RE: Trunk builds for h8300-hms

["Gary R Van Sickle" <tiberius at braemarinc dot com>]

> Hi Gary,
>
> > GCC from the cvs tip builds on Cygwin as a cross-compiler to
> > h8300-hms.  The resulting compiler compiles my non-negligible amount
> > of Hitachi code without compile errors.  The compiled code works to
> > some extent, but not fully, as it did when compiled with 2.9x.  I
> > don't know if that's the compiler itself, or perhaps something my
> > code is doing 'wrong' that made it through 2.9x and not 3.x.
>
> I am interested in hearing what's possibly going wrong.

Yeah, you and me both. ;-)

>  h8300 port of
> gcc from CVS is fairly in good shape even though it is not considered
> stable. :-)
>

Well that's partly what I was trying to get across, that it does seem to be
in pretty good shape overall.  My code is quite extensive, and of course
embedded, so there's a lot of low-level stuff that could be wrong on one end
or the other ('*(volatile unsigned char *const)' craziness etc). I've been
looking at the assembly and haven't come up with any obvious problems yet
(e.g. it seems to handle the above sorts of constructs fine).

As an idea of how well it is working, I have a fairly full-featured
FAT-compatible file system in this code, complete with 'scandisk'-like
functionality and fault-tolerance, and it appears to work flawlessly.

So anyway, I'll keep beating on it until I figure out what one of us is
doing wrong.

> Kazu Hirata

Gary R. Van Sickle
Braemar Inc.
11481 Rupp Dr.
Burnsville, MN 55337