mfd: twl4030-power: Enable off-idle configuration when booted with device tree

Looks like we can still hit the issue of wrong load order of twl4030 configuration. If we have a sleep configuration loaded, and do a warm reset, the device can hang while initializing the wakeup12 sequence. We do have a warning message about wrong order of twl4030 configuration, but in this case it does not help as the sleep configuration was loaded during the previous boot and the state of twl4030 is maintained throughout the warm reset.

Fix the issue by clearing any existing sleep configuration before we load the warm reset configuration.

Cc Peter Ujfalusi <[email protected]>

Signed-off-by Tony Lindgren <[email protected]>

---
 drivers/mfd/twl4030-power.c | 6 ++++++
 1 file changed, 6 insertions(+)

Quoted 8 lines · Tony Lindgren

Looks like we can still hit the issue of wrong load order of twl4030 configuration. If we have a sleep configuration loaded, and do a warm reset, the device can hang while initializing the wakeup12 sequence. We do have a warning message about wrong order of twl4030 configuration, but in this case it does not help as the sleep configuration was loaded during the previous boot and the state of twl4030 is maintained throughout the warm reset.

Quoted 8 lines

Fix the issue by clearing any existing sleep configuration before we load the warm reset configuration.

Cc Peter Ujfalusi <[email protected]>

Signed-off-by Tony Lindgren <[email protected]>

---
 drivers/mfd/twl4030-power.c | 6 ++++++
 1 file changed, 6 insertions(+)

Can't vouch for the semantics of the patch, but the code looks good.

Acked-by Lee Jones <[email protected]>

Can I just apply this, or would you like Peter to review it first?

Quoted 19 lines

Quoted 8 lines

Looks like we can still hit the issue of wrong load order of twl4030 configuration. If we have a sleep configuration loaded, and do a warm reset, the device can hang while initializing the wakeup12 sequence. We do have a warning message about wrong order of twl4030 configuration, but in this case it does not help as the sleep configuration was loaded during the previous boot and the state of twl4030 is maintained throughout the warm reset.

Quoted 8 lines

Fix the issue by clearing any existing sleep configuration before we load the warm reset configuration.

Cc Peter Ujfalusi <[email protected]>

Signed-off-by Tony Lindgren <[email protected]>

---
 drivers/mfd/twl4030-power.c | 6 ++++++
 1 file changed, 6 insertions(+)

Can't vouch for the semantics of the patch, but the code looks good.

Quoted 3 lines

Acked-by Lee Jones <[email protected]>

Can I just apply this, or would you like Peter to review it first?

As far as I'm concerned you can go ahead an apply, it's been floating on the list for a while and Peter is probably busy with other things.

Quoted 21 lines

Quoted 21 lines

Quoted 8 lines

Looks like we can still hit the issue of wrong load order of twl4030 configuration. If we have a sleep configuration loaded, and do a warm reset, the device can hang while initializing the wakeup12 sequence. We do have a warning message about wrong order of twl4030 configuration, but in this case it does not help as the sleep configuration was loaded during the previous boot and the state of twl4030 is maintained throughout the warm reset.

Quoted 8 lines

Fix the issue by clearing any existing sleep configuration before we load the warm reset configuration.

Cc Peter Ujfalusi <[email protected]>

Signed-off-by Tony Lindgren <[email protected]>

---
 drivers/mfd/twl4030-power.c | 6 ++++++
 1 file changed, 6 insertions(+)

Can't vouch for the semantics of the patch, but the code looks good.

Acked-by Lee Jones <[email protected]>

Quoted 5 lines

Quoted 1 line

Can I just apply this, or would you like Peter to review it first?

As far as I'm concerned you can go ahead an apply, it's been floating on the list for a while and Peter is probably busy with other things.

No problem. Just a couple of things for you to look at in the other patches, then we can go ahead and apply for v3.16.

Quoted 27 lines

Quoted 26 lines

Quoted 17 lines

Quoted 8 lines

Looks like we can still hit the issue of wrong load order of twl4030 configuration. If we have a sleep configuration loaded, and do a warm reset, the device can hang while initializing the wakeup12 sequence. We do have a warning message about wrong order of twl4030 configuration, but in this case it does not help as the sleep configuration was loaded during the previous boot and the state of twl4030 is maintained throughout the warm reset.

Quoted 8 lines

Fix the issue by clearing any existing sleep configuration before we load the warm reset configuration.

Cc Peter Ujfalusi <[email protected]>

Signed-off-by Tony Lindgren <[email protected]>

---
 drivers/mfd/twl4030-power.c | 6 ++++++
 1 file changed, 6 insertions(+)

Quoted 5 lines

Can't vouch for the semantics of the patch, but the code looks good.

Acked-by Lee Jones <[email protected]>

Can I just apply this, or would you like Peter to review it first?

As far as I'm concerned you can go ahead an apply, it's been floating on the list for a while and Peter is probably busy with other things.

Quoted 2 lines

No problem. Just a couple of things for you to look at in the other patches, then we can go ahead and apply for v3.16.

OK thanks yeah saw those. Will fix them up and test today. Then maybe you can apply them to the immutable branch we set earlier or I can send you a pull request against that branch whichever you prefer.

We have these bits partially defined in two different places, so let's fix them up and add defines for the missing bits. These bits are the same for P1_SW_EVENTS, P2_SW_EVENTS and P3_SW_EVENTS.

Cc Peter Ujfalusi <[email protected]>

Signed-off-by Tony Lindgren <[email protected]>

---
 drivers/mfd/twl4030-power.c | 23 +++++++++++++----------
 1 file changed, 13 insertions(+), 10 deletions(-)

drivers/mfd/twl4030-power.c +0 -0

index 1b30d8a..c0e4fc3 100644

Quoted 18 lines

We have these bits partially defined in two different places, so let's fix them up and add defines for the missing bits. These bits are the same for P1_SW_EVENTS, P2_SW_EVENTS and P3_SW_EVENTS.

Cc Peter Ujfalusi <[email protected]>

Signed-off-by Tony Lindgren <[email protected]>

---
 drivers/mfd/twl4030-power.c | 23 +++++++++++++----------
 1 file changed, 13 insertions(+), 10 deletions(-)

drivers/mfd/twl4030-power.c +0 -0

index 1b30d8a..c0e4fc3 100644
@@ -35,7 +35,14 @@
 static u8 twl4030_start_script_address = 0x2b;
 

drivers/mfd/twl4030-power.c line 37

Quoted 1 line

drivers/mfd/twl4030-power.c +0 -0

 #define PWR_P1_SW_EVENTS	0x10

BIT(4)?

drivers/mfd/twl4030-power.c line 46

Quoted 9 lines

drivers/mfd/twl4030-power.c +7 -0

+#define PWR_STOPON_PRWON	(1 << 6)
+#define PWR_STOPON_SYSEN	(1 << 5)
+#define PWR_ENABLE_WARMRESET	(1 << 4)
+#define PWR_LVL_WAKEUP		(1 << 3)
+#define PWR_DEVACT		(1 << 2)
+#define PWR_DEVSLP		(1 << 1)
 #define PWR_DEVOFF		(1 << 0)
+
 #define SEQ_OFFSYNC		(1 << 0)

Please use the BIT() macro for these.

[...]

Once fixed, apply my Acked-by on re-submission.

The twl4030 PMIC needs to be configured properly for things like warm reset and deeper idle states so the PMIC manages the regulators properly based on the hardware triggers from the SoC.

For example, when rebooting an OMAP3530 at 125 MHz, it hangs. With this patch, TWL4030 will be reset when a warm reset occures. This way the OMAP3530 does not hang on reboot.

Let's use this as the default when compatible = "ti,twl4030-power". Other more complicated configurations can be added to the driver based on other compatible flags.

Based on earlier patch by Matthias Brugger <[email protected]>:

http://lists.infradead.org/pipermail/linux-arm-kernel/2013-January/144165.html

And Lesly A M <[email protected]>:

https://github.com/openembedded/openembedded/blob/master/recipes/linux/linux-omap-2.6.39/mfd/0010-MFD-TWL4030-power-scripts-for-OMAP3-boards.patch

For more information about twl4030 configuration for the "power scripts" see:

http://www.omappedia.com/wiki/TWL4030_power_scripts

Cc Matthias Brugger <[email protected]>

Cc Robert Nelson <[email protected]>

Cc Peter De Schrijver <[email protected]>

Cc Samuel Ortiz <[email protected]>

Cc Lee Jones <[email protected]>

Signed-off-by Tony Lindgren <[email protected]>

---
 .../devicetree/bindings/mfd/twl4030-power.txt      |  7 +-
 drivers/mfd/twl4030-power.c                        | 99 +++++++++++++++++++---
 include/linux/i2c/twl.h                            |  3 +
 3 files changed, 95 insertions(+), 14 deletions(-)

Documentation/devicetree/bindings/mfd/twl4030-power.txt +6 -1

index 8e15ec3..b906116 100644
@@ -5,7 +5,12 @@ to control the power resources, including power scripts. For now, the
 binding only supports the complete shutdown of the system after poweroff.
 
 Required properties:
-- compatible : must be "ti,twl4030-power"
+- compatible : must be one of the following
+	"ti,twl4030-power"
+	"ti,twl4030-power-reset"
+
+The use of ti,twl4030-power-reset is recommended at least on
+3530 that needs a special configuration for warm reset to work.
 
 Optional properties:
 - ti,use_poweroff: With this flag, the chip will initiates an ACTIVE-to-OFF or

drivers/mfd/twl4030-power.c +3 -0

index c0e4fc3..b61b725 100644
@@ -29,6 +29,7 @@
 #include <linux/i2c/twl.h>
 #include <linux/platform_device.h>
 #include <linux/of.h>
+#include <linux/of_device.h>
 
 #include <asm/mach-types.h>
 

@@ -128,6 +129,30 @@ static u8 res_config_addrs[] = {
 	[RES_MAIN_REF]	= 0x94,
 };
 
+/*
+ * Usable values for .remap_sleep and .remap_off

drivers/mfd/twl4030-power.c +12 -0

+#define TWL_RESOURCE_OFF(res)						\
+	{ MSG_SINGULAR(DEV_GRP_NULL, (res), RES_STATE_OFF), 0x02 }
+#define TWL_RESOURCE_RESET(res)						\
+	{ MSG_SINGULAR(DEV_GRP_NULL, (res), RES_STATE_WRST), 0x02 }
+#define TWL_RESOURCE_GROUP_RESET(group, type1, type2)			\
+	{ MSG_BROADCAST(DEV_GRP_NULL, (group), (type1), (type2),	\
+		RES_STATE_WRST), 0x02 }
+#define TWL_REMAP_SLEEP(res, devgrp, typ, typ2)				\
+	{ .resource = (res), .devgroup = (devgrp),			\
+	  .type = (typ), .type2 = (typ2),				\
+	  .remap_off = TWL_REMAP_OFF, .remap_sleep = TWL_REMAP_SLEEP, }
+
 static int twl4030_write_script_byte(u8 address, u8 byte)
 {
 	int err;

Quoted 2 lines

http://www.omappedia.com/wiki/TWL4030_power_scripts

Do you really want this bumph in the commit log?

Review

Quoted 2 lines

(quoted diff) +2 -0

+#define TWL_RESOURCE_ON(res)						\
+	{ MSG_SINGULAR(DEV_GRP_NULL, (res), RES_STATE_ACTIVE), 0x02 }

Was ist das? Magic, floating, fluffy numbers.

Review

Quoted 3 lines

(quoted diff) +2 -1

-static int twl4030_power_configure_scripts(struct twl4030_power_data *pdata)
+static int
+twl4030_power_configure_scripts(const struct twl4030_power_data *pdata)

Is this doing anything besides some sneaky clean-up? If not, it might be worth mentioning that you're taking advantage of these changes to conduct such activities.

Review static int twl4030_power_configure_scripts(struct twl4030_power_data *pdata) · line 548

Quoted 10 lines

{

(quoted diff) +2 -1

 	int err;
 	int i;

@@ -518,7 +544,8 @@ static int twl4030_power_configure_scripts(struct twl4030_power_data *pdata)
 	return 0;
 }
 
-static int twl4030_power_configure_resources(struct twl4030_power_data *pdata)
+static int
+twl4030_power_configure_resources(const struct twl4030_power_data *pdata)

Ditto.

Quoted 7 lines

(quoted diff) +0 -0

 {
 	struct twl4030_resconfig *resconfig = pdata->resource_config;
 	int err;

@@ -550,7 +577,7 @@ void twl4030_power_off(void)
 		pr_err("TWL4030 Unable to power off\n");
 }
 

Review void twl4030_power_off(void) · line 581

Quoted 3 lines

(quoted diff) +1 -1

-static bool twl4030_power_use_poweroff(struct twl4030_power_data *pdata,
+static bool twl4030_power_use_poweroff(const struct twl4030_power_data *pdata,
 					struct device_node *node)

Etc.

These settings are based on the "Recommended Sleep Sequences for the Zoom Platform" pdf at:

http://omappedia.com/wiki/File:Recommended_Sleep_Sequences_Zoom.pdf

These settings assume most of the regulators are under control of Linux, and twl4030 only cuts off VDD1 and VDD2 during off-idle as Linux cannot do it.

For any board specific changes to these, let's patch them in as changes to the generic data in the follow-up patches. This keeps the board specific changes small.

Note that this does not consider the twl5030 errata 27 and 28. That can be added later on after it has been tested. For more information about errata 27 and 28, please see:

https://github.com/openembedded/openembedded/blob/master/recipes/linux/linux-omap-2.6.39/mfd/0012-MFD-TWL4030-workaround-changes-for-Erratum-27.patch

Cc Peter De Schrijver <[email protected]>

Cc Samuel Ortiz <[email protected]>

Acked-by Lee Jones <[email protected]>

Signed-off-by Tony Lindgren <[email protected]>

---
 .../devicetree/bindings/mfd/twl4030-power.txt      |   4 +
 drivers/mfd/twl4030-power.c                        | 103 +++++++++++++++++++++
 2 files changed, 107 insertions(+)

Assuming the magic numbers are replaced by transparent #defines:

Acked-by Lee Jones <[email protected]>

With the recommended twl4030 configuration added, we can now add board specific changes as modifications to the recommended configuration.

Note that the data is private to this driver, and the data must always have a NULL resource in the sentinel.

Cc Peter De Schrijver <[email protected]>

Cc Samuel Ortiz <[email protected]>

Cc Lee Jones <[email protected]>

Signed-off-by Tony Lindgren <[email protected]>

---
 drivers/mfd/twl4030-power.c | 21 +++++++++++++++++++++
 include/linux/i2c/twl.h     |  1 +
 2 files changed, 22 insertions(+)

Ah, I forgot to Ack this last time, didn't I?

Acked-by Lee Jones <[email protected]>

Quoted 4 lines · Lee Jones

Ah, I forgot to Ack this last time, didn't I?

Acked-by Lee Jones <[email protected]>

Yeah, here's this one updated with your ack.

8< ---------------------------
From: Tony Lindgren <[email protected]>
Date: Tue, 20 May 2014 11:17:54 -0700
Subject: [PATCH] mfd: twl4030-power: Add support for board specific configuration

With the recommended twl4030 configuration added, we can now add board specific changes as modifications to the recommended configuration.

Note that the data is private to this driver, and the data must always have a NULL resource in the sentinel.

Cc Peter De Schrijver <[email protected]>

Cc Samuel Ortiz <[email protected]>

Acked-by Lee Jones <[email protected]>

Signed-off-by Tony Lindgren <[email protected]>

drivers/mfd/twl4030-power.c +21 -0

@@ -567,13 +567,34 @@ twl4030_power_configure_scripts(const struct twl4030_power_data *pdata)
 	return 0;
 }
 
+static void twl4030_patch_rconfig(struct twl4030_resconfig *common,
+				  struct twl4030_resconfig *board)
+{
+	while (common->resource) {
+		struct twl4030_resconfig *b = board;
+
+		while (b->resource) {
+			if (b->resource == common->resource) {
+				*common = *b;
+				break;
+			}
+			b++;
+		}
+		common++;
+	}
+}
+
 static int
 twl4030_power_configure_resources(const struct twl4030_power_data *pdata)
 {
 	struct twl4030_resconfig *resconfig = pdata->resource_config;
+	struct twl4030_resconfig *boardconf = pdata->board_config;
 	int err;
 
 	if (resconfig) {
+		if (boardconf)
+			twl4030_patch_rconfig(resconfig, boardconf);
+
 		while (resconfig->resource) {
 			err = twl4030_configure_resource(resconfig);
 			if (err)

include/linux/i2c/twl.h +1 -0

@@ -674,6 +674,7 @@ struct twl4030_power_data {
 	struct twl4030_script **scripts;
 	unsigned num;
 	struct twl4030_resconfig *resource_config;
+	struct twl4030_resconfig *board_config;
 #define TWL4030_RESCONFIG_UNDEF	((u8)-1)
 	bool use_poweroff;	/* Board is wired for TWL poweroff */
 };

Quoted 2 lines · Tony Lindgren

Quoted 1 line · Lee Jones

Quoted 5 lines

Quoted 3 lines

Ah, I forgot to Ack this last time, didn't I?

Acked-by Lee Jones <[email protected]>

Yeah, here's this one updated with your ack.

Has the code changed at all, or are you just reposting with my Ack applied?

Quoted 6 lines

Quoted 6 lines · Lee Jones

Quoted 4 lines

Ah, I forgot to Ack this last time, didn't I?

Acked-by Lee Jones <[email protected]>

Quoted 4 lines

Quoted 1 line

Yeah, here's this one updated with your ack.

Has the code changed at all, or are you just reposting with my Ack applied?

No changes, I just reposted this with your ack as you seemed to prefer for some of the other patches :)

Here's an updated set of patches to enable low-power idle modes for some omap3 boards when booted with device tree.

This series when applied on top of the patches in tread "[PATCH 00/11] Fixes for omap PM for making omap3 DT only", or omap-for-v3.16/pm branch.

Lee, if these patches look OK to you, please feel free to pick them, preferrably to the immutable branch you already have set up so I can also merge them in to be able to have things working in my branch properly for PM.

Tony Lindgren (7):
mfd: twl4030-power: Fix hang on reboot if sleep configuration was
loaded earlier
mfd: twl4030-power: Fix some defines for SW_EVENTS
mfd: twl4030-power: Add generic reset configuration
mfd: twl4030-power: Add recommended idle configuration
mfd: twl4030-power: Add support for board specific configuration
mfd: twl4030power: Add a configuration to turn off oscillator during
off-idle
ARM: dts: Enable twl4030 off-idle configuration for selected omaps

 .../devicetree/bindings/mfd/twl4030-power.txt      |  17 +-
 arch/arm/boot/dts/omap3-beagle-xm.dts              |   6 +
 arch/arm/boot/dts/omap3-evm-common.dtsi            |   7 +
 arch/arm/boot/dts/omap3-n900.dts                   |   5 +
 drivers/mfd/twl4030-power.c                        | 269 +++++++++++++++++++--
 include/linux/i2c/twl.h                            |   4 +
 6 files changed, 284 insertions(+), 24 deletions(-)

Quoted 7 lines

Here's an updated set of patches to enable low-power idle modes for some omap3 boards when booted with device tree.

This series when applied on top of the patches in tread "[PATCH 00/11] Fixes for omap PM for making omap3 DT only", or omap-for-v3.16/pm branch.

Quoted 4 lines

Lee, if these patches look OK to you, please feel free to pick them, preferrably to the immutable branch you already have set up so I can also merge them in to be able to have things working in my branch properly for PM.

Okay, it looks like this series is good to go now. If you believe so too, give me the nod and I'll apply.

Here's an updated set of patches to enable low-power idle modes for some omap3 boards when booted with device tree.

This series when applied on top of the patches in tread "[PATCH 00/11] Fixes for omap PM for making omap3 DT only", or omap-for-v3.16/pm branch.

Lee, if these patches look OK to you, please feel free to pick them, preferrably to the immutable branch you already have set up so I can also merge them in to be able to have things working in my branch properly for PM.

Tony Lindgren (7):
mfd: twl4030-power: Fix hang on reboot if sleep configuration was
loaded earlier
mfd: twl4030-power: Fix some defines for SW_EVENTS
mfd: twl4030-power: Add generic reset configuration
mfd: twl4030-power: Add recommended idle configuration
mfd: twl4030-power: Add support for board specific configuration
mfd: twl4030power: Add a configuration to turn off oscillator during
off-idle
ARM: dts: Enable twl4030 off-idle configuration for selected omaps

 .../devicetree/bindings/mfd/twl4030-power.txt      |  17 +-
 arch/arm/boot/dts/omap3-beagle-xm.dts              |   6 +
 arch/arm/boot/dts/omap3-evm-common.dtsi            |   7 +
 arch/arm/boot/dts/omap3-n900.dts                   |   5 +
 drivers/mfd/twl4030-power.c                        | 269 +++++++++++++++++++--
 include/linux/i2c/twl.h                            |   4 +
 6 files changed, 284 insertions(+), 24 deletions(-)

Quoted 11 lines

Here's an updated set of patches to enable low-power idle modes for some omap3 boards when booted with device tree.

This series when applied on top of the patches in tread "[PATCH 00/11] Fixes for omap PM for making omap3 DT only", or omap-for-v3.16/pm branch.

Lee, if these patches look OK to you, please feel free to pick them, preferrably to the immutable branch you already have set up so I can also merge them in to be able to have things working in my branch properly for PM.

Whoops, nearly forgot about this patchset.

Tony, can you resend it with all your fix-ups please. The 8< doesn't work well with my patch application 'system'.

Quoted 12 lines

Quoted 11 lines

Here's an updated set of patches to enable low-power idle modes for some omap3 boards when booted with device tree.

This series when applied on top of the patches in tread "[PATCH 00/11] Fixes for omap PM for making omap3 DT only", or omap-for-v3.16/pm branch.

Lee, if these patches look OK to you, please feel free to pick them, preferrably to the immutable branch you already have set up so I can also merge them in to be able to have things working in my branch properly for PM.

Quoted 4 lines

Whoops, nearly forgot about this patchset.

Tony, can you resend it with all your fix-ups please. The 8< doesn't work well with my patch application 'system'.

Sure I'll do that. FYI, these kind of patches should apply just fine when piped to git-am --scissors.