2.6.26-rc1 regression since 2.6.25

2.6.26-rc1 regression since 2.6.25

[Lukas Hejtmanek]

Hello,

as of some high versions of 2.6.25-gitXX and 2.6.26-rc1, USB devices pluged
into a dock (minidock for Lenovo ThinkPad T61) are not discoverred until
I reload ehci-hcd module with already plugged device.

Also minor omnipresent bug, most of the time, I got garbage message instead of
the message "machine restart" when restarting. I think, the message is OK if
I boot up and then immediatelly restart. If I restart after some time, only
garbage is displayed (ASCII chars with value above 127 mostly).

-- 
Lukáš Hejtmánek

Re: 2.6.26-rc1 regression since 2.6.25

[Greg KH]

On Mon, May 05, 2008 at 10:02:27AM +0200, Lukas Hejtmanek wrote:
> Hello,
> 
> as of some high versions of 2.6.25-gitXX and 2.6.26-rc1, USB devices pluged
> into a dock (minidock for Lenovo ThinkPad T61) are not discoverred until
> I reload ehci-hcd module with already plugged device.

Can you please post the kernel log file from when this happens?

Added the linux-usb list on the CC: for this as well...

thanks,

greg k-h

Re: 2.6.26-rc1 regression since 2.6.25

[Lukas Hejtmanek]

On Thu, May 08, 2008 at 01:29:21PM -0700, Greg KH wrote:
> On Mon, May 05, 2008 at 10:02:27AM +0200, Lukas Hejtmanek wrote:
> > Hello,
> > 
> > as of some high versions of 2.6.25-gitXX and 2.6.26-rc1, USB devices pluged
> > into a dock (minidock for Lenovo ThinkPad T61) are not discoverred until
> > I reload ehci-hcd module with already plugged device.
> 
> Can you please post the kernel log file from when this happens?

I created an attachment to this bug.
http://bugzilla.kernel.org/show_bug.cgi?id=10630
 
-- 
Lukáš Hejtmánek

Re: 2.6.26-rc1 regression since 2.6.25

[Greg KH]

On Fri, May 09, 2008 at 11:20:06AM +0200, Lukas Hejtmanek wrote:
> On Thu, May 08, 2008 at 01:29:21PM -0700, Greg KH wrote:
> > On Mon, May 05, 2008 at 10:02:27AM +0200, Lukas Hejtmanek wrote:
> > > Hello,
> > > 
> > > as of some high versions of 2.6.25-gitXX and 2.6.26-rc1, USB devices pluged
> > > into a dock (minidock for Lenovo ThinkPad T61) are not discoverred until
> > > I reload ehci-hcd module with already plugged device.
> > 
> > Can you please post the kernel log file from when this happens?
> 
> I created an attachment to this bug.
> http://bugzilla.kernel.org/show_bug.cgi?id=10630

I'd prefer to do this in email, not bugzilla, if you can.

That log file has a lot of suspend/resume cycles (showing some acpi
problems as well...)

Can you just boot the box and then plug it into the dock and show the
log file for that?

thanks,

greg k-h

Re: 2.6.26-rc1 regression since 2.6.25

[Lukas Hejtmanek]

On Fri, May 09, 2008 at 08:19:24AM -0700, Greg KH wrote:
> I'd prefer to do this in email, not bugzilla, if you can.
> 
> That log file has a lot of suspend/resume cycles (showing some acpi
> problems as well...)
> 
> Can you just boot the box and then plug it into the dock and show the
> log file for that?

sure, but not till Monday, I do not have the dock at home.

-- 
Lukáš Hejtmánek

Re: 2.6.26-rc1 regression since 2.6.25

[Alan Stern]

On Fri, 9 May 2008, Greg KH wrote:

> Can you just boot the box and then plug it into the dock and show the
> log file for that?

If you can, use a kernel that has CONFIG_USB_DEBUG turned on.

Alan Stern

Re: 2.6.26-rc1 regression since 2.6.25

[Lukas Hejtmanek]

On Fri, May 09, 2008 at 02:47:45PM -0400, Alan Stern wrote:
> On Fri, 9 May 2008, Greg KH wrote:
> 
> > Can you just boot the box and then plug it into the dock and show the
> > log file for that?
> 
> If you can, use a kernel that has CONFIG_USB_DEBUG turned on.

I did. The dmesg is attached to the bug:
http://bugzilla.kernel.org/show_bug.cgi?id=10630

-- 
Lukáš Hejtmánek

Re: 2.6.26-rc1 regression since 2.6.25

[Alan Stern]

On Mon, 12 May 2008, Lukas Hejtmanek wrote:

> On Fri, May 09, 2008 at 02:47:45PM -0400, Alan Stern wrote:
> > On Fri, 9 May 2008, Greg KH wrote:
> > 
> > > Can you just boot the box and then plug it into the dock and show the
> > > log file for that?
> > 
> > If you can, use a kernel that has CONFIG_USB_DEBUG turned on.
> 
> I did. The dmesg is attached to the bug:
> http://bugzilla.kernel.org/show_bug.cgi?id=10630

It looks as though the dock's USB hub may not be working right.

Just as a test, can you try applying gregkh-04-usb-2.6.26-rc2.patch 
(it applies to 2.6.26-rc2, obviously) from

	http://www.kernel.org/pub/linux/kernel/people/gregkh/gregkh-2.6/

and attaching the dmesg log from that?  It contains extra debugging not 
present in the vanilla kernel -- make sure that you keep 
CONFIG_USB_DEBUG enabled.

Alan Stern

Re: 2.6.26-rc1 regression since 2.6.25

[Lukas Hejtmanek]

On Tue, May 13, 2008 at 04:06:20PM -0400, Alan Stern wrote:
> > I did. The dmesg is attached to the bug:
> > http://bugzilla.kernel.org/show_bug.cgi?id=10630
> 
> It looks as though the dock's USB hub may not be working right.

Maybe, Lenovo HW is a piece of crap. Anyway, it worked up to 2.6.25
(including).

> Just as a test, can you try applying gregkh-04-usb-2.6.26-rc2.patch 
> (it applies to 2.6.26-rc2, obviously) from
> 
> 	http://www.kernel.org/pub/linux/kernel/people/gregkh/gregkh-2.6/
> 
> and attaching the dmesg log from that?  It contains extra debugging not 
> present in the vanilla kernel -- make sure that you keep 
> CONFIG_USB_DEBUG enabled.

I did another attachement of the dmesg to the bugzilla.

-- 
Lukáš Hejtmánek

Re: 2.6.26-rc1 regression since 2.6.25

[Alan Stern]

On Mon, 19 May 2008, Lukas Hejtmanek wrote:

> On Tue, May 13, 2008 at 04:06:20PM -0400, Alan Stern wrote:
> > > I did. The dmesg is attached to the bug:
> > > http://bugzilla.kernel.org/show_bug.cgi?id=10630
> > 
> > It looks as though the dock's USB hub may not be working right.
> 
> Maybe, Lenovo HW is a piece of crap.

That's certainly possible.

>  Anyway, it worked up to 2.6.25
> (including).

Can you attach an equivalent dmesg log for 2.6.25 (with
CONFIG_USB_DEBUG enabled, of course)?

> > Just as a test, can you try applying gregkh-04-usb-2.6.26-rc2.patch 
> > (it applies to 2.6.26-rc2, obviously) from
> > 
> > 	http://www.kernel.org/pub/linux/kernel/people/gregkh/gregkh-2.6/
> > 
> > and attaching the dmesg log from that?  It contains extra debugging not 
> > present in the vanilla kernel -- make sure that you keep 
> > CONFIG_USB_DEBUG enabled.
> 
> I did another attachement of the dmesg to the bugzilla.

Let me get this straight.  You booted without the dock, and then at 
this line:

> [14315425.975807] ACPI: \_SB_.GDCK - docking

you plugged the computer into the dock, right?

Then apparently here:

> [14315508.451197] ehci_hcd 0000:00:1a.7: port 4 remote wakeup

you plugged the USB drive into the dock, but it wasn't detected.  Or
maybe I've got it backward, the drive was already plugged in when you
docked the computer, and 14315508.451197 is when you unplugged the
drive.  Which was it?

AFAICS, the dock never reported the existence of the USB drive.

Okay, let's see some more debugging information on this.  Get a usbmon
trace (see Documentation/usb/usbmon.txt) for bus 3, starting just 
before you dock the computer.  That will tell us exactly what the 
dock's USB hub is reporting.

By the way, are you entirely certain that 2.6.25 works correctly in
both cases: drive already attached when you dock the computer, and
drive plugged in after the computer is docked?  I guess it would be
good to see equivalent usbmon traces for both 2.6.25 and 2.6.26-rc2.

Alan Stern

Re: 2.6.26-rc1 regression since 2.6.25

[Lukas Hejtmanek]

On Mon, May 19, 2008 at 11:47:11AM -0400, Alan Stern wrote:
> > > > http://bugzilla.kernel.org/show_bug.cgi?id=10630
> 
> Can you attach an equivalent dmesg log for 2.6.25 (with
> CONFIG_USB_DEBUG enabled, of course)?

attached.

> Let me get this straight.  You booted without the dock, and then at 
> this line:
> 
> > [14315425.975807] ACPI: \_SB_.GDCK - docking
> 
> you plugged the computer into the dock, right?

Yup.

> 
> Then apparently here:
> 
> > [14315508.451197] ehci_hcd 0000:00:1a.7: port 4 remote wakeup
> 
> you plugged the USB drive into the dock, but it wasn't detected.  Or
> maybe I've got it backward, the drive was already plugged in when you
> docked the computer, and 14315508.451197 is when you unplugged the
> drive.  Which was it?

More precisly, it was already plugged but the external power of the USB drive 
was switched off and I just switched it on after I had docked. 

> Okay, let's see some more debugging information on this.  Get a usbmon
> trace (see Documentation/usb/usbmon.txt) for bus 3, starting just 
> before you dock the computer.  That will tell us exactly what the 
> dock's USB hub is reporting.

Moreover, I discovered that removing ehci module before I put the machine into
the dock causes the uhci module to discover the USB device. Dmesg traces are
attached.

Do you still need the usbmon traces?

-- 
Lukáš Hejtmánek