[PATCH] acpi: Unconditionally set SCI_EN on resume

[PATCH] acpi: Unconditionally set SCI_EN on resume

[Matthew Garrett]

The ACPI spec tells us that the firmware will reenable SCI_EN on resume.
Reality disagrees in some cases. The ACPI spec tells us that the only way
to set SCI_EN is via an SMM call.
https://bugzilla.kernel.org/show_bug.cgi?id=13745 shows us that doing so
may break machines. Tracing the ACPI calls made by Windows shows that it
unconditionally sets SCI_EN on resume with a direct register write, and
therefore the overwhelming probability is that everything is fine with
this behaviour.

Signed-off-by: Matthew Garrett <mjg@redhat.com>
---
 arch/x86/kernel/acpi/sleep.c |    2 -
 drivers/acpi/sleep.c         |  157 +-----------------------------------------
 include/linux/acpi.h         |    1 -
 3 files changed, 2 insertions(+), 158 deletions(-)

diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c
index f996103..82e5086 100644
--- a/arch/x86/kernel/acpi/sleep.c
+++ b/arch/x86/kernel/acpi/sleep.c
@@ -162,8 +162,6 @@ static int __init acpi_sleep_setup(char *str)
 #endif
 		if (strncmp(str, "old_ordering", 12) == 0)
 			acpi_old_suspend_ordering();
-		if (strncmp(str, "sci_force_enable", 16) == 0)
-			acpi_set_sci_en_on_resume();
 		str = strchr(str, ',');
 		if (str != NULL)
 			str += strspn(str, ", \t");
diff --git a/drivers/acpi/sleep.c b/drivers/acpi/sleep.c
index baa76bb..4ab2275 100644
--- a/drivers/acpi/sleep.c
+++ b/drivers/acpi/sleep.c
@@ -80,22 +80,6 @@ static int acpi_sleep_prepare(u32 acpi_state)
 
 #ifdef CONFIG_ACPI_SLEEP
 static u32 acpi_target_sleep_state = ACPI_STATE_S0;
-/*
- * According to the ACPI specification the BIOS should make sure that ACPI is
- * enabled and SCI_EN bit is set on wake-up from S1 - S3 sleep states.  Still,
- * some BIOSes don't do that and therefore we use acpi_enable() to enable ACPI
- * on such systems during resume.  Unfortunately that doesn't help in
- * particularly pathological cases in which SCI_EN has to be set directly on
- * resume, although the specification states very clearly that this flag is
- * owned by the hardware.  The set_sci_en_on_resume variable will be set in such
- * cases.
- */
-static bool set_sci_en_on_resume;
-
-void __init acpi_set_sci_en_on_resume(void)
-{
-	set_sci_en_on_resume = true;
-}
 
 /*
  * ACPI 1.0 wants us to execute _PTS before suspending devices, so we allow the
@@ -253,11 +237,8 @@ static int acpi_suspend_enter(suspend_state_t pm_state)
 		break;
 	}
 
-	/* If ACPI is not enabled by the BIOS, we need to enable it here. */
-	if (set_sci_en_on_resume)
-		acpi_write_bit_register(ACPI_BITREG_SCI_ENABLE, 1);
-	else
-		acpi_enable();
+	/* This violates the spec but is required for bug compatibility. */
+	acpi_write_bit_register(ACPI_BITREG_SCI_ENABLE, 1);
 
 	/* Reprogram control registers and execute _BFS */
 	acpi_leave_sleep_state_prep(acpi_state);
@@ -346,12 +327,6 @@ static int __init init_old_suspend_ordering(const struct dmi_system_id *d)
 	return 0;
 }
 
-static int __init init_set_sci_en_on_resume(const struct dmi_system_id *d)
-{
-	set_sci_en_on_resume = true;
-	return 0;
-}
-
 static struct dmi_system_id __initdata acpisleep_dmi_table[] = {
 	{
 	.callback = init_old_suspend_ordering,
@@ -370,22 +345,6 @@ static struct dmi_system_id __initdata acpisleep_dmi_table[] = {
 		},
 	},
 	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Apple MacBook 1,1",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "Apple Computer, Inc."),
-		DMI_MATCH(DMI_PRODUCT_NAME, "MacBook1,1"),
-		},
-	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Apple MacMini 1,1",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "Apple Computer, Inc."),
-		DMI_MATCH(DMI_PRODUCT_NAME, "Macmini1,1"),
-		},
-	},
-	{
 	.callback = init_old_suspend_ordering,
 	.ident = "Asus Pundit P1-AH2 (M2N8L motherboard)",
 	.matches = {
@@ -394,94 +353,6 @@ static struct dmi_system_id __initdata acpisleep_dmi_table[] = {
 		},
 	},
 	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Toshiba Satellite L300",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "TOSHIBA"),
-		DMI_MATCH(DMI_PRODUCT_NAME, "Satellite L300"),
-		},
-	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Hewlett-Packard HP G7000 Notebook PC",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"),
-		DMI_MATCH(DMI_PRODUCT_NAME, "HP G7000 Notebook PC"),
-		},
-	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Hewlett-Packard HP Pavilion dv3 Notebook PC",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"),
-		DMI_MATCH(DMI_PRODUCT_NAME, "HP Pavilion dv3 Notebook PC"),
-		},
-	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Hewlett-Packard Pavilion dv4",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"),
-		DMI_MATCH(DMI_PRODUCT_NAME, "HP Pavilion dv4"),
-		},
-	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Hewlett-Packard Pavilion dv7",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"),
-		DMI_MATCH(DMI_PRODUCT_NAME, "HP Pavilion dv7"),
-		},
-	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Hewlett-Packard Compaq Presario C700 Notebook PC",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"),
-		DMI_MATCH(DMI_PRODUCT_NAME, "Compaq Presario C700 Notebook PC"),
-		},
-	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Hewlett-Packard Compaq Presario CQ40 Notebook PC",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"),
-		DMI_MATCH(DMI_PRODUCT_NAME, "Compaq Presario CQ40 Notebook PC"),
-		},
-	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Lenovo ThinkPad T410",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "LENOVO"),
-		DMI_MATCH(DMI_PRODUCT_VERSION, "ThinkPad T410"),
-		},
-	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Lenovo ThinkPad T510",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "LENOVO"),
-		DMI_MATCH(DMI_PRODUCT_VERSION, "ThinkPad T510"),
-		},
-	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Lenovo ThinkPad W510",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "LENOVO"),
-		DMI_MATCH(DMI_PRODUCT_VERSION, "ThinkPad W510"),
-		},
-	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Lenovo ThinkPad X201[s]",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "LENOVO"),
-		DMI_MATCH(DMI_PRODUCT_VERSION, "ThinkPad X201"),
-		},
-	},
-	{
 	.callback = init_old_suspend_ordering,
 	.ident = "Panasonic CF51-2L",
 	.matches = {
@@ -490,30 +361,6 @@ static struct dmi_system_id __initdata acpisleep_dmi_table[] = {
 		DMI_MATCH(DMI_BOARD_NAME, "CF51-2L"),
 		},
 	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Dell Studio 1558",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
-		DMI_MATCH(DMI_PRODUCT_NAME, "Studio 1558"),
-		},
-	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Dell Studio 1557",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
-		DMI_MATCH(DMI_PRODUCT_NAME, "Studio 1557"),
-		},
-	},
-	{
-	.callback = init_set_sci_en_on_resume,
-	.ident = "Dell Studio 1555",
-	.matches = {
-		DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
-		DMI_MATCH(DMI_PRODUCT_NAME, "Studio 1555"),
-		},
-	},
 	{},
 };
 #endif /* CONFIG_SUSPEND */
diff --git a/include/linux/acpi.h b/include/linux/acpi.h
index b926afe..87ca491 100644
--- a/include/linux/acpi.h
+++ b/include/linux/acpi.h
@@ -251,7 +251,6 @@ int acpi_check_mem_region(resource_size_t start, resource_size_t n,
 void __init acpi_no_s4_hw_signature(void);
 void __init acpi_old_suspend_ordering(void);
 void __init acpi_s4_no_nvs(void);
-void __init acpi_set_sci_en_on_resume(void);
 #endif /* CONFIG_PM_SLEEP */
 
 struct acpi_osc_context {
-- 
1.7.0.1

Re: [PATCH] acpi: Unconditionally set SCI_EN on resume

[Rafael J. Wysocki]

On Tuesday 11 May 2010, Matthew Garrett wrote:
> The ACPI spec tells us that the firmware will reenable SCI_EN on resume.
> Reality disagrees in some cases. The ACPI spec tells us that the only way
> to set SCI_EN is via an SMM call.
> https://bugzilla.kernel.org/show_bug.cgi?id=13745 shows us that doing so
> may break machines. Tracing the ACPI calls made by Windows shows that it
> unconditionally sets SCI_EN on resume with a direct register write, and
> therefore the overwhelming probability is that everything is fine with
> this behaviour.
> 
> Signed-off-by: Matthew Garrett <mjg@redhat.com>
> ---
>  arch/x86/kernel/acpi/sleep.c |    2 -
>  drivers/acpi/sleep.c         |  157 +-----------------------------------------
>  include/linux/acpi.h         |    1 -
>  3 files changed, 2 insertions(+), 158 deletions(-)
> 
> diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c
> index f996103..82e5086 100644
> --- a/arch/x86/kernel/acpi/sleep.c
> +++ b/arch/x86/kernel/acpi/sleep.c
> @@ -162,8 +162,6 @@ static int __init acpi_sleep_setup(char *str)
>  #endif
>  		if (strncmp(str, "old_ordering", 12) == 0)
>  			acpi_old_suspend_ordering();
> -		if (strncmp(str, "sci_force_enable", 16) == 0)
> -			acpi_set_sci_en_on_resume();
>  		str = strchr(str, ',');
>  		if (str != NULL)
>  			str += strspn(str, ", \t");
> diff --git a/drivers/acpi/sleep.c b/drivers/acpi/sleep.c
> index baa76bb..4ab2275 100644
> --- a/drivers/acpi/sleep.c
> +++ b/drivers/acpi/sleep.c
> @@ -80,22 +80,6 @@ static int acpi_sleep_prepare(u32 acpi_state)
>  
>  #ifdef CONFIG_ACPI_SLEEP
>  static u32 acpi_target_sleep_state = ACPI_STATE_S0;
> -/*
> - * According to the ACPI specification the BIOS should make sure that ACPI is
> - * enabled and SCI_EN bit is set on wake-up from S1 - S3 sleep states.  Still,
> - * some BIOSes don't do that and therefore we use acpi_enable() to enable ACPI
> - * on such systems during resume.  Unfortunately that doesn't help in
> - * particularly pathological cases in which SCI_EN has to be set directly on
> - * resume, although the specification states very clearly that this flag is
> - * owned by the hardware.  The set_sci_en_on_resume variable will be set in such
> - * cases.
> - */
> -static bool set_sci_en_on_resume;
> -
> -void __init acpi_set_sci_en_on_resume(void)
> -{
> -	set_sci_en_on_resume = true;
> -}
>  
>  /*
>   * ACPI 1.0 wants us to execute _PTS before suspending devices, so we allow the
> @@ -253,11 +237,8 @@ static int acpi_suspend_enter(suspend_state_t pm_state)
>  		break;
>  	}
>  
> -	/* If ACPI is not enabled by the BIOS, we need to enable it here. */
> -	if (set_sci_en_on_resume)
> -		acpi_write_bit_register(ACPI_BITREG_SCI_ENABLE, 1);
> -	else
> -		acpi_enable();
> +	/* This violates the spec but is required for bug compatibility. */
> +	acpi_write_bit_register(ACPI_BITREG_SCI_ENABLE, 1);

No, this breaks my MSI Wind U100 right away.

If you called acpi_enable() before that and do that depending on its result,
_that_ would work.

Thanks,
Rafael

Re: [PATCH] acpi: Unconditionally set SCI_EN on resume

[Matthew Garrett]

On Tue, May 11, 2010 at 08:27:36PM +0200, Rafael J. Wysocki wrote:

> No, this breaks my MSI Wind U100 right away.
> 
> If you called acpi_enable() before that and do that depending on its result,
> _that_ would work.

Calling acpi_enable breaks some HPs. What version of Windows did the 
Wind ship with?

-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: [PATCH] acpi: Unconditionally set SCI_EN on resume

[Rafael J. Wysocki]

On Tuesday 11 May 2010, Matthew Garrett wrote:
> On Tue, May 11, 2010 at 08:27:36PM +0200, Rafael J. Wysocki wrote:
> 
> > No, this breaks my MSI Wind U100 right away.
> > 
> > If you called acpi_enable() before that and do that depending on its result,
> > _that_ would work.
> 
> Calling acpi_enable breaks some HPs. What version of Windows did the 
> Wind ship with?

Without any.  It shipped with SLED 10. :-)

Rafael

Re: [PATCH] acpi: Unconditionally set SCI_EN on resume

[Matthew Garrett]

On Tue, May 11, 2010 at 08:41:02PM +0200, Rafael J. Wysocki wrote:
> On Tuesday 11 May 2010, Matthew Garrett wrote:
> > On Tue, May 11, 2010 at 08:27:36PM +0200, Rafael J. Wysocki wrote:
> > 
> > > No, this breaks my MSI Wind U100 right away.
> > > 
> > > If you called acpi_enable() before that and do that depending on its result,
> > > _that_ would work.
> > 
> > Calling acpi_enable breaks some HPs. What version of Windows did the 
> > Wind ship with?
> 
> Without any.  It shipped with SLED 10. :-)

Can you dump the DSDT? It's possible that this behaviour varies 
depending on the OSI queries.

-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: [PATCH] acpi: Unconditionally set SCI_EN on resume

[Len Brown]

applied to acpi #test (replacing previous patch set on this topic)

However, s/acpi/ACPI/  :-)

thanks,
Len Brown, Intel Open Source Technology Center