usb.c: USB device not accepting new address=4 (error=-110)

usb.c: USB device not accepting new address=4 (error=-110)

[Ingo Oeser]

[-- Attachment #1: Type: text/plain, Size: 905 bytes --]

Hi there,

please CC me to replies on linux-usb-devel, since I'm not
subscribed there.

I try to use my Handspring Visor USB craddle with Linux. I've
seen, that this will work, but all the other people use a UHCI
driver for this, so this could be as well an OHCI issue.

loaded modules:
visor                  10256   0  (unused)
usbserial              18128   0  [visor]
usb-ohci               22352   0  (unused)
usbcore                58256   1  [visor usbserial usb-ohci]

Pressing the hotsync button, I get the message from the subject
line (with variing values for "address"). Even plugging into
another USB port doesn't help. Only the usb port adress changes.

lspci -vvvv (attached)
.config (attached)
kernel messages since boot (attached)

Would be nice, if we can solve/debug this ;-)

Regards

Ingo Oeser
-- 
To the systems programmer,
users and applications serve only to provide a test load.

[-- Attachment #2: pci.bz2 --]
[-- Type: application/octet-stream, Size: 1258 bytes --]

[-- Attachment #3: .config.bz2 --]
[-- Type: application/octet-stream, Size: 5091 bytes --]

[-- Attachment #4: debug.bz2 --]
[-- Type: application/octet-stream, Size: 5013 bytes --]

Re: [linux-usb-devel] usb.c: USB device not accepting new address=4 (error=-110)

[David Brownell]

Then whatever sets up your ServerWorks ServerSet III LE chipset
needs its PCI IRQ setup fixed ... I'm not sure how to do this.

Perhaps someone who's familiar with arch/i386/kernel/pci-*.c
irq setup can suggest the right patch for this problem.  I think
the "dmesg" output in your original post probably had the info
needed to figure that out.

- Dave


> From: "Ingo Oeser" <ingo.oeser@informatik.tu-chemnitz.de>
> Sent: Saturday, June 09, 2001 12:19 AM
>
> > David Brownell wrote:
> > Can you verify, using /proc/interrupts, that you're actually
> > getting interrupts on irq #30 when these timeouts happen?
>  
> I get none:
>  30:          0          0   IO-APIC-level  usb-ohci
>  
> > One possibility:  the timeout happens because the HCD
> > is not getting the interrupts it expects.  That would imply
> > that the PCI IRQ setup for this device isn't quite right.
> > Such problems have been seen before.
> 
> This seems to be my problem. How can I solve this?
> 
> My BIOS cannot set a specific IRQ for USB like other BIOSes do.
> 
> And now that you say^W it, I remember sth. like this on
> linux-kernel... I just didn't know the messages...
> 
> Thanks and Regards
> 
> Ingo Oeser

Re: [linux-usb-devel] usb.c: USB device not accepting new address=4 (error=-110)

[thunder7]

On Sat, Jun 09, 2001 at 06:14:24AM -0700, David Brownell wrote:
> Then whatever sets up your ServerWorks ServerSet III LE chipset
> needs its PCI IRQ setup fixed ... I'm not sure how to do this.
> 
> Perhaps someone who's familiar with arch/i386/kernel/pci-*.c
> irq setup can suggest the right patch for this problem.  I think
> the "dmesg" output in your original post probably had the info
> needed to figure that out.
> 
> - Dave
> 
> 
> > From: "Ingo Oeser" <ingo.oeser@informatik.tu-chemnitz.de>
> > Sent: Saturday, June 09, 2001 12:19 AM
> >
> > > David Brownell wrote:
> > > Can you verify, using /proc/interrupts, that you're actually
> > > getting interrupts on irq #30 when these timeouts happen?
> >  
> > I get none:
> >  30:          0          0   IO-APIC-level  usb-ohci
> >  
> > > One possibility:  the timeout happens because the HCD
> > > is not getting the interrupts it expects.  That would imply
> > > that the PCI IRQ setup for this device isn't quite right.
> > > Such problems have been seen before.
> > 
> > This seems to be my problem. How can I solve this?
> > 
> > My BIOS cannot set a specific IRQ for USB like other BIOSes do.
> > 
> > And now that you say^W it, I remember sth. like this on
> > linux-kernel... I just didn't know the messages...
> > 
A similar problem with USB on a VIA smp board was solved by Jeff Garzik,
so you might want to send him some info, like

dmesg
lspci -vvvxxx

Good luck,
Jurriaan
-- 
You have nothing to be afraid of. Well, that might not be true in the
larger scheme of things, but this ordeal is over, so off you go. Toodle-oo.
	Fraser, Due South
GNU/Linux 2.4.5-ac12 SMP/ReiserFS 2x1402 bogomips load av: 0.49 0.12 0.04

Re: [linux-usb-devel] usb.c: USB device not accepting new address=4 (error=-110)

[Ingo Oeser]

On Sat, Jun 09, 2001 at 06:14:24AM -0700, David Brownell wrote:
> Then whatever sets up your ServerWorks ServerSet III LE chipset
> needs its PCI IRQ setup fixed ...

Right. "noapic" helps, which is a sure sign of such bugs.

> I'm not sure how to do this.

Neither me. I boot with "noapic" now, but this sucks on SMP, as
we all know :-(

> Perhaps someone who's familiar with arch/i386/kernel/pci-*.c
> irq setup can suggest the right patch for this problem.  I think
> the "dmesg" output in your original post probably had the info
> needed to figure that out.

Maybe some of the people with good contacts to ServerWorks Inc.
can help me (and all the ServerWorks customers) out...


Thanks and Regards

Ingo oeser
-- 
To the systems programmer,
users and applications serve only to provide a test load.

Re: [linux-usb-devel] usb.c: USB device not accepting new address=4 (error=-110)

[Pete Zaitcev]

I had difficulties making it all work on ServerWorks III LE
based machines, most notably 2450 & 6450.

Dell fixed 2450 eventually, BIOS needs upgrade to level A11.

> >  30:          0          0   IO-APIC-level  usb-ohci

The line above says that MP table contained the device.
This is the situarion with 6450 - they have the MP entry,
but there is no way to make it work. I am working with their
BIOS and hw engineers on a fix.

For the moment the workaround is to run the whole thing
with "noapic". Try it, perhaps it helps the machine in question.

-- Pete

> From: David Brownell <david-b@pacbell.net>
> To: Ingo Oeser <ingo.oeser@informatik.tu-chemnitz.de>
> Cc: linux-usb-devel@lists.sourceforge.net, lkml <linux-kernel@vger.kernel.org>
> Reply-To: linux-usb-devel@lists.sourceforge.net
> Date: Sat, 09 Jun 2001 06:14:24 -0700

> Then whatever sets up your ServerWorks ServerSet III LE chipset
> needs its PCI IRQ setup fixed ... I'm not sure how to do this.
> 
> Perhaps someone who's familiar with arch/i386/kernel/pci-*.c
> irq setup can suggest the right patch for this problem.  I think
> the "dmesg" output in your original post probably had the info
> needed to figure that out.
> 
> - Dave
> 
> > From: "Ingo Oeser" <ingo.oeser@informatik.tu-chemnitz.de>
> > Sent: Saturday, June 09, 2001 12:19 AM
> >
> > > David Brownell wrote:
> > > Can you verify, using /proc/interrupts, that you're actually
> > > getting interrupts on irq #30 when these timeouts happen?
> >  
> > I get none:
> >  30:          0          0   IO-APIC-level  usb-ohci
> >  
> > > One possibility:  the timeout happens because the HCD
> > > is not getting the interrupts it expects.  That would imply
> > > that the PCI IRQ setup for this device isn't quite right.
> > > Such problems have been seen before.
> > 
> > This seems to be my problem. How can I solve this?
> > 
> > My BIOS cannot set a specific IRQ for USB like other BIOSes do.
> > 
> > And now that you say^W it, I remember sth. like this on
> > linux-kernel... I just didn't know the messages...
> > 
> > Thanks and Regards
> > 
> > Ingo Oeser