2003-07-16 15:54:14

by Philippe Gramoullé

Subject: 2.6.0-test1-mm1: bad: scheduling while atomic! after removing ohci1394 module

Hello,

With 2.6.0-test1-mm1, i have the following in my logs after i rmmod'ed the
ohci1394 module:

bad: scheduling while atomic!
Call Trace:
[<c011c958>] schedule+0x578/0x580
[<c0108ff2>] copy_thread+0x32/0x250
[<c011cd0c>] wait_for_completion+0x8c/0xf0
[<c011c9b0>] default_wake_function+0x0/0x30
[<c011c9b0>] default_wake_function+0x0/0x30
[<c012b9e1>] kill_proc_info+0x51/0x80
[<e0b87665>] nodemgr_remove_host+0x55/0xa0 [ieee1394]
[<e0b82fba>] highlevel_remove_host+0x8a/0xa0 [ieee1394]
[<e0b4163d>] ohci1394_pci_remove+0x3d/0x160 [ohci1394]
[<c018c93e>] sysfs_hash_and_remove+0x4e/0x7c
[<c022c75b>] pci_device_remove+0x3b/0x40
[<c0256916>] device_release_driver+0x66/0x70
[<c025694b>] driver_detach+0x2b/0x40
[<c0256bae>] bus_remove_driver+0x3e/0x80
[<c0256fc3>] driver_unregister+0x13/0x2a
[<c022ca56>] pci_unregister_driver+0x16/0x30
[<e0b41adf>] ohci1394_cleanup+0xf/0x11 [ohci1394]
[<c0137d89>] sys_delete_module+0x139/0x170
[<c014c1e5>] sys_munmap+0x45/0x70
[<c010b0ab>] syscall_call+0x7/0xb


Should i worry ?

Besides that, IEE1394 works fine with my DV camcorder.

Thanks,

Philippe

2003-07-16 16:22:59

by Ben Collins

Subject: Re: 2.6.0-test1-mm1: bad: scheduling while atomic! after removing ohci1394 module

On Wed, Jul 16, 2003 at 06:08:55PM +0200, Philippe Gramoull? wrote:
> Hello,
>
> With 2.6.0-test1-mm1, i have the following in my logs after i rmmod'ed the
> ohci1394 module:
>
> bad: scheduling while atomic!
> Call Trace:
> [<c011c958>] schedule+0x578/0x580
> [<c0108ff2>] copy_thread+0x32/0x250
> [<c011cd0c>] wait_for_completion+0x8c/0xf0
> [<c011c9b0>] default_wake_function+0x0/0x30
> [<c011c9b0>] default_wake_function+0x0/0x30
> [<c012b9e1>] kill_proc_info+0x51/0x80
> [<e0b87665>] nodemgr_remove_host+0x55/0xa0 [ieee1394]
> [<e0b82fba>] highlevel_remove_host+0x8a/0xa0 [ieee1394]
> [<e0b4163d>] ohci1394_pci_remove+0x3d/0x160 [ohci1394]
> [<c018c93e>] sysfs_hash_and_remove+0x4e/0x7c
> [<c022c75b>] pci_device_remove+0x3b/0x40
> [<c0256916>] device_release_driver+0x66/0x70
> [<c025694b>] driver_detach+0x2b/0x40
> [<c0256bae>] bus_remove_driver+0x3e/0x80
> [<c0256fc3>] driver_unregister+0x13/0x2a
> [<c022ca56>] pci_unregister_driver+0x16/0x30
> [<e0b41adf>] ohci1394_cleanup+0xf/0x11 [ohci1394]
> [<c0137d89>] sys_delete_module+0x139/0x170
> [<c014c1e5>] sys_munmap+0x45/0x70
> [<c010b0ab>] syscall_call+0x7/0xb

So is kill_proc() bad while atomic, e.g. when a module is being removed?
I'll look into it.

--
Debian - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/

2003-07-16 16:30:16

by Jeff Garzik

Subject: Re: 2.6.0-test1-mm1: bad: scheduling while atomic! after removing ohci1394 module

On Wed, Jul 16, 2003 at 12:36:48PM -0400, Ben Collins wrote:
> So is kill_proc() bad while atomic, e.g. when a module is being removed?
> I'll look into it.

hmmm, that's a big strange. During ->remove and module unload, you can
schedule and sleep.

Jeff

2003-07-16 16:39:48

by Ben Collins

Subject: Re: 2.6.0-test1-mm1: bad: scheduling while atomic! after removing ohci1394 module

On Wed, Jul 16, 2003 at 12:45:06PM -0400, Jeff Garzik wrote:
> On Wed, Jul 16, 2003 at 12:36:48PM -0400, Ben Collins wrote:
> > So is kill_proc() bad while atomic, e.g. when a module is being removed?
> > I'll look into it.
>
> hmmm, that's a big strange. During ->remove and module unload, you can
> schedule and sleep.

I'd have to check to make sure the remove callbacks are doing something
stupid in the ieee1394 subsystem, like causing this problem.

--
Debian - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/