Re: Linux 2.6.27-rc3: kernel BUG at mm/vmalloc.c - bisected

["Björn Steinbrink" <B.Steinbrink@gmx.de>]

On 2008.08.22 14:16:51 -0700, Andrew Morton wrote:
> It's pretty simple.  If git-bisect tells us that the regression was
> introduced by a merge commit, we want to perform a bisection within
> that merge's individual commits.

bisect already did that. It asked for the left side and the right side,
both were good. What you can still do is creating a _new_ history where
the commits are not in parallel but linearized, like Jeff described. But
that's (in general) not a trivial task as you need to reapply individual
commit patches which can cause conflicts that were already solved in the
existing merges to show up again. Or, it can even produce new conflicts,
for example when a commit on one side was reverted before the merge
happened. In that case, you get into a state with changes that were
never visible before.

Björn