Re: debian-installer and kernel status (pre- and post-sarge) on sparc

[Horms <horms@debian.org>]

On Wed, Apr 06, 2005 at 12:55:48PM -0700, Steve Langasek wrote:
> Horms,
> 
> On Wed, Apr 06, 2005 at 04:34:34PM +0900, Horms wrote:
> > I have been looking into the Blade1500 clock chip problem on request
> > from Dave Miller. It seems that while we have a solution in the pipeline
> > for 2.4.27, that is getting kernel-image-sparc 2.4.27-9 into testing and
> > d-i once gluck comes back,
> 
> No, this is certainly not in the cards; kernel-image-sparc 2.4.27-9
> build-depends on kernel-tree-2.4.27-9, which is not in testing and can't be
> updated without also updating kernel-image packages for all those
> architectures that use kernel-source without build-depending on
> kernel-tree-2.4.27-x.  Getting this fix in for 2.4 on sparc would mean
> destabilizing d-i on a number of unrelated archs (m68k, arm, mips).

I must say that while I understand the reasons for this,
I find it highly frustrating that what is clearly a sparc-specific
fix has to update all the other architectures as well,
making it to hard for sarge.

> > the same cannot be said for 2.6.8.
> 
> Actually, 2.6.8 is in much better shape for being updated, because after the
> last accidental acceptance of kernel-source-2.6.8 into testing, powerpc and
> m68k have been fixed up so that it's now possible to update
> kernel-source-2.6.8 in testing without disturbing the kernel-image packages,
> and to update kernel-image packages on one arch without disturbing the
> other.  So I'm in favor of getting a kernel-image-2.6.8-sparc update into
> sarge to fix 288180; it's up to Joey to decide whether this is really going
> to happen, though, and to set limits on what other changes are allowed in
> with it.  (For instance, I personally think that doing an ABI update at the
> same time and getting the security fixes in is ok, but that rearranging the
> set of modules being built is not ok.)
> 
> > The current d-i kernel image is linux-kernel-di-sparc-2.6 0.05.
> > I am not sure which version of kernel-source this is based on,
> > as a quick glance of the .dsc seems to indicate that the
> > source dependancy is not versioned. However it too seems
> > to be missing "m5823" and thus seems to not have the patch.
> 
> Yes, and unfortunately Josh's changelog entry is exceedingly vague; but from
> the date, it would have to have been built against k-i-2.6.8-sparc 2.6.8-6
> AFAICT.
> 
> > From the previous messaage I understand that there is some likely
> > hood of 2.6.8 being updated for sparc because of #288180.
> > It seems like that would be a good chance to get this patch in.
> 
> Yah, agreed.

Ok, so at least people with Blade's with this chip will be able to use
2.6. Its probably the recommended kernel for sarge on sparc anyway.

-- 
Horms