Re: [PATCH 4/7] iio: adc: qcom-spmi-adc5: Add missing VCOIN/AMUX_THM3/GPIO# channels

[Jonathan Cameron <jic23@kernel.org>]

On Sun, 15 May 2022 17:57:14 +0100
Jonathan Cameron <jic23@kernel.org> wrote:

> On Sun, 15 May 2022 17:30:04 +0200
> Marijn Suijten <marijn.suijten@somainline.org> wrote:
> 
> > On 2022-05-14 17:13:12, Jonathan Cameron wrote:  
> > > On Thu, 12 May 2022 00:06:10 +0200
> > > Marijn Suijten <marijn.suijten@somainline.org> wrote:
> > >     
> > > > These channels are specified in downstream kernels [1] and actively used
> > > > by ie. the Sony Seine platform on the SM6125 SoC.    
> > > 
> > > Looking at the links, some of them are on that platform but not all.
> > > Better to make that explicit in this description.    
> > 
> > This has already been queued up for v2.  Adding these seemed easy at the
> > time but they are in fact not used, and I ended up sending the wrong
> > patch.
> > 
> > Just so that we're on the same page: only ADC5_AMUX_THM3 and
> > ADC5_GPIO2_100K_PU are unused by my platform.  It seems the first should
> > be dropped, but the latter can probably stay in the patch with an
> > explicit mention.  If you think both should stay, there are a bunch more
> > channels defined in the downstream kernel as per [1] and I'm not sure if
> > all should be added for completeness.  
> 
> Probably better to add them with a comment for platforms on which they
> apply (either in commit log or alongside the definitions in the code).
By 'them' I mean add the ones used on your platform only.  Let others
add theirs when / if boards using them are upstreamed.

(realised I'd been very unclear just after hitting send!)


> 
> Longer term we should think about whether the code can be adjusted
> to not need an explicit definition for these multi purpose channels
> though letting the dt itself describe them (under constraints of the
> hardware platform).  Not worth doing before this patch though.
> 
> >   
> > > > 
> > > > [1]: https://source.codeaurora.org/quic/la/kernel/msm-4.14/tree/drivers/iio/adc/qcom-spmi-adc5.c?h=LA.UM.7.11.r1-05200-NICOBAR.0#n688
> > > > 
> > > > Signed-off-by: Marijn Suijten <marijn.suijten@somainline.org>
> > > > Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@somainline.org>    
> > > 
> > > I'm not keen on patches with no context being
> > > sent to mailing lists. Please cc all lists (and preferably individuals)
> > > on at least the cover letter so we can see overall discussion.    
> > 
> > That can be attributed to the terrible workflow for sending
> > patch-series.  Somehow only `git send-email` supports --cc-cmd yet I'd
> > expect it on `git format-patch` for auditing and possibly copying to the
> > cover letter, if `git format-patch --cover-letter` couldn't do this from
> > the beginning.  
> 
> It would definitely be nice as an option.  Can't do it every time because
> on tree wide change the cc list can become so large the mailing lists
> reject it.
> 
> > At the same time `git send-email` has --[to/cc]-cover options to
> > propagate email addresses from the cover letter to all the individual
> > patch-replies, but not the inverse :(
> > 
> > In the end this leaves me manually running get_maintainer.pl over the
> > entire formatted patch-series, and manually copy-pasting + editing the
> > addresses into the cover letter... Which is easy to forget and is no
> > different here.
> > 
> > My apologies for (yet again) accidentally not sending at least the cover
> > letter to everyone.  That's a gross oversight, and I'm probably - no, I
> > must - be doing something wrong.  Suggestions and/or documentation
> > references are welcome.  
> 
> Andy Shevchenko has some scripts to try and help with this:
> https://github.com/andy-shev/home-bin-tools/blob/master/ge2maintainer.sh
> 
> I've not started using them myself (not gotten around to it yet!) but
> he's pointed to them when I've missed people from cover letter cc lists
> in the past.
> 
> >   
> > > If nothing else, I've no idea if intent is that the patches go through different
> > > trees or all need to merge via one route.    
> > 
> > I have no idea either, and have not yet had an answer to a similar
> > question on a different list.  Usually it seems the maintainers work out
> > amongst themselves who picks what patch, putting them on hold where
> > necessary to preseve ordering.  If not, should the sender split patches
> > across multiple series, either holding off sending part of it or linking
> > to a dependent series?  
> 
> In this case I think I can pick this patch up directly into the IIO tree
> once everyone else is happy. As you note dts patches normally wait on
> knowing the necessary support is heading in.  If you have a view on what
> makes sense as the submitter it's good to stick it in the cover letter, but
> in this case sounds like you don't. :)
> 
> Given we are near the end of this cycle, we are probably looking at next
> cycle anyway now, so plenty of time to figure it out!
> 
> > 
> > In this particular case DT has to wait for these driver patches to land,
> > otherwise they may define channels that do not exist and unnecessarily
> > fail probe.
> >   
> > > Patch itself looks fine,    
> > 
> > Thanks.
> > 
> > Looking forward to your suggestions and answers,
> > 
> > - Marijn
> >   
> > > [..]    
>