2.4 & 2.6 kernels, should sarge be 2.6 only at least for powerpc ?
On Sun, Jun 27, 2004 at 05:29:38PM +0200, Christoph Hellwig wrote:
> There's a few reports against 2.4 kernel that are fixed in 2.6 and are
> unlikely to get in 2.4 every (Examples: #146956 or #130217). How should
> we deal with them in the BTS?
The real question here is to ask ourselves what is our option for the
sarge release. Will we release with 2.4 as default, which is the track
we are on right now, or will we release with 2.6 as default, and keep
2.4 about only as backup in case there is a real problem with 2.4.
There are both advantages and problems in going with 2.6 :
advantage: it is the future, has some features and fixes which will
not be backported to 2.4, and moreover many of our new kernel team
have no interest whatsoever for 2.4, which includes benh and Christoph
among others.
problems: not all architectures support 2.6 yet (well, most of them do
not), and moreover, our userland has probably not been fully tested
with 2.6 all that much.
So, the real question, for those arches which do support 2.6, and if
those bug reports you mention are problems only on those arches where
2.6 is supported, and if we decide to go for 2.6, then it should be ok
to mark those bugs as wontfix, and put a note that it is fixed in 2.6.
If on the other hand we decide to go with 2.4 by default, or those bugs
affect arches which are not ready to go with 2.6, then not only it is
not ok to close them (even if our new kernel team doesn't care for 2.4),
but we should either backport the fix, or find another way to close it
before the sarge release.
Now, about going with 2.6, i personnally would maybe like to go with 2.6
eclusively for all the powerpc subarches, altough i am not entirely sure
we are ready for this. For this to happen we need to achieve the
following :
Have a kernel bootable on all subarches :
-> yaboot using newworld pmac & chrp-rs6k : Ok, but need testing on
chrp-rs6k
-> mkvmlinuz generated chrp : Need to find a solution for the
generation of the vmlinuz image, should be easy, once we agree on a
way to go.
-> oldworld pmac : We need to shrink the size of the kernel so it
fits on a miboot floppy and test it. This should be best achieved by
modularizing the pmac-ide driver, and other pmac stuff which could
be modularized. Benh said he scarcely has time for it, and Christoph
promised he would have a look.
-> prep : renamed pplus in the kernel code. We need to add mkvmlinuz
code for this one, not sure about the others, we did not support
them, but it should be possible to add support to mkvmlinuz easily
enough. Testing on those subarches is needed though.
-> apus : Well, a 2.6 port could be done and tested, using a
conditionally applied patch or something such, or merging the
patches. That said, since there are at most 5-10 users left, and
those are using their own kernels, maybe we should drop kernel
support for them.
Another point would be to test the 2.6 debian-installer on all those
subarches, and fi the problems if they appear.
If all this does happen before the sarge release, and if the userland
issues are solved, then i would strongly recomend going for 2.6 for
powerpc at least, especially as the members of the debian kernel team
with interest in powerpc care very little about 2.4 kernels.
Friendly,
Sven Luther
>
>
> --
> To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
>
Reply to: