usb broken in 2.5?

usb broken in 2.5?

[Felix von Leitner]

In 2.5.57 USB does not work.  The mainboard devices are still detected,
and apparently also the USB devices are found, but then the USB code
tries to remap them and fails.  It tries again with 1.5 Mb/s and fails
again.

> usb_control/bulk_msg: timeout
> usb.c: USB device not accepting new address=3 (error=-110)

The same problem also happens with the WOLK kernel patch, which appears
to also include a newer version of the USB code.

This is an excerpt from /proc/pci:

  Bus  0, device  16, function  0:
    USB Controller: VIA Technologies, Inc. USB (rev 128).
      IRQ 21.
      Master Capable.  Latency=32.
      I/O at 0xdc00 [0xdc1f].
  Bus  0, device  16, function  1:
    USB Controller: VIA Technologies, Inc. USB (#2) (rev 128).
      IRQ 21.
      Master Capable.  Latency=32.
      I/O at 0xe000 [0xe01f].
  Bus  0, device  16, function  2:
    USB Controller: VIA Technologies, Inc. USB (#3) (rev 128).
      IRQ 21.
      Master Capable.  Latency=32.
      I/O at 0xe400 [0xe41f].
  Bus  0, device  16, function  3:
    USB Controller: VIA Technologies, Inc. USB 2.0 (rev 130).
      IRQ 19.
      Master Capable.  Latency=32.
      Non-prefetchable 32 bit memory at 0xe6033000 [0xe60330ff].

Any ideas?  On 2.5 and WOLK the kernel also says that no IRQ is know and
that it is using ACPI to route the IRQs.

Felix

"PCI BIOS passed nonexistant PCI bus 0"

[Felix von Leitner]

I also get

  PCI BIOS passed nonexistant PCI bus 1

How can this be?  I configured the kernel with:

  # CONFIG_PCI_GOBIOS is not set
  CONFIG_PCI_GODIRECT=y

What is this option good for if Linux still listens to what the BIOS says?

Felix

Re: "PCI BIOS passed nonexistant PCI bus 0"

[Pete Zaitcev]

>   PCI BIOS passed nonexistant PCI bus 1

> How can this be?  I configured the kernel with:
> 
>   # CONFIG_PCI_GOBIOS is not set
>   CONFIG_PCI_GODIRECT=y
> 
> What is this option good for if Linux still listens to what the BIOS says?

How do you think we should go about IRQ routing
without reading BIOS tables? If you have a viable
solution, I'd be happy to listen to it.

Accessing PCI configuration space is trivial. Only two or
three ways actually exist on x86. The IRQ routing is different.
Every little motherboard vendor routes them differently.

-- Pete

Re: usb broken in 2.5?

[Felix von Leitner]

Thus spake Felix von Leitner (felix-linuxkernel@fefe.de):
> In 2.5.57 USB does not work.

[...]

> Any ideas?  On 2.5 and WOLK the kernel also says that no IRQ is know and
> that it is using ACPI to route the IRQs.

It turned out to be an ACPI issue.

Darn.  Not AGAIN!
ACPI is the biggest pile of crap to poison this earth.
I have yet to see it "just work".

I'll turn it off for good now, sorry for wasting your time with this.

Re: usb broken in 2.5?

[Greg KH]

On Tue, Jan 14, 2003 at 03:44:35AM +0100, Felix von Leitner wrote:
> In 2.5.57 USB does not work.  The mainboard devices are still detected,
> and apparently also the USB devices are found, but then the USB code
> tries to remap them and fails.  It tries again with 1.5 Mb/s and fails
> again.

This sounds like the old iterrupt routing problem that ACPI causes at
times, and not a USB problem (the interrupts aren't getting to the
controller, nothing the USB controller can do about that...)

> > usb_control/bulk_msg: timeout
> > usb.c: USB device not accepting new address=3 (error=-110)
> 
> The same problem also happens with the WOLK kernel patch, which appears
> to also include a newer version of the USB code.

Newer version?  Where would the WOLK people get that from, as the stuff
in the main tree is the newest.

Since you said you are using ACPI, I'd suggest taking this up with the
ACPI developers.

thanks,

greg k-h