[PATCH] /sys/firmware/efi/systab giving incorrect value for smbios

[PATCH] /sys/firmware/efi/systab giving incorrect value for smbios

[Matthew Garrett]

On my Intel imac, /sys/firmware/efi/systab is the following:

ACPI20=0x1fefd014
ACPI=0x1fefd000
SMBIOS=0x9fec9000

if I have a kernel with a 2:2 user/kernel split, and

SMBIOS=0x5fec9000

if I have a kernel with a 3:1 split. The correct value is 0x1fec9000, 
which is what the kernel prints at boot time. The following trivial 
patch seems to fix things.

Signed-off-by: Matthew Garrett <mjg59@srcf.ucam.org>

diff --git a/arch/i386/kernel/efi.c b/arch/i386/kernel/efi.c
index ecad519..6be705e 100644
--- a/arch/i386/kernel/efi.c
+++ b/arch/i386/kernel/efi.c
@@ -391,7 +391,7 @@ void __init efi_init(void)
 			printk(KERN_INFO " ACPI=0x%lx ", config_tables[i].table);
 		} else
 		    if (efi_guidcmp(config_tables[i].guid, SMBIOS_TABLE_GUID) == 0) {
-			efi.smbios = (void *) config_tables[i].table;
+			efi.smbios = __va(config_tables[i].table);
 			printk(KERN_INFO " SMBIOS=0x%lx ", config_tables[i].table);
 		} else
 		    if (efi_guidcmp(config_tables[i].guid, HCDP_TABLE_GUID) == 0) {

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

[PATCH] Remove __init from efi_get_time

[Matthew Garrett]

The comment above efi_get_time claims:

Note, this call isn't used later, so mark it __init.

Unfortunately, it's not true - it's called during suspend as well, 
leading to nasty explosions.

Signed-off-by: Matthew Garrett <mjg59@srcf.ucam.org>

diff --git a/arch/i386/kernel/efi.c b/arch/i386/kernel/efi.c
index ecad519..6be705e 100644
--- a/arch/i386/kernel/efi.c
+++ b/arch/i386/kernel/efi.c
@@ -193,9 +193,9 @@ inline int efi_set_rtc_mmss(unsigned lon
 /*
  * This should only be used during kernel init and before runtime
  * services have been remapped, therefore, we'll need to call in physical
- * mode.  Note, this call isn't used later, so mark it __init.
+ * mode.
  */
-inline unsigned long __init efi_get_time(void)
+inline unsigned long efi_get_time(void)
 {
 	efi_status_t status;
 	efi_time_t eft;
diff --git a/include/linux/efi.h b/include/linux/efi.h
index c7c5dd3..9e97bc2 100644
--- a/include/linux/efi.h
+++ b/include/linux/efi.h
@@ -295,7 +295,7 @@ extern u64 efi_mem_attributes (unsigned 
 extern int __init efi_uart_console_only (void);
 extern void efi_initialize_iomem_resources(struct resource *code_resource,
 					struct resource *data_resource);
-extern unsigned long __init efi_get_time(void);
+extern unsigned long efi_get_time(void);
 extern int __init efi_set_rtc_mmss(unsigned long nowtime);
 extern struct efi_memory_map memmap;
 
-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: [PATCH] /sys/firmware/efi/systab giving incorrect value for smbios

[Matthew Garrett]

Or, as an alternative, remove the virtual to physical mapping that 
efivars does. This requires fixing up IA64 to match. I've no idea which 
approach is right.

Signed-off-by: Matthew Garrett <mjg59@srcf.ucam.org>

diff --git a/drivers/firmware/efivars.c b/drivers/firmware/efivars.c
index bda5bce..ba598af 100644
--- a/drivers/firmware/efivars.c
+++ b/drivers/firmware/efivars.c
@@ -575,7 +575,7 @@ systab_read(struct subsystem *entry, cha
 	if (efi.acpi)
 		str += sprintf(str, "ACPI=0x%lx\n", __pa(efi.acpi));
 	if (efi.smbios)
-		str += sprintf(str, "SMBIOS=0x%lx\n", __pa(efi.smbios));
+		str += sprintf(str, "SMBIOS=0x%lx\n", efi.smbios);
 	if (efi.hcdp)
 		str += sprintf(str, "HCDP=0x%lx\n", __pa(efi.hcdp));
 	if (efi.boot_info)
diff --git a/arch/ia64/kernel/efi.c b/arch/ia64/kernel/efi.c
index a3aa45c..ff3795b 100644
--- a/arch/ia64/kernel/efi.c
+++ b/arch/ia64/kernel/efi.c
@@ -451,7 +451,7 @@ efi_init (void)
 			efi.acpi = __va(config_tables[i].table);
 			printk(" ACPI=0x%lx", config_tables[i].table);
 		} else if (efi_guidcmp(config_tables[i].guid, SMBIOS_TABLE_GUID) == 0) {
-			efi.smbios = __va(config_tables[i].table);
+			efi.smbios = config_tables[i].table;
 			printk(" SMBIOS=0x%lx", config_tables[i].table);
 		} else if (efi_guidcmp(config_tables[i].guid, SAL_SYSTEM_TABLE_GUID) == 0) {
 			efi.sal_systab = __va(config_tables[i].table);

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

Re: [Mactel-linux-devel] Re: [PATCH] /sys/firmware/efi/systab giving incorrect value for smbios

[gimli]

Hi.

Do you have any idea why the kernel crashes on machines with more then 512 MB ram ?

cu

gimli

Matthew Garrett wrote:
> Or, as an alternative, remove the virtual to physical mapping that 
> efivars does. This requires fixing up IA64 to match. I've no idea which 
> approach is right.
> 
> Signed-off-by: Matthew Garrett <mjg59@srcf.ucam.org>
> 
> diff --git a/drivers/firmware/efivars.c b/drivers/firmware/efivars.c
> index bda5bce..ba598af 100644
> --- a/drivers/firmware/efivars.c
> +++ b/drivers/firmware/efivars.c
> @@ -575,7 +575,7 @@ systab_read(struct subsystem *entry, cha
>  	if (efi.acpi)
>  		str += sprintf(str, "ACPI=0x%lx\n", __pa(efi.acpi));
>  	if (efi.smbios)
> -		str += sprintf(str, "SMBIOS=0x%lx\n", __pa(efi.smbios));
> +		str += sprintf(str, "SMBIOS=0x%lx\n", efi.smbios);
>  	if (efi.hcdp)
>  		str += sprintf(str, "HCDP=0x%lx\n", __pa(efi.hcdp));
>  	if (efi.boot_info)
> diff --git a/arch/ia64/kernel/efi.c b/arch/ia64/kernel/efi.c
> index a3aa45c..ff3795b 100644
> --- a/arch/ia64/kernel/efi.c
> +++ b/arch/ia64/kernel/efi.c
> @@ -451,7 +451,7 @@ efi_init (void)
>  			efi.acpi = __va(config_tables[i].table);
>  			printk(" ACPI=0x%lx", config_tables[i].table);
>  		} else if (efi_guidcmp(config_tables[i].guid, SMBIOS_TABLE_GUID) == 0) {
> -			efi.smbios = __va(config_tables[i].table);
> +			efi.smbios = config_tables[i].table;
>  			printk(" SMBIOS=0x%lx", config_tables[i].table);
>  		} else if (efi_guidcmp(config_tables[i].guid, SAL_SYSTEM_TABLE_GUID) == 0) {
>  			efi.sal_systab = __va(config_tables[i].table);
> 

Re: [Mactel-linux-devel] Re: [PATCH] /sys/firmware/efi/systab giving incorrect value for smbios

[Matthew Garrett]

On Sat, Mar 04, 2006 at 08:04:08PM +0100, gimli wrote:
> Hi.
> 
> Do you have any idea why the kernel crashes on machines with more then 512 MB ram ?

As yet, absolutely none. I don't have access to a 1GB machine - if it's 
practical to add more, then I'll see what I can do.

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