Semantics aside...
Yes, you need to be aware of pipeline latencies. That being said,
you certainly can use it in C code display in a minimally-optimized
project. If you do have reason to question the location of the
break, changing the view to mixed source/asm for a moment will clear
up any doubts.
As far as what changes they may or may not have made to the family
over time wrt this feature, I don't know.
Gary
--- In c..., Jeff Brower wrote: > To add to the discussion, SPRA841 refers to
"watchpoint triggers" and not data > breakpoints. Watchpoint triggers are subject to
pipeline latency and apparently need > to be monitored in asm instruction display, not C
code display.
>
> SPRA841 makes reference to data-matching improvements in future 55x devices: >
> Watchpoint Limitations
> The present implementation is the first step on the roadmap to extensive > debugging capabilities within the C55x family of
DSPs. It still has some > limitations as can be seen from the previous
sections.
>
> although I'm not sure that TI has actually added data breakpoints for the 55x family. >
> -Jeff
>
Reply by Jeff Brower●October 3, 20072007-10-03
Gary-
> Anyone can do this with 55x under CCS 2.1 (I would
assume it is in CCS3
> as well). Under Tools, select C55xx Emulator Analysis then setup
> breakpoints based on the accesses you want trapped.
>
> Look on the TI website for the following.
> SPRA841 Debugging TMS320C55x Applications Using Emulation Analysis
To add to the discussion, SPRA841 refers to "watchpoint triggers" and not
data
breakpoints. Watchpoint triggers are subject to pipeline latency and apparently
need
to be monitored in asm instruction display, not C code display.
SPRA841 makes reference to data-matching improvements in future 55x devices:
Watchpoint Limitations
The present implementation is the first step on the roadmap to extensive
debugging capabilities within the C55x family of DSPs. It still has some
limitations as can be seen from the previous sections.
although I'm not sure that TI has actually added data breakpoints for the
55x family.
-Jeff
Reply by iewil2000●October 3, 20072007-10-03
Thanks for the input. SPRA841 is the one.
--- In c..., "smee642003" wrote: >
> --- In c..., Jeff Brower wrote:
> >
> > Wei-
> >
> > > Is there any way to be notified by CCS in debug mode when a
> particular
> > > address is written? One of my buffer seems be polluted by other
> data,
> > > but I don't know which line of code did it and in what
> circumstances.
>
> Anyone can do this with 55x under CCS 2.1 (I would assume it is in CCS3 > as well). Under Tools, select C55xx Emulator
Analysis then setup
> breakpoints based on the accesses you want trapped.
>
> Look on the TI website for the following.
> SPRA841 Debugging TMS320C55x Applications Using Emulation Analysis
>
> Hope this helps,
> Gary
>
Reply by smee642003●September 30, 20072007-09-30
--- In c..., Jeff Brower wrote: >
> Wei-
>
> > Is there any way to be notified by CCS in debug mode when a particular > > address is written? One of my buffer seems be
polluted by other data, > > but I don't know which line of code did it and
in what circumstances.
Anyone can do this with 55x under CCS 2.1 (I would assume it is in CCS3
as well). Under Tools, select C55xx Emulator Analysis then setup
breakpoints based on the accesses you want trapped.
Look on the TI website for the following.
SPRA841 Debugging TMS320C55x Applications Using Emulation Analysis
Hope this helps,
Gary
Reply by Jeff Brower●September 27, 20072007-09-27
Wei-
> Is there any way to be notified by CCS in debug mode
when a particular
> address is written? One of my buffer seems be polluted by other data,
> but I don't know which line of code did it and in what circumstances.
"...the UBM allows programmers to set watchpoints or data breakpoints..."
So evidently you need CCS 3.3. As a new feature, my guess would be that 64x+
devices
are supported first... whether 55x is supported would be a question for TI.
-Jeff
Reply by iewil2000●September 27, 20072007-09-27
Dear group,
Is there any way to be notified by CCS in debug mode when a particular
address is written? One of my buffer seems be polluted by other data,
but I don't know which line of code did it and in what circumstances.