Bug 7568 - logitech cordless desktop receiver causes laptop to hang during boot
logitech cordless desktop receiver causes laptop to hang during boot
Status: REJECTED INSUFFICIENT_DATA
Product: Drivers
Classification: Unclassified
Component: USB
i386 Linux
: P2 blocking
Assigned To: Greg Kroah-Hartman
:
Depends on:
Blocks: USB
  Show dependency treegraph
 
Reported: 2006-11-22 12:32 UTC by Hans Dembinski
Modified: 2008-09-24 10:11 UTC (History)
6 users (show)

See Also:
Kernel Version: 2.6.18
Tree: Mainline
Regression: ---


Attachments

Description Hans Dembinski 2006-11-22 12:32:48 UTC
Distribution:
Kubuntu Dapper Drake

Hardware Environment:
Gericom 14" Laptop, 1.5 GHz Pentium M, USB 2.0
Logitech Cordless Desktop (mouse and keyboard, one usb receiver with two 
connectors)

Problem Description:
The kernel hangs on boot if the mentioned receiver is connected to the laptop.  
The bootup stops right after grub loaded the kernel, even before the 
bootsplash in Kubuntu is activated. The text cursor is still blinking, 
however.

The same receiver can be used normally and without any problems, if it is 
detached on boot, and reattached later after the login.

Steps to reproduce:
plug Logitech Cordless Desktop Receiver in any USB port and reboot -> hang

Comments:
Kubuntu loads 3 usb drivers simulataneous as modules: uhci_hcd, ohci_hcd and 
ehci_hcd. In my setup, ehci_hcd is the driver, which has the actual control of 
the devices later.

In the ubuntu forums, some people were mentioning problems with ehci_hcd. As a 
work around, is there a way to turn disable this driver on bootup?

Best regards!
Comment 1 Andrew Morton 2006-11-22 12:51:00 UTC
2.6.15 is very old.  Can you test 2.6.18 or 2.6.19-rc6?
Comment 2 Hans Dembinski 2006-12-04 14:53:40 UTC
Hi,

I reproduced the same behaviour on a 2.6.18.3 kernel.
The problems seems actually related to the ACPI system. I can say that, 
because I disabled the quiet boot and the bootsplash and could see, that the 
kernel stops right at the first steps, when the ACPI <-> BIOS communication is 
done - sorry, if I am telling complete bogus, that's the way I would describe 
it from my limited knowledge.

I will write the messages down and post them later.
Comment 3 Hans Dembinski 2006-12-07 13:49:26 UTC
hi,

the last two lines before the kernel stops, are:
[17179574.188000] ACPI: Looking for DSDT ... not found
[17179574.196000] ACPI: setting ELCR to 0200 (from 0e28)

and then a seemingly endlessly blinking cursor (at least several minutes).

By the way: what do the numbers in the brackets mean?

Best regards...
Comment 4 Hans Dembinski 2006-12-16 10:59:54 UTC
Anything new on this field? Can I supply you with more information?
Comment 5 Jiri Kosina 2007-02-21 12:48:22 UTC
Does your BIOS have "USB Legacy" setting? If so, does it make any difference
when you turn it off?

Also, does booting with acpi=off work?
Comment 6 Hans Dembinski 2007-02-21 15:38:02 UTC
Hi,

I checked: my bios does not have a Legacy USB setting.

Adding "acpi=off" to the kernel parameters works, it does not hang, but I get 
an error from the kernel, it is complaining about expected faulty behavior and 
the recommendation to also set "pnpbios=off". I did not strictly followed that, 
but tried this: just turning pngbios=off, but not acpi=off. That setting again 
reproduced the bug.

So it's really acpi-related.

If I can supply you with more information, please drop me a line.

Best regards!
Comment 7 Natalie Protasevich 2007-06-13 18:50:10 UTC
To proceed further, could you please
  try the latest 2.6.22-rc4
  collect the boot trace with and without the receiver, preferably enable debug for ACPI and USB in the kernel and with "debug" boot parameter.
Thanks,
--Natalie
Comment 8 Andrew Morton 2007-07-25 17:21:51 UTC
Len, do you agree that this is acpi-related?

If so, could you please assign it appropriately?

Thanks.
Comment 9 Hans Dembinski 2007-07-26 03:49:06 UTC
Dear developers,

I will try 2.6.22-rc4 collect the required data as soon as possible!

Maybe, if you know, can you point out, how I can save the kernel output? At 
this early stage, there is not even the harddrive mounted. Otherwise I will 
have to write it down... and then I will only select a part, from which I 
think that it is relevant to this matter.

Best regards,
Hans

Comment 10 Hans Dembinski 2007-07-29 08:18:23 UTC
Dear Kernel Developers,

I installed kernel 2.6.22.1 and made more tests. I NOW found out, that I was wrong before, the hang ONLY happens, when the following situation is at hand:

* Cordless Desktop is plugged into a nameless passive USB hub
* USB hub is connected to laptop

Before, I claimed, that Cordless Desktop itself is the reason for the hang, but in fact, it is the combination of that device behind an USB hub! When the desktop is attached to the laptop directly, it works fine.

The last messages, I now get before the hang are the following:

[27.760822] ACPI: PCI Root Bridge [PCI0] (0000:00)
[27.761325] PCI quirk: region 0400-047f claimed by ICH4 ACPI/GPIO/TCO
[27.761391] PCI quirk: region 0500-053f claimed by ICH4 GPIO
[27.761804] PCI: Transparent Bridge - 0000:00:1e.0
[27.761912] PCI: Bus #02 (-#05) is hidden behind transparent bridge #01 (-#01) (try 'pci=assign-busses')
[27.762002] Please report the result to linux-kernel to fix this permanently
[27.762715] [ACPI Debug] String: [0x0D] "BAT_STA Start"
[27.763553] [ACPI Debug] String: [0x0D] "BAT_STA Start"
[27.764384] [ACPI Debug] String: [0x0D] "BAT_STA Start"
[27.765212] [ACPI Debug] String: [0x0D] "BAT_STA Start"
[27.766047] [ACPI Debug] String: [0x0D] "BAT_STA Start"
[27.766879] [ACPI Debug] String: [0x0D] "BAT_STA Start"
[27.767742] [ACPI Debug] String: [0x0D] "BAT_STA Start"

I hope this helps. Please don't blame me for rare small typos, because I had to write this down to paper and type it in again.

Best regards,
Hans
Comment 11 Natalie Protasevich 2008-04-05 20:38:45 UTC
Hans, do you still see this problem with recent kernel? I am copying to Bjorn to please have a look. This is pnp and acpi again...
Comment 12 Bjorn Helgaas 2008-04-08 15:38:44 UTC
Hans, can you edit your grub menu.lst file to add "debug initcall_debug" kernel parameters, boot the kernel without the Cordless Desktop receiver connected, and attach to this bug report:
  - output of "lspci -vv"
  - output of "dmesg -s64000"
  - contents of /proc/acpi/dsdt

It looks like the hang happens about the time we read the ACPI interrupt routing information, so please try booting with "acpi=noirq" and see whether that fixes the problem.  If it does, also attach the "dmesg -s64000" output for this boot.
Comment 13 Alan 2008-09-24 10:11:31 UTC
Closing as no response

Note You need to log in before you can comment on or make changes to this bug.