acpiphp makes noise on every lid close/open

acpiphp makes noise on every lid close/open

[Pavel Machek]

Hi!

With 2.6.19-rc4, acpi complains about "acpiphp_glue: cannot get bridge
info" each time I close/reopen the lid... On thinkpad x60. Any ideas?
(-mm1 behaves the same).
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: acpiphp makes noise on every lid close/open

[Kristen Carlson Accardi]

On Wed, 1 Nov 2006 12:56:18 +0100
Pavel Machek <pavel@ucw.cz> wrote:

> Hi!
> 
> With 2.6.19-rc4, acpi complains about "acpiphp_glue: cannot get bridge
> info" each time I close/reopen the lid... On thinkpad x60. Any ideas?
> (-mm1 behaves the same).
> 									Pavel
> -- 
> (english) http://www.livejournal.com/~pavelmachek
> (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Looks like acpi is sending a BUS_CHECK notification to acpiphp on the 
PCI Root Bridge whenever the lid opens up.

There is a bug here in that acpiphp shouldn't even be used on the X60 -
it has no hotpluggable slots.  This problem only occurs when acpiphp is
built in, as when a module it just doesn't load.  It appears to not clean
up after itself properly when it finds no ejectable slots and leaves the
acpi notifier installed for the PCI Root Bridge.  The message is printing
"cannot get bridge info" because it partially cleaned some stuff up (without
actually removing the notifier).  I'll put this bug into bugzilla since
I won't have time to fix right away:

http://bugzilla.kernel.org/show_bug.cgi?id=7452

Feel free to add yourself to the CC list if you are interested in being
notified when it is fixed.

Kristen
 

Re: acpiphp makes noise on every lid close/open

[David Weinehall]

On Thu, Nov 02, 2006 at 05:54:03PM -0800, Kristen Carlson Accardi wrote:
> On Wed, 1 Nov 2006 12:56:18 +0100
> Pavel Machek <pavel@ucw.cz> wrote:
> 
> > Hi!
> > 
> > With 2.6.19-rc4, acpi complains about "acpiphp_glue: cannot get bridge
> > info" each time I close/reopen the lid... On thinkpad x60. Any ideas?
> > (-mm1 behaves the same).
> > 									Pavel
> > -- 
> > (english) http://www.livejournal.com/~pavelmachek
> > (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
> 
> Looks like acpi is sending a BUS_CHECK notification to acpiphp on the 
> PCI Root Bridge whenever the lid opens up.
> 
> There is a bug here in that acpiphp shouldn't even be used on the X60 -
> it has no hotpluggable slots.

How about the docking station?


Regards: David
-- 
 /) David Weinehall <tao@acc.umu.se> /) Northern lights wander      (\
//  Maintainer of the v2.0 kernel   //  Dance across the winter sky //
\)  http://www.acc.umu.se/~tao/    (/   Full colour fire           (/

Re: acpiphp makes noise on every lid close/open

[Stefan Seyfried]

On Mon, Nov 06, 2006 at 12:29:44AM +0100, David Weinehall wrote:
> On Thu, Nov 02, 2006 at 05:54:03PM -0800, Kristen Carlson Accardi wrote:
> 
> > There is a bug here in that acpiphp shouldn't even be used on the X60 -
> > it has no hotpluggable slots.
> 
> How about the docking station?

Hm, i haven't seen one - i only saw the "media slice" (or however it is
called this week) which gives you a drive bay, but no additional PCI
devices, so it does not count.
-- 
Stefan Seyfried
QA / R&D Team Mobile Devices        |              "Any ideas, John?"
SUSE LINUX Products GmbH, Nürnberg  | "Well, surrounding them's out." 

Re: acpiphp makes noise on every lid close/open

[Pavel Machek]

Hi!

> > > With 2.6.19-rc4, acpi complains about "acpiphp_glue: cannot get bridge
> > > info" each time I close/reopen the lid... On thinkpad x60. Any ideas?
> > > (-mm1 behaves the same).
> > 
> > Looks like acpi is sending a BUS_CHECK notification to acpiphp on the 
> > PCI Root Bridge whenever the lid opens up.
> > 
> > There is a bug here in that acpiphp shouldn't even be used on the X60 -
> > it has no hotpluggable slots.
> 
> How about the docking station?

"Dock" for x60 only contains cdrom slot and aditional slots, no PCI or
PCMCIA slots.
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: acpiphp makes noise on every lid close/open

[Henrique de Moraes Holschuh]

> > > There is a bug here in that acpiphp shouldn't even be used on the X60 -
> > > it has no hotpluggable slots.

What about the internal mini PCI express slots (where the Intel 3945ABG
device and EVDO wwwan cards are plugged)?

Also, ThinkWiki lists this machine as one that has a CardBus slot.  A photo
I found also shows this cardbus slot, left side near palm rest, right above
a SD card slot.  According to Lenovo, it appears that this cardbus slot is
also a 34mm express card slot, which would account for a PCIe x1 hotplug bus
and a USB2 hotplug bus in addition to the PCI (pccard) hotplug bus.

And it is extremely likely that this notebook exposes either PCI or PCIe
lanes to the dock station.  Whether they implemented PCIe or PCI slots in
the dock is another story, of course...

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Re: acpiphp makes noise on every lid close/open

[Tomasz Torcz]

[-- Attachment #1: Type: text/plain, Size: 645 bytes --]

On Mon, Nov 06, 2006 at 11:29:03AM -0200, Henrique de Moraes Holschuh wrote:
> > > > There is a bug here in that acpiphp shouldn't even be used on the X60 -
> > > > it has no hotpluggable slots.
> 
> What about the internal mini PCI express slots (where the Intel 3945ABG
> device and EVDO wwwan cards are plugged)?

  EVDO is connected to USB. At least it works with usb-serial (after
manually binding IDs). Works == responds to Hayes commands, I didn't
tried with SIM Card yet.

-- 
Tomasz Torcz               RIP is irrevelant. Spoofing is futile.
zdzichu@irc.-nie.spam-.pl     Your routes will be aggreggated. -- Alex Yuriev


[-- Attachment #2: Type: application/pgp-signature, Size: 229 bytes --]

Re: acpiphp makes noise on every lid close/open

[Stefan Seyfried]

On Mon, Nov 06, 2006 at 11:29:03AM -0200, Henrique de Moraes Holschuh wrote:
> > > > There is a bug here in that acpiphp shouldn't even be used on the X60 -
> > > > it has no hotpluggable slots.
> 
> What about the internal mini PCI express slots (where the Intel 3945ABG
> device and EVDO wwwan cards are plugged)?

These are not in the media bay. I'd also guess that they are not
hotpluggable.
 
> Also, ThinkWiki lists this machine as one that has a CardBus slot.  A photo

The cardbus slot is also not in the media bay and AFAIK not handled by
acpiphp.

-- 
Stefan Seyfried
QA / R&D Team Mobile Devices        |              "Any ideas, John?"
SUSE LINUX Products GmbH, Nürnberg  | "Well, surrounding them's out." 

Re: acpiphp makes noise on every lid close/open

[Henrique de Moraes Holschuh]

On Mon, 06 Nov 2006, Stefan Seyfried wrote:
> On Mon, Nov 06, 2006 at 11:29:03AM -0200, Henrique de Moraes Holschuh wrote:
> > > > > There is a bug here in that acpiphp shouldn't even be used on the X60 -
> > > > > it has no hotpluggable slots.
> > 
> > What about the internal mini PCI express slots (where the Intel 3945ABG
> > device and EVDO wwwan cards are plugged)?
> 
> These are not in the media bay. I'd also guess that they are not
> hotpluggable.

They are hotplug buses.  But the buses themselves are always there (and not
hotplugged. i.e. the bus doesn't appear and disappear).  Whether they should
be bothering acpiphp, I don't know.  

You have a lot of buses that support hotplugging in the X60, and at least
one is certainly exported to the dock unless IBM/Lenovo changed completely
the way they make docks, which I very much doubt.  That was the information
I was trying to convey.   As I said, I don't know if *these* buses should be
bothering acpiphp or not.

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Re: acpiphp makes noise on every lid close/open

[David Weinehall]

On Mon, Nov 06, 2006 at 10:21:17AM +0100, Pavel Machek wrote:
> Hi!
> 
> > > > With 2.6.19-rc4, acpi complains about "acpiphp_glue: cannot get bridge
> > > > info" each time I close/reopen the lid... On thinkpad x60. Any ideas?
> > > > (-mm1 behaves the same).
> > > 
> > > Looks like acpi is sending a BUS_CHECK notification to acpiphp on the 
> > > PCI Root Bridge whenever the lid opens up.
> > > 
> > > There is a bug here in that acpiphp shouldn't even be used on the X60 -
> > > it has no hotpluggable slots.
> > 
> > How about the docking station?
> 
> "Dock" for x60 only contains cdrom slot and aditional slots, no PCI or
> PCMCIA slots.

Well, when I press the undock button on the dock without the acpiphp
module loaded, I never get the green light that confirms that removing
the laptop is safe.  If acpiphp is loaded, things work just fine.


Regards: David
-- 
 /) David Weinehall <tao@acc.umu.se> /) Northern lights wander      (\
//  Maintainer of the v2.0 kernel   //  Dance across the winter sky //
\)  http://www.acc.umu.se/~tao/    (/   Full colour fire           (/

Re: acpiphp makes noise on every lid close/open

[Kristen Carlson Accardi]

On Tue, 7 Nov 2006 21:44:09 +0100
David Weinehall <tao@acc.umu.se> wrote:

> On Mon, Nov 06, 2006 at 10:21:17AM +0100, Pavel Machek wrote:
> > Hi!
> > 
> > > > > With 2.6.19-rc4, acpi complains about "acpiphp_glue: cannot get bridge
> > > > > info" each time I close/reopen the lid... On thinkpad x60. Any ideas?
> > > > > (-mm1 behaves the same).
> > > > 
> > > > Looks like acpi is sending a BUS_CHECK notification to acpiphp on the 
> > > > PCI Root Bridge whenever the lid opens up.
> > > > 
> > > > There is a bug here in that acpiphp shouldn't even be used on the X60 -
> > > > it has no hotpluggable slots.
> > > 
> > > How about the docking station?
> > 
> > "Dock" for x60 only contains cdrom slot and aditional slots, no PCI or
> > PCMCIA slots.
> 
> Well, when I press the undock button on the dock without the acpiphp
> module loaded, I never get the green light that confirms that removing
> the laptop is safe.  If acpiphp is loaded, things work just fine.
> 
> 
> Regards: David
> -- 
>  /) David Weinehall <tao@acc.umu.se> /) Northern lights wander      (\
> //  Maintainer of the v2.0 kernel   //  Dance across the winter sky //
> \)  http://www.acc.umu.se/~tao/    (/   Full colour fire           (/

David,
What kernel version are you using?  You should not need acpiphp to do
docking on the X60.  If you are using a recent kernel, do you mind sending
the dmesg output so we can figure out why this doesn't work for you?

Thanks,
kristen

Re: acpiphp makes noise on every lid close/open

[David Weinehall]

On Tue, Nov 07, 2006 at 01:44:39PM -0800, Kristen Carlson Accardi wrote:
> On Tue, 7 Nov 2006 21:44:09 +0100
> David Weinehall <tao@acc.umu.se> wrote:
> 
> > On Mon, Nov 06, 2006 at 10:21:17AM +0100, Pavel Machek wrote:
> > > Hi!
> > > 
> > > > > > With 2.6.19-rc4, acpi complains about "acpiphp_glue: cannot get bridge
> > > > > > info" each time I close/reopen the lid... On thinkpad x60. Any ideas?
> > > > > > (-mm1 behaves the same).
> > > > > 
> > > > > Looks like acpi is sending a BUS_CHECK notification to acpiphp on the 
> > > > > PCI Root Bridge whenever the lid opens up.
> > > > > 
> > > > > There is a bug here in that acpiphp shouldn't even be used on the X60 -
> > > > > it has no hotpluggable slots.
> > > > 
> > > > How about the docking station?
> > > 
> > > "Dock" for x60 only contains cdrom slot and aditional slots, no PCI or
> > > PCMCIA slots.
> > 
> > Well, when I press the undock button on the dock without the acpiphp
> > module loaded, I never get the green light that confirms that removing
> > the laptop is safe.  If acpiphp is loaded, things work just fine.
> > 
> > 
> > Regards: David
> > -- 
> >  /) David Weinehall <tao@acc.umu.se> /) Northern lights wander      (\
> > //  Maintainer of the v2.0 kernel   //  Dance across the winter sky //
> > \)  http://www.acc.umu.se/~tao/    (/   Full colour fire           (/
> 
> David,
> What kernel version are you using?  You should not need acpiphp to do
> docking on the X60.  If you are using a recent kernel, do you mind sending
> the dmesg output so we can figure out why this doesn't work for you?

That was with 2.6.17; with 2.7.19-pre? (don't remember right now),
docking seems to work without acpiphp.  It still would be nice to be
able to undock when the laptop is sleeping though; how do I achieve
that?


Regards: David
-- 
 /) David Weinehall <tao@acc.umu.se> /) Northern lights wander      (\
//  Maintainer of the v2.0 kernel   //  Dance across the winter sky //
\)  http://www.acc.umu.se/~tao/    (/   Full colour fire           (/

Re: acpiphp makes noise on every lid close/open

[Matthew Garrett]

On Fri, Nov 17, 2006 at 11:22:38AM +0100, David Weinehall wrote:

> That was with 2.6.17; with 2.7.19-pre? (don't remember right now),
> docking seems to work without acpiphp.  It still would be nice to be
> able to undock when the laptop is sleeping though; how do I achieve
> that?

My experience of most laptops is that they'll fire off a bus check 
notification when you resume, so as long as nothing actually tries to 
access the hardware before that's handled, everything should be fine. 
What currently breaks when you undock while asleep?

-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: acpiphp makes noise on every lid close/open

[Pavel Machek]

On Fri 2006-11-17 15:13:41, Matthew Garrett wrote:
> On Fri, Nov 17, 2006 at 11:22:38AM +0100, David Weinehall wrote:
> 
> > That was with 2.6.17; with 2.7.19-pre? (don't remember right now),
> > docking seems to work without acpiphp.  It still would be nice to be
> > able to undock when the laptop is sleeping though; how do I achieve
> > that?
> 
> My experience of most laptops is that they'll fire off a bus check 
> notification when you resume, so as long as nothing actually tries to 
> access the hardware before that's handled, everything should be fine. 
> What currently breaks when you undock while asleep?

(At least some versions of dock can't undock -- physically -- when not
powered. Something is locked inside, and won't unlock without electricity).
								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: acpiphp makes noise on every lid close/open

[David Weinehall]

On Fri, Nov 17, 2006 at 03:13:41PM +0000, Matthew Garrett wrote:
> On Fri, Nov 17, 2006 at 11:22:38AM +0100, David Weinehall wrote:
> 
> > That was with 2.6.17; with 2.7.19-pre? (don't remember right now),
> > docking seems to work without acpiphp.  It still would be nice to be
> > able to undock when the laptop is sleeping though; how do I achieve
> > that?
> 
> My experience of most laptops is that they'll fire off a bus check 
> notification when you resume, so as long as nothing actually tries to 
> access the hardware before that's handled, everything should be fine. 
> What currently breaks when you undock while asleep?

The fact that the dock starts to beep annoyingly. It has a button that
you should press before undocking, and wait for a green light to light
up before removing the laptop.  If you remove the computer without
doing so, the dock starts beeping, and it doesn't stop (AFAIK, haven't
managed to stand the beeping for more than 30 seconds or so) until you
replug the laptop.

My guess is that there is some wait to trigger an undock event from
software as well, and that it would be nice to send that signal to the
dock before suspending...


Regards: David
-- 
 /) David Weinehall <tao@acc.umu.se> /) Northern lights wander      (\
//  Maintainer of the v2.0 kernel   //  Dance across the winter sky //
\)  http://www.acc.umu.se/~tao/    (/   Full colour fire           (/

Re: acpiphp makes noise on every lid close/open

[Matthew Garrett]

On Fri, Nov 17, 2006 at 04:37:17PM +0100, David Weinehall wrote:

> The fact that the dock starts to beep annoyingly. It has a button that
> you should press before undocking, and wait for a green light to light
> up before removing the laptop.  If you remove the computer without
> doing so, the dock starts beeping, and it doesn't stop (AFAIK, haven't
> managed to stand the beeping for more than 30 seconds or so) until you
> replug the laptop.

Ah, hm. Interesting. Maybe it does want OS support, then. Have you tried 
it in the Leading Brand OS?

> My guess is that there is some wait to trigger an undock event from
> software as well, and that it would be nice to send that signal to the
> dock before suspending...

You possibly don't want to do that if there's a mounted bay device in 
the dock.

We really need to determine some sort of policy when it comes to mounted 
devices that will potentially be removed by the user over 
suspend/resume. Do we support this configuration (by not killing the 
mount point), or do we prevent users from shooting themselves in the 
foot?

-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: acpiphp makes noise on every lid close/open

[David Weinehall]

On Fri, Nov 17, 2006 at 03:46:27PM +0000, Matthew Garrett wrote:
> On Fri, Nov 17, 2006 at 04:37:17PM +0100, David Weinehall wrote:
> 
> > The fact that the dock starts to beep annoyingly. It has a button that
> > you should press before undocking, and wait for a green light to light
> > up before removing the laptop.  If you remove the computer without
> > doing so, the dock starts beeping, and it doesn't stop (AFAIK, haven't
> > managed to stand the beeping for more than 30 seconds or so) until you
> > replug the laptop.
> 
> Ah, hm. Interesting. Maybe it does want OS support, then. Have you tried 
> it in the Leading Brand OS?

Nope.

> > My guess is that there is some wait to trigger an undock event from
> > software as well, and that it would be nice to send that signal to the
> > dock before suspending...
> 
> You possibly don't want to do that if there's a mounted bay device in 
> the dock.

Well, right now I don't have a bay device in the dock, but refusing to
suspend in that case would be reasonable.  I need to add some hotplug
script for the bay device anyway if I add one.

> We really need to determine some sort of policy when it comes to mounted 
> devices that will potentially be removed by the user over 
> suspend/resume. Do we support this configuration (by not killing the 
> mount point), or do we prevent users from shooting themselves in the 
> foot?

Good question.  Personally I'd say we refuse to suspend when we have
devices we *know* to be dock-devices etc mounted.


Regards: David
-- 
 /) David Weinehall <tao@acc.umu.se> /) Northern lights wander      (\
//  Maintainer of the v2.0 kernel   //  Dance across the winter sky //
\)  http://www.acc.umu.se/~tao/    (/   Full colour fire           (/

Re: acpiphp makes noise on every lid close/open

[Matthew Garrett]

On Fri, Nov 17, 2006 at 05:08:10PM +0100, David Weinehall wrote:

> Good question.  Personally I'd say we refuse to suspend when we have
> devices we *know* to be dock-devices etc mounted.

Kernel-level or userspace? IBM certainly used to sell bay-mounted hard 
drives, and while it's possible for a user to pull one out while the 
machine is suspended, I suspect that the general use case is probably 
for it to carry on being used.

Possibly what's needed is something like Apple's nullfs - force unmount 
the drive on suspend, and put a nullfs there instead. On resume, if the 
drive is still there, remount it. If not, userspace applications get 
upset about the missing drive but no data is lost. The downside to this 
approach would be trying to figure out how to get the drive remounted 
before the rest of userspace starts trying to scribble over it again...

-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: acpiphp makes noise on every lid close/open

[David Weinehall]

On Fri, Nov 17, 2006 at 04:31:28PM +0000, Matthew Garrett wrote:
> On Fri, Nov 17, 2006 at 05:08:10PM +0100, David Weinehall wrote:
> 
> > Good question.  Personally I'd say we refuse to suspend when we have
> > devices we *know* to be dock-devices etc mounted.
> 
> Kernel-level or userspace? IBM certainly used to sell bay-mounted hard 
> drives, and while it's possible for a user to pull one out while the 
> machine is suspended, I suspect that the general use case is probably 
> for it to carry on being used.

I'd say that kind of policies should be left to userspace, since it's up
to userspace to decide if we suspend at all or not.  If the user tells
the kernel to shoot him in the foot, the kernel should take aim
carefully and turn him into a cripple.

> Possibly what's needed is something like Apple's nullfs - force unmount 
> the drive on suspend, and put a nullfs there instead. On resume, if the 
> drive is still there, remount it. If not, userspace applications get 
> upset about the missing drive but no data is lost. The downside to this 
> approach would be trying to figure out how to get the drive remounted 
> before the rest of userspace starts trying to scribble over it again...

Mmmm.


Regards: David
-- 
 /) David Weinehall <tao@acc.umu.se> /) Northern lights wander      (\
//  Maintainer of the v2.0 kernel   //  Dance across the winter sky //
\)  http://www.acc.umu.se/~tao/    (/   Full colour fire           (/

Re: acpiphp makes noise on every lid close/open

[Matthew Garrett]

On Fri, Nov 17, 2006 at 04:31:28PM +0000, Matthew Garrett wrote:

> Possibly what's needed is something like Apple's nullfs 

deadfs, rather.

-- 
Matthew Garrett | mjg59@srcf.ucam.org