Kernel numbering query
davej at redhat.com
Fri Mar 5 19:56:56 UTC 2004
On Fri, Mar 05, 2004 at 03:10:53PM +0000, M A Young wrote:
> For a matter of interest can anyone tell me why we have suddenly jumped
> from kernel-2.6.3-1.118 to kernel-2.6.3-2.1.238. In particular why 238 and
> is it coincidence that 238=(118+1)*2)?
Big change of strategy. Until yesterday, we had two 2.6 kernels.
The FC2 kernel which I was maintaining, and Arjan's RHEL4-gonnabe kernel.
After a lot of discussion about it, we decided to merge the differences,
and over the last week or so, the differences between the two got less
and less, to the point where the only differences were some extra features
in the RHEL4 kernel. At this point, we realised that the amount of work
thats getting done twice (like updating to the latest upstream snapshot)
is a waste of resources, and testing for a lot of folks got a little confusing
"I see this bug in Daves 116 kernel but not in Arjan's 214 kernel" requires
quite a bit of thinking to figure out the differences.
So.. to cut a long story short, FC2 got the RHEL4 kernel.
Which means Fedora users gained..
- The 4g patch for big boxen
- The syscall auditting infrastructure
- A couple other smaller patches
And from our side of the fence..
- Consolidated bug reports (RHEL kernel bugs == Fedora kernel bugs)
- Shared fixes. For Fedora 1, it was hard at times for some of our
developers to take time away from RHEL to work on Fedora, which has meant
a lot of bugs have been lingering much longer than any of us would have liked.
For FC2, any fixes those guys come up with for RHEL are applicable to Fedora.
- Less duplication of work between Arjan and myself.
Leaving us both a lot more time to do other fun stuff like
a) fixing bugs. b) pushing bits back upstream [This is still an important
goal of the Fedora kernel].
The "lets stay close to mainline" mantra is something that both Arjan and
myself both think makes sense to try and continue.
The larger of the patches we now have in the kernel
(things like 4g & the audit code actually stand a chance of going into upstream
2.6 at some point, and apart from Tux, the other bug stuff was already in
the FC2 kernel anyway).
So, there you have it. I now expect to be duly flamed for my part in this,
but I'm hoping you'll all see the benefits of this far outweigh any downsides
that you may come up with. So, flame on 8-)
More information about the test