CCS/TMS320F28069: Global Variable Doesn't Display In Debugger Correctly After Break Point

Ki,

I thought it may be - so this is logged as: DBGTRC-4145
thought to be fixed but isn't.

OK, Thanks,
John

Yes, this issue currently remains open

Ki,

Any ETA on when this will be fixed? Puts the kabash on doing any floating point debug in the debugger right now.
I would think that's putting a lot of pressure on getting this issue resolved.

I'm shooting FP data out of the USB port so I can do debug due to this.

Thanks,
John

I just check and there is no ETA yet that I can report. I did give this issue a little "nudge". Hopefully that will move things along.

Thanks,
ki

Ki,

OK - something else that's kindof strange - I have some structures that have FP members - and those are OK in the debugger.

So, can a workaround to this be - instead of having globals - just shove these into a structure until the bug is fixed? Seems like a workaround
I can live with for the time being.

Thanks,
John

Ki,

As a workaround so I can use the debugger; I just put the variables I need access to (floating point) into a structure, so that seems to be OK.

Maybe this bug should mention this as a workaround until the bug is really fixed, tested, and verified across a few compiler/CCS releases.

Thanks,
John

Hi John,
Thanks for the tip. I will note the workaround in the bug description

Thanks
ki

Ki,

No problem - but this may have a caveat.

I was doing some debugging with a structure that started out like this:

typedef struct _pid {
float *pv; /*pointer to an <integer> that contains the process value*/
float *sp; /*pointer to an <integer> that contains the set point*/
float integral;
float pgain; // Kp: Proportional Gain
float igain;

And the debugger was having an issue with dereferencing (displaying) the first member - the pointer to float pv.

I added this as a workaround:

typedef struct _pid {
float dummy_pad;
float *pv; /*pointer to an <integer> that contains the process value*/
float *sp; /*pointer to an <integer> that contains the set point*/
float integral;
float pgain; // Kp: Proportional Gain
float igain;

And that worked OK - so, I think it's ok to use structures as long as the first member isn't a pointer to a floating point member.

Thanks,
John