Re: [PATCH] drivers/hwmon: Use pr_fmt and pr_<level>

[Jean Delvare <khali@linux-fr.org>]

Hi Guenter,

On Wed, 20 Oct 2010 07:46:08 -0700, Guenter Roeck wrote:
> On Wed, Oct 20, 2010 at 03:34:38AM -0400, Jean Delvare wrote:
> [ ... ]
> > 
> > * This isn't the kind of fixes we want to cherry-pick from. We're not
> >   fixing any bug here, are we? I certainly hope that a real bug fix
> >   wouldn't be hidden within a larger patch, but would have the separate
> >   patch it deserves. At which point we no longer care if the rest is
> >   one large patch or one patch per driver.
>
> Actually, I do this kind of thing all the time when backporting.
> It is easier to apply all patches applied to a driver than skipping
> the cleanup patches, to avoid conflicts when cherry-picking functional
> patches. The tendency for large cleanup patches affecting several drivers
> and modules created trouble for me several times already.

I see. I had stable kernel series in mind when you mentioned
backporting - obviously this was too restrictive and you have different
needs.

I didn't expect it to be difficult to limit the scope of a
cherry-picked patch to a single file. Is it?

> > * I don't see us reverting that kind of patch either. If we don't like
> >   the changes for whatever reason, we don't take them in the first
> >   place. Once in, we're not going to change our minds.
> > 
> > * 32 patches for a simple cleanup is actually a lot more work for me
> >   than a single large patch. It's cheaper for me to do minor
> >   adjustments to a large patch than to apply 32 patches individually.
>
> Good point. I am using git all the time, so it isn't that much of a
> problem for me. Our mailer problems (the tabs replacing stuff) are
> much more annoying.

I'm still using quilt so it means quite some manual work for me. Probably
I could automate part of it, and I would if it was happening
frequently. But as it stands it's rather rare.

> > * That being said, now that the hwmon subsystem maintainer is a shared
> >   duty between Guenter and myself, there's no single place where we can
> >   keep a patch touching many drivers and ensure it doesn't conflict
> >   with the changes in the other tree. But I would think   this is
> >   something for Gunter and myself to sort out, not patch contributors.

Another point in favor of splitting, which I didn't mention earlier: it
makes the Cc list way shorter :)

> > I currently have pending patches to the following hwmon drivers in my
> > tree: adt7475, ams, asc7621, hdaps, it87, k8temp, lm75, lm85, lm90,
> > pcf8591, s3c-hwmon, w83795. Two of these are affected by Joe's
> > patch(es). Guenter, what about you?
>
> coretemp, pkgtemp, via-cputemp, ltc4261 (new), lis3, hp_accel

OK. Well, only two of "my" drivers are affected by Joe's patch (it87
and pcf8591), so if you prefer the split variant, you could pick all
the individual patches except these two, and I take these two.

But first of all, I really would like the pr_fmt issue do be sorted
out. I don't like the idea of having to redefine it in every driver,
when it seems easy to avoid that. Joe?

-- 
Jean Delvare