[Regression] 2.6.37-rc1: Lenovo T61 mute button does not work

[Regression] 2.6.37-rc1: Lenovo T61 mute button does not work

[Lukas Hejtmanek]

Hi,

in 2.6.26, mute button worked as expected on my ThinkPad T61 - it toggled
mute with mixer applet notification. In 2.6.37-rc1, it only mutes without any
event to any mixer applet.

-- 
Lukáš Hejtmánek

Re: [Regression] 2.6.37-rc1: Lenovo T61 mute button does not work

[Calvin Walton]

On Tue, 2010-11-30 at 19:53 +0100, Lukas Hejtmanek wrote:
> Hi,
> 
> in 2.6.26, mute button worked as expected on my ThinkPad T61 - it toggled
> mute with mixer applet notification. In 2.6.37-rc1, it only mutes without any
> event to any mixer applet.

The mute button on this series of Thinkpads is a little weird (I have an
R61i, which is the same iirc) - the behaviour of the button depends on
the ACPI firmware. With no operating system running, it behaves as a
persistent hardware mute. Windows apparently uses a proprietary driver
to control the volume - part of this functionality has been added to the
Linux thinkpad_acpi driver, but it's not fully enabled yet.

By default on this series of laptops, Linux runs with the ACPI
_OSI(Linux) exported - when the bios of this laptop model sees that it's
running on Linux, it disables the hardware mute functionality and the
mute button acts as a standard keyboard key.

Search your dmesg for lines containing '_OSI(Linux)' - it should say
that a BIOS query was honoured. Make sure you do not have 'acpi_osi="!
Linux"' on your kernel command line.

-- 
Calvin Walton <calvin.walton@gmail.com>

Re: [Regression] 2.6.37-rc1: Lenovo T61 mute button does not work

[Lukas Hejtmanek]

On Tue, Nov 30, 2010 at 09:58:44PM -0500, Calvin Walton wrote:
> The mute button on this series of Thinkpads is a little weird (I have an
> R61i, which is the same iirc) - the behaviour of the button depends on
> the ACPI firmware. With no operating system running, it behaves as a
> persistent hardware mute. Windows apparently uses a proprietary driver
> to control the volume - part of this functionality has been added to the
> Linux thinkpad_acpi driver, but it's not fully enabled yet.

Behaviour in 2.6.36 kernel is OK as expected. Something broken it in
2.6.37-rc1.

> Search your dmesg for lines containing '_OSI(Linux)' - it should say
> that a BIOS query was honoured. Make sure you do not have 'acpi_osi="!
> Linux"' on your kernel command line.

[    0.000000] ACPI: DMI detected: Lenovo ThinkPad T61
[    0.000000] ACPI: Added _OSI(Linux)

looks good.

-- 
Lukáš Hejtmánek

Re: [Regression] 2.6.37-rc1: Lenovo T61 mute button does not work

[Lukas Hejtmanek]

On Tue, Nov 30, 2010 at 09:58:44PM -0500, Calvin Walton wrote:
> Search your dmesg for lines containing '_OSI(Linux)' - it should say
> that a BIOS query was honoured. Make sure you do not have 'acpi_osi="!
> Linux"' on your kernel command line.

the difference between working and non working seems to be in:

ACPI: BIOS _OSI(Linux) query honored via DMI

and 

ACPI: BIOS _OSI(Linux) query honored via cmdline


via DMI seems to be working as expected, via cmdline does not work. Any ideas?

-- 
Lukáš Hejtmánek

Re: [Regression] 2.6.37-rc1: Lenovo T61 mute button does not work

[Lukas Hejtmanek]

Hi,

On Tue, Nov 30, 2010 at 09:58:44PM -0500, Calvin Walton wrote:
> > in 2.6.36, mute button worked as expected on my ThinkPad T61 - it toggled
> > mute with mixer applet notification. In 2.6.37-rc1, it only mutes without any
> > event to any mixer applet.

[...]

> By default on this series of laptops, Linux runs with the ACPI
> _OSI(Linux) exported - when the bios of this laptop model sees that it's
> running on Linux, it disables the hardware mute functionality and the
> mute button acts as a standard keyboard key.
> 
> Search your dmesg for lines containing '_OSI(Linux)' - it should say
> that a BIOS query was honoured. Make sure you do not have 'acpi_osi="!
> Linux"' on your kernel command line.

The breaking commit is this one: b0ed7a915abac309fcb5a51bccd3782e3daa7417

This changes 
[    0.200241] ACPI: BIOS _OSI(Linux) query honored via DMI
to
[    0.200241] ACPI: BIOS _OSI(Linux) query honored via cmdline

which apparently does not work with Lenovo T61.

-- 
Lukáš Hejtmánek

Re: [Regression] 2.6.37-rc1: Lenovo T61 mute button does not work

[Lin Ming]

On Fri, 2010-12-03 at 22:13 +0800, Lukas Hejtmanek wrote:
> Hi,
> 
> On Tue, Nov 30, 2010 at 09:58:44PM -0500, Calvin Walton wrote:
> > > in 2.6.36, mute button worked as expected on my ThinkPad T61 - it toggled
> > > mute with mixer applet notification. In 2.6.37-rc1, it only mutes without any
> > > event to any mixer applet.
> 
> [...]
> 
> > By default on this series of laptops, Linux runs with the ACPI
> > _OSI(Linux) exported - when the bios of this laptop model sees that it's
> > running on Linux, it disables the hardware mute functionality and the
> > mute button acts as a standard keyboard key.
> > 
> > Search your dmesg for lines containing '_OSI(Linux)' - it should say
> > that a BIOS query was honoured. Make sure you do not have 'acpi_osi="!
> > Linux"' on your kernel command line.
> 
> The breaking commit is this one: b0ed7a915abac309fcb5a51bccd3782e3daa7417
> 
> This changes 
> [    0.200241] ACPI: BIOS _OSI(Linux) query honored via DMI
> to
> [    0.200241] ACPI: BIOS _OSI(Linux) query honored via cmdline
> 
> which apparently does not work with Lenovo T61.

I'm looking at this regression.

Thanks,
Lin Ming

Re: [Regression] 2.6.37-rc1: Lenovo T61 mute button does not work

[Lukas Hejtmanek]

On Mon, Dec 06, 2010 at 09:37:04AM +0800, Lin Ming wrote:
> > which apparently does not work with Lenovo T61.
> 
> I'm looking at this regression.

I did some debugging. It looks like this:
[    0.000000] ACPI: DMI detected: Lenovo ThinkPad T61
[    0.000000] ACPI: Set_osi_linux Added _OSI(Linux)
[    0.000000] Pid: 0, comm: swapper Not tainted 2.6.37-rc3+ #122
[    0.000000] Call Trace:
[    0.000000]  [<ffffffff8162c0a1>] ? set_osi_linux+0x4b/0x6b
[    0.000000]  [<ffffffff8162bbe7>] ? dmi_enable_osi_linux+0x11/0x18
[    0.000000]  [<ffffffff812f7313>] ? dmi_check_system+0x33/0x50
[    0.000000]  [<ffffffff8162bdc2>] ? acpi_blacklisted+0x16a/0x190
[    0.000000]  [<ffffffff81617b6e>] ? acpi_boot_table_init+0x3a/0x75
[    0.000000]  [<ffffffff81611cf3>] ? setup_arch+0x6d5/0x7a8
[    0.000000]  [<ffffffff814072f2>] ? printk+0x40/0x46
[    0.000000]  [<ffffffff816108de>] ? start_kernel+0xbe/0x323
[    0.000000]  [<ffffffff816103de>] ? x86_64_start_kernel+0xf9/0xff

after this call, osi_linux.dmi is set.

However, after this call:
[    0.186721] Call Trace:
[    0.186721]  [<ffffffff8120ecad>] ? acpi_install_interface_handler+0x4d/0x56
[    0.186750]  [<ffffffff8162c0a1>] ? set_osi_linux+0x4b/0x6b
[    0.186777]  [<ffffffff811edefa>] ? acpi_os_initialize1+0x10f/0x13d
[    0.186803]  [<ffffffff8162c3fa>] ? acpi_bus_init+0xa/0x1e2
[    0.186829]  [<ffffffff811c498c>] ? kobject_create_and_add+0x3c/0x80
[    0.186855]  [<ffffffff8162c63b>] ? acpi_init+0x69/0xdf
[    0.186880]  [<ffffffff810002da>] ? do_one_initcall+0x3a/0x170
[    0.186906]  [<ffffffff81610c75>] ? kernel_init+0x132/0x1be
[    0.186932]  [<ffffffff81003b94>] ? kernel_thread_helper+0x4/0x10
[    0.186958]  [<ffffffff81610b43>] ? kernel_init+0x0/0x1be
[    0.187577]  [<ffffffff81003b90>] ? kernel_thread_helper+0x0/0x10

osi_linux.cmdline is set.

-- 
Lukáš Hejtmánek

Re: [Regression] 2.6.37-rc1: Lenovo T61 mute button does not work

[Lin Ming]

On Mon, 2010-12-06 at 17:31 +0800, Lukas Hejtmanek wrote:
> On Mon, Dec 06, 2010 at 09:37:04AM +0800, Lin Ming wrote:
> > > which apparently does not work with Lenovo T61.
> > 
> > I'm looking at this regression.
> 
> I did some debugging. It looks like this:
> [    0.000000] ACPI: DMI detected: Lenovo ThinkPad T61
> [    0.000000] ACPI: Set_osi_linux Added _OSI(Linux)
> [    0.000000] Pid: 0, comm: swapper Not tainted 2.6.37-rc3+ #122
> [    0.000000] Call Trace:
> [    0.000000]  [<ffffffff8162c0a1>] ? set_osi_linux+0x4b/0x6b
> [    0.000000]  [<ffffffff8162bbe7>] ? dmi_enable_osi_linux+0x11/0x18
> [    0.000000]  [<ffffffff812f7313>] ? dmi_check_system+0x33/0x50
> [    0.000000]  [<ffffffff8162bdc2>] ? acpi_blacklisted+0x16a/0x190
> [    0.000000]  [<ffffffff81617b6e>] ? acpi_boot_table_init+0x3a/0x75
> [    0.000000]  [<ffffffff81611cf3>] ? setup_arch+0x6d5/0x7a8
> [    0.000000]  [<ffffffff814072f2>] ? printk+0x40/0x46
> [    0.000000]  [<ffffffff816108de>] ? start_kernel+0xbe/0x323
> [    0.000000]  [<ffffffff816103de>] ? x86_64_start_kernel+0xf9/0xff
> 
> after this call, osi_linux.dmi is set.
> 
> However, after this call:
> [    0.186721] Call Trace:
> [    0.186721]  [<ffffffff8120ecad>] ? acpi_install_interface_handler+0x4d/0x56
> [    0.186750]  [<ffffffff8162c0a1>] ? set_osi_linux+0x4b/0x6b
> [    0.186777]  [<ffffffff811edefa>] ? acpi_os_initialize1+0x10f/0x13d
> [    0.186803]  [<ffffffff8162c3fa>] ? acpi_bus_init+0xa/0x1e2
> [    0.186829]  [<ffffffff811c498c>] ? kobject_create_and_add+0x3c/0x80
> [    0.186855]  [<ffffffff8162c63b>] ? acpi_init+0x69/0xdf
> [    0.186880]  [<ffffffff810002da>] ? do_one_initcall+0x3a/0x170
> [    0.186906]  [<ffffffff81610c75>] ? kernel_init+0x132/0x1be
> [    0.186932]  [<ffffffff81003b94>] ? kernel_thread_helper+0x4/0x10
> [    0.186958]  [<ffffffff81610b43>] ? kernel_init+0x0/0x1be
> [    0.187577]  [<ffffffff81003b90>] ? kernel_thread_helper+0x0/0x10
> 
> osi_linux.cmdline is set.

Do you have acpi_osi=Linux in kernel boot parameters?

Please attach the dmesg.

Thanks.

Re: [Regression] 2.6.37-rc1: Lenovo T61 mute button does not work

[Lin Ming]

On Mon, 2010-12-06 at 17:31 +0800, Lukas Hejtmanek wrote:
> On Mon, Dec 06, 2010 at 09:37:04AM +0800, Lin Ming wrote:
> > > which apparently does not work with Lenovo T61.
> > 
> > I'm looking at this regression.
> 
> I did some debugging. It looks like this:
> [    0.000000] ACPI: DMI detected: Lenovo ThinkPad T61
> [    0.000000] ACPI: Set_osi_linux Added _OSI(Linux)
> [    0.000000] Pid: 0, comm: swapper Not tainted 2.6.37-rc3+ #122
> [    0.000000] Call Trace:
> [    0.000000]  [<ffffffff8162c0a1>] ? set_osi_linux+0x4b/0x6b
> [    0.000000]  [<ffffffff8162bbe7>] ? dmi_enable_osi_linux+0x11/0x18
> [    0.000000]  [<ffffffff812f7313>] ? dmi_check_system+0x33/0x50
> [    0.000000]  [<ffffffff8162bdc2>] ? acpi_blacklisted+0x16a/0x190
> [    0.000000]  [<ffffffff81617b6e>] ? acpi_boot_table_init+0x3a/0x75
> [    0.000000]  [<ffffffff81611cf3>] ? setup_arch+0x6d5/0x7a8
> [    0.000000]  [<ffffffff814072f2>] ? printk+0x40/0x46
> [    0.000000]  [<ffffffff816108de>] ? start_kernel+0xbe/0x323
> [    0.000000]  [<ffffffff816103de>] ? x86_64_start_kernel+0xf9/0xff
> 
> after this call, osi_linux.dmi is set.
> 
> However, after this call:
> [    0.186721] Call Trace:
> [    0.186721]  [<ffffffff8120ecad>] ? acpi_install_interface_handler+0x4d/0x56
> [    0.186750]  [<ffffffff8162c0a1>] ? set_osi_linux+0x4b/0x6b
> [    0.186777]  [<ffffffff811edefa>] ? acpi_os_initialize1+0x10f/0x13d
> [    0.186803]  [<ffffffff8162c3fa>] ? acpi_bus_init+0xa/0x1e2
> [    0.186829]  [<ffffffff811c498c>] ? kobject_create_and_add+0x3c/0x80
> [    0.186855]  [<ffffffff8162c63b>] ? acpi_init+0x69/0xdf
> [    0.186880]  [<ffffffff810002da>] ? do_one_initcall+0x3a/0x170
> [    0.186906]  [<ffffffff81610c75>] ? kernel_init+0x132/0x1be
> [    0.186932]  [<ffffffff81003b94>] ? kernel_thread_helper+0x4/0x10
> [    0.186958]  [<ffffffff81610b43>] ? kernel_init+0x0/0x1be
> [    0.187577]  [<ffffffff81003b90>] ? kernel_thread_helper+0x0/0x10
> 
> osi_linux.cmdline is set.

Hi, 

Would you please try below patch?

Thanks.

---
 drivers/acpi/osl.c |   55 ++++++++++++++++++++++++++++++---------------------
 1 files changed, 32 insertions(+), 23 deletions(-)

diff --git a/drivers/acpi/osl.c b/drivers/acpi/osl.c
index 966fedd..f620689 100644
--- a/drivers/acpi/osl.c
+++ b/drivers/acpi/osl.c
@@ -152,8 +152,7 @@ static struct osi_linux {
 	unsigned int	enable:1;
 	unsigned int	dmi:1;
 	unsigned int	cmdline:1;
-	unsigned int	known:1;
-} osi_linux = { 0, 0, 0, 0};
+} osi_linux = {0, 0, 0};
 
 static u32 acpi_osi_handler(acpi_string interface, u32 supported)
 {
@@ -1055,18 +1054,25 @@ static int __init acpi_os_name_setup(char *str)
 
 __setup("acpi_os_name=", acpi_os_name_setup);
 
+static void __init osi_setup(char *str)
+{
+	if (str == NULL || *str == '\0') {
+		printk(KERN_INFO PREFIX "_OSI method disabled\n");
+		acpi_gbl_create_osi_method = FALSE;
+	} else {
+		strncpy(osi_setup_string, str, OSI_STRING_LENGTH_MAX);
+	}
+}
+
 static void __init set_osi_linux(unsigned int enable)
 {
-	if (osi_linux.enable != enable) {
+	if (osi_linux.enable != enable)
 		osi_linux.enable = enable;
-		printk(KERN_NOTICE PREFIX "%sed _OSI(Linux)\n",
-			enable ? "Add": "Delet");
-	}
 
 	if (osi_linux.enable)
-		acpi_osi_setup("Linux");
+		osi_setup("Linux");
 	else
-		acpi_osi_setup("!Linux");
+		osi_setup("!Linux");
 
 	return;
 }
@@ -1088,8 +1094,6 @@ void __init acpi_dmi_osi_linux(int enable, const struct dmi_system_id *d)
 	if (enable == -1)
 		return;
 
-	osi_linux.known = 1;	/* DMI knows which OSI(Linux) default needed */
-
 	set_osi_linux(enable);
 
 	return;
@@ -1105,31 +1109,36 @@ void __init acpi_dmi_osi_linux(int enable, const struct dmi_system_id *d)
 static void __init acpi_osi_setup_late(void)
 {
 	char *str = osi_setup_string;
+	acpi_status status;
 
 	if (*str == '\0')
 		return;
 
-	if (!strcmp("!Linux", str)) {
-		acpi_cmdline_osi_linux(0);	/* !enable */
-	} else if (*str == '!') {
-		if (acpi_remove_interface(++str) == AE_OK)
+	if (*str == '!') {
+		status = acpi_remove_interface(++str);
+
+		if (ACPI_SUCCESS(status))
 			printk(KERN_INFO PREFIX "Deleted _OSI(%s)\n", str);
-	} else if (!strcmp("Linux", str)) {
-		acpi_cmdline_osi_linux(1);	/* enable */
 	} else {
-		if (acpi_install_interface(str) == AE_OK)
+		status = acpi_install_interface(str);
+
+		if (ACPI_SUCCESS(status))
 			printk(KERN_INFO PREFIX "Added _OSI(%s)\n", str);
 	}
 }
 
 int __init acpi_osi_setup(char *str)
 {
-	if (str == NULL || *str == '\0') {
-		printk(KERN_INFO PREFIX "_OSI method disabled\n");
-		acpi_gbl_create_osi_method = FALSE;
-	} else {
-		strncpy(osi_setup_string, str, OSI_STRING_LENGTH_MAX);
-	}
+	/* kernel command line overrides DMI */
+	osi_linux.dmi = 0;
+	osi_linux.enable = 0;
+
+	if (str && !strcmp("Linux", str))
+		acpi_cmdline_osi_linux(1);
+	else if (str && !strcmp("!Linux", str))
+		acpi_cmdline_osi_linux(0);
+	else
+		osi_setup(str);
 
 	return 1;
 }

Re: [Regression] 2.6.37-rc1: Lenovo T61 mute button does not work

[Lukas Hejtmanek]

On Tue, Dec 07, 2010 at 11:19:27AM +0800, Lin Ming wrote:
> Would you please try below patch?

yes, the patch fixed my problem with the mute button. Thanks.
 
> Thanks.
> 
> ---
>  drivers/acpi/osl.c |   55 ++++++++++++++++++++++++++++++---------------------
>  1 files changed, 32 insertions(+), 23 deletions(-)
> 
> diff --git a/drivers/acpi/osl.c b/drivers/acpi/osl.c
> index 966fedd..f620689 100644
> --- a/drivers/acpi/osl.c
> +++ b/drivers/acpi/osl.c
> @@ -152,8 +152,7 @@ static struct osi_linux {
>  	unsigned int	enable:1;
>  	unsigned int	dmi:1;
>  	unsigned int	cmdline:1;
> -	unsigned int	known:1;
> -} osi_linux = { 0, 0, 0, 0};
> +} osi_linux = {0, 0, 0};
>  
>  static u32 acpi_osi_handler(acpi_string interface, u32 supported)
>  {
> @@ -1055,18 +1054,25 @@ static int __init acpi_os_name_setup(char *str)
>  
>  __setup("acpi_os_name=", acpi_os_name_setup);
>  
> +static void __init osi_setup(char *str)
> +{
> +	if (str == NULL || *str == '\0') {
> +		printk(KERN_INFO PREFIX "_OSI method disabled\n");
> +		acpi_gbl_create_osi_method = FALSE;
> +	} else {
> +		strncpy(osi_setup_string, str, OSI_STRING_LENGTH_MAX);
> +	}
> +}
> +
>  static void __init set_osi_linux(unsigned int enable)
>  {
> -	if (osi_linux.enable != enable) {
> +	if (osi_linux.enable != enable)
>  		osi_linux.enable = enable;
> -		printk(KERN_NOTICE PREFIX "%sed _OSI(Linux)\n",
> -			enable ? "Add": "Delet");
> -	}
>  
>  	if (osi_linux.enable)
> -		acpi_osi_setup("Linux");
> +		osi_setup("Linux");
>  	else
> -		acpi_osi_setup("!Linux");
> +		osi_setup("!Linux");
>  
>  	return;
>  }
> @@ -1088,8 +1094,6 @@ void __init acpi_dmi_osi_linux(int enable, const struct dmi_system_id *d)
>  	if (enable == -1)
>  		return;
>  
> -	osi_linux.known = 1;	/* DMI knows which OSI(Linux) default needed */
> -
>  	set_osi_linux(enable);
>  
>  	return;
> @@ -1105,31 +1109,36 @@ void __init acpi_dmi_osi_linux(int enable, const struct dmi_system_id *d)
>  static void __init acpi_osi_setup_late(void)
>  {
>  	char *str = osi_setup_string;
> +	acpi_status status;
>  
>  	if (*str == '\0')
>  		return;
>  
> -	if (!strcmp("!Linux", str)) {
> -		acpi_cmdline_osi_linux(0);	/* !enable */
> -	} else if (*str == '!') {
> -		if (acpi_remove_interface(++str) == AE_OK)
> +	if (*str == '!') {
> +		status = acpi_remove_interface(++str);
> +
> +		if (ACPI_SUCCESS(status))
>  			printk(KERN_INFO PREFIX "Deleted _OSI(%s)\n", str);
> -	} else if (!strcmp("Linux", str)) {
> -		acpi_cmdline_osi_linux(1);	/* enable */
>  	} else {
> -		if (acpi_install_interface(str) == AE_OK)
> +		status = acpi_install_interface(str);
> +
> +		if (ACPI_SUCCESS(status))
>  			printk(KERN_INFO PREFIX "Added _OSI(%s)\n", str);
>  	}
>  }
>  
>  int __init acpi_osi_setup(char *str)
>  {
> -	if (str == NULL || *str == '\0') {
> -		printk(KERN_INFO PREFIX "_OSI method disabled\n");
> -		acpi_gbl_create_osi_method = FALSE;
> -	} else {
> -		strncpy(osi_setup_string, str, OSI_STRING_LENGTH_MAX);
> -	}
> +	/* kernel command line overrides DMI */
> +	osi_linux.dmi = 0;
> +	osi_linux.enable = 0;
> +
> +	if (str && !strcmp("Linux", str))
> +		acpi_cmdline_osi_linux(1);
> +	else if (str && !strcmp("!Linux", str))
> +		acpi_cmdline_osi_linux(0);
> +	else
> +		osi_setup(str);
>  
>  	return 1;
>  }
> 

-- 
Lukáš Hejtmánek