Git bisecting non-linear Ubuntu kernel commits Inquiry

Herton Ronaldo Krzesinski herton.krzesinski at
Mon Nov 5 20:51:42 UTC 2012

On Sun, Nov 04, 2012 at 08:35:35PM -0500, Christopher Penalver wrote:
> Dear Ubuntu Kernel Team / Ubuntu Bug Control:
> Hello. I am e-mailing you as a continuation of
> to inquire about git bisecting non-linear Ubuntu kernel commits, as
> currently documented in
> While triaging,
> it seemed the reporter fell into this pitfall.
> Naively, one might assume tag linearity is due to the order in which
> Ubuntu kernel versions are displayed top to bottom in
> . However, this seems
> incorrect. Hence, one would ask, "What does a non-linear tag mean more
> explicitly?" Is this the mainline kernel being rebased off of is
> non-linear, the downstream Ubuntu kernel versions, the tags found at
>;a=tags , etc.
> So, what is tag linearity mean specifically in this case? As well, how
> would the original reporter test around this non-linearity?

While in development, the Ubuntu kernels are usually always rebased
against Linus master, so bisect between the tags are not possible
(non-linear tags). On the contrary, when the distro is released, we
don't allow/do rebases, so any kernel stable updates are bisectable (any
tags after and including the tag of the kernel in release pocket should
bisect fine between them). It all depends on which kernel version
introduced the problem, so you have to check if you'll be able to bisect
using the Ubuntu repository.

There is no bisect between non-linear tags, you must use a linear set of
commits. When it's non-linear in Ubuntu case you should just use Linus
master tree, and pull the packaging bits, or more simpler and what I
usually do, just run "make -j<n> deb-pkg" using the same kernel config
and install the resulting deb for testing. That should work assuming the
bug is in one of the upstream changes upstream and not an Ubuntu change
or any other special case (like a config change triggering something).

> I'm happy to better document in the bisect article non-linear tags and
> testing around this, as it is a common issue for original reporters
> who end up getting to commit bisecting the Ubuntu kernel.
> Thank you for your time and consideration in this matter and I look
> forward to your response.
> --
> Christopher M. Penalver
> E-Mail: christopher.m.penalver at
> -- 
> kernel-team mailing list
> kernel-team at


More information about the kernel-team mailing list