[PATCH V2] hwmon, k10temp: Add support for F15h M60h

[PATCH V2] hwmon, k10temp: Add support for F15h M60h

[Aravind Gopalakrishnan]

This patch adds temperature monitoring support for F15h M60h processor.
 - Add new pci device id for the relevant processor
 - The functionality of REG_REPORTED_TEMPERATURE is moved to
   D0F0xBC_xD820_0CA4 [Reported Temperature Control]
   - So, use this to get CUR_TEMP value
   - Since we need an indirect register access, protect this with
     a mutex lock
 - Add Kconfig, Doc entries to indicate support for this processor.

Signed-off-by: Aravind Gopalakrishnan <aravind.gopalakrishnan@amd.com>
---
Changes in V2:
 - Prevent race with other code that may require indirect NB_SMU_REG access
 - Fix some minor style issues

 Documentation/hwmon/k10temp   |  2 +-
 arch/x86/include/asm/amd_nb.h |  2 ++
 arch/x86/kernel/amd_nb.c      | 15 +++++++++++++++
 drivers/hwmon/Kconfig         |  6 +++---
 drivers/hwmon/k10temp.c       | 22 +++++++++++++++++++---
 5 files changed, 40 insertions(+), 7 deletions(-)

diff --git a/Documentation/hwmon/k10temp b/Documentation/hwmon/k10temp
index ee6d30e..254d2f5 100644
--- a/Documentation/hwmon/k10temp
+++ b/Documentation/hwmon/k10temp
@@ -11,7 +11,7 @@ Supported chips:
   Socket S1G2: Athlon (X2), Sempron (X2), Turion X2 (Ultra)
 * AMD Family 12h processors: "Llano" (E2/A4/A6/A8-Series)
 * AMD Family 14h processors: "Brazos" (C/E/G/Z-Series)
-* AMD Family 15h processors: "Bulldozer" (FX-Series), "Trinity", "Kaveri"
+* AMD Family 15h processors: "Bulldozer" (FX-Series), "Trinity", "Kaveri", "Carrizo"
 * AMD Family 16h processors: "Kabini", "Mullins"
 
   Prefix: 'k10temp'
diff --git a/arch/x86/include/asm/amd_nb.h b/arch/x86/include/asm/amd_nb.h
index aaac3b2..e2bf43d 100644
--- a/arch/x86/include/asm/amd_nb.h
+++ b/arch/x86/include/asm/amd_nb.h
@@ -20,6 +20,8 @@ extern void amd_flush_garts(void);
 extern int amd_numa_init(void);
 extern int amd_get_subcaches(int);
 extern int amd_set_subcaches(int, unsigned long);
+/* helper function for reading from NB_SMU_IND_DATA */
+extern void amd_nb_smu_index_read(struct pci_dev *, unsigned int, int, u32 *);
 
 struct amd_l3_cache {
 	unsigned indices;
diff --git a/arch/x86/kernel/amd_nb.c b/arch/x86/kernel/amd_nb.c
index f04dbb3..6008e1d 100644
--- a/arch/x86/kernel/amd_nb.c
+++ b/arch/x86/kernel/amd_nb.c
@@ -13,6 +13,9 @@
 #include <linux/spinlock.h>
 #include <asm/amd_nb.h>
 
+/* Provide lock for writing to NB_SMU_IND_ADDR */
+DEFINE_MUTEX(nb_smu_ind_mutex);
+
 static u32 *flush_words;
 
 const struct pci_device_id amd_nb_misc_ids[] = {
@@ -221,6 +224,18 @@ int amd_set_subcaches(int cpu, unsigned long mask)
 	return 0;
 }
 
+void amd_nb_smu_index_read(struct pci_dev *pdev, unsigned int devfn,
+			   int offset, u32 *val)
+{
+	mutex_lock(&nb_smu_ind_mutex);
+	pci_bus_write_config_dword(pdev->bus, devfn,
+				   0xb8, offset);
+	pci_bus_read_config_dword(pdev->bus, devfn,
+				  0xbc, val);
+	mutex_unlock(&nb_smu_ind_mutex);
+}
+EXPORT_SYMBOL_GPL(amd_nb_smu_index_read);
+
 static int amd_cache_gart(void)
 {
 	u16 i;
diff --git a/drivers/hwmon/Kconfig b/drivers/hwmon/Kconfig
index 02d3d85..0234a08 100644
--- a/drivers/hwmon/Kconfig
+++ b/drivers/hwmon/Kconfig
@@ -275,13 +275,13 @@ config SENSORS_K8TEMP
 
 config SENSORS_K10TEMP
 	tristate "AMD Family 10h+ temperature sensor"
-	depends on X86 && PCI
+	depends on X86 && PCI && AMD_NB
 	help
 	  If you say yes here you get support for the temperature
 	  sensor(s) inside your CPU. Supported are later revisions of
 	  the AMD Family 10h and all revisions of the AMD Family 11h,
-	  12h (Llano), 14h (Brazos), 15h (Bulldozer/Trinity/Kaveri) and
-	  16h (Kabini/Mullins) microarchitectures.
+	  12h (Llano), 14h (Brazos), 15h (Bulldozer/Trinity/Kaveri/Carrizo)
+	  and 16h (Kabini/Mullins) microarchitectures.
 
 	  This driver can also be built as a module.  If so, the module
 	  will be called k10temp.
diff --git a/drivers/hwmon/k10temp.c b/drivers/hwmon/k10temp.c
index f7b46f6..c7422c5 100644
--- a/drivers/hwmon/k10temp.c
+++ b/drivers/hwmon/k10temp.c
@@ -24,6 +24,7 @@
 #include <linux/module.h>
 #include <linux/pci.h>
 #include <asm/processor.h>
+#include <asm/amd_nb.h>
 
 MODULE_DESCRIPTION("AMD Family 10h+ CPU core temperature monitor");
 MODULE_AUTHOR("Clemens Ladisch <clemens@ladisch.de>");
@@ -51,13 +52,27 @@ MODULE_PARM_DESC(force, "force loading on processors with erratum 319");
 #define REG_NORTHBRIDGE_CAPABILITIES	0xe8
 #define  NB_CAP_HTC			0x00000400
 
+/*
+ * For F15h M60h, functionality of REG_REPORTED_TEMPERATURE
+ * has been moved to D0F0xBC_xD820_0CA4 [Reported Temperature
+ * Control]
+ */
+#define F15H_M60H_REPORTED_TEMP_CTRL_OFFSET	0xd8200ca4
+#define PCI_DEVICE_ID_AMD_15H_M60H_NB_F3	0x1573
+
 static ssize_t show_temp(struct device *dev,
 			 struct device_attribute *attr, char *buf)
 {
 	u32 regval;
-
-	pci_read_config_dword(to_pci_dev(dev),
-			      REG_REPORTED_TEMPERATURE, &regval);
+	struct pci_dev *pdev = to_pci_dev(dev);
+
+	if (boot_cpu_data.x86 == 0x15 && boot_cpu_data.x86_model == 0x60) {
+		amd_nb_smu_index_read(pdev, PCI_DEVFN(0, 0),
+				      F15H_M60H_REPORTED_TEMP_CTRL_OFFSET,
+				      &regval);
+	} else {
+		pci_read_config_dword(pdev, REG_REPORTED_TEMPERATURE, &regval);
+	}
 	return sprintf(buf, "%u\n", (regval >> 21) * 125);
 }
 
@@ -211,6 +226,7 @@ static const struct pci_device_id k10temp_id_table[] = {
 	{ PCI_VDEVICE(AMD, PCI_DEVICE_ID_AMD_15H_NB_F3) },
 	{ PCI_VDEVICE(AMD, PCI_DEVICE_ID_AMD_15H_M10H_F3) },
 	{ PCI_VDEVICE(AMD, PCI_DEVICE_ID_AMD_15H_M30H_NB_F3) },
+	{ PCI_VDEVICE(AMD, PCI_DEVICE_ID_AMD_15H_M60H_NB_F3) },
 	{ PCI_VDEVICE(AMD, PCI_DEVICE_ID_AMD_16H_NB_F3) },
 	{ PCI_VDEVICE(AMD, PCI_DEVICE_ID_AMD_16H_M30H_NB_F3) },
 	{}
-- 
2.0.3

Re: [PATCH V2] hwmon, k10temp: Add support for F15h M60h

[Clemens Ladisch]

Aravind Gopalakrishnan wrote:
> This patch adds temperature monitoring support for F15h M60h processor.
>  - Add new pci device id for the relevant processor
>  - The functionality of REG_REPORTED_TEMPERATURE is moved to
>    D0F0xBC_xD820_0CA4 [Reported Temperature Control]
>    - So, use this to get CUR_TEMP value
>    - Since we need an indirect register access, protect this with
>      a mutex lock
>  - Add Kconfig, Doc entries to indicate support for this processor.
>
> Signed-off-by: Aravind Gopalakrishnan <aravind.gopalakrishnan@amd.com>

Acked-by: Clemens Ladisch <clemens@ladisch.de>

> ---
> Changes in V2:
>  - Prevent race with other code that may require indirect NB_SMU_REG access
>  - Fix some minor style issues
>
>  Documentation/hwmon/k10temp   |  2 +-
>  arch/x86/include/asm/amd_nb.h |  2 ++
>  arch/x86/kernel/amd_nb.c      | 15 +++++++++++++++
>  drivers/hwmon/Kconfig         |  6 +++---
>  drivers/hwmon/k10temp.c       | 22 +++++++++++++++++++---
>  5 files changed, 40 insertions(+), 7 deletions(-)
>
> diff --git a/Documentation/hwmon/k10temp b/Documentation/hwmon/k10temp
> index ee6d30e..254d2f5 100644
> --- a/Documentation/hwmon/k10temp
> +++ b/Documentation/hwmon/k10temp
> @@ -11,7 +11,7 @@ Supported chips:
>    Socket S1G2: Athlon (X2), Sempron (X2), Turion X2 (Ultra)
>  * AMD Family 12h processors: "Llano" (E2/A4/A6/A8-Series)
>  * AMD Family 14h processors: "Brazos" (C/E/G/Z-Series)
> -* AMD Family 15h processors: "Bulldozer" (FX-Series), "Trinity", "Kaveri"
> +* AMD Family 15h processors: "Bulldozer" (FX-Series), "Trinity", "Kaveri", "Carrizo"
>  * AMD Family 16h processors: "Kabini", "Mullins"
>
>    Prefix: 'k10temp'
> diff --git a/arch/x86/include/asm/amd_nb.h b/arch/x86/include/asm/amd_nb.h
> index aaac3b2..e2bf43d 100644
> --- a/arch/x86/include/asm/amd_nb.h
> +++ b/arch/x86/include/asm/amd_nb.h
> @@ -20,6 +20,8 @@ extern void amd_flush_garts(void);
>  extern int amd_numa_init(void);
>  extern int amd_get_subcaches(int);
>  extern int amd_set_subcaches(int, unsigned long);
> +/* helper function for reading from NB_SMU_IND_DATA */
> +extern void amd_nb_smu_index_read(struct pci_dev *, unsigned int, int, u32 *);
>
>  struct amd_l3_cache {
>  	unsigned indices;
> diff --git a/arch/x86/kernel/amd_nb.c b/arch/x86/kernel/amd_nb.c
> index f04dbb3..6008e1d 100644
> --- a/arch/x86/kernel/amd_nb.c
> +++ b/arch/x86/kernel/amd_nb.c
> @@ -13,6 +13,9 @@
>  #include <linux/spinlock.h>
>  #include <asm/amd_nb.h>
>
> +/* Provide lock for writing to NB_SMU_IND_ADDR */
> +DEFINE_MUTEX(nb_smu_ind_mutex);
> +
>  static u32 *flush_words;
>
>  const struct pci_device_id amd_nb_misc_ids[] = {
> @@ -221,6 +224,18 @@ int amd_set_subcaches(int cpu, unsigned long mask)
>  	return 0;
>  }
>
> +void amd_nb_smu_index_read(struct pci_dev *pdev, unsigned int devfn,
> +			   int offset, u32 *val)
> +{
> +	mutex_lock(&nb_smu_ind_mutex);
> +	pci_bus_write_config_dword(pdev->bus, devfn,
> +				   0xb8, offset);
> +	pci_bus_read_config_dword(pdev->bus, devfn,
> +				  0xbc, val);
> +	mutex_unlock(&nb_smu_ind_mutex);
> +}
> +EXPORT_SYMBOL_GPL(amd_nb_smu_index_read);
> +
>  static int amd_cache_gart(void)
>  {
>  	u16 i;
> diff --git a/drivers/hwmon/Kconfig b/drivers/hwmon/Kconfig
> index 02d3d85..0234a08 100644
> --- a/drivers/hwmon/Kconfig
> +++ b/drivers/hwmon/Kconfig
> @@ -275,13 +275,13 @@ config SENSORS_K8TEMP
>
>  config SENSORS_K10TEMP
>  	tristate "AMD Family 10h+ temperature sensor"
> -	depends on X86 && PCI
> +	depends on X86 && PCI && AMD_NB
>  	help
>  	  If you say yes here you get support for the temperature
>  	  sensor(s) inside your CPU. Supported are later revisions of
>  	  the AMD Family 10h and all revisions of the AMD Family 11h,
> -	  12h (Llano), 14h (Brazos), 15h (Bulldozer/Trinity/Kaveri) and
> -	  16h (Kabini/Mullins) microarchitectures.
> +	  12h (Llano), 14h (Brazos), 15h (Bulldozer/Trinity/Kaveri/Carrizo)
> +	  and 16h (Kabini/Mullins) microarchitectures.
>
>  	  This driver can also be built as a module.  If so, the module
>  	  will be called k10temp.
> diff --git a/drivers/hwmon/k10temp.c b/drivers/hwmon/k10temp.c
> index f7b46f6..c7422c5 100644
> --- a/drivers/hwmon/k10temp.c
> +++ b/drivers/hwmon/k10temp.c
> @@ -24,6 +24,7 @@
>  #include <linux/module.h>
>  #include <linux/pci.h>
>  #include <asm/processor.h>
> +#include <asm/amd_nb.h>
>
>  MODULE_DESCRIPTION("AMD Family 10h+ CPU core temperature monitor");
>  MODULE_AUTHOR("Clemens Ladisch <clemens@ladisch.de>");
> @@ -51,13 +52,27 @@ MODULE_PARM_DESC(force, "force loading on processors with erratum 319");
>  #define REG_NORTHBRIDGE_CAPABILITIES	0xe8
>  #define  NB_CAP_HTC			0x00000400
>
> +/*
> + * For F15h M60h, functionality of REG_REPORTED_TEMPERATURE
> + * has been moved to D0F0xBC_xD820_0CA4 [Reported Temperature
> + * Control]
> + */
> +#define F15H_M60H_REPORTED_TEMP_CTRL_OFFSET	0xd8200ca4
> +#define PCI_DEVICE_ID_AMD_15H_M60H_NB_F3	0x1573
> +
>  static ssize_t show_temp(struct device *dev,
>  			 struct device_attribute *attr, char *buf)
>  {
>  	u32 regval;
> -
> -	pci_read_config_dword(to_pci_dev(dev),
> -			      REG_REPORTED_TEMPERATURE, &regval);
> +	struct pci_dev *pdev = to_pci_dev(dev);
> +
> +	if (boot_cpu_data.x86 == 0x15 && boot_cpu_data.x86_model == 0x60) {
> +		amd_nb_smu_index_read(pdev, PCI_DEVFN(0, 0),
> +				      F15H_M60H_REPORTED_TEMP_CTRL_OFFSET,
> +				      &regval);
> +	} else {
> +		pci_read_config_dword(pdev, REG_REPORTED_TEMPERATURE, &regval);
> +	}
>  	return sprintf(buf, "%u\n", (regval >> 21) * 125);
>  }
>
> @@ -211,6 +226,7 @@ static const struct pci_device_id k10temp_id_table[] = {
>  	{ PCI_VDEVICE(AMD, PCI_DEVICE_ID_AMD_15H_NB_F3) },
>  	{ PCI_VDEVICE(AMD, PCI_DEVICE_ID_AMD_15H_M10H_F3) },
>  	{ PCI_VDEVICE(AMD, PCI_DEVICE_ID_AMD_15H_M30H_NB_F3) },
> +	{ PCI_VDEVICE(AMD, PCI_DEVICE_ID_AMD_15H_M60H_NB_F3) },
>  	{ PCI_VDEVICE(AMD, PCI_DEVICE_ID_AMD_16H_NB_F3) },
>  	{ PCI_VDEVICE(AMD, PCI_DEVICE_ID_AMD_16H_M30H_NB_F3) },
>  	{}
>

Re: [PATCH V2] hwmon, k10temp: Add support for F15h M60h

[Guenter Roeck]

On Thu, Aug 14, 2014 at 11:54:22AM -0500, Aravind Gopalakrishnan wrote:
> This patch adds temperature monitoring support for F15h M60h processor.
>  - Add new pci device id for the relevant processor
>  - The functionality of REG_REPORTED_TEMPERATURE is moved to
>    D0F0xBC_xD820_0CA4 [Reported Temperature Control]
>    - So, use this to get CUR_TEMP value
>    - Since we need an indirect register access, protect this with
>      a mutex lock
>  - Add Kconfig, Doc entries to indicate support for this processor.
> 
> Signed-off-by: Aravind Gopalakrishnan <aravind.gopalakrishnan@amd.com>
> ---
> Changes in V2:
>  - Prevent race with other code that may require indirect NB_SMU_REG access
>  - Fix some minor style issues
> 
>  Documentation/hwmon/k10temp   |  2 +-
>  arch/x86/include/asm/amd_nb.h |  2 ++
>  arch/x86/kernel/amd_nb.c      | 15 +++++++++++++++
>  drivers/hwmon/Kconfig         |  6 +++---
>  drivers/hwmon/k10temp.c       | 22 +++++++++++++++++++---
>  5 files changed, 40 insertions(+), 7 deletions(-)
> 
[ ... ]

> diff --git a/drivers/hwmon/Kconfig b/drivers/hwmon/Kconfig
> index 02d3d85..0234a08 100644
> --- a/drivers/hwmon/Kconfig
> +++ b/drivers/hwmon/Kconfig
> @@ -275,13 +275,13 @@ config SENSORS_K8TEMP
>  
>  config SENSORS_K10TEMP
>  	tristate "AMD Family 10h+ temperature sensor"
> -	depends on X86 && PCI
> +	depends on X86 && PCI && AMD_NB

Is the added dependency acceptable ?

Guenter

Re: [PATCH V2] hwmon, k10temp: Add support for F15h M60h

[Clemens Ladisch]

Guenter Roeck wrote:> On Thu, Aug 14, 2014 at 11:54:22AM -0500, Aravind Gopalakrishnan wrote:
>> This patch adds temperature monitoring support for F15h M60h processor.
>>  - Add new pci device id for the relevant processor
>>  - The functionality of REG_REPORTED_TEMPERATURE is moved to
>>    D0F0xBC_xD820_0CA4 [Reported Temperature Control]
>>    - So, use this to get CUR_TEMP value
>>    - Since we need an indirect register access, protect this with
>>      a mutex lock
>>  - Add Kconfig, Doc entries to indicate support for this processor.
>>
>> Signed-off-by: Aravind Gopalakrishnan <aravind.gopalakrishnan@amd.com>
>> ---
>> Changes in V2:
>>  - Prevent race with other code that may require indirect NB_SMU_REG access
>>  - Fix some minor style issues
>>
>>  Documentation/hwmon/k10temp   |  2 +-
>>  arch/x86/include/asm/amd_nb.h |  2 ++
>>  arch/x86/kernel/amd_nb.c      | 15 +++++++++++++++
>>  drivers/hwmon/Kconfig         |  6 +++---
>>  drivers/hwmon/k10temp.c       | 22 +++++++++++++++++++---
>>  5 files changed, 40 insertions(+), 7 deletions(-)
>>
> [ ... ]
>
>> diff --git a/drivers/hwmon/Kconfig b/drivers/hwmon/Kconfig
>> index 02d3d85..0234a08 100644
>> --- a/drivers/hwmon/Kconfig
>> +++ b/drivers/hwmon/Kconfig
>> @@ -275,13 +275,13 @@ config SENSORS_K8TEMP
>>
>>  config SENSORS_K10TEMP
>>  	tristate "AMD Family 10h+ temperature sensor"
>> -	depends on X86 && PCI
>> +	depends on X86 && PCI && AMD_NB
>
> Is the added dependency acceptable ?

Yes, it is automatically set from CPU_SUP_AMD.


Regards,
Clemens

Re: [PATCH V2] hwmon, k10temp: Add support for F15h M60h

[Borislav Petkov]

On Thu, Aug 14, 2014 at 10:22:31PM +0200, Clemens Ladisch wrote:
> >> +	depends on X86 && PCI && AMD_NB
> >
> > Is the added dependency acceptable ?
> 
> Yes, it is automatically set from CPU_SUP_AMD.

Well, we can always move that function to k10temp but I'll venture a
guess that Aravind wants to use it somewhere else too? Correct, Aravind?

-- 
Regards/Gruss,
    Boris.

Sent from a fat crate under my desk. Formatting is fine.
--

Re: [PATCH V2] hwmon, k10temp: Add support for F15h M60h

[Aravind Gopalakrishnan]

On 8/14/2014 4:08 PM, Borislav Petkov wrote:
> On Thu, Aug 14, 2014 at 10:22:31PM +0200, Clemens Ladisch wrote:
>>>> +	depends on X86 && PCI && AMD_NB
>>> Is the added dependency acceptable ?
>> Yes, it is automatically set from CPU_SUP_AMD.
> Well, we can always move that function to k10temp but I'll venture a
> guess that Aravind wants to use it somewhere else too? Correct, Aravind?
>

Actually I don't need it outside of k10temp as of now (or near future)
I added it in amd_nb as that was Clemens, Guenter's suggestion on the 
previous version;
Besides, it made sense as it's an indirect access of NB_SMU register and 
amd_nb seems a good place to put the function in case someone needs it 
in the future.

I can move it locally to k10temp and remove the dependency if that's 
more preferable. Do let me know.

Thanks,
-Aravind.

Re: [PATCH V2] hwmon, k10temp: Add support for F15h M60h

[Borislav Petkov]

On Thu, Aug 14, 2014 at 04:57:17PM -0500, Aravind Gopalakrishnan wrote:
> Actually I don't need it outside of k10temp as of now (or near future)
> I added it in amd_nb as that was Clemens, Guenter's suggestion on the
> previous version;
> Besides, it made sense as it's an indirect access of NB_SMU register and
> amd_nb seems a good place to put the function in case someone needs it in
> the future.

Then someone can move it then. But until that happens it is pretty
pointless of having the Kconfig dependency just for one small function
with a single user.

> I can move it locally to k10temp and remove the dependency if that's
> more preferable.

Yeah, it looks like a fabricated and not true dependency, which doesn't
make any sense currently.

Thanks.

-- 
Regards/Gruss,
    Boris.

Sent from a fat crate under my desk. Formatting is fine.
--

Re: [PATCH V2] hwmon, k10temp: Add support for F15h M60h

[Aravind Gopalakrishnan]

On 8/14/2014 5:05 PM, Borislav Petkov wrote:
> On Thu, Aug 14, 2014 at 04:57:17PM -0500, Aravind Gopalakrishnan wrote:
>> Actually I don't need it outside of k10temp as of now (or near future)
>> I added it in amd_nb as that was Clemens, Guenter's suggestion on the
>> previous version;
>> Besides, it made sense as it's an indirect access of NB_SMU register and
>> amd_nb seems a good place to put the function in case someone needs it in
>> the future.
> Then someone can move it then. But until that happens it is pretty
> pointless of having the Kconfig dependency just for one small function
> with a single user.
>
>> I can move it locally to k10temp and remove the dependency if that's
>> more preferable.
> Yeah, it looks like a fabricated and not true dependency, which doesn't
> make any sense currently.
>
> Thanks.
>

Ok, Will fix this and send V3.

-Aravind.