LPCXpresso IDEのトピックRe: IDE 5.2.2 Additional Debugger Hassles

IDE 5.2.2 Additional Debugger Hassles

lpcware — Thu, 16 Jun 2016 01:04:49 GMT

Content originally posted in LPCWare by MikeSimmonds on Wed Apr 24 06:37:29 MST 2013
[FONT=Tahoma][SIZE=2][XP Pro/SP3 32 Bit]
In no particular order.

1) Memory view still defaults to hex rather than traditional
and still does not remember either settings or addresses

2) Scrolling the disassembly window up over a function start
often (as in nearly always) sets the address column to [I]odd[/I] addresses.
This is nonsense as Cortex opcodes are always on even addresses.

3) The espr part of the processor status register (when in an IT block)
always seems to show a condition of 'EQ' regardless of the actual code
[Definitely seen this when it should have been 'NE']

4) Breakpoints/Move to Line/Run to Line often stop at the instruction
[I]after[/I] the one chosen.

[/SIZE][/FONT][FONT=Tahoma][SIZE=2]5) Selecting (or not) opcode bytes in the disassembly window does add
and remove a column to the list view -- unfortunately it is only about 5 pixels
wide (and blank) and there does not seem to be any way to widen it (by hand).
This one is particularly disappointing. :(

[/SIZE][/FONT][FONT=Tahoma][SIZE=2]6) There doesn't see, to be any way of getting the PC address of a breakpoint
(Properties doesn't help). Neither can I find a way to set an arbitrary bp at a
hex address (via a dialog box).

7) In one case, the debugger was 'suspended:step' but F5/F6/F8 would not
move on. Had to kill the debug session.
[Sorry, but I haven't been able to repoduce this one yet.]

[/SIZE][/FONT][FONT=Tahoma][SIZE=2]For the record, the part is LPC1778 which, as you know, is Cortex-M3.

[/SIZE][/FONT][FONT=Tahoma][SIZE=2]Cheers, Mike

[/SIZE][/FONT]

Re: IDE 5.2.2 Additional Debugger Hassles

lpcware — Thu, 16 Jun 2016 01:04:49 GMT

Content originally posted in LPCWare by MikeSimmonds on Wed Apr 24 07:15:26 MST 2013
[FONT=Tahoma][SIZE=2][COLOR=Red]OOPS! I updated the wrong post!
Please read this in the context of
http://knowledgebase.nxp.com/showthread.php?t=4445
[/COLOR]
Things just got worse!

I exited LPCXpresso, unplugged the USB of the LPC-Link, repowered my PCB.
I used Process Explorer to verify the LPCXpresso and the debug components
were gone.

Restart LPCXpresso and plug LPC-Link back in.

Now I can't run/set bp/step etc.
The only thing it lets me do is single step in instruction trace mode.

Memory Views still work; this is the FPB register area.
Where these bp's came from after restarting everything is a mystery to me.
Perhaps you could explain.

Oh, the debug profile is set to load false, attach true.

[SIZE=1][Code]
61020000 00000020 B19A0140 E99D0180 999E0180
CD9E0180 F99E0180 7D9F0140 00000000 00000000
00000000 00000000 00000000 00000000 00000000
[/Code][SIZE=2]
Actually I think the memory export has endian swapped these numbers!

[/SIZE][/SIZE] TTFN, Mike

[/SIZE][/FONT]

Re: IDE 5.2.2 Additional Debugger Hassles

lpcware — Thu, 16 Jun 2016 01:04:50 GMT

Content originally posted in LPCWare by CodeRedSupport on Wed Apr 24 10:30:22 MST 2013

Quote: MikeSimmonds
[FONT=Tahoma][SIZE=2][COLOR=Red]OOPS! I updated the wrong post!
Please read this in the context of
http://knowledgebase.nxp.com/showthread.php?t=4445
[/COLOR]
Things just got worse!

I exited LPCXpresso, unplugged the USB of the LPC-Link, repowered my PCB.
I used Process Explorer to verify the LPCXpresso and the debug components
were gone.

Restart LPCXpresso and plug LPC-Link back in.

Now I can't run/set bp/step etc.
The only thing it lets me do is single step in instruction trace mode.

Memory Views still work; this is the FPB register area.
Where these bp's came from after restarting everything is a mystery to me.
Perhaps you could explain.

Oh, the debug profile is set to load false, attach true.

[SIZE=1][Code]
61020000 00000020 B19A0140 E99D0180 999E0180
CD9E0180 F99E0180 7D9F0140 00000000 00000000
00000000 00000000 00000000 00000000 00000000
[/Code][SIZE=2]
Actually I think the memory export has endian swapped these numbers!

[/SIZE][/SIZE] TTFN, Mike
[/SIZE][/FONT]

Leave the "Load Image" debug configuration option set to 'True'. The plan is to remove this by the next release. If you want to attach to a running target, use only the 'attach' option. This option ensures the image does not get loaded, and the execution state of the target is unaffected.

Regards,

CodeRedSupport

Edit: Be certain you update the flash contents after any project/source change if you use a debug configuration that attaches to a running system. It's easy to forget the symbol/line information gets out-of-sync with the flashed image.