2.6.39.2: usb kernel panic/oops: schedule+ehci_irq+0x266/0x830

2.6.39.2: usb kernel panic/oops: schedule+ehci_irq+0x266/0x830

[Justin Piszcz]

Hi,

I plan to build a new machine in the future (one with console access) but 
in the meantime, after my PSU replacement, the host is still having 
issues, I've turned off bluetooth wireless (built into the DP55KG) and 
will see if it panics again, link the latest panic/oops/etc:

http://home.comcast.net/~jpiszcz/20110705/IMG_2875.JPG

schedule+0x266/0x830
common_interrupt+0xe/0x13
ehci_irq+0x20f/0x230

Is this a valid kernel bug?

I have another box (diff MOBO and all running 2.6.39.x no issues) but it 
does not have built-in bluetooth, I am also running a USB (non-winmodem) 
on this host as well, can anyone comment on this?

  16:      70587          0          0          0          0          0 
11675          0   IO-APIC-fasteoi   ehci_hcd:usb1

Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 001 Device 002: ID 8087:0020 Intel Corp. Integrated Rate Matching Hub
Bus 002 Device 002: ID 8087:0020 Intel Corp. Integrated Rate Matching Hub
Bus 001 Device 003: ID 2001:f103 D-Link Corp. DUB-H7 7-port USB 2.0 hub
Bus 001 Device 004: ID 0baf:0303 U.S. Robotics
Bus 001 Device 005: ID 0764:0501 Cyber Power System, Inc. CP1500 AVR UPS
Bus 001 Device 006: ID 413c:1002 Dell Computer Corp. Keyboard Hub
Bus 001 Device 007: ID 067b:2303 Prolific Technology, Inc. PL2303 Serial Port
Bus 001 Device 008: ID 413c:2002 Dell Computer Corp. SK-8125 Keyboard
Bus 001 Device 009: ID 0461:4d15 Primax Electronics, Ltd Dell Optical Mouse

I also use munin to poll the UPS for various information, could this be a 
bug in the USB driver for this chipset/motherboard?

Can polling various USB devices, running a hylafax server, etc, cause a 
bug to pop up more easily in the USB subsystem vs. just using a keyboard 
for example, I assume so?

Comments?

Thanks.

Justin.

Re: 2.6.39.2: usb kernel panic/oops: schedule+ehci_irq+0x266/0x830

[Justin Piszcz]

On Tue, 5 Jul 2011, Justin Piszcz wrote:

> Hi,
>
> I plan to build a new machine in the future (one with console access) but in 
> the meantime, after my PSU replacement, the host is still having issues, I've 
> turned off bluetooth wireless (built into the DP55KG) and will see if it 
> panics again, link the latest panic/oops/etc:
>
> http://home.comcast.net/~jpiszcz/20110705/IMG_2875.JPG
>
> schedule+0x266/0x830
> common_interrupt+0xe/0x13
> ehci_irq+0x20f/0x230
>
> Is this a valid kernel bug?
>
> I have another box (diff MOBO and all running 2.6.39.x no issues) but it does 
> not have built-in bluetooth, I am also running a USB (non-winmodem) on this 
> host as well, can anyone comment on this?
>
> 16:      70587          0          0          0          0          0 11675 
> 0   IO-APIC-fasteoi   ehci_hcd:usb1
>
> Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
> Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
> Bus 001 Device 002: ID 8087:0020 Intel Corp. Integrated Rate Matching Hub
> Bus 002 Device 002: ID 8087:0020 Intel Corp. Integrated Rate Matching Hub
> Bus 001 Device 003: ID 2001:f103 D-Link Corp. DUB-H7 7-port USB 2.0 hub
> Bus 001 Device 004: ID 0baf:0303 U.S. Robotics
> Bus 001 Device 005: ID 0764:0501 Cyber Power System, Inc. CP1500 AVR UPS
> Bus 001 Device 006: ID 413c:1002 Dell Computer Corp. Keyboard Hub
> Bus 001 Device 007: ID 067b:2303 Prolific Technology, Inc. PL2303 Serial Port
> Bus 001 Device 008: ID 413c:2002 Dell Computer Corp. SK-8125 Keyboard
> Bus 001 Device 009: ID 0461:4d15 Primax Electronics, Ltd Dell Optical Mouse
>
> I also use munin to poll the UPS for various information, could this be a bug 
> in the USB driver for this chipset/motherboard?
>
> Can polling various USB devices, running a hylafax server, etc, cause a bug 
> to pop up more easily in the USB subsystem vs. just using a keyboard for 
> example, I assume so?
>
> Comments?
>
> Thanks.
>
> Justin.
>

Any thoughts on this error?

I've run 3 memtest passes and put the box under severe CPU / disk / I/O 
load etc and there are no issues, but it appears when USB devices get 
used, the kernel panics.  Next test will probably be to remove all USB 
devices and just use PS/2 and see if the problem goes away but I was 
curious if anyone could comment on the that dump:
http://home.comcast.net/~jpiszcz/20110705/IMG_2875.JPG

As to what might be causing this otherwise? Thanks!

Memtest86 v4.20(latest): 3 passes, no problems:
http://home.comcast.net/~jpiszcz/20110706/IMG_2879.JPG

Justin.

Re: 2.6.39.2: usb kernel panic/oops: schedule+ehci_irq+0x266/0x830

[Justin Piszcz]

On Wed, 6 Jul 2011, Justin Piszcz wrote:

>
>
> On Tue, 5 Jul 2011, Justin Piszcz wrote:
>
>> Hi,
>> 
>> I plan to build a new machine in the future (one with console access) but 
>> in the meantime, after my PSU replacement, the host is still having issues, 
>> I've turned off bluetooth wireless (built into the DP55KG) and will see if 
>> it panics again, link the latest panic/oops/etc:
>>

When it recurs, I will disable the 'threadirq' boot option with 2.6.39, as
that is when all this started happening and report back if it happens without
this option enabled.

Is there a reason why threadirq would cause weird issues on one HW type
chipset vs. another (both Intel)?

Justin.

Re: 2.6.39.2: usb kernel panic/oops: schedule+ehci_irq+0x266/0x830

[Alan Stern]

On Tue, 5 Jul 2011, Justin Piszcz wrote:

> Hi,
> 
> I plan to build a new machine in the future (one with console access) but 
> in the meantime, after my PSU replacement, the host is still having 
> issues, I've turned off bluetooth wireless (built into the DP55KG) and 
> will see if it panics again, link the latest panic/oops/etc:
> 
> http://home.comcast.net/~jpiszcz/20110705/IMG_2875.JPG
> 
> schedule+0x266/0x830
> common_interrupt+0xe/0x13
> ehci_irq+0x20f/0x230

You need to enable CONFIG_FRAME_POINTER; without that it is difficult 
to interpret the stack dump.  Also you should collect more than a 
single screen's worth of information, say by using a serial console or 
netconsole.

> Is this a valid kernel bug?

What's the difference between a "valid" kernel bug and an "invalid" 
kernel bug?  :-)

Yes, clearly it is a bug in the kernel.

> I have another box (diff MOBO and all running 2.6.39.x no issues) but it 
> does not have built-in bluetooth, I am also running a USB (non-winmodem) 
> on this host as well, can anyone comment on this?
> 
>   16:      70587          0          0          0          0          0 
> 11675          0   IO-APIC-fasteoi   ehci_hcd:usb1
> 
> Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
> Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
> Bus 001 Device 002: ID 8087:0020 Intel Corp. Integrated Rate Matching Hub
> Bus 002 Device 002: ID 8087:0020 Intel Corp. Integrated Rate Matching Hub
> Bus 001 Device 003: ID 2001:f103 D-Link Corp. DUB-H7 7-port USB 2.0 hub
> Bus 001 Device 004: ID 0baf:0303 U.S. Robotics
> Bus 001 Device 005: ID 0764:0501 Cyber Power System, Inc. CP1500 AVR UPS
> Bus 001 Device 006: ID 413c:1002 Dell Computer Corp. Keyboard Hub
> Bus 001 Device 007: ID 067b:2303 Prolific Technology, Inc. PL2303 Serial Port
> Bus 001 Device 008: ID 413c:2002 Dell Computer Corp. SK-8125 Keyboard
> Bus 001 Device 009: ID 0461:4d15 Primax Electronics, Ltd Dell Optical Mouse
> 
> I also use munin to poll the UPS for various information, could this be a 
> bug in the USB driver for this chipset/motherboard?
> 
> Can polling various USB devices, running a hylafax server, etc, cause a 
> bug to pop up more easily in the USB subsystem vs. just using a keyboard 
> for example, I assume so?

This could any of a number of things.  More information is needed.

If the problem is easily reproducible, you can use git bisect to pin 
down the commit responsible for exposing the bug.

Alan Stern

Re: 2.6.39.2: usb kernel panic/oops: schedule+ehci_irq+0x266/0x830

[Alan Stern]

On Wed, 6 Jul 2011, Justin Piszcz wrote:

> On Wed, 6 Jul 2011, Justin Piszcz wrote:
> 
> >
> >
> > On Tue, 5 Jul 2011, Justin Piszcz wrote:
> >
> >> Hi,
> >> 
> >> I plan to build a new machine in the future (one with console access) but 
> >> in the meantime, after my PSU replacement, the host is still having issues, 
> >> I've turned off bluetooth wireless (built into the DP55KG) and will see if 
> >> it panics again, link the latest panic/oops/etc:
> >>
> 
> When it recurs, I will disable the 'threadirq' boot option with 2.6.39, as
> that is when all this started happening and report back if it happens without
> this option enabled.
> 
> Is there a reason why threadirq would cause weird issues on one HW type
> chipset vs. another (both Intel)?

ehci-hcd doesn't use threaded interrupts, so in theory that option 
shouldn't make any difference.  Of course, theory isn't always right.

Alan Stern

Re: 2.6.39.2: usb kernel panic/oops: schedule+ehci_irq+0x266/0x830

[Justin Piszcz]

On Wed, 6 Jul 2011, Alan Stern wrote:

> On Tue, 5 Jul 2011, Justin Piszcz wrote:
>
> You need to enable CONFIG_FRAME_POINTER; without that it is difficult
> to interpret the stack dump.  Also you should collect more than a
> single screen's worth of information, say by using a serial console or
> netconsole.
Done, I'll update when I have a new crash, thank you for the suggestion!

>> Is this a valid kernel bug?
>
> What's the difference between a "valid" kernel bug and an "invalid"
> kernel bug?  :-)
I was wondering if it could be HW related, even though I cannot seem to find
any HW fault.  Example, I attached 30+ USB devices as a test, R/W on all of
the flash and hard drives, no problems, this seems to occur spuriously when
little or no activity is occurring, hopefully w/FRAME POINTERS we will
get something useful next time.

>
> Yes, clearly it is a bug in the kernel.
>
>
> This could any of a number of things.  More information is needed.
>
> If the problem is easily reproducible, you can use git bisect to pin
> down the commit responsible for exposing the bug.
I wish it was, I tried above (30+ USB devices) no crash, its more or less
random, will wait for next crash and upadte then.

Thanks for the response.

Justin.

Re: 2.6.39.2: usb kernel panic/oops: schedule+ehci_irq+0x266/0x830

[Justin Piszcz]

On Wed, 6 Jul 2011, Alan Stern wrote:

> On Wed, 6 Jul 2011, Justin Piszcz wrote:
>
>> On Wed, 6 Jul 2011, Justin Piszcz wrote:
>>
>>>
>>>
>>> On Tue, 5 Jul 2011, Justin Piszcz wrote:
>>>
>>>> Hi,
>>>>
>>>> I plan to build a new machine in the future (one with console access) but
>>>> in the meantime, after my PSU replacement, the host is still having issues,
>>>> I've turned off bluetooth wireless (built into the DP55KG) and will see if
>>>> it panics again, link the latest panic/oops/etc:
>>>>
>>
>> When it recurs, I will disable the 'threadirq' boot option with 2.6.39, as
>> that is when all this started happening and report back if it happens without
>> this option enabled.
>>
>> Is there a reason why threadirq would cause weird issues on one HW type
>> chipset vs. another (both Intel)?
>
> ehci-hcd doesn't use threaded interrupts, so in theory that option
> shouldn't make any difference.  Of course, theory isn't always right.
>
> Alan Stern
>

Hi,

In that case I will leave it on w/ FRAME pointers enabled and we'll see 
what happens.

Justin.

Re: 2.6.39.2: usb kernel panic/oops: schedule+ehci_irq+0x266/0x830

[Alan Stern]

On Wed, 6 Jul 2011, Justin Piszcz wrote:

> > ehci-hcd doesn't use threaded interrupts, so in theory that option
> > shouldn't make any difference.  Of course, theory isn't always right.
> >
> > Alan Stern
> >
> 
> Hi,
> 
> In that case I will leave it on w/ FRAME pointers enabled and we'll see 
> what happens.

Right.  It's always possible that the threaded IRQ code causes a bug
somewhere else in the kernel, and it just happens to show up
sporadically in ehci-hcd.  Having more information will help narrow 
this down.

Alan Stern