Re: [PATCH] soc: qcom: pmic_glink: Fix boot when QRTR=m

[Johan Hovold <johan@kernel.org>]

On Thu, Dec 14, 2023 at 12:44:10PM -0800, Rob Clark wrote:
> On Thu, Dec 14, 2023 at 8:08 AM Johan Hovold <johan@kernel.org> wrote:
> >
> > On Thu, Dec 14, 2023 at 04:38:35PM +0100, Johan Hovold wrote:
> >
> > > I took a closer look at this and indeed we do have code that triggers a
> > > reprobe of a device in case there was a successful probe while the
> > > device was probing.
> > >
> > > This was introduced by commit 58b116bce136 ("drivercore: deferral race
> > > condition fix") and the workaround for the reprobe-loop bug that hack
> > > led to is to not return -EPROBE_DEFER after registering child devices as
> > > no one managed to come up with a proper fix. This was documented here:
> > >
> > >       fbc35b45f9f6 ("Add documentation on meaning of -EPROBE_DEFER")
> > >
> > > But please spell this out in some more detail in the commit message, and
> > > add a Fixes and CC stable tag.
> >
> > And please update the commit summary as I've been booting with QRTR=m
> > all along just fine. I guess the issue is if you have pmic_glink
> > built-in or in the initramfs but forgot to include qrtr or similar?
> 
> I do have both QRTR=m and QCOM_GLINK=m.  I'm honestly not sure what
> started triggering this issue for me.. it seemed to have started after
> merging msm-next + drm-misc-next on top of your
> jhovold/wip/sc8280xp-v6.7-rc5 (the merged branches were based on -rc3
> so this shouldn't have really brought in random non-drm things).
> Maybe there is a timing element to it?

Yeah, possibly, and device links may also come into play here.

> I felt like the problem was obvious enough, and the exact details of
> why I started hitting this were not important enough to spend time
> tracking down.

The patch itself is of course fine itself as a clean up (or
microoptimisation) but the claim that it solves, and is the correct fix
for, a probe loop issue is not obvious at all and requires some further
justification.

Since he last time someone suggested reverting the commit that
introduced the probe-loop issue, the result was to leave things as they
were and just document the workaround, it should be fine to just refer
to that commit:

	fbc35b45f9f6 ("Add documentation on meaning of -EPROBE_DEFER")

Perhaps you can keep the Subject if you make it clear in the commit
message that this bug isn't always hit with QRTR=m.

Johan