[news.eclipse.tools.cdt] Eclipse often hangs after debugging

[Tilmann Reh <tilmannreh@xxxxxxxxxxxxx>]

Hello,

I recently started using the YAGARTO toolchain (Eclipse, GNUARM C
compiler, Zylin CDT, OpenOCD) for developing the software for a current
ARM7 based project (OS: Win XP pro, SP2).

The Eclipse version was 3.2.0, build M20060629-1905. It contains CDT
v3.1.1.200609080933, and Eclipse platform & RCP v3.2.0.v20060609m - and
many many plugins...
The Java VM I had installed is v1.5.0_09_b03, dated 12.10.2006.
The complete toolchain can be downloaded here: <http://www.yagarto.de>.

[After having reported my problems in eclipse.newcomer, I upgraded to
the current eclipse SDK v3.2.2 and java JRE 1.6.0_01 - but that didn't
change anything, except for a popup window reporting that it waits for
something to finish... There is no newer embedded CDT from Zylin, BTW.]

Every now and then, the Eclipse IDE suddenly hangs. This happens when
trying to save after some changes have been made to a source file, or
when entering debug mode, or at some other circumstances. However, it
does not generally fail, and when I restart Eclipse it will work fine
for some time, but then suddenly it will hang again. I have read some
reports about problems during project launch, but none of these seems to
apply here. (I also included permanent memory size settings and selected
"external browser" for IDE and help system, but that also didn't change
anything.)

Every time it hangs, I need to shutdown it by the task manager. Then,
a popup message window from the java runtime module (javaw.exe or
java.exe) appears, which shows this JVM exit code: 1073807364.
"Googling" for this code reveals no useful information - this seems to
be an exit code that appears quite often, for several causes, and not
only with Eclipse. [The current eclipse 3.2.2 reports exit code 143
instead.]

However, regardless of what happened before the crash, the exit code is
always the same.

I tried starting Eclipse in debug mode, but the only error message it
shows (and that is reported in the log file) is "unable to create
variable object", which obviously happens from time to time when trying
to start the debugger. However, Eclipse does not hang when this occurs
(at least no immediately). Obviously, there are no log file entries for
the operations that finally caused the crash.

Here is a complete stack trace from a crashed session:
<http://www.autometer.de/files/Eclipse_Stack_20070424-1857.txt> (34k)

In eclipse.newcomer, I was told this looked like a CDT deadlock, so now
I'm asking here...

As it is now, I need to restart Eclipse every few minutes, which is
really boring.

What can I do to track that problem, and/or to make Eclipse run stable?

Thanks in advance,
Tilmann