Remove unused voltagedomain data for AM33xx

PATCH New Rajendra Nayak 5 participants 9 messages Jun 13, 2013 – Jun 17, 2013

Subsystems (4)

Arm (32-bit) Arm sub-architectures OMAP2+ support OMAP powerdomain SoC adaptation layer support

lore

Share & tools

lore

Rajendra Nayak

author

2013-06-13 10:08 UTC #0 lore raw

cover Remove unused voltagedomain data for AM33xx

Hi,

The powerdomain framework today expects to always have a voltagedomain associated with a given powerdomain. We already have AM33xx which has no Voltage Controller/Voltage Processor as part of PRCM. There are more SoCs' to follow starting with AM437x and DRA7xx which do not have VC/VP.

Instead of adding dummy voltage domain data files, make the powerdomain framework aware of the fact that some SoCs' might not really have scalable voltage domains.

Patches are based of 3.10-rc5 and are boot tested on am335x bone.

Rajendra Nayak (2):
ARM: OMAP2+: Powerdomain: Remove the need to always have a
voltdm associated to a pwrdm
ARM: AM33xx: Remove the unused voltagedomain data

 arch/arm/mach-omap2/Makefile                  |    1 -
 arch/arm/mach-omap2/io.c                      |    1 -
 arch/arm/mach-omap2/powerdomain.c             |    5 +++
 arch/arm/mach-omap2/powerdomain.h             |    2 ++
 arch/arm/mach-omap2/prm33xx.c                 |    7 ++++
 arch/arm/mach-omap2/voltage.h                 |    1 -
 arch/arm/mach-omap2/voltagedomains33xx_data.c |   43 -------------------------
 7 files changed, 14 insertions(+), 46 deletions(-)
 delete mode 100644 arch/arm/mach-omap2/voltagedomains33xx_data.c

Signature · 1 line

1.7.9.5

Rajendra Nayak

author

2013-06-13 10:08 UTC #1 lore raw

PATCH 1/2 ARM: OMAP2+: Powerdomain: Remove the need to always have a voltdm associated to a pwrdm

The powerdomain framework expects all powerdomains to be associated with a corresponding voltagedomain. For some SoCs' (like the already existing AM33xx family, or for the upcoming AM437x and DRA7 SoCs') which
do not have a Voltage controller/Voltage Processor (neither the SR I2C bus to communicate with the PMIC) there is no need for a Powerdomain to have a voltage domain association (since they are really non scaleable, even though the voltage domains exist in place).

Extend the arch operations to add an api which the powerdomain core can then use to identify if a voltdm lookup and association for a powerdomain is really needed.

Signed-off-by Rajendra Nayak <[email protected]>

---
 arch/arm/mach-omap2/powerdomain.c |    5 +++++
 arch/arm/mach-omap2/powerdomain.h |    2 ++
 2 files changed, 7 insertions(+)

arch/arm/mach-omap2/powerdomain.c +5 -0

index 86babd7..e233dfc 100644
@@ -102,6 +102,10 @@ static int _pwrdm_register(struct powerdomain *pwrdm)
 	if (_pwrdm_lookup(pwrdm->name))
 		return -EEXIST;
 
+	if (arch_pwrdm && arch_pwrdm->pwrdm_has_voltdm)
+		if (!arch_pwrdm->pwrdm_has_voltdm())
+			goto skip_voltdm;
+
 	voltdm = voltdm_lookup(pwrdm->voltdm.name);
 	if (!voltdm) {
 		pr_err("powerdomain: %s: voltagedomain %s does not exist\n",

@@ -111,6 +115,7 @@ static int _pwrdm_register(struct powerdomain *pwrdm)
 	pwrdm->voltdm.ptr = voltdm;
 	INIT_LIST_HEAD(&pwrdm->voltdm_node);
 	voltdm_add_pwrdm(voltdm, pwrdm);
+skip_voltdm:
 	spin_lock_init(&pwrdm->_lock);
 
 	list_add(&pwrdm->node, &pwrdm_list);

arch/arm/mach-omap2/powerdomain.h +2 -0

index 140c360..8ed89de 100644
@@ -166,6 +166,7 @@ struct powerdomain {
  * @pwrdm_disable_hdwr_sar: Disable Hardware Save-Restore feature for a pd
  * @pwrdm_set_lowpwrstchange: Enable pd transitions from a shallow to deep sleep
  * @pwrdm_wait_transition: Wait for a pd state transition to complete
+ * @pwrdm_has_voltdmi: Check if a voltdm association is needed
  *
  * Regarding @pwrdm_set_lowpwrstchange: On the OMAP2 and 3-family
  * chips, a powerdomain's power state is not allowed to directly

@@ -196,6 +197,7 @@ struct pwrdm_ops {
 	int	(*pwrdm_disable_hdwr_sar)(struct powerdomain *pwrdm);
 	int	(*pwrdm_set_lowpwrstchange)(struct powerdomain *pwrdm);
 	int	(*pwrdm_wait_transition)(struct powerdomain *pwrdm);
+	int	(*pwrdm_has_voltdm)(void);
 };
 
 int pwrdm_register_platform_funcs(struct pwrdm_ops *custom_funcs);

Signature · 1 line

1.7.9.5

Kevin Hilman

2013-06-14 13:59 UTC #2 lore raw

PATCH 1/2 ARM: OMAP2+: Powerdomain: Remove the need to always have a voltdm associated to a pwrdm

Quoted 1 line · Rajendra Nayak

The powerdomain framework expects all powerdomains to be associated with

s/expects/currently expects/

Quoted 6 lines

a corresponding voltagedomain. For some SoCs' (like the already existing AM33xx family, or for the upcoming AM437x and DRA7 SoCs') which
do not have a Voltage controller/Voltage Processor (neither the SR I2C bus to communicate with the PMIC) there is no need for a Powerdomain to have a voltage domain association (since they are really non scaleable, even though the voltage domains exist in place).

This last phrase inside the parentheses doesn't make sense to me. Reading that makes me think that the lack of an on-chip voltage domain means that evn an external regulators can't scale the voltage, which I don't believe is the case.

Quoted 3 lines

Extend the arch operations to add an api which the powerdomain core can then use to identify if a voltdm lookup and association for a powerdomain is really needed.

Yes, this idea looks right to me.

In addition to the wording above, a minor nit below...

[...]

arch/arm/mach-omap2/powerdomain.h struct powerdomain { · line 169

Quoted 9 lines

arch/arm/mach-omap2/powerdomain.h +1 -0

index 140c360..8ed89de 100644
@@ -166,6 +166,7 @@ struct powerdomain {
  * @pwrdm_disable_hdwr_sar: Disable Hardware Save-Restore feature for a pd
  * @pwrdm_set_lowpwrstchange: Enable pd transitions from a shallow to deep sleep
  * @pwrdm_wait_transition: Wait for a pd state transition to complete
+ * @pwrdm_has_voltdmi: Check if a voltdm association is needed

s/voltdmi/voltdm/

Kevin

Rajendra Nayak

author

2013-06-13 10:08 UTC #3 lore raw

PATCH 2/2 ARM: AM33xx: Remove the unused voltagedomain data

In reply to Rajendra Nayak

Now that there is a way to tell the powerdomain core about missing voltage domain control in SoCs', get rid of the dummy voltage domain data populated for AM33xx devices.

Signed-off-by Rajendra Nayak <[email protected]>

---
 arch/arm/mach-omap2/Makefile                  |    1 -
 arch/arm/mach-omap2/io.c                      |    1 -
 arch/arm/mach-omap2/prm33xx.c                 |    7 ++++
 arch/arm/mach-omap2/voltage.h                 |    1 -
 arch/arm/mach-omap2/voltagedomains33xx_data.c |   43 -------------------------
 5 files changed, 7 insertions(+), 46 deletions(-)
 delete mode 100644 arch/arm/mach-omap2/voltagedomains33xx_data.c

arch/arm/mach-omap2/Makefile +0 -1

index 55a9d67..a4f1d2a 100644
@@ -125,7 +125,6 @@ obj-$(CONFIG_ARCH_OMAP3)		+= voltagedomains3xxx_data.o
 obj-$(CONFIG_ARCH_OMAP4)		+= $(voltagedomain-common)
 obj-$(CONFIG_ARCH_OMAP4)		+= voltagedomains44xx_data.o
 obj-$(CONFIG_SOC_AM33XX)		+= $(voltagedomain-common)
-obj-$(CONFIG_SOC_AM33XX)                += voltagedomains33xx_data.o
 obj-$(CONFIG_SOC_OMAP5)			+= $(voltagedomain-common)
 
 # OMAP powerdomain framework

arch/arm/mach-omap2/io.c +0 -1

index 09abf99..50b93df 100644
@@ -577,7 +577,6 @@ void __init am33xx_init_early(void)
 	omap2_set_globals_cm(AM33XX_L4_WK_IO_ADDRESS(AM33XX_PRCM_BASE), NULL);
 	omap3xxx_check_revision();
 	ti81xx_check_features();
-	am33xx_voltagedomains_init();
 	am33xx_powerdomains_init();
 	am33xx_clockdomains_init();
 	am33xx_hwmod_init();

arch/arm/mach-omap2/prm33xx.c +7 -0

index 44c0d72..7204407 100644
@@ -320,6 +320,12 @@ static int am33xx_pwrdm_wait_transition(struct powerdomain *pwrdm)
 	return 0;
 }
 
+static int am33xx_check_vcvp(void)
+{
+	/* No VC/VP on am33xx devices */
+	return 0;
+}
+
 struct pwrdm_ops am33xx_pwrdm_operations = {
 	.pwrdm_set_next_pwrst		= am33xx_pwrdm_set_next_pwrst,
 	.pwrdm_read_next_pwrst		= am33xx_pwrdm_read_next_pwrst,

@@ -335,4 +341,5 @@ struct pwrdm_ops am33xx_pwrdm_operations = {
 	.pwrdm_set_mem_onst		= am33xx_pwrdm_set_mem_onst,
 	.pwrdm_set_mem_retst		= am33xx_pwrdm_set_mem_retst,
 	.pwrdm_wait_transition		= am33xx_pwrdm_wait_transition,
+	.pwrdm_has_voltdm		= am33xx_check_vcvp,
 };

arch/arm/mach-omap2/voltage.h +0 -1

index a0ce4f1..9bf796d 100644
@@ -169,7 +169,6 @@ int omap_voltage_late_init(void);
 
 extern void omap2xxx_voltagedomains_init(void);
 extern void omap3xxx_voltagedomains_init(void);
-extern void am33xx_voltagedomains_init(void);
 extern void omap44xx_voltagedomains_init(void);
 
 struct voltagedomain *voltdm_lookup(const char *name);

arch/arm/mach-omap2/voltagedomains33xx_data.c deleted +0 ~~-43~~

43 lines not on this page · raw message

Signature · 1 line

1.7.9.5

Paul Walmsley

maintainer

2013-06-14 02:46 UTC #4 lore raw

cover Remove unused voltagedomain data for AM33xx

In reply to Rajendra Nayak

cc Kevin, Vaibhav

Quoted 9 lines · Rajendra Nayak

The powerdomain framework today expects to always have a voltagedomain associated with a given powerdomain. We already have AM33xx which has no Voltage Controller/Voltage Processor as part of PRCM. There are more SoCs' to follow starting with AM437x and DRA7xx which do not have VC/VP.
Instead of adding dummy voltage domain data files, make the powerdomain framework aware of the fact that some SoCs' might not really have scalable voltage domains.

Fine with me in principle if AM335x doesn't support voltage scaling. Vaibhav, if this is okay for you, please ack it.

Then, in terms of merging, probably Kevin would be the right person for this since he's done much of the voltagedomain work.

- Paul

Nishanth Menon

2013-06-14 13:14 UTC #5 lore raw

cover Remove unused voltagedomain data for AM33xx

Quoted 10 lines · Paul Walmsley

cc Kevin, Vaibhav
Quoted 6 lines · Rajendra Nayak
The powerdomain framework today expects to always have a voltagedomain associated with a given powerdomain. We already have AM33xx which has no Voltage Controller/Voltage Processor as part of PRCM. There are more SoCs' to follow starting with AM437x and DRA7xx which do not have VC/VP.

Quoted 6 lines

Quoted 3 lines
Instead of adding dummy voltage domain data files, make the powerdomain framework aware of the fact that some SoCs' might not really have scalable voltage domains.
Fine with me in principle if AM335x doesn't support voltage scaling. Vaibhav, if this is okay for you, please ack it.

just a nitpick :)
There is no VFSM auto scaling to retention voltage, nor is VC-VP IP available here, but that does not mean DVFS voltage scaling does not exist - it is done with traditional regulator framework using regular shared I2C.

We now seem to be moving to a generation of SoCs where VC-VP has been dumped in favor of allowing integration with PMIC which may talk SPI/I2C over standard interfaces - this allows us to reach better product options, but gives up on extreme low power scenarios we had seen in the past on OMAP.
Overall, I love this idea.

Acked-by Nishanth Menon <[email protected]>

Quoted 3 lines

Then, in terms of merging, probably Kevin would be the right person for this since he's done much of the voltagedomain work.

Signature · 2 lines

Regards,
Nishanth Menon

Kevin Hilman

2013-06-14 14:01 UTC #6 lore raw

cover Remove unused voltagedomain data for AM33xx

In reply to Paul Walmsley

Quoted 14 lines · Paul Walmsley

cc Kevin, Vaibhav
Quoted 9 lines · Rajendra Nayak
The powerdomain framework today expects to always have a voltagedomain associated with a given powerdomain. We already have AM33xx which has no Voltage Controller/Voltage Processor as part of PRCM. There are more SoCs' to follow starting with AM437x and DRA7xx which do not have VC/VP.
Instead of adding dummy voltage domain data files, make the powerdomain framework aware of the fact that some SoCs' might not really have scalable voltage domains.

Quoted 5 lines

Fine with me in principle if AM335x doesn't support voltage scaling. Vaibhav, if this is okay for you, please ack it.
Then, in terms of merging, probably Kevin would be the right person for this since he's done much of the voltagedomain work.

Yeah, I'll take this series after the minor issues I commented on are fixed, and Vaivhav ack's the AM33xx parts.

Kevin

Hiremath, Vaibhav

2013-06-17 05:08 UTC #7 lore raw

cover Remove unused voltagedomain data for AM33xx

Quoted 31 lines

-----Original Message-----
From: linux-omap-owner at vger.kernel.org [mailto:linux-omap- owner at vger.kernel.org] On Behalf Of Kevin Hilman
Sent: Friday, June 14, 2013 7:31 PM
To: Paul Walmsley
Cc: Nayak, Rajendra; Hiremath, Vaibhav; linux-omap at vger.kernel.org; linux-arm-kernel at lists.infradead.org
Subject: Re: [PATCH 0/2] Remove unused voltagedomain data for AM33xx
Quoted 5 lines · Paul Walmsley
cc Kevin, Vaibhav
Quoted 1 line · Rajendra Nayak
The powerdomain framework today expects to always have a
voltagedomain
Quoted 6 lines
Quoted 6 lines
associated with a given powerdomain. We already have AM33xx which has no Voltage Controller/Voltage Processor as part of PRCM. There are more SoCs' to follow starting with AM437x and DRA7xx which do not have VC/VP.
Instead of adding dummy voltage domain data files, make the
powerdomain
Quoted 6 lines
Quoted 2 lines
framework aware of the fact that some SoCs' might not really have scalable voltage domains.
Fine with me in principle if AM335x doesn't support voltage scaling. Vaibhav, if this is okay for you, please ack it.

Quoted 7 lines

Quoted 1 line
Then, in terms of merging, probably Kevin would be the right person
for
Quoted 1 line
this since he's done much of the voltagedomain work.
Yeah, I'll take this series after the minor issues I commented on are fixed, and Vaivhav ack's the AM33xx parts.

Yeup, Feel free to add my Ack here.

Thanks,
Vaibhav

Rajendra Nayak

author

2013-06-17 05:12 UTC #8 lore raw

cover Remove unused voltagedomain data for AM33xx

Quoted 32 lines · Vaibhav

Quoted 32 lines
-----Original Message-----
From: linux-omap-owner at vger.kernel.org [mailto:linux-omap- owner at vger.kernel.org] On Behalf Of Kevin Hilman
Sent: Friday, June 14, 2013 7:31 PM
To: Paul Walmsley
Cc: Nayak, Rajendra; Hiremath, Vaibhav; linux-omap at vger.kernel.org; linux-arm-kernel at lists.infradead.org
Subject: Re: [PATCH 0/2] Remove unused voltagedomain data for AM33xx
Quoted 5 lines · Paul Walmsley
cc Kevin, Vaibhav
Quoted 1 line · Rajendra Nayak
The powerdomain framework today expects to always have a
voltagedomain
Quoted 6 lines
Quoted 6 lines
associated with a given powerdomain. We already have AM33xx which has no Voltage Controller/Voltage Processor as part of PRCM. There are more SoCs' to follow starting with AM437x and DRA7xx which do not have VC/VP.
Instead of adding dummy voltage domain data files, make the
powerdomain
Quoted 7 lines
Quoted 2 lines
framework aware of the fact that some SoCs' might not really have scalable voltage domains.
Fine with me in principle if AM335x doesn't support voltage scaling. Vaibhav, if this is okay for you, please ack it.
Then, in terms of merging, probably Kevin would be the right person

Quoted 8 lines

Quoted 6 lines
for
Quoted 1 line
this since he's done much of the voltagedomain work.
Yeah, I'll take this series after the minor issues I commented on are fixed, and Vaivhav ack's the AM33xx parts.
Yeup, Feel free to add my Ack here.

Thanks Vaibhav, I will add your and Nishanths' ack for this series and repost a v2 addressing Kevins comments.

regards,
Rajendra

Quoted 5 lines

Thanks,
Vaibhav