On Wed, Feb 19, 2020 at 03:49:47PM -0800, Stephen Boyd wrote:
> Adding some Allwinner folks. Presumably there is some sort of clk that
> is failing to calculate a phase when it gets registered. Maybe that's
> because the parent isn't registered yet?

It's simpler than that :)

> Quoting Guillaume Tucker (2020-02-17 23:45:41)
> > Hi Stephen,
> >
> > Please see the bisection report below about a boot failure.
> >
> > Reports aren't automatically sent to the public while we're
> > trialing new bisection features on kernelci.org but this one
> > looks valid.
> >
> > There's nothing in the serial console log, probably because it's
> > crashing too early during boot.  I'm not sure if other platforms
> > on kernelci.org were hit by this in the same way, it's tricky to
> > tell partly because there is no output.  It should possible to
> > run it again with earlyprintk enabled in BayLibre's test lab
> > though.
> >
> > Thanks,
> > Guillaume
> >
> >
> > On 18/02/2020 02:14, kernelci.org bot wrote:
> > > * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
> > > * This automated bisection report was sent to you on the basis  *
> > > * that you may be involved with the breaking commit it has      *
> > > * found.  No manual investigation has been done to verify it,   *
> > > * and the root cause of the problem may be somewhere else.      *
> > > *                                                               *
> > > * If you do send a fix, please include this trailer:            *
> > > *   Reported-by: "kernelci.org bot" <bot@kernelci.org>          *
> > > *                                                               *
> > > * Hope this helps!                                              *
> > > * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
> > >
> > > next/master bisection: baseline.login on sun8i-h2-plus-orangepi-zero
> > >
> > > Summary:
> > >   Start:      c25a951c50dc Add linux-next specific files for 20200217
> > >   Plain log:  https://storage.kernelci.org//next/master/next-20200217/arm/multi_v7_defconfig/gcc-8/lab-baylibre/baseline-sun8i-h2-plus-orangepi-zero.txt
> > >   HTML log:   https://storage.kernelci.org//next/master/next-20200217/arm/multi_v7_defconfig/gcc-8/lab-baylibre/baseline-sun8i-h2-plus-orangepi-zero.html
> > >   Result:     2760878662a2 clk: Bail out when calculating phase fails during clk registration
> > >
> > > Checks:
> > >   revert:     PASS
> > >   verify:     PASS
> > >
> > > Parameters:
> > >   Tree:       next
> > >   URL:        git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git
> > >   Branch:     master
> > >   Target:     sun8i-h2-plus-orangepi-zero
> > >   CPU arch:   arm
> > >   Lab:        lab-baylibre
> > >   Compiler:   gcc-8
> > >   Config:     multi_v7_defconfig
> > >   Test case:  baseline.login
> > >
> > > Breaking commit found:
> > >
> > > -------------------------------------------------------------------------------
> > > commit 2760878662a290ac57cff8a5a8d8bda8f4dddc37
> > > Author: Stephen Boyd <sboyd@kernel.org>
> > > Date:   Wed Feb 5 15:28:02 2020 -0800
> > >
> > >     clk: Bail out when calculating phase fails during clk registration
> > >
> > >     Bail out of clk registration if we fail to get the phase for a clk that
> > >     has a clk_ops::get_phase() callback. Print a warning too so that driver
> > >     authors can easily figure out that some clk is unable to read back phase
> > >     information at boot.
> > >
> > >     Cc: Douglas Anderson <dianders@chromium.org>
> > >     Cc: Heiko Stuebner <heiko@sntech.de>
> > >     Suggested-by: Jerome Brunet <jbrunet@baylibre.com>
> > >     Signed-off-by: Stephen Boyd <sboyd@kernel.org>
> > >     Link: https://lkml.kernel.org/r/20200205232802.29184-5-sboyd@kernel.org
> > >     Acked-by: Jerome Brunet <jbrunet@baylibre.com>
> > >
> > > diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
> > > index dc8bdfbd6a0c..ed1797857bae 100644
> > > --- a/drivers/clk/clk.c
> > > +++ b/drivers/clk/clk.c
> > > @@ -3457,7 +3457,12 @@ static int __clk_core_init(struct clk_core *core)
> > >        * Since a phase is by definition relative to its parent, just
> > >        * query the current clock phase, or just assume it's in phase.
> > >        */
> > > -     clk_core_get_phase(core);
> > > +     ret = clk_core_get_phase(core);
> > > +     if (ret < 0) {
> > > +             pr_warn("%s: Failed to get phase for clk '%s'\n", __func__,
> > > +                     core->name);
> > > +             goto out;
> > > +     }

The thing is, clk_core_get_phase actually returns the phase on success :)

So, when you actually have a phase returned, and not an error, you end
up with a positive, non-zero, value for ret.

And since it's the latest assignment of that value, and that we return
ret all the time, even on success, we end up returning that positive,
non-zero value to __clk_register, which in turn tests whether it's
non-zero for success (it's not), and then proceeds to garbage collect
everything.

I guess we're just the odd ones actually returning non-zero phases at
init time and in kernelci.

I'll send a patch

Maxime