CW10.x freezes on debugger-launch

キャンセル
次の結果を表示 
表示  限定  | 次の代わりに検索 
もしかして: 

CW10.x freezes on debugger-launch

ソリューションへジャンプ
1,842件の閲覧回数
WadeH
Contributor III

I've been using CW10.5 for three weeks, and have used it on both 9S08 and MCFMM256 projects. Everything went well until yesterday, when the 'debugger-launch' froze with no explanation.

Looking at the 'progress' tab shows this:

16449_16449.pngCW_Fault001.png

Unless one looks at the 'progress'-tab, it's not obvious what;s going on, and any number of launches can be attempted, each one leaving a 'Loading symbolic information' operation that can only be deleted by closing CW.

That "40%" occasionally gets as high as 60%, but never finishes. And clicking the red button does not affect the results, except to note that 'Cancel Requested.' No cancellation occurs

 

- Working with TWR-MCF51MM demo-board, using on-board OSBDM

- Windows 7, 64-bit

- Re-installed 10.5 with no change

- Installed 10.6 ditto

 

Does anyone have an idea on this?

 

Wade Hassler

ラベル(1)
0 件の賞賛
返信
1 解決策
1,662件の閲覧回数
BlackNight
NXP Employee
NXP Employee

That's definitely not normal. Can you check if you Zombie tasks running, see http://mcuoneclipse.com/2012/06/01/killing-me-softly-zombies-and-debugger-engines/

How much RAM does your Windows system have?

元の投稿で解決策を見る

0 件の賞賛
返信
4 返答(返信)
1,662件の閲覧回数
WadeH
Contributor III

... and now it doesn't. "All I did was®" add a table of 12-bit values for sin(x) to the source  and the problem vanished. It works as before and I can't make it fail (not that I tried too hard.)

I couldn't really call this one 'answered,' but I am done with it.

Wade Hassler

0 件の賞賛
返信
1,662件の閲覧回数
WadeH
Contributor III

Maybe talkin' to myself, but...

The problem appears to be slow-motion response from the debugger. The 'eternal symbol load' described above will complete, but takes at least fifteen minutes.

More: clicking the 'pause execution' button takes about a minute to happen (this was always the case.).

After a pause, the 'CPU Registers' window, as an example, is populated at about one value every five seconds: this used to be nearly instant.

Editing, re-compiling and reflashing the device has a turnaround time of minimum twenty minutes:

Wade Hassler

0 件の賞賛
返信
1,663件の閲覧回数
BlackNight
NXP Employee
NXP Employee

That's definitely not normal. Can you check if you Zombie tasks running, see http://mcuoneclipse.com/2012/06/01/killing-me-softly-zombies-and-debugger-engines/

How much RAM does your Windows system have?

0 件の賞賛
返信
1,662件の閲覧回数
WadeH
Contributor III

Thanks for the response.

Your explanation of zombie-processes makes sense, and I had already checked Task Manager several times looking for suspects: at one time there were two running copies of DE.exe, but only once, of a dozen or so instances.

It still sits at "Loading symbolic information: 40%" for a reliable 20 minutes. What exactly is going on in that operation?

I have 8GB of memory and have turned off all the big consumers ( big apps, multiple .pdf's etc.)

Wade Hassler

0 件の賞賛
返信