HELP - Redirect output of already running process?

[gaf at blu.org (Jerry Feldman)]

I am aware of what Patrick wrote:
freopen(3) essentially closes the file descriptor associated with the 
stream. SO, you can use this to connect stdout to a file. The FILE 
structure is simply a C data structure. The underlying file is always 
represented by a file descriptor. So, what freopen does is to close(2) the 
open file descriptor, open(2) a new one associated with the pathname and 
associates that with the existing FILE structure.

The problem is how the user can communicate things to a running program. 
Here's a scenario, which I think was presented:
1. start the program, let it run in the background.
2. log out. 
3. log back in, and capture the output. 
The simple solution without program mods is just to let the program write 
to a file and use tail with the -f option to monitor the output. You can 
log in and out many times this way.

So, you need to invent a way to tell the running program where to send its 
output (you can use one of many methods, such as signals, files, SystemV 
message queues, System V shared memory, or even mmap. If you want it to 
direct output to the current terminal then  simply give it the name of the 
pseudo terminal. Most of this is pretty basic programming techniques.
"Derek D. Martin" wrote:

> I'm unfamiliar with a lot of the more advanced uses of GDB, but in
> addition to the author's response, I can think of one more case where
> I believe this will not work: where there are no streams involved.
> 
> freopen() opens a file, and reassociates the stream given as the last
> argument with that file.  If the file was opened with open(), there is
> no stream to associate.
> 
> What I'm not clear on is whether or not some trickery can be done to
> do fdopen() followed by freopen().  If this is possible, it might
> work.  But I don't know how one would take the resulting stream
> pointer (from the call to fdopen()) and pass it to the subsequent call
> to freopen()...  First off, the pointer would (I think) have to
> already exist in the program space, which is probably unlikely, or in
> many cases impossible to determine.  Secondly, whatever value it
> contained would have to be irrelevant to the rest of the execution of
> the program, since you're going to modify it.  Then, maybe, you could
> pass it as an argument to freopen()...  Yes?
> 
> 
> - -- 
> Derek Martin               ddm at pizzashack.org    
> - ---------------------------------------------
> I prefer mail encrypted with PGP/GPG!
> GnuPG Key ID: 0x81CFE75D
> Retrieve my public key at http://pgp.mit.edu
> Learn more about it at http://www.gnupg.org
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.0.6 (GNU/Linux)
> Comment: For info see http://www.gnupg.org
> 
> iD8DBQE8/pHYdjdlQoHP510RAsD9AJ4xknfNpuiU31k0XdjmxmW+E9bCgACdHvyp
uTUL1QCPkitSu4wDsuL54Xc=
> =DIJ1
> -----END PGP SIGNATURE-----
> _______________________________________________
> Discuss mailing list
> Discuss at blu.org
> http://www.blu.org/mailman/listinfo/discuss
> 

-- 
Jerry Feldman <gaf at blu.org>
Boston Linux and Unix user group
http://www.blu.org PGP key id:C5061EA9
PGP Key fingerprint:053C 73EC 3AC1 5C44 3E14 9245 FB00 3ED5 C506 1EA9