On 01/07/2016 12:25 PM, Josh Boyer wrote:
Adding the kernel list on CC because this is a good question and
there's no reason to not answer a broader number of people.
On Thu, Jan 7, 2016 at 3:17 PM, Andrew Lutomirski <andy(a)luto.us> wrote:
> Other than the bug I filed, 4.3 is missing two things for full Dell Skylake
> The Broadcom driver, available in 4.4, needed for WiFi on some models.
Have to think about this. We tend to avoid full driver backports and
4.4.y will be released fairly soon. I don't believe 4.3 will live
very long in F23, perhaps a month at most so doing a full backport
might not be worthwhile.
> The intel-hid driver, queued for 4.5, needed for the WiFi button on all
Similar answer. We'd be more likely to backport it to 4.4 than 4.3.
And in both this and the above case, it would probably have to go into
rawhide and live there for a while to get some testing.
> I don't know about the former, but the latter is self-contained and trivial
> to backport. What's your/Fedora's opinion on driver backports?
So we have no fast and hard rules, but as I said we tend to avoid full
driver backports. Feature additions to existing driver and bugfixes
are much easier to carry. In either case, the code needs to be
upstream or very clearly headed there in a not distant release. It's
probably best to think of our approach as a slightly less restrictive
version of the upstream stable process. We'll occasionally take
things from trees other than Linus' but only if it looks like it is
heading there soon anyway.
If a driver is actually present in Linus' tree, I'd encourage interested
users to test a rawhide kernel. If it's working there, there's a much
stronger case to consider pulling it into a stable branch.