RPi2 snapshot for armv7 won't boot.

[Warner Losh]

On Wed, Nov 1, 2017 at 11:28 PM, Michal Meloun <melounmichal at gmail.com>
wrote:

>
>
> On 02.11.2017 2:53, Kyle Evans wrote:
> > On Nov 1, 2017 6:24 PM, "Mark Millard" <markmi at dsl-only.net> wrote:
> >
> > On 2017-Nov-1, at 3:51 PM, Carl Johnson <carlj at peak.org> wrote:
> >
> >> I had previously tried to upgrade my RPi2 12.0-CURRENT system from armv6
> >> to armv7, but it then refused to boot.  This time I downloaded the
> >> latest snapshot (r325156 from 20171030) and it also refuses to boot for
> >> exactly the same reason.  In every case the kernel loads, but then it
> >> reports that init has died, and then panics and drops into kdb.  It
> >> gives a stack backtrace, but that doesn't mean anything to me.
> >>
> >> Is this just me, or is anybody else having this problem?  If others have
> >> the problem, is there some workaround to get it working?
> >>
> >> Thanks in advance for any ideas.
> >
> > Others are also seeing the problem, not just on
> > RPI2's.
> >
> > It looks like it broke between:
> >
> > -r324743 and -r325156
> >
> > someplace relative to whatever is causing the
> > init problem.
> >
> > Details that lead me to that conclusion, if
> > you care. . .
> >
>
> The r324938 breaks all statically linked binaries (init, /rescue/*, ...).
> I have prepared patch, but I need a day or two for more testing on other
> arches.
> https://github.com/strejda/tegra/commit/67ea3325c50770b5031ef02882b2dc
> cdd1f7611f
>
> Michal


To be extra sure, have we tried to back out this one change against the tip
of -head to confirm? It's pretty trivial.

Warner


> > One of the FreeBSD folks contacted me privately because his
> > bananapi-m3 experiments were getting the init problem and
> > he was hoping I could test independently to see if the
> > problem was local to his attempt to modernize the BPI-M3
> > support so it would again eventually be supported.
> >
> > I had to be doing other things but was able to report
> > that my context was a working variant of -r324743 and
> > showed him the two small diffs that I used. (The issues
> > needing the diffs are not tied to any init behavior.)
> >
> > He tried his experiment against -r324743 and his
> > experimental code and *.dt* based .dts booted fine.
> >
> > It looks like something after -r324743 broke things
> > such that the init problem exists for armv7.
> >
> > [I'm still not active for investigating any of this.]
> >
> > ===
> > Mark Millard
> > markmi at dsl-only.net
> >
> >
> > For any interested: I think I've got it narrowed down to somewhere
> between
> > or including r324950 - r324882 (last good I've tested).
> >
> > Looking at the logs, I don't see too many commits in there that could've
> > broken something this fundamental, but I think my bisect only has three
> > more steps until I can say for sure.
> >
> > Thanks,
> >
> > Kyle Evans
> > _______________________________________________
> > freebsd-arm at freebsd.org mailing list
> > https://lists.freebsd.org/mailman/listinfo/freebsd-arm
> > To unsubscribe, send any mail to "freebsd-arm-unsubscribe at freebsd.org"
> >
> _______________________________________________
> freebsd-arm at freebsd.org mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-arm
> To unsubscribe, send any mail to "freebsd-arm-unsubscribe at freebsd.org"
>