Re: [linux-sh:03299] Re: Weird behaviour, possible compiler bug?

[Alex Bennee <kernel-hacker at bennee dot com>]

On Thu, 2004-08-19 at 16:52, Dan Kegel wrote:
> Alex Bennee wrote:
> > On Wed, 2004-08-18 at 20:33, Daniel Kegel wrote:
> > 
> >>Have you tried building a kernel with gcc-3.4.1 yet?
> > 
> > Is there a more upto date .dat/patches dir for gcc3.4? 
> 
> Nope.

Well I built it anyway without patches and built a booting kernel. So
its a start ;-)

> That's why you ran into trouble with gcc-3.3.3.  You hacked
> the .dat file to use a snapshot, but since the patches are
> taken from a directory named after the gcc version, you
> didn't get the patches.

Mmm, I'm sure I didn't - but I could be wrong. I was using the 0.27
crosstool to build it.

> If you are using something like gcc-3.3.4, but that is named gcc-3.3-foo,
> and you want the patches for gcc-3.3.4, then you need to do
>    cd patches
>    ln -s gcc-3.3.4 gcc-3.3-foo
> or something like that.  Ditto for 3.4.1.

Does crosstool throw errors out if the patches don't apply cleanly? I
assume they need to be re-diffed every now and again.

Are the gcc team taking the alt-arch patches and integrating them into
the mainline faster these days?

> So go back to your 3.3-ish snapshot, add the missing symlink
> in the crosstool/patches directory, and try again.  No need
> to play with gcc-3.4 if all you wanted was to get 3.3 working.

Ahh, I went ahead because I kinda assumed 3.4 was stable. So is 3.3 the
latest greatest stable gcc these days. I kinda lost track of there
versioning scheme.

> Have a look at crosstool/getandpatch.sh if you want to see
> how the patch stuff works.

Thanks for the pointers.

-- 
Alex, Kernel Hacker: http://www.bennee.com/~alex/

Man is the best computer we can put aboard a spacecraft ... and the
only one that can be mass produced with unskilled labor.
		-- Wernher von Braun


------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sources.redhat.com