Linux Power Management development
 help / color / mirror / Atom feed
* [PATCH 0/2] Series short description
@ 2017-06-20  7:02 Sherry Hurwitz
  2017-06-20  7:07 ` [PATCH 1/2] cpupower: Fix bug where return value was not used Sherry Hurwitz
  2017-06-20  7:08 ` [PATCH 2/2] cpupower: Add support for new AMD family 0x17 Sherry Hurwitz
  0 siblings, 2 replies; 6+ messages in thread
From: Sherry Hurwitz @ 2017-06-20  7:02 UTC (permalink / raw)
  To: trenn; +Cc: linux-pm

The following series fixes a bug caused by not saving the 
function return value but using the value from a previous function.
It also implements support for the new AMD Family 0x17: including
the current operating frequency calculation particular to AMD 
Family 0x17, detecting core performance boost support and whether
it is currently active.

---

Sherry Hurwitz (2):
      cpupower: Check the return from amd_pci_get_num_boost_states
      cpupower: Add support for new AMD family 0x17


 tools/power/cpupower/utils/helpers/amd.c     |   31 +++++++++++++++++++-------
 tools/power/cpupower/utils/helpers/helpers.h |    2 ++
 tools/power/cpupower/utils/helpers/misc.c    |   23 ++++++++++++++++---
 3 files changed, 44 insertions(+), 12 deletions(-)

--
Signature

^ permalink raw reply	[flat|nested] 6+ messages in thread

* [PATCH 1/2] cpupower: Fix bug where return value was not used
  2017-06-20  7:02 [PATCH 0/2] Series short description Sherry Hurwitz
@ 2017-06-20  7:07 ` Sherry Hurwitz
  2017-06-23 14:00   ` Thomas Renninger
  2017-06-20  7:08 ` [PATCH 2/2] cpupower: Add support for new AMD family 0x17 Sherry Hurwitz
  1 sibling, 1 reply; 6+ messages in thread
From: Sherry Hurwitz @ 2017-06-20  7:07 UTC (permalink / raw)
  To: trenn; +Cc: linux-pm

Save return value from amd_pci_get_num_boost_states
and remove redundant setting of *support

Signed-off-by: Sherry Hurwitz <sherry.hurwitz@amd.com>
---
 tools/power/cpupower/utils/helpers/misc.c |    5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/tools/power/cpupower/utils/helpers/misc.c b/tools/power/cpupower/utils/helpers/misc.c
index 1609243..6952a6a 100644
--- a/tools/power/cpupower/utils/helpers/misc.c
+++ b/tools/power/cpupower/utils/helpers/misc.c
@@ -16,10 +16,9 @@ int cpufreq_has_boost_support(unsigned int cpu, int *support, int *active,
 
 	if (cpupower_cpu_info.caps & CPUPOWER_CAP_AMD_CBP) {
 		*support = 1;
-		amd_pci_get_num_boost_states(active, states);
-		if (ret <= 0)
+		ret = amd_pci_get_num_boost_states(active, states);
+		if (ret)
 			return ret;
-		*support = 1;
 	} else if (cpupower_cpu_info.caps & CPUPOWER_CAP_INTEL_IDA)
 		*support = *active = 1;
 	return 0;

^ permalink raw reply related	[flat|nested] 6+ messages in thread

* [PATCH 2/2] cpupower: Add support for new AMD family 0x17
  2017-06-20  7:02 [PATCH 0/2] Series short description Sherry Hurwitz
  2017-06-20  7:07 ` [PATCH 1/2] cpupower: Fix bug where return value was not used Sherry Hurwitz
@ 2017-06-20  7:08 ` Sherry Hurwitz
  2017-06-23 14:04   ` Thomas Renninger
  1 sibling, 1 reply; 6+ messages in thread
From: Sherry Hurwitz @ 2017-06-20  7:08 UTC (permalink / raw)
  To: trenn; +Cc: linux-pm

Add support for new AMD family 0x17
- Add bit field changes to the msr_pstate structure
- Add the new formula for the  calculation of cof
- Changed method to access to CpbDis

Signed-off-by: Sherry Hurwitz <sherry.hurwitz@amd.com>
---
 tools/power/cpupower/utils/helpers/amd.c     |   31 +++++++++++++++++++-------
 tools/power/cpupower/utils/helpers/helpers.h |    2 ++
 tools/power/cpupower/utils/helpers/misc.c    |   22 ++++++++++++++++--
 3 files changed, 44 insertions(+), 11 deletions(-)

diff --git a/tools/power/cpupower/utils/helpers/amd.c b/tools/power/cpupower/utils/helpers/amd.c
index 6437ef3..5fd5c5b 100644
--- a/tools/power/cpupower/utils/helpers/amd.c
+++ b/tools/power/cpupower/utils/helpers/amd.c
@@ -26,6 +26,15 @@
 		unsigned res3:21;
 		unsigned en:1;
 	} bits;
+	struct {
+		unsigned fid:8;
+		unsigned did:6;
+		unsigned vid:8;
+		unsigned iddval:8;
+		unsigned idddiv:2;
+		unsigned res1:30;
+		unsigned en:1;
+	} fam17h_bits;
 	unsigned long long val;
 };
 
@@ -35,6 +44,8 @@ static int get_did(int family, union msr_pstate pstate)
 
 	if (family == 0x12)
 		t = pstate.val & 0xf;
+	else if (family == 0x17)
+		t = pstate.fam17h_bits.did;
 	else
 		t = pstate.bits.did;
 
@@ -44,16 +55,20 @@ static int get_did(int family, union msr_pstate pstate)
 static int get_cof(int family, union msr_pstate pstate)
 {
 	int t;
-	int fid, did;
+	int fid, did, cof;
 
 	did = get_did(family, pstate);
-
-	t = 0x10;
-	fid = pstate.bits.fid;
-	if (family == 0x11)
-		t = 0x8;
-
-	return (100 * (fid + t)) >> did;
+	if (family == 0x17) {
+		fid = pstate.fam17h_bits.fid;
+		cof = 200 * fid / did;
+	} else {
+		t = 0x10;
+		fid = pstate.bits.fid;
+		if (family == 0x11)
+			t = 0x8;
+		cof = (100 * (fid + t)) >> did;
+	}
+	return cof;
 }
 
 /* Needs:
diff --git a/tools/power/cpupower/utils/helpers/helpers.h b/tools/power/cpupower/utils/helpers/helpers.h
index afb66f8..799a18b 100644
--- a/tools/power/cpupower/utils/helpers/helpers.h
+++ b/tools/power/cpupower/utils/helpers/helpers.h
@@ -70,6 +70,8 @@ enum cpupower_cpu_vendor {X86_VENDOR_UNKNOWN = 0, X86_VENDOR_INTEL,
 #define CPUPOWER_CAP_IS_SNB		0x00000020
 #define CPUPOWER_CAP_INTEL_IDA		0x00000040
 
+#define CPUPOWER_AMD_CPBDIS		0x02000000
+
 #define MAX_HW_PSTATES 10
 
 struct cpupower_cpu_info {
diff --git a/tools/power/cpupower/utils/helpers/misc.c b/tools/power/cpupower/utils/helpers/misc.c
index 6952a6a..601d719 100644
--- a/tools/power/cpupower/utils/helpers/misc.c
+++ b/tools/power/cpupower/utils/helpers/misc.c
@@ -2,11 +2,14 @@
 
 #include "helpers/helpers.h"
 
+#define MSR_AMD_HWCR	0xc0010015
+
 int cpufreq_has_boost_support(unsigned int cpu, int *support, int *active,
 			int *states)
 {
 	struct cpupower_cpu_info cpu_info;
 	int ret;
+	unsigned long long val;
 
 	*support = *active = *states = 0;
 
@@ -16,9 +19,22 @@ int cpufreq_has_boost_support(unsigned int cpu, int *support, int *active,
 
 	if (cpupower_cpu_info.caps & CPUPOWER_CAP_AMD_CBP) {
 		*support = 1;
-		ret = amd_pci_get_num_boost_states(active, states);
-		if (ret)
-			return ret;
+
+		/* AMD Family 0x17 does not utilize PCI D18F4 like prior
+		 * families and has no fixed discrete boost states but
+		 * has Hardware determined variable increments instead.
+		 */
+
+		if (cpu_info.family == 0x17) {
+			if (!read_msr(cpu, MSR_AMD_HWCR, &val)) {
+				if (!(val & CPUPOWER_AMD_CPBDIS))
+					*active = 1;
+			}
+		} else {
+			ret = amd_pci_get_num_boost_states(active, states);
+			if (ret)
+				return ret;
+		}
 	} else if (cpupower_cpu_info.caps & CPUPOWER_CAP_INTEL_IDA)
 		*support = *active = 1;
 	return 0;

^ permalink raw reply related	[flat|nested] 6+ messages in thread

* Re: [PATCH 1/2] cpupower: Fix bug where return value was not used
  2017-06-20  7:07 ` [PATCH 1/2] cpupower: Fix bug where return value was not used Sherry Hurwitz
@ 2017-06-23 14:00   ` Thomas Renninger
  2017-06-23 23:03     ` Rafael J. Wysocki
  0 siblings, 1 reply; 6+ messages in thread
From: Thomas Renninger @ 2017-06-23 14:00 UTC (permalink / raw)
  To: Sherry Hurwitz; +Cc: linux-pm, Rafael J. Wysocki

On Tuesday, June 20, 2017 02:07:37 AM Sherry Hurwitz wrote:
> Save return value from amd_pci_get_num_boost_states
> and remove redundant setting of *support
> 
> Signed-off-by: Sherry Hurwitz <sherry.hurwitz@amd.com>

Acked-by: Thomas Renninger <trenn@suse.com>
Reviewed-by: Thomas Renninger <trenn@suse.com>
whatever is more appropriate or both...

Thanks, good catch!

Rafael can you queue this an the next one, please?

Thanks,

     Thomas

> ---
>  tools/power/cpupower/utils/helpers/misc.c |    5 ++---
>  1 file changed, 2 insertions(+), 3 deletions(-)
> 
> diff --git a/tools/power/cpupower/utils/helpers/misc.c 
b/tools/power/cpupower/utils/helpers/misc.c
> index 1609243..6952a6a 100644
> --- a/tools/power/cpupower/utils/helpers/misc.c
> +++ b/tools/power/cpupower/utils/helpers/misc.c
> @@ -16,10 +16,9 @@ int cpufreq_has_boost_support(unsigned int cpu, int 
*support, int *active,
>  
>  	if (cpupower_cpu_info.caps & CPUPOWER_CAP_AMD_CBP) {
>  		*support = 1;
> -		amd_pci_get_num_boost_states(active, states);
> -		if (ret <= 0)
> +		ret = amd_pci_get_num_boost_states(active, states);
> +		if (ret)
>  			return ret;
> -		*support = 1;
>  	} else if (cpupower_cpu_info.caps & CPUPOWER_CAP_INTEL_IDA)
>  		*support = *active = 1;
>  	return 0;
> 
> 

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH 2/2] cpupower: Add support for new AMD family 0x17
  2017-06-20  7:08 ` [PATCH 2/2] cpupower: Add support for new AMD family 0x17 Sherry Hurwitz
@ 2017-06-23 14:04   ` Thomas Renninger
  0 siblings, 0 replies; 6+ messages in thread
From: Thomas Renninger @ 2017-06-23 14:04 UTC (permalink / raw)
  To: Sherry Hurwitz; +Cc: linux-pm, Rafael J. Wysocki

On Tuesday, June 20, 2017 02:08:42 AM Sherry Hurwitz wrote:
> Add support for new AMD family 0x17
> - Add bit field changes to the msr_pstate structure
> - Add the new formula for the  calculation of cof
> - Changed method to access to CpbDis
> 
> Signed-off-by: Sherry Hurwitz <sherry.hurwitz@amd.com>

I have to trust you with the HW bits and get_did() func fiddling
with them.

Rest looks like properly embedded code, no risk for other families.

Thanks!

Acked-by: Thomas Renninger <trenn@suse.com>


> ---
>  tools/power/cpupower/utils/helpers/amd.c     |   31 
+++++++++++++++++++-------
>  tools/power/cpupower/utils/helpers/helpers.h |    2 ++
>  tools/power/cpupower/utils/helpers/misc.c    |   22 ++++++++++++++++--
>  3 files changed, 44 insertions(+), 11 deletions(-)
> 
> diff --git a/tools/power/cpupower/utils/helpers/amd.c 
b/tools/power/cpupower/utils/helpers/amd.c
> index 6437ef3..5fd5c5b 100644
> --- a/tools/power/cpupower/utils/helpers/amd.c
> +++ b/tools/power/cpupower/utils/helpers/amd.c
> @@ -26,6 +26,15 @@
>  		unsigned res3:21;
>  		unsigned en:1;
>  	} bits;
> +	struct {
> +		unsigned fid:8;
> +		unsigned did:6;
> +		unsigned vid:8;
> +		unsigned iddval:8;
> +		unsigned idddiv:2;
> +		unsigned res1:30;
> +		unsigned en:1;
> +	} fam17h_bits;
>  	unsigned long long val;
>  };
>  
> @@ -35,6 +44,8 @@ static int get_did(int family, union msr_pstate pstate)
>  
>  	if (family == 0x12)
>  		t = pstate.val & 0xf;
> +	else if (family == 0x17)
> +		t = pstate.fam17h_bits.did;
>  	else
>  		t = pstate.bits.did;
>  
> @@ -44,16 +55,20 @@ static int get_did(int family, union msr_pstate pstate)
>  static int get_cof(int family, union msr_pstate pstate)
>  {
>  	int t;
> -	int fid, did;
> +	int fid, did, cof;
>  
>  	did = get_did(family, pstate);
> -
> -	t = 0x10;
> -	fid = pstate.bits.fid;
> -	if (family == 0x11)
> -		t = 0x8;
> -
> -	return (100 * (fid + t)) >> did;
> +	if (family == 0x17) {
> +		fid = pstate.fam17h_bits.fid;
> +		cof = 200 * fid / did;
> +	} else {
> +		t = 0x10;
> +		fid = pstate.bits.fid;
> +		if (family == 0x11)
> +			t = 0x8;
> +		cof = (100 * (fid + t)) >> did;
> +	}
> +	return cof;
>  }
>  
>  /* Needs:
> diff --git a/tools/power/cpupower/utils/helpers/helpers.h 
b/tools/power/cpupower/utils/helpers/helpers.h
> index afb66f8..799a18b 100644
> --- a/tools/power/cpupower/utils/helpers/helpers.h
> +++ b/tools/power/cpupower/utils/helpers/helpers.h
> @@ -70,6 +70,8 @@ enum cpupower_cpu_vendor {X86_VENDOR_UNKNOWN = 0, 
X86_VENDOR_INTEL,
>  #define CPUPOWER_CAP_IS_SNB		0x00000020
>  #define CPUPOWER_CAP_INTEL_IDA		0x00000040
>  
> +#define CPUPOWER_AMD_CPBDIS		0x02000000
> +
>  #define MAX_HW_PSTATES 10
>  
>  struct cpupower_cpu_info {
> diff --git a/tools/power/cpupower/utils/helpers/misc.c 
b/tools/power/cpupower/utils/helpers/misc.c
> index 6952a6a..601d719 100644
> --- a/tools/power/cpupower/utils/helpers/misc.c
> +++ b/tools/power/cpupower/utils/helpers/misc.c
> @@ -2,11 +2,14 @@
>  
>  #include "helpers/helpers.h"
>  
> +#define MSR_AMD_HWCR	0xc0010015
> +
>  int cpufreq_has_boost_support(unsigned int cpu, int *support, int *active,
>  			int *states)
>  {
>  	struct cpupower_cpu_info cpu_info;
>  	int ret;
> +	unsigned long long val;
>  
>  	*support = *active = *states = 0;
>  
> @@ -16,9 +19,22 @@ int cpufreq_has_boost_support(unsigned int cpu, int 
*support, int *active,
>  
>  	if (cpupower_cpu_info.caps & CPUPOWER_CAP_AMD_CBP) {
>  		*support = 1;
> -		ret = amd_pci_get_num_boost_states(active, states);
> -		if (ret)
> -			return ret;
> +
> +		/* AMD Family 0x17 does not utilize PCI D18F4 like prior
> +		 * families and has no fixed discrete boost states but
> +		 * has Hardware determined variable increments instead.
> +		 */
> +
> +		if (cpu_info.family == 0x17) {
> +			if (!read_msr(cpu, MSR_AMD_HWCR, &val)) {
> +				if (!(val & CPUPOWER_AMD_CPBDIS))
> +					*active = 1;
> +			}
> +		} else {
> +			ret = amd_pci_get_num_boost_states(active, states);
> +			if (ret)
> +				return ret;
> +		}
>  	} else if (cpupower_cpu_info.caps & CPUPOWER_CAP_INTEL_IDA)
>  		*support = *active = 1;
>  	return 0;
> 

-- 
Thomas Renninger
Software Developer

+++++++
3 important things to remember:

OPENSOURCE is going to rule the world
LINUX was, is and will stay the most important OpenSource Software stack
SUSE is a steady part of Linux ... from the very beginnings and forever.
+++++++

SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Dilip Upmanyu, Graham 
Norton, HRB 21284 (AG Nürnberg) 

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH 1/2] cpupower: Fix bug where return value was not used
  2017-06-23 14:00   ` Thomas Renninger
@ 2017-06-23 23:03     ` Rafael J. Wysocki
  0 siblings, 0 replies; 6+ messages in thread
From: Rafael J. Wysocki @ 2017-06-23 23:03 UTC (permalink / raw)
  To: Thomas Renninger; +Cc: Sherry Hurwitz, linux-pm

On Friday, June 23, 2017 04:00:57 PM Thomas Renninger wrote:
> On Tuesday, June 20, 2017 02:07:37 AM Sherry Hurwitz wrote:
> > Save return value from amd_pci_get_num_boost_states
> > and remove redundant setting of *support
> > 
> > Signed-off-by: Sherry Hurwitz <sherry.hurwitz@amd.com>
> 
> Acked-by: Thomas Renninger <trenn@suse.com>
> Reviewed-by: Thomas Renninger <trenn@suse.com>
> whatever is more appropriate or both...
> 
> Thanks, good catch!
> 
> Rafael can you queue this an the next one, please?

I'll do that.

Thanks,
Rafael

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2017-06-23 23:10 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2017-06-20  7:02 [PATCH 0/2] Series short description Sherry Hurwitz
2017-06-20  7:07 ` [PATCH 1/2] cpupower: Fix bug where return value was not used Sherry Hurwitz
2017-06-23 14:00   ` Thomas Renninger
2017-06-23 23:03     ` Rafael J. Wysocki
2017-06-20  7:08 ` [PATCH 2/2] cpupower: Add support for new AMD family 0x17 Sherry Hurwitz
2017-06-23 14:04   ` Thomas Renninger

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox