APM and ACPI sleep issues with 2.6

APM and ACPI sleep issues with 2.6

[Ross Burton]

[-- Attachment #1: Type: text/plain, Size: 1156 bytes --]

Hi,

I've just built 2.6.1-rc1-mm1, and am using it on my IBM ThinkPad X22. 
However, it not won't sleep when I shut the lid.

With 2.4.22 (from debian sid's kernel package) APM works, and will
suspend-to-RAM when I shut the lid, and wake up again cleanly.  As that
worked well I never got around to configuring ACPI.

With 2.6.1-rc1-mm, when I shut the lid with APM enabled nothing
happens.  No messages on the console, nothing.

With ACPI and "echo 3 > /proc/bus/acpi/sleep" in the relevant acpid
file, shutting the lid results in at least a partial sleep, but it won't
resume.  The power button was also non-responsive at this point, and I
had to remove the battery to reboot.

So, what can I do to debug this problem?  APM not working is a
regression over 2.4.22, which is sad.

Thanks for any help,
Ross
(please CC: me, I'm not on this list)
-- 
Ross Burton                                 mail: ross@burtonini.com
                                          jabber: ross@burtonini.com
                                     www: http://www.burtonini.com./
 PGP Fingerprint: 1A21 F5B0 D8D0 CFE3 81D4 E25A 2D09 E447 D0B4 33DF


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: APM and ACPI sleep issues with 2.6

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 397 bytes --]

Hi Ross,

On Sun, 04 Jan 2004 16:05:51 +0000 Ross Burton <ross@burtonini.com> wrote:
>
> With 2.6.1-rc1-mm, when I shut the lid with APM enabled nothing
> happens.  No messages on the console, nothing.

Can you try booting with apm=debug and see if you get any messages
when you shut the lid.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: APM and ACPI sleep issues with 2.6

[Ross Burton]

On Mon, 2004-01-05 at 03:00, Stephen Rothwell wrote:
> > With 2.6.1-rc1-mm, when I shut the lid with APM enabled nothing
> > happens.  No messages on the console, nothing.
> 
> Can you try booting with apm=debug and see if you get any messages
> when you shut the lid.

Sorry for the long delay...

hda: start_power_step(step: 0)
hda: start_power_step(step: 1)
hda: complete_power_step(step: 1, stat: 50, err: 0)
hda: completing PM request, suspend
PCI: Setting latency timer of device 0000:00:1d.0 to 64
PCI: Setting latency timer of device 0000:00:1d.1 to 64
PCI: Setting latency timer of device 0000:00:1d.2 to 64
hda: Wakeup request inited, waiting for !BSY...
e100: eth0: e100_watchdog: link up, 100Mbps, full-duplex
hda: start_power_step(step: 1000)
hda: completing PM request, resume
PCI: Found IRQ 5 for device 0000:00:1f.5
PCI: Sharing IRQ 5 with 0000:00:1f.3
PCI: Sharing IRQ 5 with 0000:02:03.1
PCI: Setting latency timer of device 0000:00:1f.5 to 64
intel8x0_measure_ac97_clock: measured 49401 usecs
intel8x0: clocking to 48000
request_module: failed /sbin/modprobe -- snd-card-1. error = 256
request_module: failed /sbin/modprobe -- snd-card-1. error = 256
atkbd.c: Unknown key released (translated set 2, code 0x7a on
isa0060/serio0).
atkbd.c: Unknown key released (translated set 2, code 0x7a on

Looks like e100 (network) and intel8x0 (sound) recovered okay.  The
second time I shut the lid, when I opened it the keyboard had locked up.

I've been told that building 2.6.1-mm4, making i8042 and atkdb modules
and unloading them before sleeping should fix this problem.  Is that the
blessed solution?  Unloading the modules for the keyboard controller
does seem a little too much like brute-force for me, especially since
2.4.x managed fine. :)

Thanks,
Ross
-- 
Ross Burton                                 mail: ross@burtonini.com
                                          jabber: ross@burtonini.com
                                     www: http://www.burtonini.com./
 PGP Fingerprint: 1A21 F5B0 D8D0 CFE3 81D4 E25A 2D09 E447 D0B4 33DF

Re: APM and ACPI sleep issues with 2.6

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1244 bytes --]

On Mon, 19 Jan 2004 13:47:45 +0000 Ross Burton <ross@burtonini.com> wrote:
>
> On Mon, 2004-01-05 at 03:00, Stephen Rothwell wrote:
> > > With 2.6.1-rc1-mm, when I shut the lid with APM enabled nothing
> > > happens.  No messages on the console, nothing.
> > 
> > Can you try booting with apm=debug and see if you get any messages
> > when you shut the lid.
> 
> Sorry for the long delay...

I was actually wondering if there are any messages beginning with "apm:"

> Looks like e100 (network) and intel8x0 (sound) recovered okay.  The
> second time I shut the lid, when I opened it the keyboard had locked up.
> 
> I've been told that building 2.6.1-mm4, making i8042 and atkdb modules
> and unloading them before sleeping should fix this problem.  Is that the
> blessed solution?  Unloading the modules for the keyboard controller
> does seem a little too much like brute-force for me, especially since
> 2.4.x managed fine. :)

I am not sure if you need to build i8042 and atkb as modules amy more, I
thought there was a fix applied (in 2.6.1?).  However it would be
interesting to the results of removing the modules before suspending.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: APM and ACPI sleep issues with 2.6

[Ross Burton]

On Tue, 2004-01-20 at 14:59, Stephen Rothwell wrote:
> On Mon, 19 Jan 2004 13:47:45 +0000 Ross Burton <ross@burtonini.com> wrote:
> >
> > On Mon, 2004-01-05 at 03:00, Stephen Rothwell wrote:
> > > > With 2.6.1-rc1-mm, when I shut the lid with APM enabled nothing
> > > > happens.  No messages on the console, nothing.
> > > 
> > > Can you try booting with apm=debug and see if you get any messages
> > > when you shut the lid.
> > 
> > Sorry for the long delay...
> 
> I was actually wondering if there are any messages beginning with "apm:"

Hm, nothing was produced on the console, in dmesg, or in syslog.

> > I've been told that building 2.6.1-mm4, making i8042 and atkdb modules
> > and unloading them before sleeping should fix this problem.  Is that the
> > blessed solution?  Unloading the modules for the keyboard controller
> > does seem a little too much like brute-force for me, especially since
> > 2.4.x managed fine. :)
> 
> I am not sure if you need to build i8042 and atkb as modules amy more, I
> thought there was a fix applied (in 2.6.1?).  However it would be
> interesting to the results of removing the modules before suspending.

Exactly the same messages...

:(

Ross
-- 
Ross Burton                                 mail: ross@burtonini.com
                                          jabber: ross@burtonini.com
                                     www: http://www.burtonini.com./
 PGP Fingerprint: 1A21 F5B0 D8D0 CFE3 81D4 E25A 2D09 E447 D0B4 33DF

Re: APM and ACPI sleep issues with 2.6 (2.6.2pre1-mm1 vs mm2)

[Ross Burton]

On Tue, 2004-01-20 at 14:59, Stephen Rothwell wrote:
> > I've been told that building 2.6.1-mm4, making i8042 and atkdb modules
> > and unloading them before sleeping should fix this problem.  Is that the
> > blessed solution?  Unloading the modules for the keyboard controller
> > does seem a little too much like brute-force for me, especially since
> > 2.4.x managed fine. :)
> 
> I am not sure if you need to build i8042 and atkb as modules amy more, I
> thought there was a fix applied (in 2.6.1?).  However it would be
> interesting to the results of removing the modules before suspending.

After some frantic kernel building I've more interesting data (with
i8042 and atkbd built into the kernel).

2.6.2-rc1-mm1 will APM suspend and resume fine at first with the
following log:

atkbd.c: Unknown key released (translated set 2, code 0x7a on isa0060/serio0).
atkbd.c: This is an XFree86 bug. It shouldn't access hardware directly.
atkbd.c: Unknown key released (translated set 2, code 0x7a on isa0060/serio0).
atkbd.c: This is an XFree86 bug. It shouldn't access hardware directly.
hda: start_power_step(step: 0)
hda: start_power_step(step: 1)
hda: complete_power_step(step: 1, stat: 50, err: 0)
hda: completing PM request, suspend
PCI: Setting latency timer of device 0000:00:1d.0 to 64
PCI: Setting latency timer of device 0000:00:1d.1 to 64
PCI: Setting latency timer of device 0000:00:1d.2 to 64
PCI: Found IRQ 5 for device 0000:00:1f.5
PCI: Sharing IRQ 5 with 0000:00:1f.3
PCI: Sharing IRQ 5 with 0000:02:03.1
PCI: Setting latency timer of device 0000:00:1f.5 to 64
hda: Wakeup request inited, waiting for !BSY...
hda: start_power_step(step: 1000)
hda: completing PM request, resume
MCE: The hardware reports a non fatal, correctable incident occurred on CPU 0.
Bank 1: f200000000000095

The atkbd lines are interesting as I was not running X at the time.  The
MCE line slightly worries me, is this normal?

However, I cannot sleep once I've used PCMCIA (with yenta_socket). 
Instead of sleeping it locks up.  I didn't notice this the first time
and came back to a very warm laptop, it felt like it had been
busy-looping.


I was very excited by the changelog for 2.6.2-rc1-mm2 with its "updated
ACPI" so I tried that.  ACPI still hard-crashes (grey screen, power
button dead, I had to remove the battery) and APM also crashes: it looks
like it is suspending (screen turns off and there is some disk activity)
but it never sleeps and I can't resume without holding the power button
(so at least *something* is still running).

Summary: something regressed over mm1 to mm2 with APM suspending, and
yenta_socket or cs isn't handling the suspend correctly.  Does anyone
have any ideas where I can look for these?

However, I did manage to use 2.6.2-pre1-mm1 for a day before I used my
wifi card, and did like what I saw -- the new scheduler makes GNOME more
responsive when I've a compile going.

Ross
-- 
Ross Burton                                 mail: ross@burtonini.com
                                          jabber: ross@burtonini.com
                                     www: http://www.burtonini.com./
 PGP Fingerprint: 1A21 F5B0 D8D0 CFE3 81D4 E25A 2D09 E447 D0B4 33DF