CCS: Non #0 BKPT still causing debugger step issue

Peter,

If I look up the bug id it shows it having been fixed years ago.  However that was in our driver for Cortex A.  It is possible that our Cortex M driver still has the same issue.  I will setup a Cortex M device and give this a try.

Regards,

John

Peter,

I have tried this out on MSP432E401Y and TM4C1294.

Which compiler are you using? Using the TI compiler I can't build without the #.  However even if I do BKPT #123 I am not able to step over the instruction.

When working with the Cortex A device in the other thread I would have been using GCC.

Regardless I think there is a bug here.

John

Hi John,

thanks for looking into this so promptly. I guess this is an advantage of the time zone difference. :-D

I'm using TI v 18.1.3 LTS on TM4C129 family processors.

Cheers,

Peter

No problem, sometimes the timezone difference works in your favour.  I will switch my compiler to that tomorrow and see if I can make it work without the #.

I have filed DBGTRC-4992 to track the issue.

John

Thanks John. I'm not bothered by the need or not for # (it's documented in the "Cortex-M3/M4F Instruction Set" as being there anyway).

Fixing the step over issue in the debugger however would be a definite plus.

Now, if you wanted to have the stack unwound correctly in exception handlers that would be a major boon! :-)

Peter

Here is a summary of what I see.  I changed my compiler to be the same as yours.

    bkpt 123 ;does not build, must be an absolute constant error

    bkpt #0  ;builds, unable to step past

    bkpt #123 ;builds, unable to step past

Are you able to step past bkpt #0?

I am using the latest CCS I am going to setup 8.1.0 to see if I can step past bkpt #0 in that.  

John

In CCSv8.1.0 I see the same as in v9.  I can't step past any bkpt instruction.

I will update the bug report as to me this is worse.

Hi John,

No, I can't step past a BKPT #0.

I'm also interested to note that CCS has gotten up to version 9 while I wasn't looking. In the past it had been pretty good at letting me know when a new version was available, but it looks like that hasn't been working for quite a while - since 8.1.0.11 anyway and maybe before that.

For major versions the update check won't pick them up as it requires a new installation.  However for the 8.x versions the update check should be bugging you that there are updates available.  Maybe you disabled the update check?  It can be a bit annoying as there is almost always some sort of update available (compiler, device support, ccs service release...).

If you do a "check for updates" from the help menu does it show anything?

9.1 will be out around the end of the month if you want to wait for that one.

I've updated to the current 9.x and I see that it is prompting me with updates already :-D.