Re: 2.0.25 - as errors STILL stop compile

Rob Glover (
Mon, 11 Nov 1996 14:59:30 -0800 (PST)

On Mon, 11 Nov 1996, Robey Pointer wrote:

> Lindsay Haisley said this, and I just had to respond:
> >
> > I also tried upgrading my gcc from 2.7.2 to as was suggested (using
> > the bin file from ftp://sunsite/, but it was a
> > disaster! gcc quit on the first file with the following complaints:
> >
> > > gateway:/usr/src/linux# make zImage
> > > gcc -D__KERNEL__ -I/usr/src/linux-2.0.25/include -Wall -Wstrict-prototypes
> > > -O2 -fomit-frame-pointer -fno-strength-reduce -pipe -m486 -DCPU=486 -c -o
> > > init/main.o init/main.c
> > > stdin:1: numeric constant with no digits
> > > stdin:1: numeric constant with no digits
> > > stdin:1: invalid #line
> I had this same problem with, and had to spend a huge amount of time
> reverting. This is such a fundamental problem that I'm really surprised that
> anyone has gotten it to work, much less is advising people to upgrade to it!
> Possibly it should have been marked as a beta version. Definitely it should
> not be required for new kernels. :/
> What seems to be happening is that gcc can't understand the output of gcc -E.
> I'm not kidding.
> Can I suggest that the kernel's gcc requirement remain at merely the cutting
> edge (2.7.2) and not require beta versions that don't work yet? This is
> getting ludicrous.
> Robey

Pardon me? Baking Soda? This GCC works just fine.. that error above is
from installing it incorrectly.. I have had it installed <the the
gcc- from tsx-11> and it works flawlessly. The ONLY
difference between and 2.7.2 is the strength reduce bug is taken
out <the whole strength reduce optimization was removed>. That is it :>