shragi
asked on
fork+exec in c++, linux
I have 2 processes (actually working programs) in 2 different directories, I want to spawn the second process from the first.
Now I am running both of them in two different terminals like client and server and able to communicate. I donot know much about child process creation. can some one explain using c++ code. Now I need to start the client in my program or spawn the child from my program (server) in different location using only fork()+ ecec**().
I need both of them need to execute concurrently...and want to see the outputs in 2 tabs of terminals...can some one help me regarding this.
Thanks in advance.
Now I am running both of them in two different terminals like client and server and able to communicate. I donot know much about child process creation. can some one explain using c++ code. Now I need to start the client in my program or spawn the child from my program (server) in different location using only fork()+ ecec**().
I need both of them need to execute concurrently...and want to see the outputs in 2 tabs of terminals...can some one help me regarding this.
Thanks in advance.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
>> ...and want to see the outputs in 2 tabs of terminals
That's the tricky bit of your requirement. It is easy to achieve with a simple command (in a shell script eg.). In Gnome eg., you could do :
gnome-terminal --tab-with-profile=Default -e "program1" --tab-with-profile=Default -e "program2"
where program1 and program2 are your two programs (with path, parameters, etc.).
I wouldn't do something like this in C.
That's the tricky bit of your requirement. It is easy to achieve with a simple command (in a shell script eg.). In Gnome eg., you could do :
gnome-terminal --tab-with-profile=Default
where program1 and program2 are your two programs (with path, parameters, etc.).
I wouldn't do something like this in C.
The second requirement can be achieved with a different way: in the second program, write the output on a named pipe, then spawn (manually or by a script) a terminal with a "cat /named/pipe" command. This is simpler than executing the terminal and the program inside it, in my opinion.
On the other side, pay attention that when you exec the second program by using fork(), the second program is a child of the first; this means that the second program must terminate before the first one, or you'll end with a zombie process.
To prevent this you have two options: 1) make the second (child) program end before the first one, and call waitpid() to cleanup the ended process, or 2) use a double fork() (+ setsid()) technique to unlink the second process form the first one.
On the other side, pay attention that when you exec the second program by using fork(), the second program is a child of the first; this means that the second program must terminate before the first one, or you'll end with a zombie process.
To prevent this you have two options: 1) make the second (child) program end before the first one, and call waitpid() to cleanup the ended process, or 2) use a double fork() (+ setsid()) technique to unlink the second process form the first one.
ASKER
Initially about my first question, For example my executable names are prg1 and prg2. i.e ./prg1 and ./prg2.
Please specify me ./prg2 isvused and where I can specify the exact path. I am little confused about this line.
execl("/bin/ls", "/bin/ls", "-l", (char *) 0); help me understand the parameters better.
Please specify me ./prg2 isvused and where I can specify the exact path. I am little confused about this line.
execl("/bin/ls", "/bin/ls", "-l", (char *) 0); help me understand the parameters better.
The first parameter is the path to the executable, intended as "complete file name of the executable". The second and the following parameters are command line arguments. The first (argv[0]) should be the name of the executable. This list must be terminated by a null pointer. See [1].
So, if you need to execute ./prg2:
execl("./prg2", "./prg2", (char *)0);
[1] http://linux.die.net/man/3/exec
So, if you need to execute ./prg2:
execl("./prg2", "./prg2", (char *)0);
[1] http://linux.die.net/man/3/exec
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
I want to do something programatically, I hope above script executes the above executables in 2 different terminals.
I dont understand How I can use this
gnome-terminal --tab-with-profile=Default -e "./prg1" --tab-with-profile=Default -e "./prg2" in my program.
Sorry If I do not understand that properly.
I dont understand How I can use this
gnome-terminal --tab-with-profile=Default
Sorry If I do not understand that properly.
If you are using gnome, try running the command I mentioned in a terminal window. Make sure that the paths to both executables are correct. And it will run both executables in a separate tab of a new terminal window.
Which is exactly what you wanted :)
>> I want to do something programatically
You can place the command in a bash script, and then run the bash script.
Which is exactly what you wanted :)
>> I want to do something programatically
You can place the command in a bash script, and then run the bash script.
ASKER
I am already using fork() and execl() to run by 2 programs as a parent and child.
I am seeing the outputs in the same terminal, my outputs are pretty large and continues...So I wanted to know How I can separate the output of a child from parent.
The above bash script will make my 2 executables execute. But I am already running them in my single program.
If this task is pretty difficult, let me know know how I can store my child output (displayed on the terminal) to store to a file.
Because by output is continues, I though displaying it in a tab of terminal is easy to debug and observe. Otherwise suggest me how I can append the output in a file.
I am seeing the outputs in the same terminal, my outputs are pretty large and continues...So I wanted to know How I can separate the output of a child from parent.
The above bash script will make my 2 executables execute. But I am already running them in my single program.
If this task is pretty difficult, let me know know how I can store my child output (displayed on the terminal) to store to a file.
Because by output is continues, I though displaying it in a tab of terminal is easy to debug and observe. Otherwise suggest me how I can append the output in a file.
redirect the output to a named pipe
then (manually or by a script) open a terminal performing "cat named_pipe"
then (manually or by a script) open a terminal performing "cat named_pipe"
>> The above bash script will make my 2 executables execute. But I am already running them in my single program.
Right, but you can undo the changes that made them run in a single program, and have two separate programs which you can then easily run in separate tabs.
It's the logical approach.
Right, but you can undo the changes that made them run in a single program, and have two separate programs which you can then easily run in separate tabs.
It's the logical approach.
http://www.yolinux.com/TUTORIALS/ForkExecProcesses.html
http://www-h.eng.cam.ac.uk/help/tpl/unix/fork.html
Wish I had more time to help but tied up for a couple of days.
Here is a related question that may also help:
https://www.experts-exchange.com/questions/26533583/About-system-call-and-managing-processes-in-linux.html