Re: Should drm2 be based on Linux kernel 3.14 since it is LTS?

["David H. Gutteridge" <dhgutteridge%sympatico.ca@localhost>]

On Wed, 16 Jul 2014 22:04:40 -0700, Waitman Gobble wrote:
>On Wed, 16 Jul 2014 21:26:21 -0700 David Shao wrote:
>> Great thanks should be given to Taylor R Campbell for the work
>> updating drm2 to Linux kernel 3.15.  Perhaps too much has been done in
>> that direction, but is it possible that Linux kernel 3.14 should be
>> considered as the baseline since it is LTS, which to my understanding
>> means updates for 2 years including updates to select drivers.
>> Whereas if 3.15 is not LTS it could be rapidly superseded.  If the
>> plan is to make drm2 part of NetBSD 7, using 3.14 LTS is perhaps the
>> easiest way to support the drm2 drivers at least a couple of the years
>> NetBSD 7 will be supported.
>> 
>> Furthermore I believe both Debian unstable and Fedora 19 have 3.14 as
>> their most recent kernels so that people debugging drm2 right now
>> before the NetBSD 7 beta announcement could have systems actually
>> running 3.14 even if it is a different OS.
>
>When I grep the change logs it looks like the same drm fixes in 3.15 are
>happening in 3.14, so I'm not sure it's really an issue? I get
>your point but I think you are talking about the Linux kernel as a
>whole. Except there seem to be i2c changes in 3.15.. is there i2c in
>NetBSD? It's not in FreeBSD, deprecated years ago AFIK. One big part of
>porting the drivers is changing the i2c code. So i2c stuff probablyl
>not relevant.

Fedora 20 pushed 3.15 kernels to their user base a couple of weeks
ago, e.g. 3.15.5 was pushed out yesterday. So the code is in active
use now. (I'm sure Fedora isn't alone in this.) Linux users have
found DRM regressions in 3.15 that aren't in 3.14, e.g.:

https://bugzilla.redhat.com/show_bug.cgi?id=1117008

3.15 is getting attention. (I can't claim any special knowledge of
this, mind you, I'm just a user sitting in the peanut gallery, as
I am with NetBSD.)

Regards,

Dave