Re: 2.6.17-rc4-mm2 (hard lockup after resume from disk on AMD64)

[Andrew Morton <akpm@osdl.org>]

"Rafael J. Wysocki" <rjw@sisk.pl> wrote:
>
> On Saturday 20 May 2006 14:41, Andrew Morton wrote:
>  > 
>  > ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.17-rc4/2.6.17-rc4-mm2/
> 
>  My box (Asus L5D, x86_64 kernel) locks up hard after the resume from suspend
>  to disk.  This happens right after the console has been restored and only if
>  all modules are loaded before suspend (eg. it doesn't lock up when booted with
>  init=/bin/bash).  Also it's caught by the softlockup watchdog which produces
>  traces similar to the appended one.
> 
>  Greetings,
>  Rafael
> 
> 
>  BUG: soft lockup detected on CPU#0!
> 
>  Call Trace: <IRQ> <ffffffff8025b2f1>{softlockup_tick+193}
>         <ffffffff802360a3>{run_local_timers+19} <ffffffff802362c7>{update_process_times+87}
>         <ffffffff8020e174>{main_timer_handler+548} <ffffffff8020e3c5>{timer_interrupt+21}
>         <ffffffff8025b4e3>{handle_IRQ_event+51} <ffffffff8025b5d2>{__do_IRQ+162}
>         <ffffffff8020c3b1>{do_IRQ+65} <ffffffff80209d5e>{ret_from_intr+0}
>         <ffffffff803cb8d0>{urb_destroy+0} <ffffffff803ca84a>{usb_hcd_poll_rh_status+250}
>         <ffffffff880dafa2>{:ohci_hcd:ohci_irq+194} <ffffffff803c9ce2>{usb_hcd_irq+50}
>         <ffffffff8025b4e3>{handle_IRQ_event+51} <ffffffff8025b5d2>{__do_IRQ+162}
>         <ffffffff8020c3b1>{do_IRQ+65} <ffffffff80209d5e>{ret_from_intr+0}
>         <ffffffff8025b4d4>{handle_IRQ_event+36} <ffffffff8025b5d2>{__do_IRQ+162}
>         <ffffffff8020c3b1>{do_IRQ+65} <ffffffff80209d5e>{ret_from_intr+0}
>         <ffffffff8025b4d4>{handle_IRQ_event+36} <ffffffff8025b5d2>{__do_IRQ+162}
>         <ffffffff8020c3b1>{do_IRQ+65} <ffffffff80209d5e>{ret_from_intr+0} <EOI>
>         <ffffffff803cb8d0>{urb_destroy+0} <ffffffff803cb788>{hcd_submit_urb+2072}
>         <ffffffff80463d4d>{_spin_unlock_irqrestore+29} <ffffffff803cb9a5>{usb_free_urb+21}
>         <ffffffff803ca0ab>{usb_hcd_giveback_urb+315} <ffffffff80463d4d>{_spin_unlock_irqrestore+29}
>         <ffffffff803cbd2b>{usb_submit_urb+859} <ffffffff803cc66a>{usb_start_wait_urb+122}
>         <ffffffff8027c58d>{dbg_redzone1+29} <ffffffff8027d804>{cache_alloc_debugcheck_after+532}
>         <ffffffff8027c58d>{dbg_redzone1+29} <ffffffff803cca40>{usb_control_msg+240}
>         <ffffffff803c5e74>{set_port_feature+68} <ffffffff803c61f9>{hub_port_reset+57}
>         <ffffffff803c64d1>{hub_port_init+129} <ffffffff803c8b31>{hub_thread+2241}
>         <ffffffff80242140>{autoremove_wake_function+0} <ffffffff803c8270>{hub_thread+0}
>         <ffffffff80241f29>{kthread+217} <ffffffff80463d84>{_spin_unlock_irq+20}
>         <ffffffff80228d99>{schedule_tail+73} <ffffffff8020a38a>{child_rip+8}
>         <ffffffff80241e50>{kthread+0} <ffffffff8020a382>{child_rip+0}
>  BUG: soft lockup detected on CPU#0!

What a revolting backtrace.  There were some x86_64 precise-backtrace
patches floating around on the list this week.

Still, it would appear that USB has become confused.  If you have the time
to do a bisection search, I'd start in on gregkh-usb-*.patch.