CW6.3/MCF5772 Sporadic inability to break execution with USB TAP

[Tim_SS]

Using CW 6.3 with a USB TAP, I'm trying to debug what I believe is a sporadic problem with my application on an MCF5272 based board. Normally, I have no problem with the TAP, I can set breakpoints, pause execution, etc. However, when I observe externally that the problem with my application may have occurred, and try to break execution of the processor to see what is happening, I receive the error "CFCSProtocol Plugin : Failed to stop process". The USB TAP continues to show that the CPU is running (LED RUN/PAUSE continuously green) and that there is communication via USB (LEG TX/RX flashing). If I perform a Kill Thread, the debugger is then able to reset the board and reload the application, with no need to cycle power or perform any other external reset.

Could a failure in the application modify the state of the CPU in such a way that it becomes impossible for the TAP to interrupt execution, for example entering HALT due to a fault-on-fault exception scenario? If that's the case, is there something I can trap on before getting into this unbreakable state?

Regards,
Tim Butler

[ZiglioNZ]

I've got a similar, maybe the same, problem with the latest version of CW:
http://forums.freescale.com/freescale/board/message?board.id=CFCOMM&message.id=5928#M5928

(I got no replies yet but maybe the admins would find useful to merge the two threads)

[J2MEJediMaster]

For the first poster: Try upgrading CW to a version higher than 6.3. There have been some reports of issues with CW 6.3 and USB in these forums.
For the second poster: Try placing a powered USB hub between your workstation and the target board. That has cured the intermittent problem on occasion.

---Tom

[Tim_SS]

Tom:

Thanks for your suggestion. I downloaded CodeWarrior 7.1 for Coldfire, updated it with the 7.1.1 patch, updated the firmware of the USB-TAP and updated the USB driver. Unfortunately, I still have the exact same problem: at some point during execution of my application it becomes impossible for the USB TAP to break in.

I'd still be grateful for any kind of suggestion as to what else to try. At the moment, the only conclusion I'm able to draw is that the CPU gets into some state which is un-interruptable by the debugger, although I don't know if that is theoretically possible.

Tim
P.S. I just noticed that I had a typo in the message subject, it should read "MCF5272"

[J2MEJediMaster]

Sorry that the upgrade route didn't fix the problem. It's time to file a service request on this. Click here to do that.

---Tom