Re: git: e779891327b1 - main - sys/power: Sleep type reporting by PM backends

Guido,

> The two reviews to fix this breakage don't address the problem completely.
>
> I am formally requesting the revert of the four ACPI commits that have
> caused this regression. Revert them, please.

Can you elaborate on how they don't address the problem completely?
None of my machines have this issue so I can't really reproduce it,
but from the panic logs I've seen D52598 should completely fix that,
unless there's yet another issue.

> My amd64 context is one of those getting the very late crash
> during "shutdown -r now" or "poweroff". (PS2 keyboard use
> required, not USB keyboard --but it is too late for dump to
> work as well.)

Does D52598 fix this for you?

On Thu, 18 Sept 2025 at 08:22, Mark Millard <marklmi@yahoo.com> wrote:
>
> Cy Schubert <Cy.Schubert_at_cschubert.com> wrote on
> Date: Thu, 18 Sep 2025 04:59:32 UTC :
>
> > In message <9dc5eff9-3612-4bb8-be96-637bb4257a5f@FreeBSD.org>, Guido Falsi
> > writ
> > es:
> > > On 9/17/25 22:50, Guido Falsi wrote:
> > > > On 9/17/25 11:15, Nuno Teixeira wrote:
> > > >>
> > > >> Hello Cy,
> > > >>
> > > >>     One of this series of ACPI commits has broken power management on m
> > > y
> > > >>     machines.
> > > >>     One of which managed to capture a dump from a kernel panic after
> > > >>     poweroff(8)
> > > >>     was issued.
> > > >>
> > > >>
> > > >> I can confirm same crash on reboot on a Intel laptop, after upgrading
> > > >> from main 2025-09-06 -> 2025-09-16.
> > > >>
> > > >
> > > > Hi,
> > > >
> > > > A me too.
> > > >
> > > > Just noticed this while investigating why my machines with head (laptop
> > > > and two desktops) were not switching off anymore.
> > > >
> > > > So looks like I'm also experiencing this.
> > > >
> > > > All machines behave the same, the almost turns off, monitor turns off,
> > > > but power light stays on.
> > > >
> > > > I'm testing reverting commits to see if it goes back to working.
> > > >
> > >
> > > I can confirm reverting this commit and the previous one "fixes" it.
> > >
> > > --
> > > Guido Falsi <madpilot@FreeBSD.org>
> >
> > The two reviews to fix this breakage don't address the problem completely.
> >
> > I am formally requesting the revert of the four ACPI commits that have
> > caused this regression. Revert them, please.
>
> The above request was apparently not sent to the Author/Committer
> of "sys/power: Sleep type reporting by PM backends" ( e779891327b1 )
> and related: Aymeric Wibo ( obiwac@FreeBSD.org ) --at least there
> is no reference to the name or Email address in the above quoted
> text.
>
> So this CC's the author of that and the other commits.
>
> My amd64 context is one of those getting the very late crash
> during "shutdown -r now" or "poweroff". (PS2 keyboard use
> required, not USB keyboard --but it is too late for dump to
> work as well.)
>
> ===
> Mark Millard
> marklmi at yahoo.com
>