This is the mail archive of the ecos-discuss@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]

Re: Creating a drop in replacement for the pcmb package


Ian Campbell wrote:
> 
> On Fri, 2001-09-28 at 18:52, Jonathan Larmour wrote:
> > The idea is more that the PCMB package collects together those parts common
> > to PC-like boards. It doesn't itself provide an abstraction - instead it
> > provides those facilities which the underlying platform HAL (the PC HAL)
> > would use.
> >
> > In your case, by the sound of it, you would still use the pcmb HAL _with_ a
> > new merlin platform HAL. That merlin platform HAL would do the equivalent
> > of the PC HAL (i.e. hal/i386/pc) for your merlin board.
> 
> Aaah, I see. I got it backwards. Happily, once I started thinking this
> way round things were much simpler ;-)

Phew :-).

> Most of the abstraction seems to be in place already and the attached
> 'pc-dependency-fixes.patch' includes two tiny little patches which
> remove the two dependencies from pcmb and arch on the pc package which I
> tripped over. There might be others but they haven't bitten me as yet.
> There's even ChangeLog entries this time..

Cool! I've applied this.
 
> The other attached patch (memsize-fixes.patch - which I am posting more
> for comment than inclusion since it may not be the correct approach) is
> to allow the memory size to be hardcoded. The display attributes in the
> CDL have FIXMEs because I am not entirely sure how to describe what they
> are. I'm also not convinced by my choice of option names.

That's very good for a CDL newbie though! I think CYGNUM_HAL_MEMSIZE would
be better as CYGPKG_HAL_I386_PCMB_MEMSIZE. 

I suppose there's one bit of rationale for the pcmb package I'm not 100%
about, which is that a few of the things it does are tied to the existence
of a BIOS, which I would have thought is rather PC specific. I strongly
suspect the idea really is to make common those bits that may be common
between many PC-like bits of hardware, but for specific cases, we would
need to make the pcmb package configurable (as you've done).  Hopefully
Nick (who is on holiday right now) might give more insight, but for now,
work on the assumption that this is the case :-). 

> I would like to be able to change the value of CYGNUM_HAL_RTC_PERIOD in
> hal_i386_pcmb.cdl, however it is defined using a 'calculated' statement.
> Is there any problem with making this into a 'default_value'?

Yeah, for now; although I'd be tempted just to move it down into the
platform. I take it the merlin doesn't use an 8253 compatible PIT then?
 
> My last change is to move romboot.S + associated CDL into the pc package
> but I've not gotten round to it yet. I'll post a patch when I have.

> > In either case you should probably consider looking at the
> > various legal technicalities for contributing to the main tree:
> > http://sources.redhat.com/ecos/faq.html#contrib_assign
> 
> This is something which I will need to discuss with my boss at some
> point. Hopefully the attached patches are trivial enough not to warrant
> it (He said with his fingers crossed...)

Yes, these ones are.

> > > This message has been checked for all viruses by MessageLabs Virus Control Centre.
> >
> > I'm always amused by these lines.
> 
> Our mail server just started adding them recently, I wonder how many
> it's caught so far...

Except obviously any new virii it doesn't recognise would also have this
line appended :-).
 
Jifl
-- 
Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine


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