Skip to contentSkip to conversation

twl6030: regulator: Fix vsel calculations in set/get voltage apis

PATCH Applied Rajendra Nayak 3 participants 13 messages Feb 17, 2010 – Feb 19, 2010

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/2 twl6030: regulator: Fix vsel calculations in set/get voltage apis

TWL6030 has a formula to be used to calculate the vsel values to be programmed in the VREG_VOLTAGE registers.

Voltage(in mV) = 1000mv + 100mv * (vsel - 1)

Ex: if vsel = 0x9, mV = 1000 + 100 * (9 -1) = 1800mV.

Cc Liam Girdwood <[email protected]>
Cc Samuel Ortiz <[email protected]>
Cc Mark Brown <[email protected]>
---
 drivers/regulator/twl-regulator.c |   23 +++++++++++++++++++----
 1 files changed, 19 insertions(+), 4 deletions(-)
drivers/regulator/twl-regulator.c +19 -4
index 7e67485..e7871d6 100644
@@ -367,11 +367,17 @@ twlldo_set_voltage(struct regulator_dev *rdev, int min_uV, int max_uV)
 		/* REVISIT for VAUX2, first match may not be best/lowest */
 
 		/* use the first in-range value */
-		if (min_uV <= uV && uV <= max_uV)
+		if (min_uV <= uV && uV <= max_uV) {
+			if (twl_class_is_6030())
+				/*
+				 * Use the below formula to calculate vsel
+				 * mV = 1000mv + 100mv * (vsel - 1)
+				 */
+				vsel = (LDO_MV(mV) - 1000)/100 + 1;
 			return twlreg_write(info, TWL_MODULE_PM_RECEIVER,
 							VREG_VOLTAGE, vsel);
+		}
 	}
-
 	return -EDOM;
 }
 
@@ -384,8 +390,17 @@ static int twlldo_get_voltage(struct regulator_dev *rdev)
 	if (vsel < 0)
 		return vsel;
 
-	vsel &= info->table_len - 1;
-	return LDO_MV(info->table[vsel]) * 1000;
+	if (twl_class_is_4030()) {
+		vsel &= info->table_len - 1;
+		return LDO_MV(info->table[vsel]) * 1000;
+	} else if (twl_class_is_6030()) {
+		/*
+		 * Use the below formula to calculate vsel
+		 * mV = 1000mv + 100mv * (vsel - 1)
+		 */
+		return (1000 + (100 * (vsel - 1))) * 1000;
+	}
+	return -EDOM;
 }
 
 static struct regulator_ops twlldo_ops = {
Signature · 1 line
1.5.4.7
#1 lore raw

PATCH 2/2 twl6030: regulator: Configure STATE register instead of REMAP

This is no REMAP register on twl6030, instead there is a STATE register to drive a resource to a given state.
The state register can be used to specify what state the resource should enter when its associated with a GRP.
Register Bit field description is as below. The patch programmes the corresponding STATE registers for all LDO's to turn ON when assocaited with GRP_P1.

STATE REG:

Bit7   |Bit6   |Bit5   |Bit4  |Bit3  |Bit2  |Bit1   |Bit0
P3_GRP |P2_GRP |P1_GRP |RES   |RES   |RES   |State1 |State0

State can be specified as below
00: OFF
01: ON
10: OFF
11: SLEEP

Cc Liam Girdwood <[email protected]>
Cc Samuel Ortiz <[email protected]>
Cc Mark Brown <[email protected]>
---
 drivers/regulator/twl-regulator.c |   20 ++++++++++----------
 1 files changed, 10 insertions(+), 10 deletions(-)
drivers/regulator/twl-regulator.c +10 -10
index e7871d6..f4ccccc 100644
@@ -534,16 +534,16 @@ static struct twlreg_info twl_regs[] = {
 	/* 6030 REG with base as PMC Slave Misc : 0x0030 */
 	/* Turnon-delay and remap configuration values for 6030 are not
 	   verified since the specification is not public */
-	TWL6030_ADJUSTABLE_LDO(VAUX1_6030, 0x54, 1, 0, 0x08),
-	TWL6030_ADJUSTABLE_LDO(VAUX2_6030, 0x58, 2, 0, 0x08),
-	TWL6030_ADJUSTABLE_LDO(VAUX3_6030, 0x5c, 3, 0, 0x08),
-	TWL6030_ADJUSTABLE_LDO(VMMC, 0x68, 4, 0, 0x08),
-	TWL6030_ADJUSTABLE_LDO(VPP, 0x6c, 5, 0, 0x08),
-	TWL6030_ADJUSTABLE_LDO(VUSIM, 0x74, 7, 0, 0x08),
-	TWL6030_FIXED_LDO(VANA, 0x50, 2100, 15, 0, 0x08),
-	TWL6030_FIXED_LDO(VCXIO, 0x60, 1800, 16, 0, 0x08),
-	TWL6030_FIXED_LDO(VDAC, 0x64, 1800, 17, 0, 0x08),
-	TWL6030_FIXED_LDO(VUSB, 0x70, 3300, 18, 0, 0x08)
+	TWL6030_ADJUSTABLE_LDO(VAUX1_6030, 0x54, 1, 0, 0x21),
+	TWL6030_ADJUSTABLE_LDO(VAUX2_6030, 0x58, 2, 0, 0x21),
+	TWL6030_ADJUSTABLE_LDO(VAUX3_6030, 0x5c, 3, 0, 0x21),
+	TWL6030_ADJUSTABLE_LDO(VMMC, 0x68, 4, 0, 0x21),
+	TWL6030_ADJUSTABLE_LDO(VPP, 0x6c, 5, 0, 0x21),
+	TWL6030_ADJUSTABLE_LDO(VUSIM, 0x74, 7, 0, 0x21),
+	TWL6030_FIXED_LDO(VANA, 0x50, 2100, 15, 0, 0x21),
+	TWL6030_FIXED_LDO(VCXIO, 0x60, 1800, 16, 0, 0x21),
+	TWL6030_FIXED_LDO(VDAC, 0x64, 1800, 17, 0, 0x21),
+	TWL6030_FIXED_LDO(VUSB, 0x70, 3300, 18, 0, 0x21)
 };
 
 static int twlreg_probe(struct platform_device *pdev)
Signature · 1 line
1.5.4.7
Acked-by on v1 2/2
#2 lore raw

Re: [PATCH 2/2] twl6030: regulator: Configure STATE register instead of REMAP

Quoted 7 lines · Rajendra Nayak

This is no REMAP register on twl6030, instead there is a STATE register to drive a resource to a given state.
The state register can be used to specify what state the resource should enter when its associated with a GRP.
Register Bit field description is as below. The patch programmes the corresponding STATE registers for all LDO's to turn ON when assocaited with GRP_P1.

#3 lore raw

Re: [PATCH 2/2] twl6030: regulator: Configure STATE register instead of REMAP

In reply to Rajendra Nayak
Quoted 18 lines · Rajendra Nayak

This is no REMAP register on twl6030, instead there is a STATE register to drive a resource to a given state.
The state register can be used to specify what state the resource should enter when its associated with a GRP.
Register Bit field description is as below. The patch programmes the corresponding STATE registers for all LDO's to turn ON when assocaited with GRP_P1.

STATE REG:

Bit7   |Bit6   |Bit5   |Bit4  |Bit3  |Bit2  |Bit1   |Bit0
P3_GRP |P2_GRP |P1_GRP |RES   |RES   |RES   |State1 |State0

State can be specified as below
00: OFF
01: ON
10: OFF
11: SLEEP

Quoted 5 lines
Cc Liam Girdwood <[email protected]>
Cc Samuel Ortiz <[email protected]>
Cc Mark Brown <[email protected]>

---

Applied.

Thanks

Liam

Signature · 3 lines
Freelance Developer, SlimLogic Ltd
ASoC and Voltage Regulator Maintainer.
http://www.slimlogic.co.uk
#4 lore raw
In reply to Rajendra Nayak

drivers/regulator/twl-regulator.c twlldo_set_voltage(struct regulator_dev *rdev, int min_uV, int max_uV) · line 378

Quoted 11 lines · Rajendra Nayak
drivers/regulator/twl-regulator.c +7 -1
 		/* use the first in-range value */
-		if (min_uV <= uV && uV <= max_uV)
+		if (min_uV <= uV && uV <= max_uV) {
+			if (twl_class_is_6030())
+				/*
+				 * Use the below formula to calculate vsel
+				 * mV = 1000mv + 100mv * (vsel - 1)
+				 */
+				vsel = (LDO_MV(mV) - 1000)/100 + 1;
 			return twlreg_write(info, TWL_MODULE_PM_RECEIVER,
 							VREG_VOLTAGE, vsel);

This looks wrong - this code is inside a loop over vsel->voltage mappings, if that mapping is wrong and needs to be ignored (which is what this code is doing) then we shouldn't be looking at it at all.

#5 lore raw
Quoted 25 lines

-----Original Message-----
From: Mark Brown [mailto:[email protected]]
Sent: Wednesday, February 17, 2010 9:45 PM
To: Nayak, Rajendra
Cc: [email protected]; Liam Girdwood; Samuel Ortiz
Subject: Re: [PATCH 1/2] twl6030: regulator: Fix vsel calculations in set/get voltage apis

drivers/regulator/twl-regulator.c twlldo_set_voltage(struct regulator_dev *rdev, int min_uV, int max_uV)

Quoted 6 lines · Rajendra Nayak
drivers/regulator/twl-regulator.c +4 -1
 		/* use the first in-range value */
-		if (min_uV <= uV && uV <= max_uV)
+		if (min_uV <= uV && uV <= max_uV) {
+			if (twl_class_is_6030())
+				/*
+				 * Use the below formula to 

calculate vsel

drivers/regulator/twl-regulator.c twlldo_set_voltage(struct regulator_dev *rdev, int min_uV, int max_uV)

Quoted 4 lines
drivers/regulator/twl-regulator.c +3 -0
+				 * mV = 1000mv + 100mv * (vsel - 1)
+				 */
+				vsel = (LDO_MV(mV) - 1000)/100 + 1;
 			return twlreg_write(info, 

TWL_MODULE_PM_RECEIVER,

Quoted 1 line
Quoted 5 lines

VREG_VOLTAGE, vsel);

This looks wrong - this code is inside a loop over vsel->voltage mappings, if that mapping is wrong and needs to be ignored (which is what this code is doing) then we shouldn't be looking at it at all.

Mark,

Not sure if I understood your comment completely, but here's how I think it will work.

The for loop runs through all the valid voltages supported by the regulator and and when it finds the first in-range value programs the VOLTAGE register. For twl4030 the VOLTAGE register can be programmed with just the index of the table since the mapping was something like
1800mV = 0x0
2800mV = 0x1
2900mV = 0x2
3000mV = 0x3

Now for twl6030 I just have added an additional vsel calulation part after the in-range value is found, since using the table index directly as vsel would not work.

regards,
Rajendra

Quoted 1 line
#6 lore raw
Quoted 12 lines · Rajendra
Quoted 3 lines

This looks wrong - this code is inside a loop over vsel->voltage mappings, if that mapping is wrong and needs to be ignored (which is what this code is doing) then we shouldn't be looking at it at all.

Not sure if I understood your comment completely, but here's how I think it will work.

The for loop runs through all the valid voltages supported by the regulator and and when it finds the first in-range value programs the VOLTAGE register. For twl4030 the VOLTAGE register can be programmed with just the index of the table since the mapping was something like

Quoted 8 lines

1800mV = 0x0
2800mV = 0x1
2900mV = 0x2
3000mV = 0x3

Now for twl6030 I just have added an additional vsel calulation part after the in-range value is found, since using the table index directly as vsel would not work.

Right, but this means that the code is looking at one table of values to find a voltage it likes and then picking the actual selector from another set of values. This means that depending on the actual differences it may either pick a value which isn't valid for the TWL6030 or omit values which the TWL6030 can actually support. Either way from a code inspection point of view the code doesn't look good due to the mismatch between the two sets of selectors.

The TWL6030 case should be doing something more like other regulator drivers with straightforward mappings between selector and voltage (most of them) and not using the TWL4030 value table.

#7 lore raw
Quoted 52 lines

-----Original Message-----
From: Mark Brown [mailto:[email protected]]
Sent: Thursday, February 18, 2010 3:18 PM
To: Nayak, Rajendra
Cc: [email protected]; Liam Girdwood; Samuel Ortiz
Subject: Re: [PATCH 1/2] twl6030: regulator: Fix vsel calculations in set/get voltage apis

Quoted 2 lines · Rajendra
Quoted 2 lines

This looks wrong - this code is inside a loop over vsel->voltage mappings, if that mapping is wrong and needs to be

ignored (which is

Quoted 1 line
Quoted 1 line

what this code is doing) then we shouldn't be looking at

it at all.

Quoted 1 line

Not sure if I understood your comment completely, but

here's how I think

Quoted 3 lines

it will work.

The for loop runs through all the valid voltages supported

by the regulator

Quoted 1 line

and and when it finds the first in-range value programs the

VOLTAGE register.

Quoted 1 line

For twl4030 the VOLTAGE register can be programmed with

just the index

Quoted 8 lines

of the table since the mapping was something like

1800mV = 0x0
2800mV = 0x1
2900mV = 0x2
3000mV = 0x3

Now for twl6030 I just have added an additional vsel

calulation part after

Quoted 1 line

the in-range value is found, since using the table index

directly as vsel

Quoted 1 line

would not work.

Right, but this means that the code is looking at one table of values to
find a voltage it likes and then picking the actual selector from another set of values. This means that depending on the actual differences it may either pick a value which isn't valid for the TWL6030
or omit values which the TWL6030 can actually support. Either way from
a code inspection point of view the code doesn't look good due to the mismatch between the two sets of selectors.

Quoted 4 lines

The TWL6030 case should be doing something more like other regulator drivers with straightforward mappings between selector and voltage (most
of them) and not using the TWL4030 value table.

The code is'nt looking into twl4030 tables for finding the valid voltage for twl6030. There are seperate tables for twl4030 and twl6030 regulators with all the valid supported voltages.

The difference is, in case of twl4030 once you find a valid voltage from the right table you could just use the table index to program the register on twl4030.
In case of twl6030, once you find a valid voltage, again from the right table (meant for twl6030) you still need to derive
what value needs to be programmed in the register, so twl6030 understands it. Using the table index does not help.

Example:
In case of TWL4030

static const u16 VAUX1_VSEL_table[] = {
        2500, 2800, 3000, 3000, 3000, 3000,
};

If the request is to set 2800mv, look up the table, once found program the VOLTAGE register with the table index. In this case 0x1.

So the mapping is
VOLTAGE_REG=0x0, VAUX1=2500mv
VOLTAGE_REG=0x1, VAUX1=2800mv
VOLTAGE_REG=0x2, VAUX1=3000mv

In case of TWL6030

static const u16 VAUX1_6030_VSEL_table[] = {
        1000, 1300, 1800, 2500,
        2800, 2900, 3000, 3000,
};

if the request is to set 2800mv, look up the table, once found calulate the value to be programmed value = (2800 - 1000)/100 + 1 = 0x13

So here the mapping is
VOLTAGE_REG=0x10, VAUX1=2500mv
VOLTAGE_REG=0x13, VAUX1=2800mv
VOLTAGE_REG=0x15, VAUX1=3000mv

Without this patch the code goes ahead and programs the table index (In this case 0x4) which is wrong

Quoted 1 line
#8 lore raw
Quoted 9 lines · Rajendra
Quoted 4 lines

The TWL6030 case should be doing something more like other regulator drivers with straightforward mappings between selector and voltage (most
of them) and not using the TWL4030 value table.

The code is'nt looking into twl4030 tables for finding the valid voltage for twl6030. There are seperate tables for twl4030 and twl6030 regulators with all the valid supported voltages.

Quoted 7 lines

The difference is, in case of twl4030 once you find a valid voltage from the right table you could just use the table index to program the register on twl4030.
In case of twl6030, once you find a valid voltage, again from the right table (meant for twl6030) you still need to derive
what value needs to be programmed in the register, so twl6030 understands it. Using the table index does not help.

In that case the tables for TWL6030 are not ideal as-is and should either be fixed somehow, have a different function accessing them which makes their meaning clear or removed and replaced with just the formula.

I know what the code is trying to do, it just seems to be trying to do it at the wrong abstraction level and so looks buggy on code inspection. My point here is that the code looks wrong since it's iterating over a table giving voltage to selector mappings but then ignoring the selector value it comes up with and instead using a formula which at least gives way more options than most of the selector tables do. It's readability that's my focus here, the code possibly does do exactly what it ought to but it looks wrong.

Quoted 9 lines

In case of TWL6030

static const u16 VAUX1_6030_VSEL_table[] = {
        1000, 1300, 1800, 2500,
        2800, 2900, 3000, 3000,
};

if the request is to set 2800mv, look up the table, once found calulate the value to be programmed value = (2800 - 1000)/100 + 1 = 0x13

Are these really the only supported values (obviously you can set other selectors)? Note also that the name says "_VSEL_table" which is definitely no longer true, this isn't helping clarity at all.

#9 lore raw
Quoted 38 lines

-----Original Message-----
From: Mark Brown [mailto:[email protected]]
Sent: Thursday, February 18, 2010 4:11 PM
To: Nayak, Rajendra
Cc: [email protected]; Liam Girdwood; Samuel Ortiz
Subject: Re: [PATCH 1/2] twl6030: regulator: Fix vsel calculations in set/get voltage apis

Quoted 1 line · Rajendra
Quoted 1 line

The TWL6030 case should be doing something more like

other regulator

Quoted 6 lines
Quoted 3 lines

drivers with straightforward mappings between selector and voltage (most
of them) and not using the TWL4030 value table.

The code is'nt looking into twl4030 tables for finding the valid voltage for twl6030. There are seperate tables for twl4030

and twl6030

Quoted 6 lines

regulators with all the valid supported voltages.

The difference is, in case of twl4030 once you find a valid voltage from the right table you could just use the table index to program the register on twl4030.
In case of twl6030, once you find a valid voltage, again

from the right

Quoted 2 lines

table (meant for twl6030) you still need to derive what value needs to be programmed in the register, so

twl6030 understands

Quoted 1 line

it. Using the table index does not help.

In that case the tables for TWL6030 are not ideal as-is and should either be fixed somehow, have a different function accessing them which
makes their meaning clear or removed and replaced with just the formula.

Quoted 9 lines

I know what the code is trying to do, it just seems to be trying to do it at the wrong abstraction level and so looks buggy on code inspection.
My point here is that the code looks wrong since it's iterating over a table giving voltage to selector mappings but then ignoring the selector
value it comes up with and instead using a formula which at least gives
way more options than most of the selector tables do.

Mark, I get your point. The table values were derived from what the spec suggests, but I agree that its confusing that in theory the formula suggests there could be more selectors than whats present in the table.

I will check if its indeed possible to support all possible selectors between say a min/max range. And then have maybe seperate kind of tables for twl6030 which only specify min/max range and then use the formula to generate the right selector.

Quoted 21 lines

It's
readability
that's my focus here, the code possibly does do exactly what it ought to
but it looks wrong.

Quoted 9 lines

In case of TWL6030

static const u16 VAUX1_6030_VSEL_table[] = {
        1000, 1300, 1800, 2500,
        2800, 2900, 3000, 3000,
};

if the request is to set 2800mv, look up the table, once found calulate the value to be programmed value = (2800 - 1000)/100 + 1 = 0x13

Are these really the only supported values (obviously you can set other
selectors)? Note also that the name says "_VSEL_table" which is definitely no longer true, this isn't helping clarity at all.

#10 lore raw
Quoted 8 lines · Rajendra

Mark, I get your point. The table values were derived from what the spec suggests, but I agree that its confusing that in theory the formula suggests there could be more selectors than whats present in the table.

I will check if its indeed possible to support all possible selectors between say a min/max range. And then have maybe seperate kind of tables for twl6030 which only specify min/max range and then use the formula to generate the right selector.

Another option if you do need the supported voltage list would be to rename the table and have a separate function to do set on twl6030 which makes the intended usage clear.

#11 lore raw
Quoted 22 lines

-----Original Message-----
From: Mark Brown [mailto:[email protected]]
Sent: Thursday, February 18, 2010 5:13 PM
To: Nayak, Rajendra
Cc: [email protected]; Liam Girdwood; Samuel Ortiz
Subject: Re: [PATCH 1/2] twl6030: regulator: Fix vsel calculations in set/get voltage apis

Quoted 1 line · Rajendra

Mark, I get your point. The table values were derived from

what the spec

Quoted 2 lines

suggests, but I agree that its confusing that in theory the formula suggests there could be more selectors than whats present

in the table.

Quoted 1 line

I will check if its indeed possible to support all possible

selectors

Quoted 1 line

between say a min/max range. And then have maybe seperate

kind of tables

Quoted 8 lines
Quoted 1 line

for twl6030 which only specify min/max range and then use

the formula to

Quoted 1 line

generate the right selector.

Another option if you do need the supported voltage list would be to rename the table and have a separate function to do set on twl6030 which
makes the intended usage clear.

Mark, It turns out, all twl6030 regulators can be programmed from 1.0v to 3.3v with 100mV steps. The spec I was referring to was specfic to the OMAP4 platform I am working on. Hence, it does not prevent someone using the PMIC on a different platform to have different voltage requirements.
So I was thinking of removing all the twl6030 tables and just use the formula to get the selector and make sure they fall in the allowed range (1.0v to 3.3v).
If that sounds good then I can post a patch for this.

Quoted 1 line
#12 lore raw
Quoted 4 lines · Rajendra

So I was thinking of removing all the twl6030 tables and just use the formula to get the selector and make sure they fall in the allowed range (1.0v to 3.3v).
If that sounds good then I can post a patch for this.

Yes, that sounds good to me.