dim question: How to revert patches?

Daniel Vetter daniel at ffwll.ch
Thu Apr 28 12:12:25 UTC 2022


On Wed, Apr 27, 2022 at 09:20:59PM +0200, Helge Deller wrote:
> Hi Daniel,
> 
> On 4/27/22 16:21, Daniel Vetter wrote:
> > On Thu, Apr 14, 2022 at 10:37:55PM +0200, Helge Deller wrote:
> >> Hello dri-devel & dim users,
> >
> > Apologies for late reply, I'm way behind on stuff.
> >
> >> I committed this patch to the drm-misc-next branch:
> >>
> >> commit d6cd978f7e6b6f6895f8d0c4ce6e5d2c8e979afe
> >>     video: fbdev: fbmem: fix pointer reference to null device field
> >>
> >> then I noticed that it was fixed already in another branch which led to this error:
> >>
> >> Merging drm-misc/drm-misc-next... dim:
> >> dim: FAILURE: Could not merge drm-misc/drm-misc-next
> >> dim: See the section "Resolving Conflicts when Rebuilding drm-tip"
> >> dim: in the drm-tip.rst documentation for how to handle this situation.
> >>
> >> I fixed it by reverting that patch above with this new commit in the drm-misc-next branch:
> >>
> >> commit cabfa2bbe617ddf0a0cc4d01f72b584dae4939ad (HEAD -> drm-misc-next, drm-misc/for-linux-next, drm-misc/drm-misc-next)
> >> Author: Helge Deller <deller at gmx.de>
> >>     Revert "video: fbdev: fbmem: fix pointer reference to null device field"
> >>
> >> My question (as "dim" newbie):
> >> Was that the right solution?
> >
> > The patch wasn't really broken, so revert feels a bit silly. The hint was
> > to look at the documentation referenced by the error message - the issue
> > was only in rebuilding the integration tree:
> >
> > https://drm.pages.freedesktop.org/maintainer-tools/drm-tip.html#resolving-conflicts-when-rebuilding-drm-tip
> >
> > This should cover you even for really rare conflict situations.
> >
> >> Is there a possibility to drop those two patches from the drm-misc-next branch before it gets pushed upstream?
> >
> > It's a shared tree, mistakes are forever. The only time we did a forced
> > push ever is when someone managed to push their local pile of hacks or
> > something, and we're catching those pretty well now with a server-side
> > test to make sure you're using dim to push.
> >
> > It's also no big deal, and next time you get a conflict just resolve it
> > in drm-tip per the docs and it's all fine.
> 
> Thanks for the feedback!
> So, basically I think I did the right thing (although a revert isn't nice).
> There was no other useful fixup I could have come up with, because the other conflicting
> patch had the right & better solution already pushed.

Yeah if the other patch is strictly better then reverting sounds like the
right approach. Otherwise just resolving the conflict in drm-tip is the
way to go.

Cheers, Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch


More information about the dri-devel mailing list