"Undocumented Error" while Debugging

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: 

"Undocumented Error" while Debugging

1,183件の閲覧回数
lpcware
NXP Employee
NXP Employee
Content originally posted in LPCWare by Sal Ammoniac on Thu Aug 21 17:41:21 MST 2014
I'm trying to debug some code on an LPC4357, but I hit a point in the code where the debugger keeps trowing the following errors and won't let me continue:

10: Could not start from stop: Ee(FF). Undocumented error.
15: Target error from Register access: Ee(FF). Undocumented error.

This is very frustrating and not letting me debug the problem.

Any ideas?

Once this happens, I have to terminate the debug session and start over, but I always hit this point and there doesn't seem to be any reasonable reason why this always occurs. There is nothing more frustrating than a debugger that won't let you debug.
0 件の賞賛
返信
8 返答(返信)

1,163件の閲覧回数
lpcware
NXP Employee
NXP Employee
Content originally posted in LPCWare by lpcxpresso-support on Tue Sep 02 16:47:00 MST 2014
Sal,

This will be fixed in the forthcoming LPCXpresso 7.4.0 release later this month.

Regards,

LPCXpresso Support
0 件の賞賛
返信

1,163件の閲覧回数
lpcware
NXP Employee
NXP Employee
Content originally posted in LPCWare by lpcxpresso-support on Mon Aug 25 16:35:05 MST 2014

Sal,

We've reproduced your problem, and we'll let you know when a fix is available.

Regards,

LPCXpresso-Support
0 件の賞賛
返信

1,163件の閲覧回数
lpcware
NXP Employee
NXP Employee
Content originally posted in LPCWare by Sal Ammoniac on Sat Aug 23 16:56:27 MST 2014
Here's an example project cut down to the bare minimum that exhibits this behavior.

If a breakpoint is set in the PendSV handler, the debugger hangs with and error dialog and will not go any further without resetting the board.

If the breakpoint is not set, then everything runs fine.

For some reason, setting a breakpoint in the PendSV handler causes this strange behavior.

If I turn off hardware floating point (by changing "FPv4-SP (Soft ABI)"  to "None") in the properties dialog, this problem does not occur when the breakpoint in the PendSV handler is hit.
0 件の賞賛
返信

1,163件の閲覧回数
lpcware
NXP Employee
NXP Employee
Content originally posted in LPCWare by lpcxpresso-support on Sat Aug 23 08:44:33 MST 2014
Please provide an example project, that we can try here, that causes the problem. To reiterate, unless we can reproduce the problem here, we cannot help.
0 件の賞賛
返信

1,163件の閲覧回数
lpcware
NXP Employee
NXP Employee
Content originally posted in LPCWare by Sal Ammoniac on Fri Aug 22 10:28:34 MST 2014
Here's more info related to this issue:

The problem happens right after my code does a context switch between two tasks that use the FPU. The context switch code saves and restores task context (CPU registers and FPU registers) on the task's stacks. The actual error is a Usage Fault since the debugger stops on the first instruction of my usage fault handler. At that point, the debugger is useless. I can't look at CPU registers and I can't read memory, so I can't determine which instruction caused the fault or determine which of the several possible usage faults caused this.

I suspect what is happening in my code is a misalignment of the stack when popping registers so that the PSR ends up with the thumb bit not set, or something like that. With the debugger being hung at that point means that I have to guess what's wrong rather than being able to examine the stack and registers and determine the cause directly.
0 件の賞賛
返信

1,163件の閲覧回数
lpcware
NXP Employee
NXP Employee
Content originally posted in LPCWare by lpcxpresso-support on Fri Aug 22 05:47:21 MST 2014

Quote: Sal Ammoniac

I want to reiterate:…..



But equally, unless we can reproduce, there is little we can do. Thus the more information you provide, the better the chance we can get to the cause of the problem.

Regards,
LPCXpresso Support
0 件の賞賛
返信

1,163件の閲覧回数
lpcware
NXP Employee
NXP Employee
Content originally posted in LPCWare by Sal Ammoniac on Fri Aug 22 05:28:07 MST 2014
Yes, I am using LPCXpresso 7.3.0.

I'm debugging an RTOS, specifically the context switch code that saves/restores the floating point registers.

I want to reiterate: no matter how buggy my code, that should not cause the debugger to lock up and throw errors the way it does.
0 件の賞賛
返信

1,163件の閲覧回数
lpcware
NXP Employee
NXP Employee
Content originally posted in LPCWare by lpcxpresso-support on Fri Aug 22 01:56:30 MST 2014
Can you provide some more actual details of what you have been doing in the lead up to  encountering this error?

Can you also confirm that you are using the latest version of LPCXpresso (currently 7.3.0) ?

Regards,
LPCXpresso Support
0 件の賞賛
返信