On Thu, 6 Sep 2001, Jarom=EDr Dolecek wrote:

> Bill Studenmund wrote:
>
> Yeah, though I wonder if it's actually needed for native ECOFF
> binaries? It has been suggested to me to ask on port-alpha for testers,
> which I'd do (with a patch switching to regular 'setregs'
> for native ECOFF).

Early mips stuff (pmax) used ecoff too.

If the build lab has either a pmax or an alpha, you might be able to find
an *OLD* NetBSD archive, and play with some of the files from it. Like a
cp from 1.2 or so (I don't know if 1.2 used ECOFF or ELF, but something of
that vintage is what I think you'd need). That way you could do the
testing (I've always found being able to do the testing myself makes for
shorter debug cycles - I'm not just trying to dump work in your plate).
:-)

> > What functional difference does the change make?
>
> None.  It's matter of code organization, and proper layering. The propose=
d
> way would also be less error prone, and one thing less to think of
> when adding execution support for emulations. The fact that this
> would help LKMs is by-product; if I'd aim for happy exec LKMs only,
> I'd split the emul setregs to separate files and make every
> exec LKM include appropriate setregs.

I guess what I had in mind was looking at what the different routines do.
By inspection, can't you tell if things are similar or not?

Take care,

Bill