2.6.15 EHCI hang on boot

2.6.15 EHCI hang on boot

[Marc Haber]

Hi,

I have rolled out 2.6.15 on a number of test hosts. On one of my
boxes, which is by far the most recent one, has an i865 chipset, hangs
on boot when the EHCI driver is loaded. USB is not compiled as module,
so the system doesn't come up at all:

ACPI: PCI Interrupt 0000:00:1d.7[D] -> GSI 23 (level, low) -> IRQ 18
ehci_hcd 0000:00:1d.7: EHCI Host Controller
ehci_hcd 0000:00:1d.7: debug port 1

These are the last lines of the boot log (which I have completely
captured via serial console and can submit on request).

The EHCI controller's lspci output (obtained with 2.6.14.3):
0000:00:1d.7 USB Controller: Intel Corporation 82801EB/ER (ICH5/ICH5R) USB2 EHCI
 Controller (rev 02) (prog-if 20 [EHCI])
        Subsystem: Micro-Star International Co., Ltd. 865PE Neo2 (MS-6728)
        Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
        Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort - <MAbort- >SERR- <PERR-
        Latency: 0
        Interrupt: pin D routed to IRQ 18
        Region 0: Memory at febffc00 (32-bit, non-prefetchable) [size=1K]
        Capabilities: [50] Power Management version 2
                Flags: PMEClk- DSI- D1- D2- AuxCurrent=375mA PME(D0+,D1-,D2-,D3hot+,D3cold+)
                Status: D0 PME-Enable- DSel=0 DScale=0 PME-
        Capabilities: [58] #0a [20a0]
(complete lspci output available on request)

The 2.6.14.3 kernel which was installed on that box before works fine.
The 2.6.15 configuration is the result of make oldconfig over that
2.6.14.3 kernel, so I suspect that the configurations are sufficiently
similiar, and the same 2.6.15 binary works fine on other systems which
have their EHCI as PCI cards.

I suspect an incompatibility with the i865 chipset. Is there anything
I can do to help debugging?

Greetings
Marc

-- 
-----------------------------------------------------------------------------
Marc Haber         | "I don't trust Computers. They | Mailadresse im Header
Mannheim, Germany  |  lose things."    Winona Ryder | Fon: *49 621 72739834
Nordisch by Nature |  How to make an American Quilt | Fax: *49 621 72739835

Re: 2.6.15 EHCI hang on boot

[Ben Dooks]

On Wed, Jan 04, 2006 at 05:18:44PM +0100, Marc Haber wrote:
> Hi,
> 
> I have rolled out 2.6.15 on a number of test hosts. On one of my
> boxes, which is by far the most recent one, has an i865 chipset, hangs
> on boot when the EHCI driver is loaded. USB is not compiled as module,
> so the system doesn't come up at all:
> 
> ACPI: PCI Interrupt 0000:00:1d.7[D] -> GSI 23 (level, low) -> IRQ 18
> ehci_hcd 0000:00:1d.7: EHCI Host Controller
> ehci_hcd 0000:00:1d.7: debug port 1
> 
> These are the last lines of the boot log (which I have completely
> captured via serial console and can submit on request).
> 
> The EHCI controller's lspci output (obtained with 2.6.14.3):
> 0000:00:1d.7 USB Controller: Intel Corporation 82801EB/ER (ICH5/ICH5R) USB2 EHCI
>  Controller (rev 02) (prog-if 20 [EHCI])
>         Subsystem: Micro-Star International Co., Ltd. 865PE Neo2 (MS-6728)
>         Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
>         Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort - <MAbort- >SERR- <PERR-
>         Latency: 0
>         Interrupt: pin D routed to IRQ 18
>         Region 0: Memory at febffc00 (32-bit, non-prefetchable) [size=1K]
>         Capabilities: [50] Power Management version 2
>                 Flags: PMEClk- DSI- D1- D2- AuxCurrent=375mA PME(D0+,D1-,D2-,D3hot+,D3cold+)
>                 Status: D0 PME-Enable- DSel=0 DScale=0 PME-
>         Capabilities: [58] #0a [20a0]
> (complete lspci output available on request)
> 
> The 2.6.14.3 kernel which was installed on that box before works fine.
> The 2.6.15 configuration is the result of make oldconfig over that
> 2.6.14.3 kernel, so I suspect that the configurations are sufficiently
> similiar, and the same 2.6.15 binary works fine on other systems which
> have their EHCI as PCI cards.
> 
> I suspect an incompatibility with the i865 chipset. Is there anything
> I can do to help debugging?

I have the same (but only intermittently) on an Intel i875 based
board by MSI. In about 25% of the cases it manages to boot past
these lines, and initialise the mouse connected. 

The usb bus has a low-speed Microsoft mouse on it, and a
USB 1.1 Hub, with a card reader and serial dongle connected.

-- 
Ben (ben@fluff.org, http://www.fluff.org/)

  'a smiley only costs 4 bytes'

Re: 2.6.15 EHCI hang on boot

[Marc Haber]

On Wed, Jan 04, 2006 at 04:27:43PM +0000, Ben Dooks wrote:
> I have the same (but only intermittently) on an Intel i875 based
> board by MSI. In about 25% of the cases it manages to boot past
> these lines, and initialise the mouse connected. 

After trying a few more times, I can now confirm that the issue is
indeed intermittent. My mainboard in question is an MSI as well.

> The usb bus has a low-speed Microsoft mouse on it, and a
> USB 1.1 Hub, with a card reader and serial dongle connected.

The only device on my USB is a Logitech Receiver:
Bus 002 Device 002: ID 046d:c504 Logitech, Inc. Cordless Mouse+Keyboard Receiver

Greetings
Marc

-- 
-----------------------------------------------------------------------------
Marc Haber         | "I don't trust Computers. They | Mailadresse im Header
Mannheim, Germany  |  lose things."    Winona Ryder | Fon: *49 621 72739834
Nordisch by Nature |  How to make an American Quilt | Fax: *49 621 72739835

Re: 2.6.15 EHCI hang on boot

[Greg KH]

On Wed, Jan 04, 2006 at 05:18:44PM +0100, Marc Haber wrote:
> Hi,
> 
> I have rolled out 2.6.15 on a number of test hosts. On one of my
> boxes, which is by far the most recent one, has an i865 chipset, hangs
> on boot when the EHCI driver is loaded. USB is not compiled as module,
> so the system doesn't come up at all:
> 
> ACPI: PCI Interrupt 0000:00:1d.7[D] -> GSI 23 (level, low) -> IRQ 18
> ehci_hcd 0000:00:1d.7: EHCI Host Controller
> ehci_hcd 0000:00:1d.7: debug port 1
> 
> These are the last lines of the boot log (which I have completely
> captured via serial console and can submit on request).
> 
> The EHCI controller's lspci output (obtained with 2.6.14.3):
> 0000:00:1d.7 USB Controller: Intel Corporation 82801EB/ER (ICH5/ICH5R) USB2 EHCI
>  Controller (rev 02) (prog-if 20 [EHCI])
>         Subsystem: Micro-Star International Co., Ltd. 865PE Neo2 (MS-6728)
>         Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
>         Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort - <MAbort- >SERR- <PERR-
>         Latency: 0
>         Interrupt: pin D routed to IRQ 18
>         Region 0: Memory at febffc00 (32-bit, non-prefetchable) [size=1K]
>         Capabilities: [50] Power Management version 2
>                 Flags: PMEClk- DSI- D1- D2- AuxCurrent=375mA PME(D0+,D1-,D2-,D3hot+,D3cold+)
>                 Status: D0 PME-Enable- DSel=0 DScale=0 PME-
>         Capabilities: [58] #0a [20a0]
> (complete lspci output available on request)
> 
> The 2.6.14.3 kernel which was installed on that box before works fine.
> The 2.6.15 configuration is the result of make oldconfig over that
> 2.6.14.3 kernel, so I suspect that the configurations are sufficiently
> similiar, and the same 2.6.15 binary works fine on other systems which
> have their EHCI as PCI cards.
> 
> I suspect an incompatibility with the i865 chipset. Is there anything
> I can do to help debugging?

I don't know, David, any ideas?

thanks,

greg k-h

Re: 2.6.15 EHCI hang on boot

[David Brownell]

On Wednesday 04 January 2006 2:04 pm, Greg KH wrote:

> > I suspect an incompatibility with the i865 chipset. Is there anything
> > I can do to help debugging?
> 
> I don't know, David, any ideas?

Such things are more often BIOS than chipset, so check for updates there.
Plus, you commented that the issue only shows up with your newest board,
and most folk don't have such hang-on-boot issues.

There were updates in 2.6.15 to how all the PCI based usb host controller
drivers handle their reset/init logic ... basically "handoff" from BIOS to
the OS kernel (Linux) is always done "early" now.  Previously it wasn't
always done until late (too late for the input subsystem, given usb keyboards
and mice that the BIOS uses to emulate ps/2 ones), and wasn't done very
consistently.  (The "early handoff" code didn't act identically to the
code inside the HCDs that did the bios handoff "later".)

If there's no BIOS update, one experiment would be to #ifdef out the call
to bios_handoff() in drivers/usb/host/ehci-pci.c ... the version of that
code in the .../usb/hosts/pci-quirks.c file _should_ eventually suffice,
though I think there are still differences between that version and the
one (in ehci-pci.c) that's better tested.

- Dave