I am considering writing a BBS-like program in C and thinking about exactly how the I/O architecture would work with such a program. I'm familiar with sockets programming already, more specifically the master/remote model (not sure if there's a more official name for it) where a master process running as a daemon runs the vast majority of the application in a main process. When remote TTYs connect, they do so in a separate process that communicates with the main process via a Unix domain socket, and there's a thread on the main process for each remote TTY's I/O. All the modules and functionality are running in the main process.
This works well for things like CLIs for some kind of process, but I don't think it's as well suited for a significantly richer/more interactive program, where I think it'd make much more sense for all the TTYs to be managed in the same process rather than communicating over a socket. For example, you can't run ncurses over a socket, since the termios that we care about is in that remote process, not in our main process or usable over the socket. So taking the master/remote model further, you'd need to move a lot of logic from the main program to the remote processes.
The problem I'm a little stuck on is exactly how you can have the main process handling all the TTYs without itself handling all of the network socket traffic. For example, say we want to allow telnet and SSH connections. With the master/remote model, it might look like this:
Telnet:
Inbound telnet connection
Telnet server launches /usr/sbin/remote_process (custom login shell)
remote_process (a C program, shell script, etc.) begins executing, communicating with main_process
SSH:
Inbound SSH connection
Authentication
SSH server launches /usr/sbin/remote_process (custom login shell)
remote_process (a C program, shell script, etc.) begins executing, communicating with main_process
Importantly, with the master/remote model we consider above, the telnet/SSH protocol is abstracted away from the program in question. It doesn't care if the incoming connection is from Telnet, SSH, a serial port, etc. We don't need to handle the details of these protocols ourselves.
Naively trying to apply this to the single-process model, handling all the TTYs directly, I would think the thing to do would be that step # 3/4 somehow needs to have the main process take over its terminal/PTY. main_process can't be called directly though, since it's already running, and I'm not sure if anything like that would be possible since somehow it would be moving the master/slave for the pty between processes, but the goal would be to have main_process doing everything remote_process was doing in the other model, directly handling the I/O from the Telnet server, SSH server, etc.
The standard way of doing this kind of thing seems to be having the main_process directly run its own listeners - that is, instead of listening for UNIX domain socket connections, directly accept Telnet/SSH traffic, etc. But then, the program is now responsible for handling the details of each individual protocol.
You can see an example of this with SyncrhonetBBS: https://github.com/SynchronetBBS/sbbs/tree/b35365c2e470bde58838cbb7445fe7e8c4bc1beb/src/syncterm
The BBS program itself has code to handle each supported protocol: SSH, TELNET, TELNETS, etc.
(I suppose there is a third model: have the main daemon process itself be quite minimal in what it does, and just have each individual TTY process contain the bulk of all the logic, and just use the daemon process for IPC between the TTYs... but then that gets tricky if you want to do stuff like dynamically loadable and unloadable modules that are really at a "system" level as opposed to per-TTY... so I'm not really considering this other extreme).
Is there any way to have the best of both worlds - be able to control all the different TTYs from a single process, but without having to directly implement protocol-specific handling? And if so, how does the TTY setup occur? I'm not looking for code examples here so much as a general high-level explanation/guidance of what this would likely look and how the different components - processes, sockets, TTYs - would interact.
Related
I am implementing a very basic C server that allows clients to chat. Right now I am using fork(), but I am having trouble having clients see each others' messages.
It also seems that all clients get the same file descriptor from accept(). Basically, I have a while loop where I test if someone wants to connect with select(), accept() their connection, and fork(). After that I read input and try to pass them to all users (whom I am keeping in a list). I can copy/paste my code if necessary.
So, is it possible to have the clients communicate with processes, or do I have to use pthreads?
Inter-process communication -IPC- (in general) don't care about client vs server (except at connect phase). A given process can have both a client and a server role (on different sockets), and would use poll(2) or the older select on several sockets in some event loop.
Notice that processes have each their own virtual address space, while threads share the same virtual address space (the one of their containing process). Read some pthread tutorial, and some book on POSIX programming (perhaps the old ALP). Be aware that a lot of information regarding processes can be queried on Linux thru /proc/ (see proc(5) for more). In particular, the virtual address space of process of pid 1234 can be obtained thru /proc/1234/maps and its opened file descriptors thru /proc/1234/fd/ and /proc/1234/fdinfo/ etc....
However, it is simpler to code a common server keeping the shared state, and dispatching messages to clients.
You could design a protocol where the clients have some way to initiate that IPC. For example, if all the processes are on the same machine, you could have a protocol which transmits a file path used as unix(7) socket address, or as fifo(7), and each "client" process later initiate (with some connect) a direct communication with another "client". It might be unwise to do so.
Look also into libraries like 0mq. They often are free software, so you can study their source code.
On a Linux machine, you have a daemon that listens on TCP port A. However, it is usually stopped because it is rarely used and takes away a large amount of system resources. Instead, I want to do something like this:
Code an application that listens on port B and does the following as soon as a connection is established: If the daemon is stopped, start it and wait until it listens on port A. Now the difficult part: Connect the client to the daemon in a completely transparent way, i.e. without the client having to reconnect on port A. Also, but this is irrelevant for this question, the application will shut down the daemon when there are no connections for a certain amount of time.
Of course, I could have my application connect to the daemon and pipe all communication. I do not want that. I want some way to forward the established connection to the daemon and then get rid of the connected socket, while the client is now happily connected with the daemon. In some way, I want to give the daemon's process my already connected socket. Is there any way to do something like this?
I'm running Debian, if that's important. I would want to code the application in C/C++, and it's okay to have OS-specific solutions (i.e. use syscalls). Forgive me though, I am not much of a Linux coder, so I am not very familiar with Linux system programming. If there is some obvious way to do it, I simply didn't know.
Of course, I am open for any kind of suggestion.
This problem has a pre-existing standard solution, generically known as inetd. It has been around for a long time, first in Unix systems and then Linux.
The more modern implementation is xinetd
I would like to inject a shared library into a process (I'm using ptrace() to do that part) and then be able to get output from the shared library back into the debugger I'm writing using some form of IPC. My instinct is to use a pipe, but the only real requirements are:
I don't want to store anything on the filesystem to facilitate the communication as it will only last as long as the debugger is running.
I want a portable Unix solution (so Unix-standard syscalls would be ideal).
The problem I'm running into is that as far as I can see, if I call pipe() in the debugger, there is no way to pass the "sending" end of the pipe to the target process, and vice versa with the receiving end. I could set up shared memory, but I think that would require creating a file somewhere so I could reference the memory segment from both processes. How do other debuggers capture output when they attach to a process after it has already begun running?
I assume that you are in need of a debugging system for your business logic code (I mean application). From my experience, this kind of problem is tackled with below explained system design. (My experience is in C++, I think the same must hold good for the C based system also.)
Have a logger system (a separate process). This will contain - logger manager and the logging code - which will take the responsibility of dumping the log into hard disk.
Each application instance (process running in Unix) will communicate to this process with sockets. So you can have your own messaging protocol and communicate with the logger system with socket based communication.
Later, for each of this application - have a switch which can switch off/on the log.So that you can have a tool - to send signal to this process to switch on/off the message logging.
At a high level, this is the most generic way to develop a logging system. In case you need any information - Do comment it. I will try to answer.
Using better search terms showed me this question is a dup of these guys:
Can I share a file descriptor to another process on linux or are they local to the process?
Can I open a socket and pass it to another process in Linux
How to use sendmsg() to send a file-descriptor via sockets between 2 processes?
The top answers were what I was looking for. You can use a Unix-domain socket to hand a file descriptor off to a different process. This could work either from debugger to library or vice versa, but is probably easier to do from debugger to library because the debugger can write the socket's address into the target process while it injects the library.
However, once I pass the socket's address into the target process, I might as well just use the socket itself instead of using a pipe in addition.
I've been working on a complex server-client system in C and I'm not sure how to implement the socket communication.
In a nutshell, the system is a server application which communicates with a database and uses a UNIX socket to communicate with one or more child processes created with fork(). The purpose of the children is to run game servers. The process of launching a game server is like this:
The server/"manager" identifies a game server in the database that is to be made. (Assume database communication is already sorted.)
The manager forks a child (the "game controller").
The game controller sets up two pipe pairs, then forks, replacing its child's stdin with a pipe, and it's stdout and stderr with another pipe.
The game controller's child then runs execlp() to begin running the actual game server executable.
My experience with sockets is fairly minimal. I have used select() on a server application before to 'multiplex' numerous clients, as demonstrated by the simple example in the GNU C documentation here.
I now have a new challenge, as the system must be able to do more: the manager needs to be able to arbitrarily send commands to the game controller children (that it will find by periodically checking the database) and get replies, but also expect incoming arbitrary commands/errors from them and send replies back.
So, I need a sort-of "context" system, where sockets are meaningful only between themselves. In other words, when a command is sent from the manager to the game controller, each party needs to be aware of who is asking and know what the reply is (and, therefore, which command it is a reply to).
Because select() is only useful for knowing when we have incoming data, and a thread should block on it, would I need another thread that sends data and gets the replies? Will this require each game controller, although technically a 'client', to use a listening socket and use select() as well?
I hope I've explained the system and the problem concisely; I will add more detail if required. Thanks!
Ok, I am still not really sure I understand exactly where your trouble is, so I will just spout off some things about writing a client/server app. If I am off track, just let me know.
The way that the server will know which clients corresponds to which socket is that the clients will tell the server. Essentially, you need to have a log-in protocol. When the game controller connects to the server, it will send a message that says "Hi, i am registering as controller foo1 on host xyz, port abc..." and whatever else the server needs to know about its clients. The server will keep a data structure that maps sockets to client metadata, state, etc. Whenever it gets a new message, it can easily map from the incoming host/port to its metadata. Or your protocol can require that on each incoming message, the will client send the name it registered with as a field.
Handling the request/response can be done several ways. First lets deal with the networking part of it on the server side. One way to manage this, as you mentioned, is by using select (or poll, or epoll) to multiplex the sockets. This is actually usually considered the more complicated way to do things. Another way is to spawn off a thread (or fork a process, which is less common these days) for each incoming client. Each spawned thread can read its own assigned socket, responding to messages one at a time without worrying about the fact that there are other clients besides the own it is dealing with. This simple one to one thread to socket model breaks down if there are many clients, but if that is not the case, then it is worth consideration.
Part 2 really covers only the client sending the server a message, and the server replying. What happens when the server wants to initiate communication? How does it do it and how does the client handle it? Also, how do you model the model the communication at the application level, meaning assuming we have the read/write part down, how do we know what to send? You will probably want to model things in terms of state machines. There is also a lot more to deal with like what happens when a client crashes? What about when the server crashes? Also, what if you really have your heart set of using select, perhaps because you expect many client? I will try to add more to this answer tomorrow.
What are they and how do they work?
Context happens to be SQL Server
Both on Windows and POSIX systems, named-pipes provide a way for inter-process communication to occur among processes running on the same machine. What named pipes give you is a way to send your data without having the performance penalty of involving the network stack.
Just like you have a server listening to a IP address/port for incoming requests, a server can also set up a named pipe which can listen for requests. In either cases, the client process (or the DB access library) must know the specific address (or pipe name) to send the request. Often, a commonly used standard default exists (much like port 80 for HTTP, SQL server uses port 1433 in TCP/IP; \\.\pipe\sql\query for a named pipe).
By setting up additional named pipes, you can have multiple DB servers running, each with its own request listeners.
The advantage of named pipes is that it is usually much faster, and frees up network stack resources.
--
BTW, in the Windows world, you can also have named pipes to remote machines -- but in that case, the named pipe is transported over TCP/IP, so you will lose performance. Use named pipes for local machine communication.
Unix and Windows both have things called "Named pipes", but they behave differently. On Unix, a named pipe is a one-way street which typically has just one reader and one writer - the writer writes, and the reader reads, you get it?
On Windows, the thing called a "Named pipe" is an IPC object more like a TCP socket - things can flow both ways and there is some metadata (You can obtain the credentials of the thing on the other end etc).
Unix named pipes appear as a special file in the filesystem and can be accessed with normal file IO commands including the shell. Windows ones don't, and need to be opened with a special system call (after which they behave mostly like a normal win32 handle).
Even more confusing, Unix has something called a "Unix socket" or AF_UNIX socket, which works more like (but not completely like) a win32 "named pipe", being bidirectional.
Linux Pipes
First In First Out (FIFO) interproccess communication mechanism.
Unnamed Pipes
On the command line, represented by a "|" between two commands.
Named Pipes
A FIFO special file. Once created, you can use the pipe just like a normal file(open, close, write, read, etc).
To create a named pipe, called "myPipe", from the command line (man page):
mkfifo myPipe
To create a named pipe from c, where "pathname" is the name you would like the pipe to have and "mode" contains the permissions you want the pipe to have (man page):
#include <sys/types.h>
#include <sys/stat.h>
int mkfifo(const char *pathname, mode_t mode);
According to Wikipedia:
[...] A traditional pipe is "unnamed" because it exists anonymously and persists only for as long as the process is running. A named pipe is system-persistent and exists beyond the life of the process and must be "unlinked" or deleted once it is no longer being used. Processes generally attach to the named pipe (usually appearing as a file) to perform IPC (inter-process communication).
Compare
echo "test" | wc
to
mkdnod apipe p
wc apipe
wc will block until
echo "test" > apipe
executes
This is an exeprt from Technet (so not sure why the marked answer says named pipes are faster??):
Named Pipes vs. TCP/IP Sockets
In a fast local area network (LAN) environment, Transmission Control Protocol/Internet Protocol (TCP/IP) Sockets and Named Pipes clients are comparable with regard to performance. However, the performance difference between the TCP/IP Sockets and Named Pipes clients becomes apparent with slower networks, such as across wide area networks (WANs) or dial-up networks. This is because of the different ways the interprocess communication (IPC) mechanisms communicate between peers.
For named pipes, network communications are typically more interactive. A peer does not send data until another peer asks for it using a read command. A network read typically involves a series of peek named pipes messages before it starts to read the data. These can be very costly in a slow network and cause excessive network traffic, which in turn affects other network clients.
It is also important to clarify if you are talking about local pipes or network pipes. If the server application is running locally on the computer that is running an instance of SQL Server, the local Named Pipes protocol is an option. Local named pipes runs in kernel mode and is very fast.
For TCP/IP Sockets, data transmissions are more streamlined and have less overhead. Data transmissions can also take advantage of TCP/IP Sockets performance enhancement mechanisms such as windowing, delayed acknowledgements, and so on. This can be very helpful in a slow network. Depending on the type of applications, such performance differences can be significant.
TCP/IP Sockets also support a backlog queue. This can provide a limited smoothing effect compared to named pipes that could lead to pipe-busy errors when you are trying to connect to SQL Server.
Generally, TCP/IP is preferred in a slow LAN, WAN, or dial-up network, whereas named pipes can be a better choice when network speed is not the issue, as it offers more functionality, ease of use, and configuration options.
Pipes are a way of streaming data between applications. Under Linux I use this all the time to stream the output of one process into another. This is anonymous because the destination app has no idea where that input-stream comes from. It doesn't need to.
A named pipe is just a way of actively hooking onto an existing pipe and hoovering-up its data. It's for situations where the provider doesn't know what clients will be eating the data.
Inter-process communication (mostly) for Windows Applications. Similar to using sockets to communicate between applications in Unix.
MSDN
Named pipes in a unix/linux context can be used to make two different shells to communicate since a shell just can't share anything with another.
Furthermore, one script instantiated twice in the same shell can't share anything through the two instances. I found a use for named pipes when coding a daemon that contains the start() and stop() function, and I wanted to use the same script to perform the two actions.
Without named pipes (or any kind of semaphore) starting the script in the background is not a problem. The thing is when it finishes you just can't access the instance in background.
So when you want to send him the stop command you just can't: running the same script without named pipes and calling the stop() function won't do anything since you are actually running another instance.
The solution was to implement two pipes, one READ and the other WRITE when you start the daemon. Then make him, among its other tasks, listen to the READ pipe. Then the Stop() function contains a command that will write a message in the pipe, that will be handled by the background running script that will perform an exit 0. This way our second instance of the same script has only on task to do: tell the first instance to stop.
This way one and only one script can start and stop itself.
Of course you have different ways to do it by triggering the stop via a touch for example. But this one is nice and interesting to code.
Named pipes is a windows system for inter-process communication. In the case of SQL server, if the server is on the same machine as the client, then it is possible to use named pipes to tranfer the data, as opposed to TCP/IP.