Santiago Ruano Rincón [off-list ref] writes:
El 02/09/20 a las 17:45, Greg KH escribió:
quoted
On Wed, Sep 02, 2020 at 03:27:28PM +0200, Santiago Ruano Rincón wrote:
quoted
This:
37a2ebdd9e597ae1a0270ac747883ea8f6f767b6
e10dcb1b6ba714243ad5a35a11b91cc14103a9a9
e506addeff844237d60545ef4f6141de21471caf
0226009ce0f6089f9b31211f7a2703cf9a327a01
These do not look like bugfixes, but a new feature being added for this
driver. So why not just use a newer kernel version for this feature?
From my point of view as user these are bugfixes, since IPv6 NDP or any
other protocol relying on multicast do not work without them. In other
words, my computer's networking is broken.
I was in doubt when I submitted these, but ended up specfying net-next
instead of net+stable as the target for a reason. This is a new feature
as Greg says. Even if the feature is essential for your use case, it is
still new. "Has never been supported" isn't really a bug.
And I am still convinced that my decision was correct. The patches are
a bit more intrusive than I'd be comfortable submitting to stable, as
was demonstrated by the stupid build bug I added... Fixed by commit
5fd99b5d9950 ("net: cdc_ncm: Fix build error") BTW.
I want to have them in linux stable releases because that would make
easier to include them in Debian stable release.
This has not been an absolute requirement in the past. Distros tend to
have a more relaxed stable policy.
Using a newer kernel until Debian moves on is obviously also an option.
Bjørn