refcounting problems with ibm_acpi but acpi=off

refcounting problems with ibm_acpi but acpi=off

[Pavel Machek]

Hi!

I get this nastyness in syslog if I compile ibm_acpi into kernel, then
boot with acpi=off.

floppy0: no floppy controllers found
loop: module loaded
WARNING: at lib/kref.c:33 kref_get()
 [<c025bd80>] kref_get+0x40/0x50
 [<c025aeff>] kobject_get+0xf/0x20
 [<c0325eee>] get_driver+0xe/0x20
 [<c0325f13>] driver_remove_file+0x13/0x40
 [<c0337812>] thinkpad_acpi_module_exit+0x92/0x100
 [<c083a192>] thinkpad_acpi_module_init+0x252/0x810
 [<c025a71a>] idr_get_new+0xa/0x30
 [<c01ad15d>] proc_register+0x8d/0x100
 [<c081e570>] kernel_init+0x140/0x320
 [<c01040ca>] ret_from_fork+0x6/0x1c
 [<c081e430>] kernel_init+0x0/0x320
 [<c081e430>] kernel_init+0x0/0x320
 [<c0104d8f>] kernel_thread_helper+0x7/0x18
 =======================
WARNING: at lib/kref.c:33 kref_get()
 [<c025bd80>] kref_get+0x40/0x50
 [<c025aeff>] kobject_get+0xf/0x20
 [<c0325eee>] get_driver+0xe/0x20
 [<c0325f13>] driver_remove_file+0x13/0x40
 [<c0337812>] thinkpad_acpi_module_exit+0x92/0x100
 [<c083a192>] thinkpad_acpi_module_init+0x252/0x810
 [<c025a71a>] idr_get_new+0xa/0x30
 [<c01ad15d>] proc_register+0x8d/0x100
 [<c081e570>] kernel_init+0x140/0x320
 [<c01040ca>] ret_from_fork+0x6/0x1c
 [<c081e430>] kernel_init+0x0/0x320
 [<c081e430>] kernel_init+0x0/0x320
 [<c0104d8f>] kernel_thread_helper+0x7/0x18
 =======================
WARNING: at lib/kref.c:33 kref_get()
 [<c025bd80>] kref_get+0x40/0x50
 [<c025aeff>] kobject_get+0xf/0x20
 [<c0325eee>] get_driver+0xe/0x20
 [<c0325f13>] driver_remove_file+0x13/0x40
 [<c0337812>] thinkpad_acpi_module_exit+0x92/0x100
 [<c083a192>] thinkpad_acpi_module_init+0x252/0x810
 [<c025a71a>] idr_get_new+0xa/0x30
 [<c01ad15d>] proc_register+0x8d/0x100
 [<c081e570>] kernel_init+0x140/0x320
 [<c01040ca>] ret_from_fork+0x6/0x1c
 [<c081e430>] kernel_init+0x0/0x320
 [<c081e430>] kernel_init+0x0/0x320
 [<c0104d8f>] kernel_thread_helper+0x7/0x18
 =======================
Intel(R) PRO/1000 Network Driver - version 7.3.20-k2
Copyright (c) 1999-2006 Intel Corporation.
PCI: Found IRQ 11 for device 0000:02:00.0
PCI: Sharing IRQ 11 with 0000:00:1c.0

...

Perhaps this is needed?
								Pavel

diff --git a/drivers/acpi/ibm_acpi.c b/drivers/acpi/ibm_acpi.c
index a664f2b..654a124 100644
--- a/drivers/acpi/ibm_acpi.c
+++ b/drivers/acpi/ibm_acpi.c
@@ -2695,6 +2695,9 @@ static void acpi_ibm_exit(void)
 {
 	int i;
 
+	if (acpi_disabled)
+		return;
+
 	for (i = ARRAY_SIZE(ibms) - 1; i >= 0; i--)
 		ibm_exit(&ibms[i]);
 

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: refcounting problems with ibm_acpi but acpi=off

[Pavel Machek]

Hi!

> I get this nastyness in syslog if I compile ibm_acpi into kernel, then
> boot with acpi=off.

> Perhaps this is needed?

Indeed the patch fixes it. Signed-off-by: Pavel Machek <pavel@suse.cz>

> 
> diff --git a/drivers/acpi/ibm_acpi.c b/drivers/acpi/ibm_acpi.c
> index a664f2b..654a124 100644
> --- a/drivers/acpi/ibm_acpi.c
> +++ b/drivers/acpi/ibm_acpi.c
> @@ -2695,6 +2695,9 @@ static void acpi_ibm_exit(void)
>  {
>  	int i;
>  
> +	if (acpi_disabled)
> +		return;
> +
>  	for (i = ARRAY_SIZE(ibms) - 1; i >= 0; i--)
>  		ibm_exit(&ibms[i]);
>  
> 

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: refcounting problems with ibm_acpi but acpi=off

[Henrique de Moraes Holschuh]

On Tue, 07 Aug 2007, Pavel Machek wrote:
> Perhaps this is needed?
> 								Pavel
> 
> diff --git a/drivers/acpi/ibm_acpi.c b/drivers/acpi/ibm_acpi.c
> index a664f2b..654a124 100644
> --- a/drivers/acpi/ibm_acpi.c
> +++ b/drivers/acpi/ibm_acpi.c
> @@ -2695,6 +2695,9 @@ static void acpi_ibm_exit(void)
>  {
>  	int i;
>  
> +	if (acpi_disabled)
> +		return;
> +
>  	for (i = ARRAY_SIZE(ibms) - 1; i >= 0; i--)
>  		ibm_exit(&ibms[i]);
>  
> 

Indeed it is required, my bad.

Acked-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>

Do you want me to forward a git patch to Len brown?

-- 
  "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: refcounting problems with ibm_acpi but acpi=off

[Pavel Machek]

Hi!

> > --- a/drivers/acpi/ibm_acpi.c
> > +++ b/drivers/acpi/ibm_acpi.c
> > @@ -2695,6 +2695,9 @@ static void acpi_ibm_exit(void)
> >  {
> >  	int i;
> >  
> > +	if (acpi_disabled)
> > +		return;
> > +
> >  	for (i = ARRAY_SIZE(ibms) - 1; i >= 0; i--)
> >  		ibm_exit(&ibms[i]);
> >  
> > 
> 
> Indeed it is required, my bad.
> 
> Acked-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
> 
> Do you want me to forward a git patch to Len brown?

Yes, please. Hmm, seems like at least asus_acpi is affected by same problem.
							Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: refcounting problems with ibm_acpi but acpi=off

[Henrique de Moraes Holschuh]

On Tue, 07 Aug 2007, Pavel Machek wrote:
> > > --- a/drivers/acpi/ibm_acpi.c
> > > +++ b/drivers/acpi/ibm_acpi.c
> > > @@ -2695,6 +2695,9 @@ static void acpi_ibm_exit(void)
> > >  {
> > >  	int i;
> > >  
> > > +	if (acpi_disabled)
> > > +		return;
> > > +
> > >  	for (i = ARRAY_SIZE(ibms) - 1; i >= 0; i--)
> > >  		ibm_exit(&ibms[i]);
> > >  
> > > 
> > 
> > Indeed it is required, my bad.
> > 
> > Acked-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
> > 
> > Do you want me to forward a git patch to Len brown?
> 
> Yes, please. Hmm, seems like at least asus_acpi is affected by same problem.

I just tested it, and thinkpad-acpi with all accepted patches by Len is not
affected by this problem, only some versions of ibm-acpi is.  So there's
nothing to send to Len.

-- 
  "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