Reply-To: akpm@linux-foundation.org


(switched to email.  Please respond via emailed reply-to-all, not via the
bugzilla web interface).

On Fri, 27 Jun 2008 16:54:09 -0700 (PDT)
bugme-daemon@bugzilla.kernel.org wrote:

> http://bugzilla.kernel.org/show_bug.cgi?id=10997
> 
>            Summary: Module ehci_hcd behaves nasty on device use and/or
>                     device removal.
>            Product: Drivers
>            Version: 2.5
>      KernelVersion: 2.6.26-rc8 (any version >= 2.6.25)
>           Platform: All
>         OS/Version: Linux
>               Tree: Mainline
>             Status: NEW
>           Severity: normal
>           Priority: P1
>          Component: USB
>         AssignedTo: greg@kroah.com
>         ReportedBy: valentin.quequet@orange.fr
>                 CC: valentin.quequet@orange.fr
> 
> 
> Latest working kernel version: 2.6.24.7
> Earliest failing kernel version: 2.6.25.6
> Distribution: Debian over PRISTINE Linux 2.6.26-rc8
> Hardware Environment:
> 
> Linux debian 2.6.26-rc8-custom-0001 #1 SMP Fri Jun 27 20:45:19 CEST 2008
> x86_64
> GNU/Linux
> 
> Software Environment:
> 
> Linux debian 2.6.26-rc8-custom-0001 #1 SMP Fri Jun 27 20:45:19 CEST 2008
> x86_64
> GNU/Linux under Debian Lenny/Sid.
> 
> 
> Problem Description:
> 
> Hello ;-)
> 
> Troubles (Oopses) often (see details after Oopses below) happen when I try to
> scan a document via S.A.N.E. with my combo Printer/Scanner Canon PIXMA MP150.
> 
> I always have at most 1 scanner plugged (the above mentionned).
> 
> The same trouble happens when I type the command 'scanimage -T' which is a
> test
> consisting effectively in scanning a small part of the document.
> 
> Contrary to 'scanimage -T', the command 'scanimage -L', which lists available
> scanners (1 scanner here if plugged and ON), initiate dialogue with the
> scanner
> and report it as available for susequent scanning operations.
> 
> 
> Example of an unsuccessful scanning/test attempt, where module ehci_hcd is
> loaded:
> 
> Jun 27 22:39:10 debian kernel: [  245.365633] usb 2-5: unlink
> qh128-0001/ffff81003e6e9320 start 127 [2/0 us]
> Jun 27 22:39:10 debian kernel: [  245.365740] ehci_hcd 0000:00:0b.1: reused
> qh
> ffff81003e6e9320 schedule
> Jun 27 22:39:10 debian kernel: [  245.365746] usb 2-5: link
> qh128-0001/ffff81003e6e9320 start 127 [2/0 us]
> Jun 27 22:39:10 debian kernel: [  245.820940] usb 2-5: unlink
> qh128-0001/ffff81003e6e9320 start 127 [2/0 us]
> Jun 27 22:39:10 debian kernel: [  245.822675] ehci_hcd 0000:00:0b.1: HC died;
> cleaning up
> Jun 27 22:39:10 debian kernel: [  245.824940] hub 2-0:1.0: state 0 ports 8
> chg
> 0000 evt fe00
> Jun 27 22:39:10 debian kernel: [  245.824940] usb 2-5: USB disconnect,
> address
> 4
> Jun 27 22:39:10 debian kernel: [  245.824940] usb 2-5: unregistering device
> Jun 27 22:39:10 debian kernel: [  245.824940] usb 2-5: usb_disable_device
> nuking all URBs
> Jun 27 22:39:10 debian kernel: [  245.824940] usb 2-5: unregistering
> interface
> 2-5:1.0
> Jun 27 22:39:10 debian kernel: [  245.824940] PM: Removing info for
> usb:2-5:1.0
> 
> I never get to scan a document with module ehci_hcd loaded.
> 
> If I power off my scanner while ehci_hcd is loaded, I get that:
> 
> Jun 27 22:40:48 debian kernel: [  342.999039] irq 7: nobody cared (try
> booting
> with the "irqpoll" option)
> Jun 27 22:40:48 debian kernel: [  342.999358] Pid: 0, comm: swapper Not
> tainted
> 2.6.26-rc8-custom-0001 #1
> Jun 27 22:40:48 debian kernel: [  342.999533] 
> Jun 27 22:40:48 debian kernel: [  342.999534] Call Trace:
> Jun 27 22:40:48 debian kernel: [  342.999868]  <IRQ>  [<ffffffff8026bb3b>]
> __report_bad_irq+0x30/0x72
> Jun 27 22:40:48 debian kernel: [  343.000084]  [<ffffffff8026bd68>]
> note_interrupt+0x1eb/0x253
> Jun 27 22:40:48 debian kernel: [  343.000262]  [<ffffffff8026c429>]
> handle_fasteoi_irq+0xa5/0xc8
> Jun 27 22:40:48 debian kernel: [  343.000440]  [<ffffffff8020ec6a>]
> do_IRQ+0xf2/0x160
> Jun 27 22:40:48 debian kernel: [  343.000612]  [<ffffffff8020a000>]
> default_idle+0x0/0x3b
> Jun 27 22:40:48 debian kernel: [  343.000787]  [<ffffffff8020c331>]
> ret_from_intr+0x0/0xa
> Jun 27 22:40:48 debian kernel: [  343.000959]  <EOI>  [<ffffffff8020a000>]
> default_idle+0x0/0x3b
> Jun 27 22:40:48 debian kernel: [  343.001162]  [<ffffffff8020a027>]
> default_idle+0x27/0x3b
> Jun 27 22:40:48 debian kernel: [  343.001337]  [<ffffffff8020afeb>]
> cpu_idle+0x77/0xbf
> Jun 27 22:40:48 debian kernel: [  343.001518] 
> Jun 27 22:40:48 debian kernel: [  343.001685] handlers:
> Jun 27 22:40:48 debian kernel: [  343.001853] [<ffffffff80373b82>]
> (usb_hcd_irq+0x0/0x58)
> Jun 27 22:40:48 debian kernel: [  343.002078] Disabling IRQ #7
> 
> I even sometimes get this later Oops when I just try to scan a document (or
> test with 'scanimage -T').
> 
> Steps to reproduce:
> 
> When ehci_hcd is NOT loaded, then both 'scanimage -L' and 'scanimage -T'
> ALWAYS
> SUCCEED at dialoging with the scanner.
> 
> When ehci_hcd is loaded, then 'scanimage -T' ALWAYS FAILS, outputting Oopses
> or
> saying (if already oopsed) that there is no scanner.
> 
> When ehci_hcd is loaded, then 'scanimage -L' SUCCEED ONLY IF there has been
> no
> Oopses (of the 2 sorts described above).
> 
> More information:
> 
>     I enabled some debugging/logging facilities at .config time for the
>     purpose
> of this report.
>     Troubles showed even without those options anyway.
>     Troubles equally showed with kernel versions >= 2.6.25.6 
>     Troubles did not show with kernel version == 2.6.24.7 
> 
>     (syslog, .config, ...) on my WEB site:
>         http://pagesperso-orange.fr/mandolosse/
> 
>     Initial report against Debian-packaged kernel here:
>         http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=487725
> 
> With hope my report will prove useful.
> 

can you tell us if this still happens with 2.6.27-rc6?

Hello,

I'm the original poster of this bug (bug #10997) and I just realized that this bug seems to have disappeared in pristine Linux kernel 2.6.27.7 though my test case is not very intensive.

I cannot however tell you which was the first kernel version where this bug (bug #10997) was already solved.

I'm sorry for being (possibly) late.

I'm going to do a bit more testing in order to:
  - Stress my system
  - Check if this bug (bug #10997) was solved before 2.7.27.7

Sincerely,
Valentin.
( v.quequet-techniques@orange.fr )

Hello,

I am proud to tell you that this bug (bug #10997) has disappeared on my system since pristine Linux kernel 2.6.27 °.

° To be accurate, by 2.6.27 I meant the exact predecessor release of 2.6.27.1 , not to be confused with some 2.6.27.x release ( i.e. 2.6.27.7 ).

I hope my report will prove useful and still appologize for having been so long to test these few releases.

Sincerely,
Valentin

Thanks for letting us know (and for reporting it)