regulator: of: simplifing the parsing code
PATCH 1/1 regulator: of: simplifing the parsing code
in case of_property_read_u32 fails, it keeps the parameter unchanged so no need to test if its success and then assign the value
---
Hi Mark,
I also have concern related to how we are passing 'regulator-mode' and 'regulator-initial-mode'. Currently this require a extra function to be set in 'of_map_mode', which can be avoided.
These two parameters can be set directly from the device tree as in below patch:
https://lkml.org/lkml/2014/1/16/263
All drivers can have only out of 4 predefined values for these parameters, define in linux/iregulator/consumer.h, its not driver specific. Please let me know your comments, if this suits you I will send a one more patch on top of this it will further simplify this code.
Regards,
Saurabh
drivers/regulator/of_regulator.c | 41 +++++++++++++++++----------------------- 1 file changed, 17 insertions(+), 24 deletions(-)
drivers/regulator/of_regulator.c +17 -24
@@ -51,16 +51,14 @@ static void of_get_regulation_constraints(struct device_node *np,if (min_uV && max_uV && constraints->min_uV == constraints->max_uV) constraints->apply_uV = true; -if (!of_property_read_u32 (np,"regulator-microvolt-offset" , &pval)) - constraints->uV_offset = pval; -if (!of_property_read_u32 (np,"regulator-min-microamp" , &pval)) - constraints->min_uA = pval; -if (!of_property_read_u32 (np,"regulator-max-microamp" , &pval)) - constraints->max_uA = pval; - -if (!of_property_read_u32 (np,"regulator-input-current-limit-microamp" , - &pval)) - constraints->ilim_uA = pval; +of_property_read_u32 (np,"regulator-microvolt-offset" , + &constraints->uV_offset); +of_property_read_u32 (np,"regulator-min-microamp" , + &constraints->min_uA); +of_property_read_u32 (np,"regulator-max-microamp" , + &constraints->max_uA); +of_property_read_u32 (np,"regulator-input-current-limit-microamp" , + &constraints->ilim_uA);/* Current change possible? */ if (constraints->min_uA != constraints->max_uA)
@@ -79,17 +77,13 @@ static void of_get_regulation_constraints(struct device_node *np,if (of_property_read_bool (np,"regulator-allow-set-load" )) constraints->valid_ops_mask |=REGULATOR_CHANGE_DRMS ; - ret =of_property_read_u32 (np,"regulator-ramp-delay" , &pval); -if (!ret) { -if (pval) - constraints->ramp_delay = pval; -else +if (!of_property_read_u32 (np,"regulator-ramp-delay" , + &constraints->ramp_delay)) +if (!constraints->ramp_delay) constraints->ramp_disable = true; - } - ret =of_property_read_u32 (np,"regulator-enable-ramp-delay" , &pval); -if (!ret) - constraints->enable_time = pval; +of_property_read_u32 (np,"regulator-enable-ramp-delay" , + &constraints->enable_time); constraints->soft_start =of_property_read_bool (np,"regulator-soft-start" );
@@ -107,8 +101,8 @@ static void of_get_regulation_constraints(struct device_node *np, } } -if (!of_property_read_u32 (np,"regulator-system-load" , &pval)) - constraints->system_load = pval; +of_property_read_u32 (np,"regulator-system-load" , + &constraints->system_load); constraints->over_current_protection =of_property_read_bool (np,"regulator-over-current-protection" );
@@ -154,9 +148,8 @@ static void of_get_regulation_constraints(struct device_node *np, "regulator-off-in-suspend")) suspend_state->disabled = true; -if (!of_property_read_u32 (suspend_np, -"regulator-suspend-microvolt" , &pval)) - suspend_state->uV = pval; +of_property_read_u32 (suspend_np, +"regulator-suspend-microvolt" , &suspend_state->uV);of_node_put (suspend_np); suspend_state =NULL ;
Signature · 1 line
1.9.1
Mark Brown
maintainerQuoted 2 lines · Saurabh Sengar
in case
of_property_read_u32fails, it keeps the parameter unchanged so no need to test if its success and then assign the value
On the other hand the current code doesn't require us to know that detail... it's really not entirely clear to me that this is a great win.
Quoted 5 lines
I also have concern related to how we are passing 'regulator-mode' and 'regulator-initial-mode'. Currently this require a extra function to be set in '
of_map_mode', which can be avoided.
These two parameters can be set directly from the device tree as in below patch:
https://lkml.org/lkml/2014/1/16/263
No, they can't - the values we set for modes in the DT are defined per regulator and we need to translate these into the regulator API definitions.
drivers/regulator/of_regulator.c static void of_get_regulation_constraints(struct device_node *np, · line 55
Quoted 2 lines
drivers/regulator/of_regulator.c +2
-0+of_property_read_u32 (np,"regulator-microvolt-offset" , + &constraints->uV_offset);
The indentation for these is very strange - the second line is indented a really long way.
Quoted 8 lines · Mark Brown
Quoted 5 lines · Saurabh Sengar
I also have concern related to how we are passing 'regulator-mode' and 'regulator-initial-mode'. Currently this require a extra function to be set in '
of_map_mode', which can be avoided.
These two parameters can be set directly from the device tree as in below patch:
https://lkml.org/lkml/2014/1/16/263
Quoted 4 lines
No, they can't - the values we set for modes in the DT are defined per regulator and we need to translate these into the regulator API definitions.
What I want to say is there are only 4 possible values which can be set for initial_mode and suspend_state modes:
#define REGULATOR_MODE_FAST 0x1 #define REGULATOR_MODE_NORMAL 0x2 #define REGULATOR_MODE_IDLE 0x4 #define REGULATOR_MODE_STANDBY 0x8
these are defined in include/linux/regulator/consumer.h (core regulator file). 'of_map_modes' of each regulator driver ultimately set these values only. I want to suggest that these values can directly be pass from dts. Different dts can pass different values according to the regulator they are using,
no need to decode them from of_map_modes function.
Let me know your comments on this.
Regards,
Saurabh
Mark Brown
maintainerQuoted 7 lines · Saurabh Sengar
Quoted 5 lines · Mark Brown
On Fri, Nov 20, 2015 at 01:27:27PM +0530, Saurabh Sengar wrote:
No, they can't - the values we set for modes in the DT are defined per regulator and we need to translate these into the regulator API definitions.
Quoted 7 lines
What I want to say is there are only 4 possible values which can be set for
initial_modeandsuspend_statemodes:#define REGULATOR_MODE_FAST 0x1 #define REGULATOR_MODE_NORMAL 0x2 #define REGULATOR_MODE_IDLE 0x4 #define REGULATOR_MODE_STANDBY 0x8
No, the regulator can specify *any* value in their bindings - they don't have to be those values, they should be values that are relevant to the hardware rather than values that mean something in the current regulator API abstractions. Things like the values that are specified in the datasheet for example.