Bug#665493: linux-image-3.2.0-2-amd64: kernel fails to boot completely (timeout on modprobe -b pci:xxxx)
 

Package: linux-2.6
Followup-For: Bug #665493

Dear Maintainer,

I'm seeing the same symptoms as the original reporter on a free install of wheezy on an HP Folio 13-2000.

Since I seem to be too stupid to compile a booting custom kernel I installed linux-image-3.2.0-0.bpo.2-amd64 (3.2.20-1~bpo60+1) – and it works.

Cheers,
Stefan Nagy.



-- System Information:
Debian Release: wheezy/sid
APT prefers testing
APT policy: (500, 'testing')
Architecture: amd64 (x86_64)

Kernel: Linux 3.2.0-3-amd64 (SMP w/4 CPU cores)
Locale: LANG=de_AT.UTF-8, LC_CTYPE=de_AT.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash


--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/20120725144749.14433.35820.reportbug@rosa

Bug#665493: linux-image-3.2.0-2-amd64: kernel fails to boot completely (timeout on modprobe -b pci:xxxx)
 

Sorry, I forgot to attach my output of dmesg & acpidump last time. BTW,
I was talking about a fresh install of wheezy (weekly build 2012-07-16).

@ Jeroen Nijhof:
> Is there a kernel boot parameter, maybe, with which I can disable the
> built-in ACPI hotplug?

I use option pci=noacpi with linux-image-3.2.0-3-amd64 (3.2.21-3) and it
works without any problems.

Cheers,
Stefan.

Bug#665493: linux-image-3.2.0-2-amd64: kernel fails to boot completely (timeout on modprobe -b pci:xxxx)
 

As the original reporter pointed out, this bug is triggered by a change
in the debian kernel configuration regarding the ACPI PCI Hotplug driver
[1]. Since kernel update linux-image-3.2.0-2-amd64 (3.2.10-1) this
driver is build in (CONFIG_HOTPLUG_PCI_ACPI=y) instead of being compiled
as a module (CONFIG_HOTPLUG_PCI_ACPI=m).

The kernel changelog lists Ben Hutchings as author for this change:
- [x86] Change HOTPLUG_PCI_ACPI to built-in (Closes: #663433) [2]

The reason for this decision was that the ACPI PCI Hotplug driver didn't
get loaded on machines which need it for ExpressCard hotplugging to work
[3].

Since this bug got no attention at all on the kernel-pci mailinglist for
three months now [4] I wanted to file a bugreport upstream against
driver: pci – but I'm not sure what to do now since it doesn't seem to
be very safe to build-in this driver and maybe the debian kernel team
wants to reconsider this decision…?

Please tell me if this decision is final – even though the kernel fails
to load on at least two recent machines with the ACPI PCI Hotplug driver
build in. If it is I'll file a bugreport upstream.

Thanks,
Stefan.



P.S. The default kernel configuration for this item is
CONFIG_HOTPLUG_PCI_ACPI=n. I also had a look at this configuration item
in Ubuntu – it's set to CONFIG_HOTPLUG_PCI_ACPI=m.

[1] http://cateee.net/lkddb/web-lkddb/HOTPLUG_PCI_ACPI.html
[2]
http://packages.debian.org/changelogs/pool/main/l/linux/linux_3.2.23-1/changelog#version3.2.10-1
[3] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=663433
[4] http://thread.gmane.org/gmane.linux.kernel.pci/15559

Bug#665493: linux-image-3.2.0-2-amd64: kernel fails to boot completely (timeout on modprobe -b pci:xxxx)
 

On Sun, 2012-08-26 at 21:37 +0200, Stefan Nagy wrote:
> As the original reporter pointed out, this bug is triggered by a change
> in the debian kernel configuration regarding the ACPI PCI Hotplug driver
> [1]. Since kernel update linux-image-3.2.0-2-amd64 (3.2.10-1) this
> driver is build in (CONFIG_HOTPLUG_PCI_ACPI=y) instead of being compiled
> as a module (CONFIG_HOTPLUG_PCI_ACPI=m).
>
> The kernel changelog lists Ben Hutchings as author for this change:
> - [x86] Change HOTPLUG_PCI_ACPI to built-in (Closes: #663433) [2]
>
> The reason for this decision was that the ACPI PCI Hotplug driver didn't
> get loaded on machines which need it for ExpressCard hotplugging to work
> [3].
>
> Since this bug got no attention at all on the kernel-pci mailinglist for
> three months now [4] I wanted to file a bugreport upstream against
> driver: pci – but I'm not sure what to do now since it doesn't seem to
> be very safe to build-in this driver and maybe the debian kernel team
> wants to reconsider this decision…?

I think it is the correct decision. Building it as a module effectively
meant it was disabled on most machine, since it doesn't get auto-loaded.

It may be preferable to return to building it as a module *if* we can
arrange for it to be loaded automatically where it's useful. But that
presumably would not fix this bug.

> Please tell me if this decision is final – even though the kernel fails
> to load on at least two recent machines with the ACPI PCI Hotplug driver
> build in. If it is I'll file a bugreport upstream.

Please test Linux 3.5 (packaged in experimental) and if that has the
same problem then open an upstream bug report.

> P.S. The default kernel configuration for this item is
> CONFIG_HOTPLUG_PCI_ACPI=n.

The default kernel configuration has a lot of features disabled that
many users need.

> I also had a look at this configuration item
> in Ubuntu – it's set to CONFIG_HOTPLUG_PCI_ACPI=m.

Have you checked whether they add it to /proc/modules?

Ben.

> [1] http://cateee.net/lkddb/web-lkddb/HOTPLUG_PCI_ACPI.html
> [2]
> http://packages.debian.org/changelogs/pool/main/l/linux/linux_3.2.23-1/changelog#version3.2.10-1
> [3] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=663433
> [4] http://thread.gmane.org/gmane.linux.kernel.pci/15559

--
Ben Hutchings
Experience is what causes a person to make new mistakes instead of old ones.

Bug#665493: linux-image-3.2.0-2-amd64: kernel fails to boot completely (timeout on modprobe -b pci:xxxx)
 

Am Sonntag, den 26.08.2012, 13:37 -0700 schrieb Ben Hutchings:
> Please test Linux 3.5 (packaged in experimental) and if that has the
> same problem then open an upstream bug report.

OK – I already tested linux-image-3.5-trunk-amd64
(3.5.2-1~experimental.1) & the bug is still present, so I'll file the
upstream bug report soon.

> > P.S. The default kernel configuration for this item is
> > CONFIG_HOTPLUG_PCI_ACPI=n.
>
> The default kernel configuration has a lot of features disabled that
> many users need.

I see.

> > I also had a look at this configuration item
> > in Ubuntu – it's set to CONFIG_HOTPLUG_PCI_ACPI=m.
>
> Have you checked whether they add it to /proc/modules?

You mean if they add it to /etc/modules, right? I don't know, I haven't
checked this yet.

But this is maybe interesting: When I compile my kernel with
CONFIG_HOTPLUG_PCI_ACPI=m and put acpiphp in /etc/modules I can't
reproduce this bug even though I'm sure the module is loaded:

[ 19.299137] acpiphp: ACPI Hot Plug PCI Controller Driver version: 0.5
[ 19.299341] acpiphp: Slot [1] registered

$ lsmod | grep acpiphp
acpiphp 22567 0


Stefan.

Bug#665493: linux-image-3.2.0-2-amd64: kernel fails to boot completely (timeout on modprobe -b pci:xxxx)
 

[Re-sending with the bug address cc'd.]

On Sun, 2012-08-26 at 23:31 +0200, Stefan Nagy wrote:
> Am Sonntag, den 26.08.2012, 13:37 -0700 schrieb Ben Hutchings:
> > Please test Linux 3.5 (packaged in experimental) and if that has the
> > same problem then open an upstream bug report.
>
> OK – I already tested linux-image-3.5-trunk-amd64
> (3.5.2-1~experimental.1) & the bug is still present, so I'll file the
> upstream bug report soon.

Thanks.

> > > P.S. The default kernel configuration for this item is
> > > CONFIG_HOTPLUG_PCI_ACPI=n.
> >
> > The default kernel configuration has a lot of features disabled that
> > many users need.
>
> I see.
>
> > > I also had a look at this configuration item
> > > in Ubuntu – it's set to CONFIG_HOTPLUG_PCI_ACPI=m.
> >
> > Have you checked whether they add it to /proc/modules?
>
> You mean if they add it to /etc/modules, right?

Hah, yes!

> I don't know, I haven't checked this yet.
>
> But this is maybe interesting: When I compile my kernel with
> CONFIG_HOTPLUG_PCI_ACPI=m and put acpiphp in /etc/modules I can't
> reproduce this bug even though I'm sure the module is loaded:

Yes, that is interesting - please include this information in the
upstream bug report.

Ben.

> [ 19.299137] acpiphp: ACPI Hot Plug PCI Controller Driver version: 0.5
> [ 19.299341] acpiphp: Slot [1] registered
>
> $ lsmod | grep acpiphp
> acpiphp 22567 0
>
>
> Stefan.

--
Ben Hutchings
One of the nice things about standards is that there are so many of them.

Bug#665493: linux-image-3.2.0-2-amd64: kernel fails to boot completely (timeout on modprobe -b pci:xxxx)
 

Here's the upstrem report:
https://bugzilla.kernel.org/show_bug.cgi?id=46481

Since I filed the bug against the ACPI PCI Hotplug driver I didn't want
to decide if we (me and the original reporter Jeroen Nijhof) are
affected by the same bug – all I know is that the build-in driver causes
similar symptoms. So I wrote the bug report for my machine now.

Cheers,
Stefan.