Is the Fedora kernel git tree suitable for bisects?

Josh Boyer jwboyer at fedoraproject.org
Thu Oct 2 19:58:33 UTC 2014


On Thu, Oct 2, 2014 at 3:48 PM, Bruno Wolff III <bruno at wolff.to> wrote:
> If one wanted to bisect a kernel with Fedora changes instead of Linus'
> kernel, would this likely work well?

Not really.

You could run git bisect on it well enough, but it would treat the
Fedora patches as just another set of linear commits on top of Linus'
tree.  Which means that if you were having problems with e.g. 3.17-rc7
but 3.17-rc5 works, the first bisect on the rawhide branch is going to
land you in the middle of those releases, which means none of the
Fedora patches would be applied at that point.  That's somewhat
valuable, because more often than not it's an upstream change and not
a Fedora change that causes and issue, but it really doesn't buy you
anything over just using Linus' tree to begin with.  It would be rare
to have a bisect land somewhere in only the Fedora commits.

You could always export them as a patchset before you started your
bisect and apply/unapply them every bisect step, but that can be
tedious.

josh


More information about the kernel mailing list