08-11 14:56 ` Henrik Nordstrom
2005-08-11 16:24 ` Oliver Gerlich
2005-08-11 16:56 ` Jim C. Brown
2005-08-12 10:02 ` Henrik Nordstrom
2005-08-12 18:07 ` Jim C. Brown
2005-08-11 17:00 ` Paul Brook
2005-08-12 0:11 ` Herbert Poetzl
2005-08-12 9:53 ` Henrik Nordstrom
2005-07-10 17:48 ` Oliver Gerlich
2005-07-11 1:36 ` Henrik Nordstrom
2005-07-12 19:43 ` Ross Kendall Axe
2005-07-12 20:31 ` Oliver Gerlich
2005-07-13 3:02 ` Ross Kendall Axe
2005-07-10 18:27 ` Bakul Shah
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20050711150204.GA7933@jbrown.mylinuxbox.org \
--to=jma5@umd.edu \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).