I am trying to do a simple communication between a server (running on Ubuntu from Qemu with Cortex-A53 cpu) and a client (running on CentOS from my pc), using sockets.
If I run the C++ code only from Centos (both client.c and server.c) it works fine. Same if I run both from Ubuntu. But if I start the server.c from Ubuntu and client.c from CentOS the communication doesn't work.
The C code I'm using is from this tutorial:
https://www.geeksforgeeks.org/socket-programming-cc/
The Qemu command:
qemu-system-aarch64 -m 2048 -smp 2 -cpu cortex-a53 -M virt -nographic \
-pflash flash0.img \
-pflash flash1.img \
-drive if=none,file=${IMAGE},format=qcow2,id=hd0 -device virtio-blk-device,drive=hd0 \
-drive if=none,id=cloud,file=cloud.img,format=qcow2 \
-device virtio-blk-device,drive=cloud \
-device virtio-net-device,netdev=user0 \
-netdev user,id=user0,hostfwd=tcp::10022-:22 \
-device virtio-serial \
-chardev socket,id=foo,host=localhost,port=8080,server,nowait \
-device virtserialport,chardev=foo,name=test0
When I run server.c from Qemu and client.c from my pc. I see that the server.c it's blocked at accept() function and client.c it's blocked at read() function.
If I run the following command on Ubuntu:
$ sudo lsof -i -P -n | grep LISTEN I get this:
systemd-r 644 systemd-resolve 13u IPv4 15994 0t0 TCP 127.0.0.53:53 (LISTEN)
sshd 745 root 3u IPv4 18696 0t0 TCP *:22 (LISTEN)
sshd 745 root 4u IPv6 18699 0t0 TCP *:22 (LISTEN)
server 14164 root 3u IPv4 74481 0t0 TCP *:8080 (LISTEN)
If I run the same command on CentOS I get this:
qemu-syst 57073 ibestea 10u IPv4 2648807035 0t0 TCP 127.0.0.1:8080 (LISTEN)
qemu-syst 57073 ibestea 13u IPv4 2648807037 0t0 TCP *:10022 (LISTEN)
Any help is welcome.
The problem was that I was trying to connect to the socket from server part by using host address and port, but to access the Qemu data I had to connect to the socket using file descriptor /dev/vport3p1.
The server.c file should look something similar to this:
#include <unistd.h>
#include <stdio.h>
#include <sys/socket.h>
#include <stdlib.h>
#include <netinet/in.h>
#include <string.h>
#include <fcntl.h>
#include <string.h>
#include <signal.h>
#include <sys/ioctl.h>
#define PORT 8080
#define DEV_PATH "/dev/vport3p1"
int main(int argc, char const *argv[])
{
int server_fd, new_socket, valread;
struct sockaddr_in address;
int opt = 1;
int addrlen = sizeof(address);
char buffer[1024] = {0};
char *hello = "Hello from server";
int dev_fd;
printf("SERVER: get an internet domain socket\n");
if ((dev_fd = open(DEV_PATH, O_RDWR)) == -1) {
perror("open");
exit(1);
}
else {
printf("SERVER: opened file descriptor first time = %d\n", dev_fd);
}
valread = read( dev_fd , buffer, 1024);
printf("%s\n",buffer );
valread = write(dev_fd , hello , strlen(hello));
printf("Hello message sent\n");
return 0;
}
Related
I am aware of bind: cannot assign requested address and Cannot assign requested address - possible causes?, I have not been able to derive a solution from these.
I am trying to create a TCP stream (specifically here a std::net::TcpListener) directly with libc. I am encountering the error Cannot assign requested address and error code: 99 when running my code.
The exact output being:
error message: error code: : Cannot assign requested address
thread 'main' panicked at 'error code: 99', src/main.rs:43:13
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
My code (playground):
use libc; // 0.2.136
use std::mem::{size_of, transmute};
fn main() {
// Create socket
let socket_fd = unsafe {
let socket = libc::socket(libc::AF_INET6, libc::SOCK_STREAM | libc::SOCK_NONBLOCK, 0);
assert_ne!(socket, -1);
let optval = 1i32;
let res = libc::setsockopt(
socket,
libc::SOL_SOCKET,
libc::SO_REUSEPORT,
(&optval as *const i32).cast(),
4,
);
assert_eq!(res, 0);
socket
};
// Bind socket
// decimal 127.0.0.1 -> hexadecimal 007f.0000.0000.0001
let sin6_addr = unsafe { transmute::<_, libc::in6_addr>(*b"007f000000000001") };
let socket_address = libc::sockaddr_in6 {
sin6_family: libc::AF_INET6 as u16,
sin6_port: 8080,
sin6_flowinfo: u32::default(),
sin6_addr,
sin6_scope_id: u32::default(),
};
let socket_address_length = size_of::<libc::sockaddr_in6>() as u32;
unsafe {
let res = libc::bind(
socket_fd,
(&socket_address as *const libc::sockaddr_in6).cast(),
socket_address_length,
);
if res == -1 {
let err = errno();
print_error("error message: ");
panic!("error code: {err}");
}
}
assert_eq!(unsafe { libc::close(socket_fd) },0);
}
fn print_error(s: &str) {
unsafe {
libc::perror(s.as_ptr().cast());
}
}
fn errno() -> i32 {
unsafe { *libc::__errno_location() }
}
I set the option SO_REUSEPORT here as I need to create a stream that multiple processes can listen to.
https://lwn.net/Articles/542629/:
The basic concept of SO_REUSEPORT is simple enough. Multiple servers (processes or threads) can bind to the same port
The port does not appear to be in use:
jonathan#jonathan-pc:~$ sudo lsof -i -P -n
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
systemd-r 793 systemd-resolve 13u IPv4 19452 0t0 UDP 127.0.0.53:53
systemd-r 793 systemd-resolve 14u IPv4 19453 0t0 TCP 127.0.0.53:53 (LISTEN)
avahi-dae 855 avahi 12u IPv4 31377 0t0 UDP *:5353
avahi-dae 855 avahi 13u IPv6 31378 0t0 UDP *:5353
avahi-dae 855 avahi 14u IPv4 31379 0t0 UDP *:49594
avahi-dae 855 avahi 15u IPv6 31380 0t0 UDP *:58397
NetworkMa 859 root 36u IPv4 36212 0t0 UDP 192.168.0.22:68->192.168.0.1:67
NetworkMa 859 root 37u IPv4 110801 0t0 UDP 192.168.0.17:68->192.168.0.1:67
tor 1038 debian-tor 6u IPv4 31407 0t0 TCP 127.0.0.1:9050 (LISTEN)
rust-anal 4117 jonathan 46u IPv4 33176 0t0 UDP 127.0.0.1:35972->127.0.0.53:53
rust-anal 4189 jonathan 46u IPv4 33176 0t0 UDP 127.0.0.1:35972->127.0.0.53:53
firefox 4297 jonathan 3u IPv4 112786 0t0 TCP 192.168.0.22:46702->151.101.1.69:443 (ESTABLISHED)
firefox 4297 jonathan 29u IPv4 100032 0t0 TCP 192.168.0.22:55828->34.120.208.123:443 (ESTABLISHED)
firefox 4297 jonathan 56u IPv4 115182 0t0 TCP 192.168.0.22:50798->52.51.190.37:443 (ESTABLISHED)
firefox 4297 jonathan 75u IPv4 95741 0t0 TCP 192.168.0.22:48466->142.250.200.10:443 (ESTABLISHED)
firefox 4297 jonathan 81u IPv4 67879 0t0 TCP 192.168.0.22:55638->34.214.17.205:443 (ESTABLISHED)
firefox 4297 jonathan 116u IPv4 111739 0t0 TCP 192.168.0.22:37986->172.217.169.68:443 (ESTABLISHED)
firefox 4297 jonathan 121u IPv4 95751 0t0 TCP 192.168.0.22:46054->198.252.206.25:443 (ESTABLISHED)
firefox 4297 jonathan 129u IPv4 102073 0t0 TCP 192.168.0.22:51478->34.120.237.76:443 (ESTABLISHED)
firefox 4297 jonathan 136u IPv4 96742 0t0 TCP 192.168.0.22:36606->34.120.115.102:443 (ESTABLISHED)
firefox 4297 jonathan 139u IPv4 97943 0t0 TCP 192.168.0.22:36626->172.217.169.68:443 (ESTABLISHED)
firefox 4297 jonathan 144u IPv4 99520 0t0 TCP 192.168.0.22:49264->198.252.206.25:443 (ESTABLISHED)
firefox 4297 jonathan 157u IPv4 104859 0t0 TCP 192.168.0.22:58042->93.184.220.29:80 (ESTABLISHED)
jonathan#jonathan-pc:~$
You are transmuting b"007f000000000001" into a libc::in6_addr. But I don't thing that is the proper thing to do.
Looking at the man page, that struct is:
struct in6_addr {
uint8_t s6_addr[16];
};
That is, just 16 bytes, the proper thing for an IPv6. But your bytes are actually the ASCII values of that string: 30303766..., that while technically an IPv6 address, it is not a local one of yours so you cannot bind to it.
Moreover, in IPv6 the proper localhost address is ::1, not 127.0.0.1. That is 15 zeros followed by a single one.
If you want to bind to the IPv6 localhost by using a transmute that would be:
let sin6_addr = unsafe { transmute::<_, libc::in6_addr>([0_u8,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1]) };
Or if you insist on using a literal string (why?):
let sin6_addr = unsafe { transmute::<_, libc::in6_addr>(*b"\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\x01") };
If Powershell Core on Linux, using dotnet, can return the FQDN, then how does mono, also using dotnet, get the same result?
Here's powershell finding the local system FQDN:
thufir#dur:~/powershell$
thufir#dur:~/powershell$ pwsh
PowerShell v6.0.1
Copyright (c) Microsoft Corporation. All rights reserved.
https://aka.ms/pscore6-docs
Type 'help' to get help.
PS /home/thufir/powershell>
PS /home/thufir/powershell> [System.Net.Dns]::GetHostByName((hostname)).HostName
dur.bounceme.net
PS /home/thufir/powershell>
and mono:
thufir#dur:~$
thufir#dur:~$ mono Projects/console/console/bin/Debug/console.exe
Hello Mono World
dur
dur
localhost
localhost
localhost
dur
done
thufir#dur:~$
code:
using System;
using System.Net;
using System.Net.Sockets;
using System.Globalization;
using System.Text;
namespace console
{
class MainClass
{
public static void Main(string[] args)
{
Console.WriteLine ("Hello Mono World");
String name;
name = Environment.MachineName;
Console.WriteLine(name);
name = System.Net.Dns.GetHostName();
Console.WriteLine(name);
name = System.Environment.GetEnvironmentVariable("COMPUTERNAME");
Console.WriteLine(name);
name =System.Net.Dns.GetHostEntry("localhost").HostName;
Console.WriteLine(name);
name = Dns.GetHostEntry("LocalHost").HostName;
Console.WriteLine(name);
// var ipProperties = IPGlobalProperties.GetIPGlobalProperties();
// name = string.Format("{0}.{1}", ipProperties.HostName, ipProperties.DomainName);
// Console.WriteLine(name);
name = Dns.GetHostEntry("LocalHost").HostName;
Console.WriteLine(name);
name = System.Net.Dns.GetHostEntry(Environment.MachineName).HostName;
Console.WriteLine(name);
Console.WriteLine("done");
}
}
}
thanks to jjw for pointing me in the right direction.
This is tantalizingly close, because dur, which is how the prompt is shown, is listed. Just not the whole name. Also:
thufir#dur:~$
thufir#dur:~$ hostname
dur
thufir#dur:~$
thufir#dur:~$ hostname --fqdn
dur.bounceme.net
thufir#dur:~$
thufir#dur:~$ cat /etc/hosts
127.0.0.1 localhost
127.0.1.1 dur.bounceme.net dur
# The following lines are desirable for IPv6 capable hosts
::1 ip6-localhost ip6-loopback
fe00::0 ip6-localnet
ff00::0 ip6-mcastprefix
ff02::1 ip6-allnodes
ff02::2 ip6-allrouters
thufir#dur:~$
thufir#dur:~$ cat /etc/hostname
dur
thufir#dur:~$
Looking for the whole enchilada here, not just the prepend of dur but dur.bounceme.net if possible. Because Powershell Core can find the FQDN shouldn't mono be able to do the same?
I am running a node.js server on port 5403. I can telent to the private ip on this port but cannot telnet to the public ip on the same port.
I assume the cause of this is because node.js is only listening on ipv6. This is the result of
netstat -tpln
(Not all processes could be identified, non-owned process info
will not be shown, you would have to be root to see it all.)
Active Internet connections (only servers)
Proto Recv-Q Send-Q Local Address Foreign Address State
PID/Program name
tcp 0 0 127.0.0.1:6379 0.0.0.0:* LISTEN
-
tcp 0 0 0.0.0.0:22 0.0.0.0:* LISTEN
-
tcp 0 0 127.0.0.1:631 0.0.0.0:* LISTEN
-
tcp 0 0 127.0.0.1:5432 0.0.0.0:* LISTEN
-
tcp6 0 0 :::5611 :::* LISTEN
25715/node
tcp6 0 0 :::22 :::* LISTEN
-
tcp6 0 0 ::1:631 :::* LISTEN
-
tcp6 0 0 :::5403 :::* LISTEN
25709/node
How do I make the node server listen on ipv4
You need to specify an IPV4 address when you call the listen(), I had the same issue with the http module. If I use this:
var http = require('http');
var server = http.createServer(function(request, response) {
...
});
server.listen(13882, function() { });
It only listen on IPV6, as you can see from netstat output:
$ netstat -lntp
Proto Recv-Q Send-Q Local Address Foreign Address State
tcp6 0 0 :::13882 :::* LISTEN
However, if I specify an IPV4 address like this:
var http = require('http');
var server = http.createServer(function(request, response) {
...
});
server.listen(13882, "0.0.0.0", function() { });
netstat will report the server as listening on IPV4:
$ netstat -lntp
Proto Recv-Q Send-Q Local Address Foreign Address State
tcp 0 0 0 0.0.0.0:13882 0 0.0.0.0:13882 LISTEN
I'm using Ubuntu 16.04 and npm 5.3.0.
HTH
I don't have much knowledge about the IPv6 protocol, so sorry if the question sounds stupid.
When I retrieve the list of all IPv6 addresses in my network, I get a field named scope, as shown below :
inet6 addr: 2001:470:1:82::11/64 Scope:Global
inet6 addr: 2001:470:1:82::10/64 Scope:Global
inet6 addr: 2001:470:1:82::13/64 Scope:Global
inet6 addr: fe80::21d:9ff:fe69:2c50/64 Scope:Link
inet6 addr: 2001:470:1:82::12/64 Scope:Global
inet6 addr: 2001:470:1:82::15/64 Scope:Global
inet6 addr: 2001:470:1:82::14/64 Scope:Global
inet6 addr: 2001:470:1:82::5/64 Scope:Global
inet6 addr: 2001:470:1:82::17/64 Scope:Global
inet6 addr: 2001:470:1:82::6/64 Scope:Global
inet6 addr: 2001:470:1:82::16/64 Scope:Global
inet6 addr: 2001:470:1:82::7/64 Scope:Global
inet6 addr: 2001:470:1:82::19/64 Scope:Global
inet6 addr: 2001:470:1:82::8/64 Scope:Global
inet6 addr: 2001:470:1:82::18/64 Scope:Global
inet6 addr: 2001:470:1:82::9/64 Scope:Global
inet6 addr: 2001:470:1:82::1b/64 Scope:Global
inet6 addr: 2001:470:1:82::a/64 Scope:Global
inet6 addr: 2001:470:1:82::1a/64 Scope:Global
inet6 addr: 2001:470:1:82::b/64 Scope:Global
inet6 addr: 2001:470:1:82::1d/64 Scope:Global
inet6 addr: 2001:470:1:82::c/64 Scope:Global
inet6 addr: 2001:470:1:82::1c/64 Scope:Global
inet6 addr: 2001:470:1:82::d/64 Scope:Global
inet6 addr: 2001:470:1:82::1f/64 Scope:Global
inet6 addr: 2001:470:1:82::e/64 Scope:Global
inet6 addr: 2001:470:1:82::1e/64 Scope:Global
inet6 addr: 2001:470:1:82::f/64 Scope:Global
inet6 addr: ::1/128 Scope:Host
In my application, I need to get those addresses for which the scope is 'Link'. I could have used a system call to ifconfig and then parsed the output to extract corresponding addresses. But the problem is, I'm using the call to getifaddrs(), which returns a linked list of structure ifaddr, given as :
struct ifaddrs {
struct ifaddrs *ifa_next; /* Next item in list */
char *ifa_name; /* Name of interface */
unsigned int ifa_flags; /* Flags from SIOCGIFFLAGS */
struct sockaddr *ifa_addr; /* Address of interface */
struct sockaddr *ifa_netmask; /* Netmask of interface */
union {
struct sockaddr *ifu_broadaddr;
/* Broadcast address of interface */
struct sockaddr *ifu_dstaddr;
/* Point-to-point destination address */
} ifa_ifu;
#define ifa_broadaddr ifa_ifu.ifu_broadaddr
#define ifa_dstaddr ifa_ifu.ifu_dstaddr
void *ifa_data; /* Address-specific data */
};
The question is : how to get the addresses with 'Link' scope from this list ?
There are helper macros to assist:
struct sockaddr_in6 s6;
if (IN6_IS_ADDR_LINKLOCAL(&s6.sin6_addr)) {
puts ("link-local");
} else if (IN6_IS_ADDR_SITELOCAL(&s6.sin6_addr)) {
puts ("site-local");
} else if (IN6_IS_ADDR_V4MAPPED(&s6.sin6_addr)) {
puts ("v4mapped");
} else if (IN6_IS_ADDR_V4COMPAT(&s6.sin6_addr)) {
puts ("v4compat");
} else if (IN6_IS_ADDR_LOOPBACK(&s6.sin6_addr)) {
puts ("host");
} else if (IN6_IS_ADDR_UNSPECIFIED(&s6.sin6_addr)) {
puts ("unspecified");
} else {
puts ("global");
}
One way to do this would simply be to check whether the address falls within fe80::/10. The IPv6 address space is available from IANA, which details the possible scopes available.
I downloaded the source code to net-tools (the source package for ifconfig), and it looks like they parse /proc/net/if_inet6. (Comments are my own additions in the following code — also the below is extremely abridged and will most certainly not compile.)
/* some defines collected around the place: */
#define _PATH_PROCNET_IFINET6 "/proc/net/if_inet6"
#define IPV6_ADDR_LOOPBACK 0x0010U
#define IPV6_ADDR_LINKLOCAL 0x0020U
#define IPV6_ADDR_SITELOCAL 0x0040U
#define IPV6_ADDR_COMPATv4 0x0080U
int scope; /* among other stuff */
/* looks like here we parse the contents of /proc/net/if_inet6: */
if ((f = fopen(_PATH_PROCNET_IFINET6, "r")) != NULL) {
while (fscanf(f, "%4s%4s%4s%4s%4s%4s%4s%4s %02x %02x %02x %02x %20s\n",
addr6p[0], addr6p[1], addr6p[2], addr6p[3],
addr6p[4], addr6p[5], addr6p[6], addr6p[7],
&if_idx, &plen, &scope, &dad_status, devname) != EOF) {
/* slightly later: */
printf(_(" Scope:"));
switch (scope) {
case 0:
printf(_("Global"));
break;
case IPV6_ADDR_LINKLOCAL:
printf(_("Link"));
break;
case IPV6_ADDR_SITELOCAL:
printf(_("Site"));
break;
case IPV6_ADDR_COMPATv4:
printf(_("Compat"));
break;
case IPV6_ADDR_LOOPBACK:
printf(_("Host"));
break;
default:
printf(_("Unknown"));
}
So let's have a look at what the above is parsing:
$ cat /proc/net/if_inet6
20010db8000008000000000000000001 03 40 00 00 eth0
fe800000000000000000000000004321 03 40 20 80 eth0
00000000000000000000000000000001 01 80 10 80 lo
So you can see the third column from the left (0x00 Global, 0x20 Link-Local, and 0x10 Loopback) is the scope. Using the above constants from the net-tools code you can work out what they mean. Further investigation would be required to determine a more authoritative source for such constants, and also whether parsing /proc/net/if_inet6 is your best option.
I don't think the scope is stored explicitly in that data structure. However, you can infer the scope from the IP address itself. See http://en.wikipedia.org/wiki/IPv6_address#Address_scopes
telnet localhost pop3
Trying ::1...
Trying 127.0.0.1...
telnet: Unable to connect to remote host: Connection refused
netstat -l
tcp 0 0 *:www : LISTEN
tcp 0 0 localhost.localdoma:ipp : LISTEN
tcp 0 0 *:smtp : LISTEN
tcp 0 0 localhost.localdo:mysql : LISTEN
when i run this service dovecot start i got
start: Rejected send message, 1 matched rules; type="method_call", sender=":1.553" (uid=1000 pid=26250 comm="start) interface="com.ubuntu.Upstart0_6.Job" member="Start" error name="(unset)" requested_reply=0 destination="com.ubuntu.Upstart" (uid=0 pid=1 comm="/sbin/init"))
on Dovecot.conf
protocols = imap imaps pop3 pop3s
disable_plaintext_auth = no
log_timestamp = "%Y-%m-%d %H:%M:%S "
mail_location = maildir:/var/spool/mail/%d/%n
mail_access_groups = mail
first_valid_uid = 106
first_valid_gid = 106
protocol imap {
}
protocol pop3 {
listen=*:110
pop3_uidl_format = %08Xu%08Xv
}
protocol lda {
postmaster_address = samer#aiu.com
mail_plugins = quota
log_path = /var/log/dovecot-deliver.log
info_log_path = /var/log/dovecot-deliver.log
}
auth default {
mechanisms = digest-md5 plain
passdb sql {
args = /etc/dovecot/dovecot-mysql.conf
}
userdb sql {
args = /etc/dovecot/dovecot-mysql.conf
}
user = root
}
If you get this message on the command line:
start: Rejected send message, 1 matched rules; type="method_call", sender=":1.553" (uid=1000 pid=26250 comm="start) interface="com.ubuntu.Upstart0_6.Job" member="Start" error name="(unset)" requested_reply=0 destination="com.ubuntu.Upstart" (uid=0 pid=1 comm="/sbin/init"))
It indicates that you're not doing service dovecot restart as root. So make sure you do:
sudo service dovecot restart
The directive protocols = imap imaps pop3 pop3s should be sufficient to activate pop3 with dovecot. You can add
listen = *
to be sure dovecot will listen on all available interfaces. You can verify this by netstat -apn | grep 110. Are there any failures while starting dovecot? Can you post the dovecot related logs?
By default, dovecot logs to syslog, you can explicitly specify the log files:
# Log file to use for error messages, instead of sending them to syslog.
# /dev/stderr can be used to log into stderr.
log_path = /var/log/dovecot.log
# Log file to use for informational and debug messages.
# Default is the same as log_path.
info_log_path = /var/log/dovecot.info.log