[hatari-devel] Debugger breakpoints for Hatari variables too?

[Eero Tamminen]

Hi,

On Tuesday 04 August 2009, Eero Tamminen wrote:
> On Tuesday 04 August 2009, npomarede at corp.free.fr wrote:
> > > What would be utterly perfect (and is already available in STEem
> > > debug build) would be the ability to stop at a specific cycle within
> > > a frame(that is, PC relative to the current VBL/HBL in cycles), and
> > > to be able to set new breakpoints relative to the current one - in
> > > bytes from PC _or_ in cycles/NOPs from PC....
> > >
> > > This would make debugging synclock code a dream...
> >
> > in a previous mail eero talked about having some accessor functions ;
> > one possible other variable could be LineCycle, and you would get it by
> > calling Video_GetPosition(int *pFrameCycles , int *pHBL , int
> > *pLineCycles)
> >
> > This way, we would have HBL and LineCycle variables in the debugger and
> > it would be possible to do such breakpoints at a specific line
> > position.
> >
> > As for setting relative breakpoints in cycles that would mean to be
> > able to have a variable for FrameCycles (as returned by
> > Video_GetPosition) and then you could do the math youself to add the
> > desired cycles to this value and set a breakpoint where
> > FrameCycles=your_value.

Support for LineCycles and FrameCycles is now commited, I hope it helps in
debugging the video.c interrupt issues. :-)

Please test, it needed touching many places in breakcond.c so there could
be some regressions.  But after this it will be easy to add more accessors
or Hatari variables there.

	http://hg.berlios.de/repos/hatari/

I also commited the number base change support code and large update
to the manual.  I put a box around all the <pre> command line output
excerpts in the Manual, I hope you like them, see:
http://hg.berlios.de/repos/hatari/raw-file/tip/doc/manual.html#The debugger


> > I think that when we enter debug mode on a breakpoint (or maybe also
> > when pressing pause), it could be useful to display nVBLs,
> > *pFrameCycles, *pHbl and *pLineCycle by calling Video_GetPosition. This
> > way, you know at what screen position your breakpoint triggered, which
> > is very useful when you want to debug overscan or other video-sync
> > related code.

On Tuesday 04 August 2009, npomarede at corp.free.fr wrote:
> On Tue, 4 Aug 2009, Eero Tamminen wrote:
> > Btw. I forgot to mention that now "b help" will both list the supported
> > Hatari variables and show their currently values:
...
> Yes, that's a good option in case you want to see them again, but perhaps
> a 1 line summary could be nice anyway when entering the debugger ?

Do you have a proposal patch for debugui.c?  I'm not sure what it should
look like...


Once my debugger changes have been tested (and I've fixed any new bugs)
and the VDI mode interrupt assert is fixed, Hatari's ready for release from
my point of view.

(Maybe the STE screenshots and Virtual City screenshots Nicolas suggested
could also still be fixed on the www-site.)


	- Eero