On Tue, 5 Dec 2000, Philippe Moutarlier wrote:
> Michael,
>
> There are TONS of messages on the net (some from Alan Cox which seems to
> know what he is talking about) about the fact that the gcc 2.96 is NOT a
> release, that it produces broken kernel. It is easy to find on the net :
> search for "checksum.S badly punctuated".
There were, and are _TONS_ of messages in the archives of the
mailinglists (try www.moongroup.com) also discussing the use of kgcc to
build kernels in 7.0, and the reasons why they chose to release with
2.96. There are questions on both sides about the issues involved, but
Red Hat didn't release a system that couldn't build it's own kernel.
Red Hat's position basically states that it was done as a transition
state as we're moving towards a new glibc/gcc/kernel environment.
>
> The gcc web site has a special page about this and is warning that some
> distributions
> are distributing this compiler which is not, and will never produce code
> compatible with either 2.95 or 3.0 . They even claim that such release does
> NOT exist from their point of view.
Again, discussed to death on the lists. The incompatibility is at the
binary image; it's claimed that the library/source code level interfaces
are correct for 3.0, and that all that will be required is a
recompile. IIRC, a part of RH's position was that there will be a break
during the transition from 2.95 to 3.0, and that break, in many cases,
will mean that some programs will not compile with 3.0 until they have
been modified. 2.96 is supposedly compatible at the source level with
3.0, thus the theory is that we're doing a stepwise move towards 3.0.
>
> Now, why in the world RH keeps putting untested, unsupported code in their
> "stable" distribs ?
>
> How is that they didn't try to compile their own kernel with the compiler
> they distribute ?
They do. See the notes on using kgcc, which, IIRC, is actually
2.95. Kernel move towards 3.0 is incomplete and therein lies the
problems we're encountering now.
>
> Why was the glibc distributed in 6.2 bugged ?
> Why is the glibc distributed in 7.0 bugged ?
>
> It might happen you did not upgrade gcc. But the pb is very well known and
> as far as I can tell there is NO update or errata for it at the redhat site.
>
> Now, it is simple: I deinstall their compiler, go and find another STABLE
> one, recompile my kernel which was not properly installed in the first place
> (why ?) and I lost a full day "upgrading".
Unnecessarily so. Could have just used kgcc for everything.
>
> So, no I am not happy with the job RH is doing as it is not the first time
> it happens.
>
> Sorry, I am tired of this.
>
> And big thanks for your help :=)
>
> Philippe
Bit more research would have saved some time. Believe the use of kgcc
to rebuild kernels is discussed clearly in the RH 7.0 docs,
also. Which are available for free download on the net.
best
rickf
Rick Forrister <[EMAIL PROTECTED]>
Definition: Honest Politician: Once bought, stays bought."
--Robert Heinlein
_______________________________________________
Redhat-list mailing list
[EMAIL PROTECTED]
https://listman.redhat.com/mailman/listinfo/redhat-list
