m68k gcc was Re: BSP for 5307
pbarada at mail.wm.sps.mot.com
Fri Feb 15 17:45:58 UTC 2002
>> I've tried building a *straight* g++-20020211, and it dies in
>> different ways for lots of targets (arm-elf, m68k-elf, sh-elf, etc).
>> I'm now starting over by pulling out *every* stock tool (gcc-20020211,
>> newlib-1.9.0, binutils-2.11.2, gdb-5.1) in a hopeful attempt to prove
>> that *I* screwed up somewhere along the line...
>I don't know why you are having trouble. I report results on about
>3 dozen cross targets and don't have any problems that are that bad.
>Sometimes things don't build but those targets aren't that bad. :)
I don't either. That's why I'm going back to first principles and
trying it again :-(
>> I've only found patches against gcc-2.95.3 on www.fiddes.net, so I
>> don't think Dave has any 3.0.x patches...
>Do we need to find out plans for another 2.95.x release or are you
>just workign to ensure future 3.x releases are OK?
>From what I've heard on the gcc list, they're not planning on making
one, but I may be wrong. If the FSF is willing to make another
release of 2.95, I'll generate a set of changes for that, but only in
my spare time.
I used gcc-2.95.3 as a starting point for making my v3/v4/v4e changes
work, but it was with a *very* heavy hand(they aren't complete and I'm
not sure I didn't mess up the 68k side!). It will take a lot of work
to start from a clean 2.95.3 and integrate the changes all over again.
I started over again in gcc-3.x, and was a bit more careful this time.
Unfortunately the only boards I have to test against have 5407's on
them, so YMMC on the 68k side.
Peter Barada Peter.Barada at motorola.com
Wizard 781-852-2768 (direct)
WaveMark Solutions(wholly owned by Motorola) 781-270-0193 (fax)
More information about the users