This is the mail archive of the ecos-patches@sources.redhat.com mailing list for the eCos project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: V850 patches


On Sun, 30 Mar 2003, Jonathan Larmour wrote:

> > v850_p2.diff - fixes the __skipinst function in v850_stub (the
> >                breakpoint inst on v850 is a "br *" and next_pc
> >                won't skip it) and adds the 2ram section in linker
> >                script.
>
> Odd... this hasn't caused problems before, and I don't know of anything
> that would have changed between then and now.

It works fine for the GDB stubs, but it doesn't work for RedBoot.
RedBoot starts the GDB stubs with breakpoint function which uses
HAL_BREAKPOINT(_breakinst). When you issue the disconnect command
to the GDB stubs you don't get back into RedBoot, but you stay inside
the stubs, becouse the mentioned breakinst doesn't get stepped over.
If you apply the mentioned changes then the disconnect command works
fine and you can get back to RedBoot from stubs.

> > v850_p3.diff - adds the hal_delay_us function and RedBoot hal options
> >                cdl component to ceb platform (I don't have the ceb platform
> >                so this patch is untested - but should work on any
> >                v850SA1/v850SB1 platform).
>
> I think hal_delay_us should probably live in the v850 HAL, not the
> platform HAL, although you can surround it with #ifndef HAL_DELAY_US in
> var_intr.h so that plf_intr.h can have the option down the road to
> override it if it needs to.

I agree, but then it would also make sense to move hal_clock_* functions.
Becouse all of these functions need the same timer (the one chosen for
ecos hartbeat).

> > NOTE: for RedBoot to work on ceb the size of reserved RAM should be
> >       increased in MLT RAM files.
>
> Perhaps you could add a new MLT file set for RAM under RedBoot? It doesn't
> need a new startup type, it just needs a new option which
> CYGHWR_MEMORY_LAYOUT can depend on. While I'm concerned about the often
> limited memory sizes on this system, I also don't want the issue forgotten
> about :-).

What about a cdl option CYGHWR_RESERVED_RAM?
So that we don't need to make a new file set.

savin


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]