diff mbox

powercap/rapl: add support for ValleyView Soc

Message ID 1386801567-26661-1-git-send-email-jacob.jun.pan@linux.intel.com (mailing list archive)
State Accepted, archived
Headers show

Commit Message

Jacob Pan Dec. 11, 2013, 10:39 p.m. UTC
This patch adds support for RAPL on Intel ValleyView based SoC
platforms, such as Baytrail.

Besides adding CPU ID, special energy unit encoding is handled
for ValleyView.

Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com>
---
 drivers/powercap/intel_rapl.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

Comments

Rafael J. Wysocki Dec. 22, 2013, 1:06 a.m. UTC | #1
On Wednesday, December 11, 2013 02:39:27 PM Jacob Pan wrote:
> This patch adds support for RAPL on Intel ValleyView based SoC
> platforms, such as Baytrail.
> 
> Besides adding CPU ID, special energy unit encoding is handled
> for ValleyView.
> 
> Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com>

Queued up for 3.14, thanks!

> ---
>  drivers/powercap/intel_rapl.c | 13 +++++++++++--
>  1 file changed, 11 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/powercap/intel_rapl.c b/drivers/powercap/intel_rapl.c
> index 2a786c5..3c67683 100644
> --- a/drivers/powercap/intel_rapl.c
> +++ b/drivers/powercap/intel_rapl.c
> @@ -833,6 +833,11 @@ static int rapl_write_data_raw(struct rapl_domain *rd,
>  	return 0;
>  }
>  
> +static const struct x86_cpu_id energy_unit_quirk_ids[] = {
> +	{ X86_VENDOR_INTEL, 6, 0x37},/* VLV */
> +	{}
> +};
> +
>  static int rapl_check_unit(struct rapl_package *rp, int cpu)
>  {
>  	u64 msr_val;
> @@ -853,8 +858,11 @@ static int rapl_check_unit(struct rapl_package *rp, int cpu)
>  	 * time unit: 1/time_unit_divisor Seconds
>  	 */
>  	value = (msr_val & ENERGY_UNIT_MASK) >> ENERGY_UNIT_OFFSET;
> -	rp->energy_unit_divisor = 1 << value;
> -
> +	/* some CPUs have different way to calculate energy unit */
> +	if (x86_match_cpu(energy_unit_quirk_ids))
> +		rp->energy_unit_divisor = 1000000 / (1 << value);
> +	else
> +		rp->energy_unit_divisor = 1 << value;
>  
>  	value = (msr_val & POWER_UNIT_MASK) >> POWER_UNIT_OFFSET;
>  	rp->power_unit_divisor = 1 << value;
> @@ -941,6 +949,7 @@ static void package_power_limit_irq_restore(int package_id)
>  static const struct x86_cpu_id rapl_ids[] = {
>  	{ X86_VENDOR_INTEL, 6, 0x2a},/* SNB */
>  	{ X86_VENDOR_INTEL, 6, 0x2d},/* SNB EP */
> +	{ X86_VENDOR_INTEL, 6, 0x37},/* VLV */
>  	{ X86_VENDOR_INTEL, 6, 0x3a},/* IVB */
>  	{ X86_VENDOR_INTEL, 6, 0x45},/* HSW */
>  	/* TODO: Add more CPU IDs after testing */
>
Vince Weaver Dec. 22, 2013, 7:22 p.m. UTC | #2
On Sun, 22 Dec 2013, Rafael J. Wysocki wrote:
> On Wednesday, December 11, 2013 02:39:27 PM Jacob Pan wrote:

> > @@ -941,6 +949,7 @@ static void package_power_limit_irq_restore(int package_id)
> >  static const struct x86_cpu_id rapl_ids[] = {
> >  	{ X86_VENDOR_INTEL, 6, 0x2a},/* SNB */
> >  	{ X86_VENDOR_INTEL, 6, 0x2d},/* SNB EP */
> > +	{ X86_VENDOR_INTEL, 6, 0x37},/* VLV */
> >  	{ X86_VENDOR_INTEL, 6, 0x3a},/* IVB */
> >  	{ X86_VENDOR_INTEL, 6, 0x45},/* HSW */
> >  	/* TODO: Add more CPU IDs after testing */

This isn't a problem introduced in this patch, but it made me notice it...

Would it really hurt anyone to spell out the architectures here, rather
than using obscure 3-letter abbreviations?  It's bad enough trying to
map the hex model numbers to the decimal ones in /proc/cpuinfo without
also having to remember what a VLV is.

Vince
--
To unsubscribe from this list: send the line "unsubscribe linux-pm" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Rafael J. Wysocki Dec. 22, 2013, 9:21 p.m. UTC | #3
On Sunday, December 22, 2013 02:22:48 PM Vince Weaver wrote:
> On Sun, 22 Dec 2013, Rafael J. Wysocki wrote:
> > On Wednesday, December 11, 2013 02:39:27 PM Jacob Pan wrote:
> 
> > > @@ -941,6 +949,7 @@ static void package_power_limit_irq_restore(int package_id)
> > >  static const struct x86_cpu_id rapl_ids[] = {
> > >  	{ X86_VENDOR_INTEL, 6, 0x2a},/* SNB */
> > >  	{ X86_VENDOR_INTEL, 6, 0x2d},/* SNB EP */
> > > +	{ X86_VENDOR_INTEL, 6, 0x37},/* VLV */
> > >  	{ X86_VENDOR_INTEL, 6, 0x3a},/* IVB */
> > >  	{ X86_VENDOR_INTEL, 6, 0x45},/* HSW */
> > >  	/* TODO: Add more CPU IDs after testing */
> 
> This isn't a problem introduced in this patch, but it made me notice it...
> 
> Would it really hurt anyone to spell out the architectures here, rather
> than using obscure 3-letter abbreviations?  It's bad enough trying to
> map the hex model numbers to the decimal ones in /proc/cpuinfo without
> also having to remember what a VLV is.

I guess you can prepare a patch for that?

Rafael

--
To unsubscribe from this list: send the line "unsubscribe linux-pm" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/drivers/powercap/intel_rapl.c b/drivers/powercap/intel_rapl.c
index 2a786c5..3c67683 100644
--- a/drivers/powercap/intel_rapl.c
+++ b/drivers/powercap/intel_rapl.c
@@ -833,6 +833,11 @@  static int rapl_write_data_raw(struct rapl_domain *rd,
 	return 0;
 }
 
+static const struct x86_cpu_id energy_unit_quirk_ids[] = {
+	{ X86_VENDOR_INTEL, 6, 0x37},/* VLV */
+	{}
+};
+
 static int rapl_check_unit(struct rapl_package *rp, int cpu)
 {
 	u64 msr_val;
@@ -853,8 +858,11 @@  static int rapl_check_unit(struct rapl_package *rp, int cpu)
 	 * time unit: 1/time_unit_divisor Seconds
 	 */
 	value = (msr_val & ENERGY_UNIT_MASK) >> ENERGY_UNIT_OFFSET;
-	rp->energy_unit_divisor = 1 << value;
-
+	/* some CPUs have different way to calculate energy unit */
+	if (x86_match_cpu(energy_unit_quirk_ids))
+		rp->energy_unit_divisor = 1000000 / (1 << value);
+	else
+		rp->energy_unit_divisor = 1 << value;
 
 	value = (msr_val & POWER_UNIT_MASK) >> POWER_UNIT_OFFSET;
 	rp->power_unit_divisor = 1 << value;
@@ -941,6 +949,7 @@  static void package_power_limit_irq_restore(int package_id)
 static const struct x86_cpu_id rapl_ids[] = {
 	{ X86_VENDOR_INTEL, 6, 0x2a},/* SNB */
 	{ X86_VENDOR_INTEL, 6, 0x2d},/* SNB EP */
+	{ X86_VENDOR_INTEL, 6, 0x37},/* VLV */
 	{ X86_VENDOR_INTEL, 6, 0x3a},/* IVB */
 	{ X86_VENDOR_INTEL, 6, 0x45},/* HSW */
 	/* TODO: Add more CPU IDs after testing */