ACPI bay - 2.6.20-rc3-mm1 hangs on boot

ACPI bay - 2.6.20-rc3-mm1 hangs on boot

[Jiri Kosina]

Hi,

2.6.20-rc3-mm1 hangs on boot on my IBM T42p when compiled with ACPI_BAY=y. 
Below is the trace of two BUGs I get.

When compiled with ACPI_BAY=n, it boots fine.

The traces are hand-rewritten (no serial console on that machine), so I 
have omitted the code offsets in the stacktraces.

ACPI: ACPI Dock Station Driver
ACPI: \_SB_PCI0.IDE0.SCND.MSTR: found ejectable bay
ACPI: \_SB_PCI0.IDE0.SCND.MSTR: Adding notify handler
PM: Adding info for platform:bay.0
ACPI: Bay [\_SB_.PCI0.IDE0.SCND.MSTR] Added
BUG: at lib/kref.c:32 kref_get()
	kref_get+0x34/0x3b
	kobject_get+0xf/0x13
	get_bus+0xe/0x1d
	bus_add_driver+0x13/0x165
	init_waitqueue_head+0x12/0x1e
	bay_init+0x57/0x79
	find_bay+0x0/0x2c4
	init+0x88/0x16d
	restore_nocheck+0x12/0x15
	init+0x0/0x16d
	init+0x0/0x16d
	kernel_thread_helper+0x7/0x10
	==========
BUG: at lib/kref.c:32 kref_get()
	kref_get+0x34/0x3b
	kobject_get+0xf/0x13
	kobject_init+0x5d/0x70
	kobject_set_name+0x5c/0x92
	bus_add_driver+0x50/0x79
	bay_init+0x57/0x79
	find_bay+0x0/0x2c4
	init+0x88/0x16d
	init+0x88/0x16d
	kernel_thread_helper+0x7/0x10
	=========

-- 
Jiri Kosina

Re: ACPI bay - 2.6.20-rc3-mm1 hangs on boot

[Henrique de Moraes Holschuh]

On Fri, 05 Jan 2007, Jiri Kosina wrote:
> 2.6.20-rc3-mm1 hangs on boot on my IBM T42p when compiled with ACPI_BAY=y. 

I trust you have ACPI_IBM_BAY=n to use ACPI_BAY=y ?  I don't think it has
anything to do with the issue you have, but just in case...

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Re: ACPI bay - 2.6.20-rc3-mm1 hangs on boot

[Jiri Kosina]

On Fri, 5 Jan 2007, Henrique de Moraes Holschuh wrote:

> > 2.6.20-rc3-mm1 hangs on boot on my IBM T42p when compiled with ACPI_BAY=y. 
> I trust you have ACPI_IBM_BAY=n to use ACPI_BAY=y ?  I don't think it has
> anything to do with the issue you have, but just in case...

The kernel hangs with CONFIG_ACPI_BAY, with CONFIG_ACPI_IBM_BAY it works 
just fine.

-- 
Jiri Kosina

Re: ACPI bay - 2.6.20-rc3-mm1 hangs on boot

[Henrique de Moraes Holschuh]

On Fri, 05 Jan 2007, Jiri Kosina wrote:
> On Fri, 5 Jan 2007, Henrique de Moraes Holschuh wrote:
> > > 2.6.20-rc3-mm1 hangs on boot on my IBM T42p when compiled with ACPI_BAY=y. 
> > I trust you have ACPI_IBM_BAY=n to use ACPI_BAY=y ?  I don't think it has
> > anything to do with the issue you have, but just in case...
> 
> The kernel hangs with CONFIG_ACPI_BAY, with CONFIG_ACPI_IBM_BAY it works 
> just fine.

I mean don't do it with both CONFIG_ACPI_BAY and CONFIG_ACPI_IBM_BAY set to
y (i.e. loaded at the same time).

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Re: ACPI bay - 2.6.20-rc3-mm1 hangs on boot

[Jiri Kosina]

On Fri, 5 Jan 2007, Henrique de Moraes Holschuh wrote:

> > > I trust you have ACPI_IBM_BAY=n to use ACPI_BAY=y ?  I don't think it has
> > > anything to do with the issue you have, but just in case...
> > The kernel hangs with CONFIG_ACPI_BAY, with CONFIG_ACPI_IBM_BAY it works 
> > just fine.
> I mean don't do it with both CONFIG_ACPI_BAY and CONFIG_ACPI_IBM_BAY set to
> y (i.e. loaded at the same time).

Kconfig doesn't allow this, so that's not a problem.

config ACPI_IBM_BAY
        bool "Legacy Removable Bay Support"
        depends on ACPI_IBM
        depends on ACPI_BAY=n

-- 
Jiri Kosina

Re: ACPI bay - 2.6.20-rc3-mm1 hangs on boot

[Pavel Machek]

On Fri 2007-01-05 14:19:41, Jiri Kosina wrote:
> Hi,
> 
> 2.6.20-rc3-mm1 hangs on boot on my IBM T42p when compiled with ACPI_BAY=y. 
> Below is the trace of two BUGs I get.
> 
> When compiled with ACPI_BAY=n, it boots fine.

ACPI people usually prefer entries in bugzilla.kernel.org, try that if
you don't get a timely reply.

-- 
Thanks for all the (sleeping) penguins.

Re: ACPI bay - 2.6.20-rc3-mm1 hangs on boot

[Kristen Carlson Accardi]

On Fri, 5 Jan 2007 14:19:41 +0100 (CET)
Jiri Kosina <jikos@jikos.cz> wrote:

> Hi,
> 
> 2.6.20-rc3-mm1 hangs on boot on my IBM T42p when compiled with ACPI_BAY=y. 
> Below is the trace of two BUGs I get.

Hi,
I was able to duplicate this problem - it only occurs when the bay
driver is built-in, not when the bay driver is compiled as a module.
While I work on the fix for the problem, you can as a temporary workaround
just use ACPI_BAY=m.  By the way, I also found that the bay driver will
not build at all if ACPI_DOCK=n.  Oops.  I will fix that too.

thanks,
Kristen


> 
> When compiled with ACPI_BAY=n, it boots fine.
> 
> The traces are hand-rewritten (no serial console on that machine), so I 
> have omitted the code offsets in the stacktraces.
> 
> ACPI: ACPI Dock Station Driver
> ACPI: \_SB_PCI0.IDE0.SCND.MSTR: found ejectable bay
> ACPI: \_SB_PCI0.IDE0.SCND.MSTR: Adding notify handler
> PM: Adding info for platform:bay.0
> ACPI: Bay [\_SB_.PCI0.IDE0.SCND.MSTR] Added
> BUG: at lib/kref.c:32 kref_get()
> 	kref_get+0x34/0x3b
> 	kobject_get+0xf/0x13
> 	get_bus+0xe/0x1d
> 	bus_add_driver+0x13/0x165
> 	init_waitqueue_head+0x12/0x1e
> 	bay_init+0x57/0x79
> 	find_bay+0x0/0x2c4
> 	init+0x88/0x16d
> 	restore_nocheck+0x12/0x15
> 	init+0x0/0x16d
> 	init+0x0/0x16d
> 	kernel_thread_helper+0x7/0x10
> 	==========
> BUG: at lib/kref.c:32 kref_get()
> 	kref_get+0x34/0x3b
> 	kobject_get+0xf/0x13
> 	kobject_init+0x5d/0x70
> 	kobject_set_name+0x5c/0x92
> 	bus_add_driver+0x50/0x79
> 	bay_init+0x57/0x79
> 	find_bay+0x0/0x2c4
> 	init+0x88/0x16d
> 	init+0x88/0x16d
> 	kernel_thread_helper+0x7/0x10
> 	=========
> 
> -- 
> Jiri Kosina
>