I have not been using gdb for very long, and I only began to play with ddd
today (02-15-01). Consequently, I am unsure as to whether I could be doing
something (incorrectly) to trigger the behavior I have described here. If this
is the case, I apologize for any inconvenience.  Additionally, I did not
see behavior similar to that described below documented in the BUGS,
OPENBUGS, or PROBLEMS files contained in the distribution, but I hope I
have not overlooked any relevant information.

The system I am working on is a Red Hat Linux 6.1 box on which I just
installed ddd-3.3.

The behavior I've witnessed is the following:

I was attempting to perform a test using ddd to attach to a process running
a compilation of the (very simple) code in the accompanying runloop.c file.
I had compiled this code with the following syntax:

	gcc -Wall -g -o runloop runloop.c

The app's standard output was scrolling repeatedly (as expected) on the terminal in
which I had started it. When I issued the shell command

	ddd runloop 997

where 997 was the PID of the running process, ddd did manage to display the
source code correctly, and apparently managed to stop the running process
(the output of a ps command showed that the targeted process had a status
of 'T' or stopped). However, a dialog popped up telling me that GDB had
terminated abnormally, and a "Segmentation fault" message was visible beneath
the "(gdb)" prompt in the "GDB Console" area of the DDD display. The dialog
offered me the option of Restarting GDB, exiting DDD, or displaying a
simple help message. After killing the targeted process and ddd, I retried this
experiment several times. I was able to reproduce this behavior every time I
attempted to attach to the runloop process by specifying its PID as a command
line argument to ddd.

I then killed the process and ddd, restarted the process, and tried to attach
to it from within DDD via the File->Attach to Process command (I specified the
executable to ddd on the command line, as required). I was able to attach
to the process once and subsequently detach from it successfully. When I tried
to re-attach to the process for a second time, I witnessed the same behavior
that I had when trying to attach to the process via the command line argument
with one exception:

	The process had not been stopped. Even before I got a chance to specify
	what process I wished to attach to, gdb had crashed with a segment
	fault.

I restarted gdb (via the option on the dialog), and my next attempt to attach
to the process was successful. After detaching and attempting to reattach,
gdb went down again with a seg fault. I repeated this testing loop several
times, and received the same results each time. I also exited and reentered
ddd, and restarted a new runloop process several times, and I was able to reproduce
the faulty behavior every time I performed this experiment.

So, it seems like I can successfully attach to a process from within ddd
"every other" time I attempt to do so, with repeated restarts of gdb
necessary.

Incidentally, from within a non-ddd gdb session, I was able to repeatedly
attach and detach from a runloop process without incident/segment fault.
Also, before this particular issue came up, I was able to successfully
navigate the "sample" tutorial offered by the ddd info documentation
(basic ddd functionality seems to be working correctly).

This is (perhaps obviously) the first bug report I have sent. Consequently,
I am not sure what other information I should be sending. I am willing to
provide you with any requested/relevant information upon request.
Whether the  behavior I described is or is not my fault, confirmation or
rejection of this report would be greatly appreciated.

Keep up the good work.

Josh Willey