[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Reducing the number of kernels



On Fri, Jul 30, 2004 at 12:06:44PM +0200, Jens Schmalzing wrote:
> Hi,
> 
> Sven Luther writes:
> 
> > I have 2.4.26 packages, but they are utterly screwed in the
> > arch/ppc/boot stuff.
> 
> How come?

Because it is the -benh tree, and he probably never tested it apart from its
yaboot situation.

> > I will upload the current stuff to svn, and try to work on the
> > issues (or maybe ask for Christoph's help) this weekend.
> 
> Actually, I built and tested packages tonight, and they work nicely on
> the hardware I have (all PowerMacs, of course).  I checked the source
> into svn as kernel-patch-powerpc-2.4.26-2.4.26 and put the packages at
> 
> http://www.theorie.physik.uni-muenchen.de/~jens/kernel-patch-powerpc-2.4

Which patch set did you use ? 

> > One idea was to move them to your build system, but i would still
> > prefer to have them not modular, since they provide an alternative
> > for the subarches which are not happy with the initrd way, and it is
> > too late to modify the 2.4 initrd stuff by now anyway,
> [...]
> > Christoph, Benh, Jens, what do you think about reverting the 2.4
> > kernels to use the linuxppc-2.4 tree over the -benh one ?
> 
> I think the 2.4 kernels should be feature frozen.  Keep the build
> system and the package structure in place.  Leave the config files as

Ok, this makes sense.

> they are and only add drivers that enter through new upstream
> releases.  Stay with the same patch and only forward port it.

Ok.

> Optionally, split the patch for clarity and try to get some of the
> chunks into mainline.

I did that in my packages. Well, sort of, i did split the patches, and put
them in a subdirectory of powerpc, and have a script which applies them or
something such.

> > we could stay with 2.4.25 for d-i maybe.
> 
> Then powerpc would be the only arch requiring a 2.4.25 kernel.

Yeah, but d-i tentative release is tomorrow so, ...

Friendly,

Sven Luther



Reply to: