Re: linux-next: manual merge of the kunit-next tree with the apparmor tree

[David Gow]

On Tue, 13 Dec 2022 at 07:19, Shuah Khan <skhan@xxxxxxxxxxxxxxxxxxx> wrote:
>
> On 12/12/22 12:53, John Johansen wrote:
> > On 12/12/22 11:48, Shuah Khan wrote:
> >> On 12/12/22 12:20, John Johansen wrote:
> >>> On 12/12/22 10:03, Shuah Khan wrote:
> >>>> On 12/12/22 10:52, Shuah Khan wrote:
> >>>>> Hi David,
> >>>>>
> >>>>> On 12/8/22 13:10, John Johansen wrote:
> >>>>>> On 12/7/22 18:53, Stephen Rothwell wrote:
> >>>>>>> Hi all,
> >>>>>>>
> >>>>>>> Today's linux-next merge of the kunit-next tree got a conflict in:
> >>>>>>>
> >>>>>>>    security/apparmor/policy_unpack.c
> >>>>>>>
> >>>>>>> between commits:
> >>>>>>>
> >>>>>>>    371e50a0b19f ("apparmor: make unpack_array return a trianary value")
> >>>>>>>    73c7e91c8bc9 ("apparmor: Remove unnecessary size check when unpacking trans_table")
> >>>>>>>    217af7e2f4de ("apparmor: refactor profile rules and attachments")
> >>>>>>> (and probably others)
> >>>>>>>
> >>>>>>> from the apparmor tree and commit:
> >>>>>>>
> >>>>>>>    2c92044683f5 ("apparmor: test: make static symbols visible during kunit testing")
> >>>>>>>
> >>>>>>> from the kunit-next tree.
> >>>>>>>
> >>>>>>> This is somewhat of a mess ... pity there is not a shared branch (or
> >>>>>>> better routing if the patches).
> >>>>>>>
> >>>>>> sorry, there was a miscommunication/misunderstanding, probably all on me, I
> >>>>>> thought the kunit stuff that is conflicting here was going to merge next
> >>>>>> cycle.
> >>>>>>
> >>>>>
> >>>>
> >>>> How about I just drop the following for now and handle this in the next cycle?
> >>>
> >>> if you want, the other way to handle it is we coordinate our pull requests.
> >>> You go first. And then I will submit a little later in the week, with the
> >>> references to the merge conflict and a pointer to a branch with it resolved.
> >>> This isn't even a particularly tricky merge conflict, it just has the little
> >>> subtly around making sure the include symbols are conditional.
> >>>
> >>
> >> I assume Linus will not see any problems without your pull requests. In which
> >> case we can do this:
> >>
> >> - I send my pull request today
> >> - You can follow with yours with the fixes later on this week
> >>
> >
> > okay
> >
> >>> This doesn't affect me much as there is already another merge conflict with
> >>> the security tree that I need to deal with.
> >>>
> >>
> >>
> >>>> I think it might be least confusing option. Let me know. I can just do that
> >>>> and then send pull request in a day or tow once things settle down in next.
> >>>>
> >>>> 2c92044683f5 ("apparmor: test: make static symbols visible during kunit testing")
> >>>>
> >>>
> >>> that is the other option. If you go that route I can help you do the rebase/merge
> >>> fix.
> >>>
> >>
> >> Let's go with your earlier suggestion.
> >>
> >
> > ack
> >
> >>> looking back at this, there wasn't anything explicit about this not going upstream
> >>> this cycle, I must have just assumed as the final version came about after rc7. So
> >>> my bad.
> >>>
> >>
> >> Right - I ended up taking this as it looked like a patch if included could
> >> enable other changes to follow without being blocked. Also rc8 was in plan.
> >>
> >
> > yeah, my bad
> >
>
> No worries. Sent pull request with a note about apparmor and our
> coordinated pull requests with you on the cc.
>
> thanks,
> -- Shuah
>

Thanks John, Shuah for sorting this out. I confess that I hadn't
noticed the conflict before proposing this for 6.2: in retrospect I
should've checked more carefully given the amount of churn in the
patch.

If we have to drop this patch and split the series, that's not a
problem: it's really just an example. But if the conflict's resolved,
that's even better.

Thanks again!

-- David