possible USB regression with 2.6.21-rc4: iPod doesn't work

possible USB regression with 2.6.21-rc4: iPod doesn't work

[Tino Keitel]

[-- Attachment #1: Type: text/plain, Size: 804 bytes --]

Hi folks,

I plugged my iPod nano 8 GB in and got this message with 2.6.21-rc4:

hub 1-0:1.0: over-current change on port 1

along with other USB error messages. I tried a hub with own power
supply and a USB port on the computer. A full log is attached.

When I boot with 2.6.20, it works:

usb 1-5.1: new high speed USB device using ehci_hcd and address 13
usb 1-5.1: configuration #1 chosen from 2 choices
scsi5 : SCSI emulation for USB Mass Storage devices
usb-storage: device found at 13
usb-storage: waiting for device to settle before scanning
scsi 5:0:0:0: Direct-Access     Apple    iPod             1.62 PQ: 0
ANSI: 0
SCSI device sdc: 3964928 2048-byte hdwr sectors (8120 MB)

Is the over-current message an indication that the hardware does bad
things, or is this a .21-rc4 bug?

Regards,
Tino

[-- Attachment #2: kernlog-ipod.bz2 --]
[-- Type: application/octet-stream, Size: 10097 bytes --]

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Oliver Neukum]

Am Mittwoch, 21. März 2007 21:47 schrieb Tino Keitel:

> along with other USB error messages. I tried a hub with own power
> supply and a USB port on the computer. A full log is attached.

Your log basically shows a hub going berserk. Please post "lsusb -v"
and your .config

	Regards
		Oliver

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Tino Keitel]

On Thu, Mar 22, 2007 at 09:50:29 +0100, Oliver Neukum wrote:
> Am Mittwoch, 21. März 2007 21:47 schrieb Tino Keitel:
> 
> > along with other USB error messages. I tried a hub with own power
> > supply and a USB port on the computer. A full log is attached.
> 
> Your log basically shows a hub going berserk. Please post "lsusb -v"
> and your .config

I didn't tried just the external hub. I also tried a USB port on the
computer.

I'll collect the requested information this evening.

Regards,
Tino

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Tino Keitel]

[-- Attachment #1: Type: text/plain, Size: 391 bytes --]

On Thu, Mar 22, 2007 at 09:50:29 +0100, Oliver Neukum wrote:
> Am Mittwoch, 21. März 2007 21:47 schrieb Tino Keitel:
> 
> > along with other USB error messages. I tried a hub with own power
> > supply and a USB port on the computer. A full log is attached.
> 
> Your log basically shows a hub going berserk. Please post "lsusb -v"
> and your .config

Both files are attached.

Regards,
Tino

[-- Attachment #2: lsusb.txt.bz2 --]
[-- Type: application/octet-stream, Size: 5499 bytes --]

[-- Attachment #3: config-2.6.21-rc4.bz2 --]
[-- Type: application/octet-stream, Size: 8812 bytes --]

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Alan Stern]

On Thu, 22 Mar 2007, Tino Keitel wrote:

> On Thu, Mar 22, 2007 at 09:50:29 +0100, Oliver Neukum wrote:
> > Am Mittwoch, 21. März 2007 21:47 schrieb Tino Keitel:
> > 
> > > along with other USB error messages. I tried a hub with own power
> > > supply and a USB port on the computer. A full log is attached.
> > 
> > Your log basically shows a hub going berserk.

Or a bad USB transceiver.  _Something_ is generating those overcurrent 
warnings, and it sure looks like a hardware malfunction.

Alan Stern

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Oliver Neukum]

Am Donnerstag, 22. März 2007 20:17 schrieb Tino Keitel:
> On Thu, Mar 22, 2007 at 09:50:29 +0100, Oliver Neukum wrote:
> > Am Mittwoch, 21. März 2007 21:47 schrieb Tino Keitel:
> > 
> > > along with other USB error messages. I tried a hub with own power
> > > supply and a USB port on the computer. A full log is attached.
> > 
> > Your log basically shows a hub going berserk. Please post "lsusb -v"
> > and your .config
> 
> Both files are attached.

Please recompile
with CONFIG_USB_DEBUG
and without CONFIG_USB_SUSPEND

	Regards
		Oliver

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Tino Keitel]

On Thu, Mar 22, 2007 at 15:40:40 -0400, Alan Stern wrote:
> On Thu, 22 Mar 2007, Tino Keitel wrote:
> 
> > On Thu, Mar 22, 2007 at 09:50:29 +0100, Oliver Neukum wrote:
> > > Am Mittwoch, 21. März 2007 21:47 schrieb Tino Keitel:
> > > 
> > > > along with other USB error messages. I tried a hub with own power
> > > > supply and a USB port on the computer. A full log is attached.
> > > 
> > > Your log basically shows a hub going berserk.
> 
> Or a bad USB transceiver.  _Something_ is generating those overcurrent 
> warnings, and it sure looks like a hardware malfunction.

But it works with 2.6.20.

Regards,
Tino

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[David Brownell]

On Thursday 22 March 2007 12:54 pm, Tino Keitel wrote:
> On Thu, Mar 22, 2007 at 15:40:40 -0400, Alan Stern wrote:
> > 	  _Something_ is generating those overcurrent 
> > warnings, and it sure looks like a hardware malfunction.
> 
> But it works with 2.6.20.

So can you bisect to find what caused the problem?

We've been afflicted with such strange overcurrent messages
off and on for some time.  Some hardware triggers them, while
most doesn't, and the USB developers don't have any of the
hardware that triggers it (that almost goes without saying).

It's been clear to me that _something_ the software does is
making that more likely to happen.

One thing we've been lacking is anything like a usable clue
as to what kind of changes would have that failure mode ...
and thus, what kind of changes could remove it.

- Dave

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Tino Keitel]

On Thu, Mar 22, 2007 at 14:29:11 -0700, David Brownell wrote:
> On Thursday 22 March 2007 12:54 pm, Tino Keitel wrote:
> > On Thu, Mar 22, 2007 at 15:40:40 -0400, Alan Stern wrote:
> > > 	  _Something_ is generating those overcurrent 
> > > warnings, and it sure looks like a hardware malfunction.
> > 
> > But it works with 2.6.20.
> 
> So can you bisect to find what caused the problem?

I never did use git-bisect, but maybe I find the time next week.

Regards,
Tino

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Chuck Ebbert]

David Brownell wrote:
> On Thursday 22 March 2007 12:54 pm, Tino Keitel wrote:
>> On Thu, Mar 22, 2007 at 15:40:40 -0400, Alan Stern wrote:
>>> 	  _Something_ is generating those overcurrent 
>>> warnings, and it sure looks like a hardware malfunction.
>> But it works with 2.6.20.
> 
> So can you bisect to find what caused the problem?
> 
> We've been afflicted with such strange overcurrent messages
> off and on for some time.  Some hardware triggers them, while
> most doesn't, and the USB developers don't have any of the
> hardware that triggers it (that almost goes without saying).
> 

Generation 1 iPod Shuffle is notorious for high current draw.
You should be able to get the 512MB model really really cheap
by now.

Re: possible USB regression with 2.6.21-rc4: iPod doesn't work

[Pete Zaitcev]

On Thu, 22 Mar 2007 17:56:13 -0400, Chuck Ebbert <cebbert@redhat.com> wrote:

> >>> 	  _Something_ is generating those overcurrent 
> >>> warnings, and it sure looks like a hardware malfunction.

> >> But it works with 2.6.20.

> Generation 1 iPod Shuffle is notorious for high current draw.
> You should be able to get the 512MB model really really cheap
> by now.

That's true but the 2.6.20 works, right? That's what we're looking
at here.

The problem is in the device, true, but this is why Tino is the only
one who can bisect and find what we did to aggravate the problem.

-- Pete

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Adrian Bunk]

On Thu, Mar 22, 2007 at 10:46:53PM +0100, Tino Keitel wrote:
> On Thu, Mar 22, 2007 at 14:29:11 -0700, David Brownell wrote:
> > On Thursday 22 March 2007 12:54 pm, Tino Keitel wrote:
> > > On Thu, Mar 22, 2007 at 15:40:40 -0400, Alan Stern wrote:
> > > > 	  _Something_ is generating those overcurrent 
> > > > warnings, and it sure looks like a hardware malfunction.
> > > 
> > > But it works with 2.6.20.
> > 
> > So can you bisect to find what caused the problem?
> 
> I never did use git-bisect, but maybe I find the time next week.


Instructions for bisecting:


<--  snip  -->

# install git and cogito on your computer

# clone Linus' tree:
cg-clone \ 
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git

# start bisecting:
cd linux-2.6
git bisect start
git bisect bad v2.6.21-rc4
git bisect good v2.6.20

# start round
cp /path/to/.config .
make oldconfig
make
# install kernel, check whether it's good or bad, then:
git bisect [bad|good]
# start next round


After at about 12 reboots, you'll have found the guilty commit
("...  is first bad commit").


More information on git bisecting:
  man git-bisect

<--  snip  -->


> Regards,
> Tino

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Tino Keitel]

[-- Attachment #1: Type: text/plain, Size: 710 bytes --]

On Thu, Mar 22, 2007 at 20:42:44 +0100, Oliver Neukum wrote:
> Am Donnerstag, 22. März 2007 20:17 schrieb Tino Keitel:
> > On Thu, Mar 22, 2007 at 09:50:29 +0100, Oliver Neukum wrote:
> > > Am Mittwoch, 21. März 2007 21:47 schrieb Tino Keitel:
> > > 
> > > > along with other USB error messages. I tried a hub with own power
> > > > supply and a USB port on the computer. A full log is attached.
> > > 
> > > Your log basically shows a hub going berserk. Please post "lsusb -v"
> > > and your .config
> > 
> > Both files are attached.
> 
> Please recompile
> with CONFIG_USB_DEBUG
> and without CONFIG_USB_SUSPEND

With CONFIG_USB_SUSPEND disabled, the iPod works. The dmesg output is
attached.

Regards,
Tino

[-- Attachment #2: dmesg.bz2 --]
[-- Type: application/octet-stream, Size: 11306 bytes --]

[-- Attachment #3: dmesg_with_USB_SUSPEND.bz2 --]
[-- Type: application/octet-stream, Size: 11593 bytes --]

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Alan Stern]

On Sun, 25 Mar 2007, Tino Keitel wrote:

> > Please recompile
> > with CONFIG_USB_DEBUG
> > and without CONFIG_USB_SUSPEND
> 
> With CONFIG_USB_SUSPEND disabled, the iPod works. The dmesg output is
> attached.

In fact, both logs (with and without CONFIG_USB_SUSPEND) show the iPod 
working.  Can you post a log with CONFIG_USB_DEBUG turned on that shows 
any errors?

Alan Stern

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Tino Keitel]

[-- Attachment #1: Type: text/plain, Size: 838 bytes --]

On Mon, Mar 26, 2007 at 14:12:25 -0400, Alan Stern wrote:
> On Sun, 25 Mar 2007, Tino Keitel wrote:
> 
> > > Please recompile
> > > with CONFIG_USB_DEBUG
> > > and without CONFIG_USB_SUSPEND
> > 
> > With CONFIG_USB_SUSPEND disabled, the iPod works. The dmesg output is
> > attached.
> 
> In fact, both logs (with and without CONFIG_USB_SUSPEND) show the iPod 
> working.  Can you post a log with CONFIG_USB_DEBUG turned on that shows 
> any errors?

Attached is a dmesg output with CONFIG_USB_DEBUG and CONFIG_USB_SUSPEND
enabled.

There are no messages from the iPod because, well, nothing happens when
I plug it in. I tried it 3 times. The strange thing is: when I trigger
a suspend or shutdown, then the "Do not disconnect" message shows up on
the iPod, which means that the computer regognized it as a storage
device.

Regards,
Tino

[-- Attachment #2: dmesg-with-usb_suspend.bz2 --]
[-- Type: application/octet-stream, Size: 10782 bytes --]

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Alan Stern]

On Mon, 26 Mar 2007, Tino Keitel wrote:

> Attached is a dmesg output with CONFIG_USB_DEBUG and CONFIG_USB_SUSPEND
> enabled.
> 
> There are no messages from the iPod because, well, nothing happens when
> I plug it in. I tried it 3 times.

I don't understand.  The dmesg log you attached shows the iPod was
detected and recognized as scsi5 (sdc).  Then some time later (can't tell
how much later because you don't have CONFIG_PRINTK_TIME set) it was
unplugged and plugged back in.  The second time it was detected and
recognized as scsi6 (also sdc).  And then the log ends.

Why do you say there are no messages from the iPod?  Grepping for "iPod"
in the log shows that the string appears 4 times.  If nothing happened
when you plugged it in, how could the computer have known the device was
an iPod?

>  The strange thing is: when I trigger
> a suspend or shutdown, 

You mean you suspend or shutdown the computer?  Or the iPod?

>  then the "Do not disconnect" message shows up on
> the iPod, which means that the computer regognized it as a storage
> device.

What's so strange about that?  Isn't that exactly what's supposed to 
happen?

Alan Stern

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Tino Keitel]

On Mon, Mar 26, 2007 at 16:28:34 -0400, Alan Stern wrote:
> On Mon, 26 Mar 2007, Tino Keitel wrote:
> 
> > Attached is a dmesg output with CONFIG_USB_DEBUG and CONFIG_USB_SUSPEND
> > enabled.
> > 
> > There are no messages from the iPod because, well, nothing happens when
> > I plug it in. I tried it 3 times.
> 
> I don't understand.  The dmesg log you attached shows the iPod was
> detected and recognized as scsi5 (sdc).  Then some time later (can't tell
> how much later because you don't have CONFIG_PRINTK_TIME set) it was
> unplugged and plugged back in.  The second time it was detected and
> recognized as scsi6 (also sdc).  And then the log ends.

Not in the attached dmesg output from the above mail. There are not
iPod related messages it it, but I plugged it in 2 times. I rebootet
with the same kernel version without CONFIG_USB_SUSPEND right after
this and the iPod was recognized at the first try.

> 
> Why do you say there are no messages from the iPod?  Grepping for "iPod"
> in the log shows that the string appears 4 times.  If nothing happened
> when you plugged it in, how could the computer have known the device was
> an iPod?
> 
> >  The strange thing is: when I trigger
> > a suspend or shutdown, 
> 
> You mean you suspend or shutdown the computer?  Or the iPod?

The computer.

> 
> >  then the "Do not disconnect" message shows up on
> > the iPod, which means that the computer regognized it as a storage
> > device.
> 
> What's so strange about that?  Isn't that exactly what's supposed to 
> happen?

Well, it isn't recognized by a normal running system, only during
shutdown/suspend.

Regards,
Tino

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Alan Stern]

On Mon, 26 Mar 2007, Tino Keitel wrote:

> > > There are no messages from the iPod because, well, nothing happens when
> > > I plug it in. I tried it 3 times.
> > 
> > I don't understand.  The dmesg log you attached shows the iPod was
> > detected and recognized as scsi5 (sdc).  Then some time later (can't tell
> > how much later because you don't have CONFIG_PRINTK_TIME set) it was
> > unplugged and plugged back in.  The second time it was detected and
> > recognized as scsi6 (also sdc).  And then the log ends.
> 
> Not in the attached dmesg output from the above mail. There are not
> iPod related messages it it, but I plugged it in 2 times. I rebootet
> with the same kernel version without CONFIG_USB_SUSPEND right after
> this and the iPod was recognized at the first try.

Sorry, I was reading the wrong file before!  :-(

The lack of messages from the iPod seems to indicate that the hub isn't
working right.  You could try plugging the iPod into a different hub or
directly into the computer.  Or maybe into a different port of that hub.

Alan Stern

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Tino Keitel]

On Mon, Mar 26, 2007 at 17:15:53 -0400, Alan Stern wrote:

[...]

> The lack of messages from the iPod seems to indicate that the hub isn't
> working right.  You could try plugging the iPod into a different hub or
> directly into the computer.  Or maybe into a different port of that hub.

I already tried all of the above options, with the same result. Note
that all other USB devices (keyboard, mouse, hard disk with /home,
DVB-T box etc.) work fine.

I'm currently bisecting.

Regards,
Tino

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Tino Keitel]

On Mon, Mar 26, 2007 at 17:15:53 -0400, Alan Stern wrote:

[...]

> The lack of messages from the iPod seems to indicate that the hub isn't
> working right.  You could try plugging the iPod into a different hub or
> directly into the computer.  Or maybe into a different port of that hub.

Uh, I think I found the reason for the strange behaviour at
shutdown/suspend. When I unload the usblp module, then the iPod is
recognized.

Regards,
Tino

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Tino Keitel]

On Mon, Mar 26, 2007 at 23:26:14 +0200, Tino Keitel wrote:
> On Mon, Mar 26, 2007 at 17:15:53 -0400, Alan Stern wrote:
> 
> [...]
> 
> > The lack of messages from the iPod seems to indicate that the hub isn't
> > working right.  You could try plugging the iPod into a different hub or
> > directly into the computer.  Or maybe into a different port of that hub.
> 
> I already tried all of the above options, with the same result. Note
> that all other USB devices (keyboard, mouse, hard disk with /home,
> DVB-T box etc.) work fine.
> 
> I'm currently bisecting.

Hi,

this is the bisect result:

$ git bisect good
1d619f128ba911cd3e6d6ad3475f146eb92f5c27 is first bad commit
commit 1d619f128ba911cd3e6d6ad3475f146eb92f5c27
Author: Marcelo Tosatti <marcelo@kvack.org>
Date:   Sun Jan 21 19:45:59 2007 -0200

    USB: switch ehci-hcd to new polling scheme

    Switch ehci-hcd to use the new polling scheme, which reports root
    hub status changes via the interrupt handler, in an asynchronous
    fashion. Doing so disables polling for status changes (whose
    handler is
    rh_timer_func).

    Tested on a Geode GX machine, which is now capable of running at =~
5
    timer interrupts per second (in the -rt tree), resulting in
significant
    power savings.

    Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
    Cc: David Brownell <david-b@pacbell.net>
    Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

:040000 040000 f8b11b3fe3cec62063d8da0f7be807341106f494
78c5a156897b3ad7aef27823d48a546fdda2c0d2 M      drivers

Regards,
Tino

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Tino Keitel]

On Tue, Mar 27, 2007 at 00:21:24 +0200, Tino Keitel wrote:

[...]

> this is the bisect result:
> 
> $ git bisect good
> 1d619f128ba911cd3e6d6ad3475f146eb92f5c27 is first bad commit
> commit 1d619f128ba911cd3e6d6ad3475f146eb92f5c27

I just tested 2.6.21-rc5 with this commit reverted and the iPod was
regognized with CONFIG_USB_SUSPEND enabled.

Regards,
Tino

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Alan Stern]

On Tue, 27 Mar 2007, Tino Keitel wrote:

> On Tue, Mar 27, 2007 at 00:21:24 +0200, Tino Keitel wrote:
> 
> [...]
> 
> > this is the bisect result:
> > 
> > $ git bisect good
> > 1d619f128ba911cd3e6d6ad3475f146eb92f5c27 is first bad commit
> > commit 1d619f128ba911cd3e6d6ad3475f146eb92f5c27
> 
> I just tested 2.6.21-rc5 with this commit reverted and the iPod was
> regognized with CONFIG_USB_SUSPEND enabled.

It's hard to figure out what's going on with your system for several 
reasons.  For one thing, you've mentioned at least 3 different peculiar 
behaviors.  For another, you posted logs with lots of extra information 
that tends to hide the important parts.  And also you have lots of USB 
devices, which makes it hard to see what's happening with the iPod.


So here's what I want you to do.  To begin with, unplug _all_ your USB 
devices.  Then do "dmesg -c >/dev/null" to clear the kernel log buffer.

After waiting a few seconds, plug the iPod in to the computer.  After a 
few seconds, make a copy of /sys/class/usb_host/usb_host1/registers.  Then 
unplug the iPod.

After waiting a few more seconds, plug in your hub.  Wait a few seconds 
and then plug the iPod into the hub.  After a few seconds, make another 
copy of /sys/class/usb_host/usb_host1/registers.  Then unplug the iPod and 
the hub.  Finally, get a copy of the dmesg log.


After having done all that, repeat it exactly but this time using a 
version of ehci-hcd with that patch reverted.  Post everything and we'll 
see how it looks.

Alan Stern

Re: [linux-usb-devel] possible USB regression with 2.6.21-rc4: iPod doesn't work

[Bill Davidsen]

Tino Keitel wrote:
> On Mon, Mar 26, 2007 at 17:15:53 -0400, Alan Stern wrote:
> 
> [...]
> 
>> The lack of messages from the iPod seems to indicate that the hub isn't
>> working right.  You could try plugging the iPod into a different hub or
>> directly into the computer.  Or maybe into a different port of that hub.
> 
> Uh, I think I found the reason for the strange behaviour at
> shutdown/suspend. When I unload the usblp module, then the iPod is
> recognized.
> 
And that's not the case with 2.6.20?

-- 
Bill Davidsen <davidsen@tmr.com>
   "We have more to fear from the bungling of the incompetent than from
the machinations of the wicked."  - from Slashdot