XFree86 & ACPI don't like each other?

[s***@teardrop.org]

Hello list,

I've recently joined the freebsd-mobile world with 5.2.1 on
an IBM ThinkPad T22. I'm still settling into it and getting
used to the quirks of FreeBSD on a laptop.

On the whole, things are working delightfully. My Netgear
WG511T works perfectly under if_ath, and sound came up
painlessly with snd_pcm and snd_csa. On board Ethernet is
fxp and workes perfectly; used it for the FTP install.

On the other hand, X and ACPI don't seem to get along very
well. If I boot with ACPI, X will sometimes hang when
starting or stopping. Doesn't appear to be very consistent
or predictable, but I can't get through more than 1 clean
start and stop of X without it locking the machine up solid.
Have to pull the battery and AC power to reset it.

It's definitely an ACPI interaction because if I boot
without ACPI I can start and stop X all day without
issue. It's not that I go in and out of X often, but the
occassional hang on start and the fairly reliable hang at
exit have me fsck'ing the drive a bit more often than I'd
like.

I know ACPI is still a work in progress, but I was wondering
if any other T-series ThinkPad users have seen something
like this?


-Snow

[h***@lambermont.dyndns.org]

Post by s***@teardrop.org
I've recently joined the freebsd-mobile world with 5.2.1 on
an IBM ThinkPad T22.

...

Post by s***@teardrop.org
On the other hand, X and ACPI don't seem to get along very
well. If I boot with ACPI, X will sometimes hang when
starting or stopping. Doesn't appear to be very consistent
or predictable, but I can't get through more than 1 clean
start and stop of X without it locking the machine up solid.

...

Post by s***@teardrop.org
if any other T-series ThinkPad users have seen something
like this?

Not on my T30, I've been using -current on it with ACPI and X.
Maybe there's a hint in /var/log/XFree86.0.log ? Compare it with a
non-ACPI run.

Hans

[d***@roe.ch]

--ReaqsoxgOBHFXBhH
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On the other hand, X and ACPI don't seem to get along very well. If
I boot with ACPI, X will sometimes hang when starting or stopping.
Doesn't appear to be very consistent or predictable, but I can't get
through more than 1 clean start and stop of X without it locking the
machine up solid.

I have the same problem. Running a non-SMP kernel fixes the problem for
me without having to disable ACPI. Also see this open PR:
http://www.freebsd.org/cgi/query-pr.cgi?pr=3Dmisc/60512

Cheers,
Dan

--=20
Daniel Roethlisberger <***@roe.ch>
GnuPG key ID 0x804A06B1 (DSA/ElGamal)

--ReaqsoxgOBHFXBhH
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (FreeBSD)

iD8DBQFAjCjkOXQOmIBKBrERAha4AJ973y0GSkFJJtJIj0SJiNbUUB9J2gCgiNzQ
NmPvEoMnTqZ3So5AD2TIHjY=
=r2fA
-----END PGP SIGNATURE-----

--ReaqsoxgOBHFXBhH--

[s***@teardrop.org]

Post by d***@roe.ch
I have the same problem. Running a non-SMP kernel fixes the problem for
http://www.freebsd.org/cgi/query-pr.cgi?pr=misc/60512

That did the trick. Thanks very much.


-Snow

[j***@freebsd.org]

Post by d***@roe.ch

On the other hand, X and ACPI don't seem to get along very well. If
I boot with ACPI, X will sometimes hang when starting or stopping.
Doesn't appear to be very consistent or predictable, but I can't get
through more than 1 clean start and stop of X without it locking the
machine up solid.

I have the same problem. Running a non-SMP kernel fixes the problem for
http://www.freebsd.org/cgi/query-pr.cgi?pr=misc/60512

Did you remove 'device apic' as well?

[d***@langille.org]

Post by d***@roe.ch

On the other hand, X and ACPI don't seem to get along very well. If
I boot with ACPI, X will sometimes hang when starting or stopping.
Doesn't appear to be very consistent or predictable, but I can't get
through more than 1 clean start and stop of X without it locking the
machine up solid.

I have the same problem. Running a non-SMP kernel fixes the problem for
http://www.freebsd.org/cgi/query-pr.cgi?pr=misc/60512

FWIW, X under 5.2.1-RELEASE would freeze on me. Creating a non-SMP kernel
solved the problem for me. I'm still using ACPI but only just started
using sleep/resume.

[d***@langille.org]

Sorry, my previous message was incomplete.

Post by j***@freebsd.org

Post by d***@roe.ch

On the other hand, X and ACPI don't seem to get along very well. If
I boot with ACPI, X will sometimes hang when starting or stopping.
Doesn't appear to be very consistent or predictable, but I can't get
through more than 1 clean start and stop of X without it locking the
machine up solid.

I have the same problem. Running a non-SMP kernel fixes the problem for
http://www.freebsd.org/cgi/query-pr.cgi?pr=misc/60512

Did you remove 'device apic' as well?

FWIW, X under 5.2.1-RELEASE would freeze on me. Creating a non-SMP kernel
solved the problem for me. I'm still using ACPI but only just started
using sleep/resume. I did not remove apic.

[j***@freebsd.org]

Post by d***@langille.org
Sorry, my previous message was incomplete.

Post by j***@freebsd.org

Post by d***@roe.ch

On the other hand, X and ACPI don't seem to get along very well. If
I boot with ACPI, X will sometimes hang when starting or stopping.
Doesn't appear to be very consistent or predictable, but I can't
get through more than 1 clean start and stop of X without it
locking the machine up solid.

I have the same problem. Running a non-SMP kernel fixes the problem for
http://www.freebsd.org/cgi/query-pr.cgi?pr=misc/60512

Did you remove 'device apic' as well?

FWIW, X under 5.2.1-RELEASE would freeze on me. Creating a non-SMP kernel
solved the problem for me. I'm still using ACPI but only just started
using sleep/resume. I did not remove apic.

So just removing SMP, but not removing apic? What if you boot the SMP kernel
with 'kern.smp.disabled=1' set from the loader?

[s***@chereda.net]

Post by j***@freebsd.org

Post by d***@langille.org

Post by j***@freebsd.org
Did you remove 'device apic' as well?

FWIW, X under 5.2.1-RELEASE would freeze on me. Creating a non-SMP kernel
solved the problem for me. I'm still using ACPI but only just started
using sleep/resume. I did not remove apic.

So just removing SMP, but not removing apic? What if you boot the SMP kernel
with 'kern.smp.disabled=1' set from the loader?

Is removing apic necessary when removing SMP?

I have a lot of boxes without SMP and with apic, notebooks and desktops,
and everything is working ok, using APIC interrupts routing.

[d***@langille.org]

I've recently joined the freebsd-mobile world with 5.2.1 on an IBM
ThinkPad T22.

I've been on 5.2.1 with my T22 for about 2 months.

I use a T23, and my FreeBSD is 4.x, recent STABLE.

.. X and ACPI don't seem to get along very well. If I boot with ACPI, X
will sometimes hang when starting or stopping. Doesn't appear to be very
consistent or predictable, but I can't get through more than 1 clean start
and stop of X without it locking the machine up solid.
... I was wondering if any other T-series ThinkPad users have seen
something like this?

See my post today about my ACPI experiences. Not any serious problems
yet.

Under STABLE, the only XFree86 vs. power management problem comes with
hibernation. When my T23 wakes up after hibernation, XFree86 is frozen.
I have to switch to an ASCII console (ctrl-alt-f1) and back to X (alt-f12)
to get it to limber up. If the hibernation is manual rather than
automatic, I can transition to ASCII console first before hibernating, and
there is no wake-up freeze, but automatic because of flat battery is the
usual reason to hibernate.

I was hoping to get hibernation working here...

[r***@iam.unibe.ch]

Under STABLE, the only XFree86 vs. power management problem comes with
hibernation. When my T23 wakes up after hibernation, XFree86 is frozen.
I have to switch to an ASCII console (ctrl-alt-f1) and back to X (alt-f12)
to get it to limber up. If the hibernation is manual rather than
automatic, I can transition to ASCII console first before hibernating, and
there is no wake-up freeze, but automatic because of flat battery is the
usual reason to hibernate.

you do realize that this console switch can be automated?

for instance, i put vidcontrol -s 1 < /dev/ttyv0 in rc.suspend just before
the logger line and vidcontrol -s 9 < /dev/ttyv0 (12 in your case) into
rc.resume, also before the logger line.

this workaround has been mentioned quite a few times on the list.

keep in mind though that rc.suspend and rc.resume were only recently made
ready for acpi, so if you are using acpi instead of apm, you might have
to update to -current. for ibm thinkpads, this is generally a good idea
anyway, at least if you use acpi.

cheers, t.