[PATCH] power: twl4030_charger: clear IRQs after handling them

[PATCH] power: twl4030_charger: clear IRQs after handling them

[Felipe Balbi]

TRM says we *must* write 1 to each bit we're handling
in order to clear the IRQ status and bring IRQ line
low. This patch implements that.

Signed-off-by: Felipe Balbi <balbi@ti.com>
---

Russell, I don't have HW to test, but this should
solve the problem you saw when not using battery
with Zoom board. Let me know if it doesn't.

cheers

 drivers/power/twl4030_charger.c | 16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/drivers/power/twl4030_charger.c b/drivers/power/twl4030_charger.c
index f141088..b090842 100644
--- a/drivers/power/twl4030_charger.c
+++ b/drivers/power/twl4030_charger.c
@@ -301,12 +301,24 @@ static irqreturn_t twl4030_bci_interrupt(int irq, void *arg)
 	ret = twl_i2c_read_u8(TWL4030_MODULE_INTERRUPTS, &irqs1,
 			      TWL4030_INTERRUPTS_BCIISR1A);
 	if (ret < 0)
-		return IRQ_HANDLED;
+		return IRQ_NONE;
+
+	/* clear IRQs early */
+	ret = twl_i2c_write_u8(TWL4030_MODULE_INTERRUPTS, irqs1,
+			TWL4030_INTERRUPTS_BCIISR1A);
+	if (ret < 0)
+		return IRQ_NONE;
 
 	ret = twl_i2c_read_u8(TWL4030_MODULE_INTERRUPTS, &irqs2,
 			      TWL4030_INTERRUPTS_BCIISR2A);
 	if (ret < 0)
-		return IRQ_HANDLED;
+		return IRQ_NONE;
+
+	/* clear IRQs early */
+	ret = twl_i2c_write_u8(TWL4030_MODULE_INTERRUPTS, irqs2,
+			      TWL4030_INTERRUPTS_BCIISR2A);
+	if (ret < 0)
+		return IRQ_NONE;
 
 	dev_dbg(bci->dev, "BCI irq %02x %02x\n", irqs2, irqs1);
 
-- 
1.9.1.286.g5172cb3

Re: [PATCH] power: twl4030_charger: clear IRQs after handling them

[Tony Lindgren]

* Felipe Balbi <balbi@ti.com> [140416 08:18]:
> TRM says we *must* write 1 to each bit we're handling
> in order to clear the IRQ status and bring IRQ line
> low. This patch implements that.
> 
> Signed-off-by: Felipe Balbi <balbi@ti.com>
> ---
> 
> Russell, I don't have HW to test, but this should
> solve the problem you saw when not using battery
> with Zoom board. Let me know if it doesn't.

BTW, looks like we're enabling BCI automatically in twl4030.dtsi
while the legacy booting does not have TWL_COMMON_PDATA_BCI
enabled for LDP. Anyways, for LDP BCI should be enabled for
sure, that's the only way to power at least the earlier LDP
revisions.

>  drivers/power/twl4030_charger.c | 16 ++++++++++++++--
>  1 file changed, 14 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/power/twl4030_charger.c b/drivers/power/twl4030_charger.c
> index f141088..b090842 100644
> --- a/drivers/power/twl4030_charger.c
> +++ b/drivers/power/twl4030_charger.c
> @@ -301,12 +301,24 @@ static irqreturn_t twl4030_bci_interrupt(int irq, void *arg)
>  	ret = twl_i2c_read_u8(TWL4030_MODULE_INTERRUPTS, &irqs1,
>  			      TWL4030_INTERRUPTS_BCIISR1A);
>  	if (ret < 0)
> -		return IRQ_HANDLED;
> +		return IRQ_NONE;
> +
> +	/* clear IRQs early */
> +	ret = twl_i2c_write_u8(TWL4030_MODULE_INTERRUPTS, irqs1,
> +			TWL4030_INTERRUPTS_BCIISR1A);
> +	if (ret < 0)
> +		return IRQ_NONE;
>  
>  	ret = twl_i2c_read_u8(TWL4030_MODULE_INTERRUPTS, &irqs2,
>  			      TWL4030_INTERRUPTS_BCIISR2A);
>  	if (ret < 0)
> -		return IRQ_HANDLED;
> +		return IRQ_NONE;
> +
> +	/* clear IRQs early */
> +	ret = twl_i2c_write_u8(TWL4030_MODULE_INTERRUPTS, irqs2,
> +			      TWL4030_INTERRUPTS_BCIISR2A);
> +	if (ret < 0)
> +		return IRQ_NONE;
>  
>  	dev_dbg(bci->dev, "BCI irq %02x %02x\n", irqs2, irqs1);
>  
> -- 
> 1.9.1.286.g5172cb3
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: [PATCH] power: twl4030_charger: clear IRQs after handling them

[Nishanth Menon]

On 04/16/2014 11:35 AM, Tony Lindgren wrote:
> * Felipe Balbi <balbi@ti.com> [140416 08:18]:
>> TRM says we *must* write 1 to each bit we're handling
>> in order to clear the IRQ status and bring IRQ line
>> low. This patch implements that.
>>
>> Signed-off-by: Felipe Balbi <balbi@ti.com>
>> ---
>>
>> Russell, I don't have HW to test, but this should
>> solve the problem you saw when not using battery
>> with Zoom board. Let me know if it doesn't.
> 
> BTW, looks like we're enabling BCI automatically in twl4030.dtsi
> while the legacy booting does not have TWL_COMMON_PDATA_BCI
> enabled for LDP. Anyways, for LDP BCI should be enabled for
> sure, that's the only way to power at least the earlier LDP
> revisions.
> 
I picked up https://patchwork.kernel.org/patch/4002371/ for testing.

Unfortunately, it does not seem to work in my tests:

ldp-board-no-battery uses upstream u-boot in a chainloaded
configuration + does not have battery plugged in, instead, just J58
(charger) is connected - this is plugged on a remote farm setup for
developers to work with.

ldp-board-with-battery uses legacy bootloader with a battery simulator.

omap2plus_defconfig + CONFIG_CHARGER_TWL4030 enabled.

Result:
 v3-15-rc2-ldp-board-no-battery:  Boot FAIL:
http://slexy.org/raw/s20riTwu04

 v3-15-rc2-ldp-board-with-battery:  Boot PASS:
http://slexy.org/raw/s2elKdJpI7

 v3-15-rc2-with-felipe-fix-ldp-board-no-battery:  Boot FAIL:
http://slexy.org/raw/s20lKp2o0s


This patch should probably also acknowledge that Russell reported this
issue.
-- 
Regards,
Nishanth Menon

Re: [PATCH] power: twl4030_charger: clear IRQs after handling them

[Felipe Balbi]

[-- Attachment #1: Type: text/plain, Size: 1255 bytes --]

On Fri, Apr 25, 2014 at 03:58:10PM -0500, Nishanth Menon wrote:
> On 04/16/2014 11:35 AM, Tony Lindgren wrote:
> > * Felipe Balbi <balbi@ti.com> [140416 08:18]:
> >> TRM says we *must* write 1 to each bit we're handling
> >> in order to clear the IRQ status and bring IRQ line
> >> low. This patch implements that.
> >>
> >> Signed-off-by: Felipe Balbi <balbi@ti.com>
> >> ---
> >>
> >> Russell, I don't have HW to test, but this should
> >> solve the problem you saw when not using battery
> >> with Zoom board. Let me know if it doesn't.
> > 
> > BTW, looks like we're enabling BCI automatically in twl4030.dtsi
> > while the legacy booting does not have TWL_COMMON_PDATA_BCI
> > enabled for LDP. Anyways, for LDP BCI should be enabled for
> > sure, that's the only way to power at least the earlier LDP
> > revisions.
> > 
> I picked up https://patchwork.kernel.org/patch/4002371/ for testing.
> 
> Unfortunately, it does not seem to work in my tests:

alright, I'll have a look after I get this other issue out of the way.
Probably not until next week.

> This patch should probably also acknowledge that Russell reported this
> issue.

sure I can add that, although it doesn't solve anything as it is ...

-- 
balbi

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: [PATCH] power: twl4030_charger: clear IRQs after handling them

[Nishanth Menon]

On 16:00-20140425, Felipe Balbi wrote:
> On Fri, Apr 25, 2014 at 03:58:10PM -0500, Nishanth Menon wrote:
> > On 04/16/2014 11:35 AM, Tony Lindgren wrote:
> > > * Felipe Balbi <balbi@ti.com> [140416 08:18]:
> > >> TRM says we *must* write 1 to each bit we're handling
> > >> in order to clear the IRQ status and bring IRQ line
> > >> low. This patch implements that.
> > >>
> > >> Signed-off-by: Felipe Balbi <balbi@ti.com>
> > >> ---
> > >>
> > >> Russell, I don't have HW to test, but this should
> > >> solve the problem you saw when not using battery
> > >> with Zoom board. Let me know if it doesn't.
> > > 
> > > BTW, looks like we're enabling BCI automatically in twl4030.dtsi
> > > while the legacy booting does not have TWL_COMMON_PDATA_BCI
> > > enabled for LDP. Anyways, for LDP BCI should be enabled for
> > > sure, that's the only way to power at least the earlier LDP
> > > revisions.
> > > 
> > I picked up https://patchwork.kernel.org/patch/4002371/ for testing.
> > 
> > Unfortunately, it does not seem to work in my tests:
> 
> alright, I'll have a look after I get this other issue out of the way.
> Probably not until next week.

I dug a bit more into this earlier today. It makes a little sense
considering that TWL4030 was created in a world where mobile devices did
not consider battery was hot pluggable entity.. 

Long story short:
a) the BCI interrupt status register is read-on-clear (see
drivers/mfd/twl4030-irq.c set_cor is set to true for bci) - so the patch
https://patchwork.kernel.org/patch/4002371/ is practically a NOP.
b) BATSTS is a result of MADC monitoring the thermistor and is
continually generated. I tried a bunch of tricks that I could come up
with to try to do a hotplug support, but either we have infinite
interrupts (due to regeneration of event) or I loose detection ability.

So, to keep things simple I created the following patch that seems to
handle:
a) no battery connected, powered off charger - bci disabled
b) battery connected, powered off charger - bci active
c) battery connected, no charger - bci active

note: if battery was connected, and booted and we hotplug battery out,
we still have the spam of messages.. that is a seperate patch
(basically do an orderly shutdown - I cant think of anything else we
could possible do there)

Let me know your views.. if this approach is fine, I can post out formal
patches for the same.
-----8<----
>From 9474c7f37990b97c57b72eed917b9153595133ff Mon Sep 17 00:00:00 2001
From: Nishanth Menon <nm@ti.com>
Date: Tue, 6 May 2014 19:02:49 -0500
Subject: [PATCH] power: twl4030_charger: detect battery presence prior to
 enabling charger

TWL4030's Battery Charger seems to be designed for non-hotpluggable
batteries.

If battery is not present in the system, BATSTS is always set with the
expectation that software will take actions to move to a required safe
state (could be power down or disable various charger paths).

It does not seem possible even by manipulating the edge detection
of the event (using BCIEDR2 register) to have a consistent hotplug
handling. This seems to be the result of BATSTS interrupt generated
when the thermistor of the battery pack is disconnected from the
dedicated ADIN1 pin. Clearing the status just results in the status
being regenerated by the monitoring ADC(MADC) and disabling the
edges of event just makes hotplug no longer function. The only
other option is to disable the detection of the MADC by disabling
BCIMFEN4::BATSTSMCHGEN (battery presence detector) - but then, we can
never again detect battery reconnection.

So, detect battery presence based on precharge(which is hardware
automatic state) or default main charger configuration at the time of
probe and enable charger logic only if battery was present.

Reported-by: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Nishanth Menon <nm@ti.com>
---
 drivers/power/twl4030_charger.c |   43 ++++++++++++++++++++++++++++++++++++++-
 1 file changed, 42 insertions(+), 1 deletion(-)

diff --git a/drivers/power/twl4030_charger.c b/drivers/power/twl4030_charger.c
index f141088..14ec220 100644
--- a/drivers/power/twl4030_charger.c
+++ b/drivers/power/twl4030_charger.c
@@ -28,10 +28,13 @@
 #define TWL4030_BCIICHG		0x08
 #define TWL4030_BCIVAC		0x0a
 #define TWL4030_BCIVBUS		0x0c
+#define TWL4030_BCIMFSTS3	0x0F
 #define TWL4030_BCIMFSTS4	0x10
 #define TWL4030_BCICTL1		0x23
 #define TWL4030_BB_CFG		0x12
 
+#define TWL4030_BCIMFSTS1	0x01
+
 #define TWL4030_BCIAUTOWEN	BIT(5)
 #define TWL4030_CONFIG_DONE	BIT(4)
 #define TWL4030_BCIAUTOUSB	BIT(1)
@@ -52,6 +55,9 @@
 #define TWL4030_BBISEL_500uA	0x02
 #define TWL4030_BBISEL_1000uA	0x03
 
+#define TWL4030_BATSTSPCHG	BIT(2)
+#define TWL4030_BATSTSMCHG	BIT(6)
+
 /* BCI interrupts */
 #define TWL4030_WOVF		BIT(0) /* Watchdog overflow */
 #define TWL4030_TMOVF		BIT(1) /* Timer overflow */
@@ -145,6 +151,34 @@ static int twl4030bci_read_adc_val(u8 reg)
 }
 
 /*
+ * Check if Battery Pack was present
+ */
+static int twl4030_is_battery_present(struct twl4030_bci *bci)
+{
+	int ret;
+	u8 val = 0;
+
+	/* Battery presence in Main charge? */
+	ret = twl_i2c_read_u8(TWL_MODULE_MAIN_CHARGE, &val, TWL4030_BCIMFSTS3);
+	if (ret)
+		return ret;
+	if (val & TWL4030_BATSTSMCHG)
+		return 0;
+
+	/*
+	 * OK, It could be that bootloader did not enable main charger,
+	 * pre-charge is h/w auto. So, Battery presence in Pre-charge?
+	 */
+	ret = twl_i2c_read_u8(TWL4030_MODULE_PRECHARGE, &val, TWL4030_BCIMFSTS1);
+	if (ret)
+		return ret;
+	if (val & TWL4030_BATSTSPCHG)
+		return 0;
+
+	return -ENODEV;
+}
+
+/*
  * Check if VBUS power is present
  */
 static int twl4030_bci_have_vbus(struct twl4030_bci *bci)
@@ -541,8 +575,14 @@ static int __init twl4030_bci_probe(struct platform_device *pdev)
 	bci->irq_chg = platform_get_irq(pdev, 0);
 	bci->irq_bci = platform_get_irq(pdev, 1);
 
-	platform_set_drvdata(pdev, bci);
+	/* Only proceed further *IF* battery is physically present */
+	ret = twl4030_is_battery_present(bci);
+	if  (ret) {
+		dev_crit(&pdev->dev, "Battery was not detected:%d\n", ret);
+		goto fail_no_battery;
+	}
 
+	platform_set_drvdata(pdev, bci);
 	bci->ac.name = "twl4030_ac";
 	bci->ac.type = POWER_SUPPLY_TYPE_MAINS;
 	bci->ac.properties = twl4030_charger_props;
@@ -633,6 +673,7 @@ fail_chg_irq:
 fail_register_usb:
 	power_supply_unregister(&bci->ac);
 fail_register_ac:
+fail_no_battery:
 	kfree(bci);
 
 	return ret;
-- 
1.7.9.5

-- 
Regards,
Nishanth Menon

Re: [PATCH] power: twl4030_charger: clear IRQs after handling them

[Tony Lindgren]

* Nishanth Menon <nm@ti.com> [140506 17:25]:
> Subject: [PATCH] power: twl4030_charger: detect battery presence prior to
>  enabling charger
> 
> TWL4030's Battery Charger seems to be designed for non-hotpluggable
> batteries.
> 
> If battery is not present in the system, BATSTS is always set with the
> expectation that software will take actions to move to a required safe
> state (could be power down or disable various charger paths).
> 
> It does not seem possible even by manipulating the edge detection
> of the event (using BCIEDR2 register) to have a consistent hotplug
> handling. This seems to be the result of BATSTS interrupt generated
> when the thermistor of the battery pack is disconnected from the
> dedicated ADIN1 pin. Clearing the status just results in the status
> being regenerated by the monitoring ADC(MADC) and disabling the
> edges of event just makes hotplug no longer function. The only
> other option is to disable the detection of the MADC by disabling
> BCIMFEN4::BATSTSMCHGEN (battery presence detector) - but then, we can
> never again detect battery reconnection.
> 
> So, detect battery presence based on precharge(which is hardware
> automatic state) or default main charger configuration at the time of
> probe and enable charger logic only if battery was present.
> 
> Reported-by: Russell King <linux@arm.linux.org.uk>
> Signed-off-by: Nishanth Menon <nm@ti.com>

Gets rid of the errors for me if CONFIG_CHARGER_TWL4030=y.
Looks like we don't have that enabled by default in
omap2plus_defconfig which explain why it's taken so long to
notice this one:

Tested-by: Tony Lindgren <tony@atomide.com>

Re: [PATCH] power: twl4030_charger: clear IRQs after handling them

[Nishanth Menon]

On 05/08/2014 08:03 PM, Tony Lindgren wrote:
> * Nishanth Menon <nm@ti.com> [140506 17:25]:
>> Subject: [PATCH] power: twl4030_charger: detect battery presence prior to
>>  enabling charger
>>
>> TWL4030's Battery Charger seems to be designed for non-hotpluggable
>> batteries.
>>
>> If battery is not present in the system, BATSTS is always set with the
>> expectation that software will take actions to move to a required safe
>> state (could be power down or disable various charger paths).
>>
>> It does not seem possible even by manipulating the edge detection
>> of the event (using BCIEDR2 register) to have a consistent hotplug
>> handling. This seems to be the result of BATSTS interrupt generated
>> when the thermistor of the battery pack is disconnected from the
>> dedicated ADIN1 pin. Clearing the status just results in the status
>> being regenerated by the monitoring ADC(MADC) and disabling the
>> edges of event just makes hotplug no longer function. The only
>> other option is to disable the detection of the MADC by disabling
>> BCIMFEN4::BATSTSMCHGEN (battery presence detector) - but then, we can
>> never again detect battery reconnection.
>>
>> So, detect battery presence based on precharge(which is hardware
>> automatic state) or default main charger configuration at the time of
>> probe and enable charger logic only if battery was present.
>>
>> Reported-by: Russell King <linux@arm.linux.org.uk>
>> Signed-off-by: Nishanth Menon <nm@ti.com>
> 
> Gets rid of the errors for me if CONFIG_CHARGER_TWL4030=y.
> Looks like we don't have that enabled by default in
> omap2plus_defconfig which explain why it's taken so long to
> notice this one:
> 
> Tested-by: Tony Lindgren <tony@atomide.com>
> 
Thanks. I will post this out to the list as formal series.

-- 
Regards,
Nishanth Menon