Dropping i386 non-PAE as a supported kernel flavour in Precise Pangolin

Tim Gardner tim.gardner at canonical.com
Thu Nov 10 15:14:14 UTC 2011

On 11/09/2011 03:14 PM, Colin Watson wrote:
> On Wed, Nov 09, 2011 at 02:43:28PM -0700, Tim Gardner wrote:
>> Per discussion at UDS the kernel team is proposing to drop the
>> non-PAE i386 flavour. The upgrade path for non-PAE users will be the
>> PAE kernel. Those CPUs that do not have i686 and PAE support will be
>> orphaned. To the best of my knowledge, these include Intel CPUs
>> prior to Pentium II, 400Mhz Pentium M, VIA C3, and Geode LX. As far
>> as I know, there are no laptop or desktop class CPUs being produced
>> that do not meet these minimum requirements.
> Does KVM work properly with PAE kernels at the moment?  I've had trouble
> with it within the last six months, and when running server
> installations I've had to tweak them on the fly to install the generic
> kernel in order that I could boot the installed system.

This just seems like a bug. If we don't address it early in this cycle, 
then what incentive would we have to address it during the 12.10 dev cycle?

>> Before I do something that is difficult to revert, I would like to
>> hear from the development community why we should continue to
>> maintain a kernel flavour that is (in my opinion) getting
>> increasingly low utilization.
> I'd have thought we needed data here?

As far as I can tell there hasn't been a mass produced non-PAE cpu in 
over 5 years that we (as a distro) care about. The consumer grade 
electronics lifecycle is _well_ below 5 years. Furthermore, the distro 
focus has been desktop with high performance 3D graphics and servers. 
Where do non-PAE CPUs fit in that world? There are better distro choices 
to fill that niche.

> I'm worried about dropping the
> kernel that's been the default during the installer for some time, in
> one step.  If we want to switch the installer to generic-pae and then
> drop the non-PAE kernel in the next cycle if that works out well, I'd be
> happier with that approach; that gives us a much more graceful fallback
> plan in the event that our opinions are mistaken.

I want to drop the non-PAE kernel _before_ the LTS. Otherwise we have to 
deal with the complexities of LTS backported kernels _not_ having the 
same flavour set as the released LTS kernel (something I'd prefer not to 
have to do).

What do you think about dropping x86 32 bit kernels altogether for 14.04 
? By then we should have _really_ good multi-arch support, and the CPUs 
that we care about will all be 64 bit capable.

Tim Gardner tim.gardner at canonical.com

More information about the kernel-team mailing list