Craig, thank you very much for the warm heart. We have been
little further than that actually. IDE has been able to send breakpoint request
to KVM and KVM has been able to sent back breakpoint event. Currently the
problem is that JDT doesn't respond to the breakpoint event. We compare the
result from WTK and our SDK and found that the JDT send frame request to WTK but
all thread request to our SDK after the breakpoint event. There is a known issue
about thread in our SDK, so I guess the breakpoint processing in JDT has
some dependency on the thread model. Therefore I decide to resolve the thread
thing first and then go back to see if the breakpoint issue can be resolve by
itself.
Best Regards,
George
George,
Sorry for not responding again... I did receive more information from the
JDT team, but I've been swamped... Here is what they told me.
>>>>>>>>>>>> Quote
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
For a breakpoint, the java debugger asks the VM (via JDWP) whether
a class is loaded with the assocaited class name. If so, a
breakpoint request is created. If not, we ask the VM to tell us of
class loads assocaited with the class name (we use a wild card
expressions to find class loads of any inner types). For each load, we
install any breakpoints that correspond... perhaps the VM is not
telling us of aclass loads? or does not like the wild card pattern?
>>>>>>>>>>>> Quote
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
They suggested that if this didn't give you the necessary solution you
should go ahead and open up a bugzilla entry against the JDT debugger component
for further discussion. If you do that, please add me to the CC for the
bug so I can follow along.
Thanks,
craig
On Apr 7, 2009, at 8:31 AM, Lei Zhen-A21681 wrote:
Thanks Craig!
The command you mentioned is VM Version command {1:1}. We
just fixed it but it was not the root cause. So far the VM can already send
breakpoint request to our debug agent after we fixed another independent
issue. It seems that JDT debugger have a lot of internal dependencies among
the commands.
Thank you for the help anyway.
Best Regards,
George
George,
The MTJ debug support is 99.99% the JDT debug support. The only
difference is a patch from me that was applied to the JDT debugger quite a few
years back that sent one particular event (it has been a long time since I
figured this out and I don't remember the details). This particular
event is only sent if the JVM returns information saying that it is a "KVM".
Otherwise, the flows are identical.
Let me check with the debug team and see what the best approach would be
to working on this issue. I will get back to the list as soon as I know
the next steps.
Thanks,
Craig
On Apr 6, 2009, at 1:06 AM, Lei Zhen-A21681 wrote:
The breakpoint doesn't
work when running iDEN Java
SDK MTJ. According to the
packets we captured between MTJ and our debug agent, MTJ doesn't send out
the breakpoint event request. The same
features works in WTK. And it seems
that the breakpoint event
request will be sent out only after receiving the class prepare event of the
class containing breakpoints from
emulator. However, even after we sent out the class prepare event
(below), the breakpoint event request still won't arrive.
Can anybody give us some hints about this issue?
George
Lei
Project Lead, Motorola iDEN
Java SDK
+86-10-84733144
S09M 5/F MCIC,
BEIJING
_______________________________________________ dsdp-mtj-dev
mailing list dsdp-mtj-dev@xxxxxxxxxxx https://dev.eclipse.org/mailman/listinfo/dsdp-mtj-dev
_______________________________________________ dsdp-mtj-dev
mailing list dsdp-mtj-dev@xxxxxxxxxxx https://dev.eclipse.org/mailman/listinfo/dsdp-mtj-dev
|