Re: linux-next: build failure after merge of the drm-misc tree

[Daniel Vetter]

On Tue, Dec 17, 2019 at 9:22 PM Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> wrote:
>
> Hi Daniel,
>
> On Tue, 17 Dec 2019 14:19:37 +0100 Daniel Vetter <daniel@xxxxxxxx> wrote:
> >
> > On Mon, Dec 16, 2019 at 12:23:31PM +1100, Stephen Rothwell wrote:
> > >
> > > After merging the drm-misc tree, today's linux-next build (x86_64
> > > allmodconfig) failed like this:
> > >
> > > drivers/gpu/drm/bridge/analogix/analogix-anx6345.c: In function 'anx6345_i2c_probe':
> > > drivers/gpu/drm/bridge/analogix/analogix-anx6345.c:738:30: error: implicit declaration of function 'i2c_new_dummy' [-Werror=implicit-function-declaration]
> > >   738 |    anx6345->i2c_clients[i] = i2c_new_dummy(client->adapter,
> > >       |                              ^~~~~~~~~~~~~
> > > drivers/gpu/drm/bridge/analogix/analogix-anx6345.c:738:28: warning: assignment to 'struct i2c_client *' from 'int' makes pointer from integer without a cast [-Wint-conversion]
> > >   738 |    anx6345->i2c_clients[i] = i2c_new_dummy(client->adapter,
> > >       |                            ^
> > >
> > > Caused by commit
> > >
> > >   6aa192698089 ("drm/bridge: Add Analogix anx6345 support")
> > >
> > > interacting with commit
> > >
> > >   2c2f00ab1641 ("i2c: remove i2c_new_dummy() API")
> > >
> > > From Linus' tree.
> > >
> > > I have applied the following fix up patch for today:
> > >
> > > From: Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx>
> > > Date: Mon, 16 Dec 2019 12:11:19 +1100
> > > Subject: [PATCH] drm/bridge: fix up for removal of i2c_new_dummy()
> > >
> > > Signed-off-by: Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx>
> >
> > Thanks pulled into drm-next since I just processed the first drm-misc-next
> > pull.
>
> Thanks.  For the future, though, merge fixes like this should be part
> of the actual merge commit to avoid bisection problems.

I flip flop on this one, between retaining your contribution
explicitly and merging them in. Usually I squash them in, but this
felt substantial enough to retain explicitly. Worst case if someone is
unlucky they need to git bisect skip once. I guess next time I should
do a topic branch, apply it there on top of the pull, and then pull
that in so that it's both pretty history and no bisect hole.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch