USB IRQ routing problems on Via Apollo Pro 133A

USB IRQ routing problems on Via Apollo Pro 133A

[Andy Isaacson]

[Moving from private mail to linux-kernel]

On Thu, Sep 06, 2001 at 08:25:36PM +0200, thunder7@xs4all.nl wrote:
> You have missed something all right. In the later kernels, the output of
> 'dmesg' mentions:
> 
> PCI: Probing PCI hardware
[snip]
> PCI: Enabling Via external APIC routing
> PCI: Via IRQ fixup for 00:07.2, from 9 to 3
> PCI: Via IRQ fixup for 00:07.3, from 9 to 3
> 
> Those last 2 lines are the ones that are new (since 2.4.5 or so, I guess
> - it was back in June, anyway)

Hrm.  Depending on what I enable in the BIOS, I do get similar lines,
but even then my USB controller doesn't work.  On the current boot for
example, I have

PCI: Enabling Via external APIC routing
PCI: Via IRQ fixup for 00:07.2, from 9 to 3

and the usb-uhci driver shows up in /proc/interrupts:

           CPU0       CPU1
  0:     384195     341825    IO-APIC-edge  timer
  1:       4166       4032    IO-APIC-edge  keyboard
  2:          0          0          XT-PIC  cascade
  9:          0          0          XT-PIC  acpi
 12:      39181      39875    IO-APIC-edge  PS/2 Mouse
 14:       5362       5509    IO-APIC-edge  ide0
 15:          0          0    IO-APIC-edge  ide1
 17:      13441      13344   IO-APIC-level  eth0
 18:        284        304   IO-APIC-level  es1371
 19:          0          0   IO-APIC-level  usb-uhci
NMI:          0          0
LOC:     725968     725967
ERR:          0
MIS:          0

But there are still no interrupts getting from the USB controller to
the CPU.  I get the standard

hub.c: port 1 connection change
hub.c: port 1, portstatus 300, change 3, 1.5 Mb/s
hub.c: port 2 connection change
hub.c: port 2, portstatus 101, change 3, 12 Mb/s
usb-uhci.c: v1.251:USB Universal Host Controller Interface driver
hub.c: port 2, portstatus 103, change 0, 12 Mb/s
hub.c: USB new device connect on bus1/2, assigned device number 2
usb_control/bulk_msg: timeout
usb.c: USB device not accepting new address=2 (error=-110)

and /proc/interrupts continues to show 0 interrupts on line 19.

If I switch back to MPS1.1, the USB controller moves to IRQ 9 (shared
with the acpi controller) but the behavior is the same.  If I toggle
"PnP OS" in the BIOS, I see no relevant differences.  I've tried turning
off and on just about everything of relevance; no change.

Booting a non-APIC kernel makes it work, of course.

The system is a Tyan Tiger 133A, Via Apollo Pro 133A chipset, SMP,
currently running 2.4.9.  Complete dmesg, lspci -vvvvxxxx, and
/proc/interrupts are at
http://web.hexapodia.org/~adi/straum/usb/

Thanks for any help.  Feel free to CC me on replies; I am on the list,
but procmail is a harsh mistress.

(Related question:  Why the '& 0xf' in the calculation of newirq in
quirk_via_irqpic?)

-andy

Re: USB IRQ routing problems on Via Apollo Pro 133A

[Greg KH]

On Fri, Sep 07, 2001 at 03:41:29PM -0500, Andy Isaacson wrote:
> 
> Booting a non-APIC kernel makes it work, of course.
> 
> The system is a Tyan Tiger 133A, Via Apollo Pro 133A chipset, SMP,
> currently running 2.4.9.  Complete dmesg, lspci -vvvvxxxx, and
> /proc/interrupts are at
> http://web.hexapodia.org/~adi/straum/usb/

That's the only solution to enable the on board USB controller for this
motherboard, sorry.  If you can't live with noapic mode, spend $20 for
a PCI USB controller.

thanks,

greg k-h

Re: USB IRQ routing problems on Via Apollo Pro 133A

[Andy Isaacson]

On Fri, Sep 07, 2001 at 01:57:03PM -0700, Greg KH wrote:
> On Fri, Sep 07, 2001 at 03:41:29PM -0500, Andy Isaacson wrote:
> > Booting a non-APIC kernel makes it work, of course.
> > 
> > The system is a Tyan Tiger 133A, Via Apollo Pro 133A chipset, SMP,
> > currently running 2.4.9.  Complete dmesg, lspci -vvvvxxxx, and
> > /proc/interrupts are at
> > http://web.hexapodia.org/~adi/straum/usb/
> 
> That's the only solution to enable the on board USB controller for this
> motherboard, sorry.  If you can't live with noapic mode, spend $20 for
> a PCI USB controller.

Are you claiming that the USB controller IRQ line isn't routed to the
APIC?  If so, I'm curious as to any evidence you can provide to that
effect.

I'd appreciate a pointer to any discussion or whatnot.  URL?  Or even a
suggestion for where to search?  I think I've looked pretty
thoroughly...

But looking at noapic mode a bit more closely, it appears likely that I
can survive with its limitations, so thanks for the tip!

-andy

Re: USB IRQ routing problems on Via Apollo Pro 133A

[Greg KH]

On Fri, Sep 07, 2001 at 04:44:56PM -0500, Andy Isaacson wrote:
> 
> Are you claiming that the USB controller IRQ line isn't routed to the
> APIC?  If so, I'm curious as to any evidence you can provide to that
> effect.

Yes.  The only evidence is emperical (Randy Dunlap and I sat down with
the motherboard one night and tried loads of different combinations and
couldn't get any of them to work.  He knows the APIC code much better
than I and determined that the IRQ line isn't even routed to the
controller at all.)

Then there's the Win2000 errata that claimed that it would not work in
SMP mode with this controller at all.  I do think that VIA eventually
came out with a patch for Win2000 that did something like "noapic" on
Linux does.

Other than that, sorry, no "official" documentation at all.  There was a
lot of discussion on the linux-usb mailing lists for a while about this
problem a number of months ago if you want to go dig them up.

thanks,

greg k-h