Re: Endless loop with "detected XactErr"

From: Alan Stern
Date: Fri Apr 08 2011 - 11:10:41 EST

On Fri, 8 Apr 2011, Zdenek Kabelac wrote:

> >> Just with changing last number - I've got only one such line on serial console.
> >> By looking into  "drivers/usb/host/ehci-q.c"  - there seems to be
> >> endless goto loop.
> >
> > It isn't endless.  You didn't notice the test against QH_XACTERR_MAX.
> >
> Maybe I've not been patient enough - but I'd been waiting for quite some time
> and the console was just scrolling this line on display - so I need to
> turn-off the machine.
> (After like a minute of this looping)

This is because the loop ends and then starts again (you can tell
because the retry counter goes back to 1).

> > This particular error message indicates a hardware problem in the USB
> > signals.  A bad contact, a bad cable, a device failure -- something
> > like that.
> Well it's been connected to Lenovo docking station - so unsure what
> bad cable you mean here.
> (and AFAICT it seems to work quite well all the time).

It doesn't have to be a bad cable. It could be any sort of hardware
failure. Perhaps the docking or undocking operation itself messes
something up.

> >> Here is the trace from last resume:
> >
> >> [  473.873802] ACPI: \_SB_.GDCK - undocking
> >> [  473.897678] hub 2-0:1.0: state 7 ports 4 chg 0000 evt 0010
> >> [  473.904809] ehci_hcd 0000:00:1a.7: detected XactErr len 0/4 retry 1
> >
> > And presumably additional lines containing similar messages.
> >
> > There's no real reason for them ever to stop, since they are only
> > debugging messages.  If you turn off CONFIG_USB_DEBUG you'll never see
> > tham.
> You mean it's normal I get machine stuck in endless loop when I turn
> on debugging ?

Are you really sure the machine is stuck? If you disable
CONFIG_USB_DEBUG, does it hang or can you still get useful work done?

> I though debug is there to debug the problem - not to add a new one ??

It doesn't add any problems, since you can always turn the debugging
off. And the messages do aid in debugging -- if they weren't present,
you would not have been aware of the problem.

> As this happened during suspend - I had no other option than to simply
> turn off the machine.

It doesn't appear that way in the log you posted. The resume completed
at timestamp 467.8, and the debugging messages didn't start until 6
seconds later.

You might want to track down the original source of the
underlying error. Since the machine was docked the entire time, the
line saying:

[ ï473.873802] ACPI: \_SB_.GDCK - undocking

definitely looks suspicious. I bet if you can fix that then the USB
problems will go away.

Alan Stern

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at