diff mbox

[V5] ARM: shmobile: Rework the PMIC IRQ line quirk

Message ID 20180611121513.9673-1-marek.vasut+renesas@gmail.com (mailing list archive)
State New, archived
Headers show

Commit Message

Marek Vasut June 11, 2018, 12:15 p.m. UTC
Rather than hard-coding the quirk topology, which stopped scaling,
parse the information from DT. The code looks for all compatible
PMICs -- da9063 and da9210 -- and checks if their IRQ line is tied
to the same pin. If so, the code sends a matching sequence to the
PMIC to deassert the IRQ.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Geert Uytterhoeven <geert+renesas@glider.be>
Cc: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Cc: Simon Horman <horms+renesas@verge.net.au>
Cc: Wolfram Sang <wsa+renesas@sang-engineering.com>
Cc: linux-renesas-soc@vger.kernel.org
Acked-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> (on Koelsch)
---
V2: - Replace the DT shared IRQ check loop with memcmp()
    - Send the I2C message to deassert the IRQ line to all PMICs
      in the list with shared IRQ line instead of just one
    - Add comment that this works only in case all the PMICs are
      on the same I2C bus
V3: - Drop the addr = 0x00 init
    - Drop reinit of argsa in rcar_gen2_regulator_quirk
V4: - Squash regulator_quirk on single line
    - Drop !np check in for_each_matching_node_and_match()
    - Use argsa in of_irq_parse_one
V5: - Check kzalloc failure
    - Rename da...._msgs to da...._msg
    - Don't reinit quirk->shared
---
 arch/arm/mach-shmobile/regulator-quirk-rcar-gen2.c | 137 ++++++++++++++++-----
 1 file changed, 108 insertions(+), 29 deletions(-)

Comments

Simon Horman June 13, 2018, 11:06 a.m. UTC | #1
On Mon, Jun 11, 2018 at 02:15:13PM +0200, Marek Vasut wrote:
> Rather than hard-coding the quirk topology, which stopped scaling,
> parse the information from DT. The code looks for all compatible
> PMICs -- da9063 and da9210 -- and checks if their IRQ line is tied
> to the same pin. If so, the code sends a matching sequence to the
> PMIC to deassert the IRQ.
> 
> Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
> Cc: Geert Uytterhoeven <geert+renesas@glider.be>
> Cc: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
> Cc: Simon Horman <horms+renesas@verge.net.au>
> Cc: Wolfram Sang <wsa+renesas@sang-engineering.com>
> Cc: linux-renesas-soc@vger.kernel.org
> Acked-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
> Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> (on Koelsch)

Hi,

This looks fine to me but I will wait to see if there are other reviews
before applying.

Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Geert Uytterhoeven June 13, 2018, 11:21 a.m. UTC | #2
Hi Simon,

On Wed, Jun 13, 2018 at 1:06 PM Simon Horman <horms@verge.net.au> wrote:
> On Mon, Jun 11, 2018 at 02:15:13PM +0200, Marek Vasut wrote:
> > Rather than hard-coding the quirk topology, which stopped scaling,
> > parse the information from DT. The code looks for all compatible
> > PMICs -- da9063 and da9210 -- and checks if their IRQ line is tied
> > to the same pin. If so, the code sends a matching sequence to the
> > PMIC to deassert the IRQ.
> >
> > Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
> > Cc: Geert Uytterhoeven <geert+renesas@glider.be>
> > Cc: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
> > Cc: Simon Horman <horms+renesas@verge.net.au>
> > Cc: Wolfram Sang <wsa+renesas@sang-engineering.com>
> > Cc: linux-renesas-soc@vger.kernel.org
> > Acked-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
> > Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> (on Koelsch)
>
> This looks fine to me but I will wait to see if there are other reviews
> before applying.
>
> Reviewed-by: Simon Horman <horms+renesas@verge.net.au>

Note that applying it now will break gose, as its DTS doesn't describe all
regulators yet.

Gr{oetje,eeting}s,

                        Geert
Simon Horman June 13, 2018, 2:36 p.m. UTC | #3
On Wed, Jun 13, 2018 at 01:21:34PM +0200, Geert Uytterhoeven wrote:
> Hi Simon,
> 
> On Wed, Jun 13, 2018 at 1:06 PM Simon Horman <horms@verge.net.au> wrote:
> > On Mon, Jun 11, 2018 at 02:15:13PM +0200, Marek Vasut wrote:
> > > Rather than hard-coding the quirk topology, which stopped scaling,
> > > parse the information from DT. The code looks for all compatible
> > > PMICs -- da9063 and da9210 -- and checks if their IRQ line is tied
> > > to the same pin. If so, the code sends a matching sequence to the
> > > PMIC to deassert the IRQ.
> > >
> > > Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
> > > Cc: Geert Uytterhoeven <geert+renesas@glider.be>
> > > Cc: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
> > > Cc: Simon Horman <horms+renesas@verge.net.au>
> > > Cc: Wolfram Sang <wsa+renesas@sang-engineering.com>
> > > Cc: linux-renesas-soc@vger.kernel.org
> > > Acked-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
> > > Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> (on Koelsch)
> >
> > This looks fine to me but I will wait to see if there are other reviews
> > before applying.
> >
> > Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
> 
> Note that applying it now will break gose, as its DTS doesn't describe all
> regulators yet.

Ok, so old DT will break with new kernels?
Geert Uytterhoeven June 13, 2018, 4:48 p.m. UTC | #4
Hi Simon,

On Wed, Jun 13, 2018 at 4:36 PM Simon Horman <horms@verge.net.au> wrote:
> On Wed, Jun 13, 2018 at 01:21:34PM +0200, Geert Uytterhoeven wrote:
> > On Wed, Jun 13, 2018 at 1:06 PM Simon Horman <horms@verge.net.au> wrote:
> > > On Mon, Jun 11, 2018 at 02:15:13PM +0200, Marek Vasut wrote:
> > > > Rather than hard-coding the quirk topology, which stopped scaling,
> > > > parse the information from DT. The code looks for all compatible
> > > > PMICs -- da9063 and da9210 -- and checks if their IRQ line is tied
> > > > to the same pin. If so, the code sends a matching sequence to the
> > > > PMIC to deassert the IRQ.
> > > >
> > > > Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
> > > > Cc: Geert Uytterhoeven <geert+renesas@glider.be>
> > > > Cc: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
> > > > Cc: Simon Horman <horms+renesas@verge.net.au>
> > > > Cc: Wolfram Sang <wsa+renesas@sang-engineering.com>
> > > > Cc: linux-renesas-soc@vger.kernel.org
> > > > Acked-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
> > > > Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> (on Koelsch)
> > >
> > > This looks fine to me but I will wait to see if there are other reviews
> > > before applying.
> > >
> > > Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
> >
> > Note that applying it now will break gose, as its DTS doesn't describe all
> > regulators yet.
>
> Ok, so old DT will break with new kernels?

For Lager, it will break with pre-v4.2 DTS, so I think that's not a problem.
For Koelsch, the regulators were added even earlier.
Stout had all regulators from the initial support in upstream.

So Gose is the only issue. Do we care?

Gr{oetje,eeting}s,

                        Geert
Marek Vasut July 30, 2018, 11:19 a.m. UTC | #5
On 06/13/2018 01:21 PM, Geert Uytterhoeven wrote:
> Hi Simon,

Hi,

> On Wed, Jun 13, 2018 at 1:06 PM Simon Horman <horms@verge.net.au> wrote:
>> On Mon, Jun 11, 2018 at 02:15:13PM +0200, Marek Vasut wrote:
>>> Rather than hard-coding the quirk topology, which stopped scaling,
>>> parse the information from DT. The code looks for all compatible
>>> PMICs -- da9063 and da9210 -- and checks if their IRQ line is tied
>>> to the same pin. If so, the code sends a matching sequence to the
>>> PMIC to deassert the IRQ.
>>>
>>> Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
>>> Cc: Geert Uytterhoeven <geert+renesas@glider.be>
>>> Cc: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
>>> Cc: Simon Horman <horms+renesas@verge.net.au>
>>> Cc: Wolfram Sang <wsa+renesas@sang-engineering.com>
>>> Cc: linux-renesas-soc@vger.kernel.org
>>> Acked-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
>>> Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> (on Koelsch)
>>
>> This looks fine to me but I will wait to see if there are other reviews
>> before applying.
>>
>> Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
> 
> Note that applying it now will break gose, as its DTS doesn't describe all
> regulators yet.

https://patchwork.kernel.org/patch/10548919/ should fix this.
https://patchwork.kernel.org/patch/10548923/ for Silk.
Geert Uytterhoeven Sept. 5, 2018, 11:55 a.m. UTC | #6
Hi Marek,

On Mon, Jun 11, 2018 at 2:15 PM Marek Vasut <marek.vasut@gmail.com> wrote:
> Rather than hard-coding the quirk topology, which stopped scaling,
> parse the information from DT. The code looks for all compatible
> PMICs -- da9063 and da9210 -- and checks if their IRQ line is tied
> to the same pin. If so, the code sends a matching sequence to the
> PMIC to deassert the IRQ.
>
> Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
> Cc: Geert Uytterhoeven <geert+renesas@glider.be>
> Cc: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
> Cc: Simon Horman <horms+renesas@verge.net.au>
> Cc: Wolfram Sang <wsa+renesas@sang-engineering.com>
> Cc: linux-renesas-soc@vger.kernel.org
> Acked-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
> Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> (on Koelsch)
> ---
> V2: - Replace the DT shared IRQ check loop with memcmp()
>     - Send the I2C message to deassert the IRQ line to all PMICs
>       in the list with shared IRQ line instead of just one
>     - Add comment that this works only in case all the PMICs are
>       on the same I2C bus
> V3: - Drop the addr = 0x00 init
>     - Drop reinit of argsa in rcar_gen2_regulator_quirk
> V4: - Squash regulator_quirk on single line
>     - Drop !np check in for_each_matching_node_and_match()
>     - Use argsa in of_irq_parse_one
> V5: - Check kzalloc failure
>     - Rename da...._msgs to da...._msg
>     - Don't reinit quirk->shared

Thanks for the update!

> --- a/arch/arm/mach-shmobile/regulator-quirk-rcar-gen2.c
> +++ b/arch/arm/mach-shmobile/regulator-quirk-rcar-gen2.c

> @@ -122,7 +142,12 @@ static struct notifier_block regulator_quirk_nb = {
>
>  static int __init rcar_gen2_regulator_quirk(void)
>  {
> -       u32 mon;
> +       struct regulator_quirk *quirk, *pos, *tmp;
> +       struct of_phandle_args *argsa, *argsb;
> +       const struct of_device_id *id;
> +       struct device_node *np;
> +       u32 mon, addr;
> +       int ret;
>
>         if (!of_machine_is_compatible("renesas,koelsch") &&
>             !of_machine_is_compatible("renesas,lager") &&
> @@ -130,22 +155,76 @@ static int __init rcar_gen2_regulator_quirk(void)
>             !of_machine_is_compatible("renesas,gose"))
>                 return -ENODEV;
>
> +       for_each_matching_node_and_match(np, rcar_gen2_quirk_match, &id) {
> +               if (!of_device_is_available(np))
> +                       break;
> +
> +               quirk = kzalloc(sizeof(*quirk), GFP_KERNEL);
> +               if (!quirk) {
> +                       ret = -ENOMEM;
> +                       goto err_mem;
> +               }
> +
> +               argsa = &quirk->irq_args;
> +               memcpy(&quirk->i2c_msg, id->data, sizeof(quirk->i2c_msg));
> +
> +               ret = of_property_read_u32(np, "reg", &addr);
> +               if (ret)
> +                       return ret;

I think it's safer to skip this entry and continue, after calling
kfree(quirk), of course.

> +
> +               quirk->id = id;
> +               quirk->i2c_msg.addr = addr;
> +
> +               ret = of_irq_parse_one(np, 0, argsa);
> +               if (ret)
> +                       return ret;

kfree(quirk) and continue...

> +
> +               list_for_each_entry(pos, &quirk_list, list) {
> +                       argsb = &pos->irq_args;
> +
> +                       if (argsa->args_count != argsb->args_count)
> +                               continue;
> +
> +                       ret = memcmp(argsa->args, argsb->args,
> +                                    argsa->args_count *
> +                                    sizeof(argsa->args[0]));
> +                       if (!ret) {
> +                               pos->shared = true;
> +                               quirk->shared = true;
> +                       }
> +               }
> +
> +               list_add_tail(&quirk->list, &quirk_list);
> +       }
> +

With the above fixed:
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>

Gr{oetje,eeting}s,

                        Geert
Marek Vasut Sept. 18, 2018, 12:24 p.m. UTC | #7
On 09/05/2018 01:55 PM, Geert Uytterhoeven wrote:
> Hi Marek,
> 
> On Mon, Jun 11, 2018 at 2:15 PM Marek Vasut <marek.vasut@gmail.com> wrote:
>> Rather than hard-coding the quirk topology, which stopped scaling,
>> parse the information from DT. The code looks for all compatible
>> PMICs -- da9063 and da9210 -- and checks if their IRQ line is tied
>> to the same pin. If so, the code sends a matching sequence to the
>> PMIC to deassert the IRQ.
>>
>> Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
>> Cc: Geert Uytterhoeven <geert+renesas@glider.be>
>> Cc: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
>> Cc: Simon Horman <horms+renesas@verge.net.au>
>> Cc: Wolfram Sang <wsa+renesas@sang-engineering.com>
>> Cc: linux-renesas-soc@vger.kernel.org
>> Acked-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
>> Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> (on Koelsch)
>> ---
>> V2: - Replace the DT shared IRQ check loop with memcmp()
>>     - Send the I2C message to deassert the IRQ line to all PMICs
>>       in the list with shared IRQ line instead of just one
>>     - Add comment that this works only in case all the PMICs are
>>       on the same I2C bus
>> V3: - Drop the addr = 0x00 init
>>     - Drop reinit of argsa in rcar_gen2_regulator_quirk
>> V4: - Squash regulator_quirk on single line
>>     - Drop !np check in for_each_matching_node_and_match()
>>     - Use argsa in of_irq_parse_one
>> V5: - Check kzalloc failure
>>     - Rename da...._msgs to da...._msg
>>     - Don't reinit quirk->shared
> 
> Thanks for the update!
> 
>> --- a/arch/arm/mach-shmobile/regulator-quirk-rcar-gen2.c
>> +++ b/arch/arm/mach-shmobile/regulator-quirk-rcar-gen2.c
> 
>> @@ -122,7 +142,12 @@ static struct notifier_block regulator_quirk_nb = {
>>
>>  static int __init rcar_gen2_regulator_quirk(void)
>>  {
>> -       u32 mon;
>> +       struct regulator_quirk *quirk, *pos, *tmp;
>> +       struct of_phandle_args *argsa, *argsb;
>> +       const struct of_device_id *id;
>> +       struct device_node *np;
>> +       u32 mon, addr;
>> +       int ret;
>>
>>         if (!of_machine_is_compatible("renesas,koelsch") &&
>>             !of_machine_is_compatible("renesas,lager") &&
>> @@ -130,22 +155,76 @@ static int __init rcar_gen2_regulator_quirk(void)
>>             !of_machine_is_compatible("renesas,gose"))
>>                 return -ENODEV;
>>
>> +       for_each_matching_node_and_match(np, rcar_gen2_quirk_match, &id) {
>> +               if (!of_device_is_available(np))
>> +                       break;
>> +
>> +               quirk = kzalloc(sizeof(*quirk), GFP_KERNEL);
>> +               if (!quirk) {
>> +                       ret = -ENOMEM;
>> +                       goto err_mem;
>> +               }
>> +
>> +               argsa = &quirk->irq_args;
>> +               memcpy(&quirk->i2c_msg, id->data, sizeof(quirk->i2c_msg));
>> +
>> +               ret = of_property_read_u32(np, "reg", &addr);
>> +               if (ret)
>> +                       return ret;
> 
> I think it's safer to skip this entry and continue, after calling
> kfree(quirk), of course.

This can be shifted above the kzalloc() . That said, I sent V6, so
please review it one more time.
diff mbox

Patch

diff --git a/arch/arm/mach-shmobile/regulator-quirk-rcar-gen2.c b/arch/arm/mach-shmobile/regulator-quirk-rcar-gen2.c
index 93f628acfd94..cf4d260882cf 100644
--- a/arch/arm/mach-shmobile/regulator-quirk-rcar-gen2.c
+++ b/arch/arm/mach-shmobile/regulator-quirk-rcar-gen2.c
@@ -31,11 +31,12 @@ 
 #include <linux/i2c.h>
 #include <linux/init.h>
 #include <linux/io.h>
+#include <linux/list.h>
 #include <linux/notifier.h>
 #include <linux/of.h>
+#include <linux/of_irq.h>
 #include <linux/mfd/da9063/registers.h>
 
-
 #define IRQC_BASE		0xe61c0000
 #define IRQC_MONITOR		0x104	/* IRQn Signal Level Monitor Register */
 
@@ -44,34 +45,45 @@ 
 /* start of DA9210 System Control and Event Registers */
 #define DA9210_REG_MASK_A		0x54
 
+struct regulator_quirk {
+	struct list_head		list;
+	const struct of_device_id	*id;
+	struct of_phandle_args		irq_args;
+	struct i2c_msg			i2c_msg;
+	bool				shared;	/* IRQ line is shared */
+};
+
+static LIST_HEAD(quirk_list);
 static void __iomem *irqc;
 
 /* first byte sets the memory pointer, following are consecutive reg values */
 static u8 da9063_irq_clr[] = { DA9063_REG_IRQ_MASK_A, 0xff, 0xff, 0xff, 0xff };
 static u8 da9210_irq_clr[] = { DA9210_REG_MASK_A, 0xff, 0xff };
 
-static struct i2c_msg da9xxx_msgs[3] = {
-	{
-		.addr = 0x58,
-		.len = ARRAY_SIZE(da9063_irq_clr),
-		.buf = da9063_irq_clr,
-	}, {
-		.addr = 0x68,
-		.len = ARRAY_SIZE(da9210_irq_clr),
-		.buf = da9210_irq_clr,
-	}, {
-		.addr = 0x70,
-		.len = ARRAY_SIZE(da9210_irq_clr),
-		.buf = da9210_irq_clr,
-	},
+static struct i2c_msg da9063_msg = {
+	.len = ARRAY_SIZE(da9063_irq_clr),
+	.buf = da9063_irq_clr,
+};
+
+static struct i2c_msg da9210_msg = {
+	.len = ARRAY_SIZE(da9210_irq_clr),
+	.buf = da9210_irq_clr,
+};
+
+static const struct of_device_id rcar_gen2_quirk_match[] = {
+	{ .compatible = "dlg,da9063", .data = &da9063_msg },
+	{ .compatible = "dlg,da9210", .data = &da9210_msg },
+	{},
 };
 
 static int regulator_quirk_notify(struct notifier_block *nb,
 				  unsigned long action, void *data)
 {
+	struct regulator_quirk *pos, *tmp;
 	struct device *dev = data;
 	struct i2c_client *client;
 	static bool done;
+	int ret;
 	u32 mon;
 
 	if (done)
@@ -88,17 +100,20 @@  static int regulator_quirk_notify(struct notifier_block *nb,
 	client = to_i2c_client(dev);
 	dev_dbg(dev, "Detected %s\n", client->name);
 
-	if ((client->addr == 0x58 && !strcmp(client->name, "da9063")) ||
-	    (client->addr == 0x68 && !strcmp(client->name, "da9210")) ||
-	    (client->addr == 0x70 && !strcmp(client->name, "da9210"))) {
-		int ret, len;
+	/*
+	 * Send message to all PMICs that share an IRQ line to deassert it.
+	 *
+	 * WARNING: This works only if all the PMICs are on the same I2C bus.
+	 */
+	list_for_each_entry(pos, &quirk_list, list) {
+		if (!pos->shared)
+			continue;
 
-		/* There are two DA9210 on Stout, one on the other boards. */
-		len = of_machine_is_compatible("renesas,stout") ? 3 : 2;
+		dev_info(&client->dev, "clearing %s@0x%02x interrupts\n",
+			 pos->id->compatible, pos->i2c_msg.addr);
 
-		dev_info(&client->dev, "clearing da9063/da9210 interrupts\n");
-		ret = i2c_transfer(client->adapter, da9xxx_msgs, len);
-		if (ret != len)
+		ret = i2c_transfer(client->adapter, &pos->i2c_msg, 1);
+		if (ret != 1)
 			dev_err(&client->dev, "i2c error %d\n", ret);
 	}
 
@@ -111,6 +126,11 @@  static int regulator_quirk_notify(struct notifier_block *nb,
 remove:
 	dev_info(dev, "IRQ2 is not asserted, removing quirk\n");
 
+	list_for_each_entry_safe(pos, tmp, &quirk_list, list) {
+		list_del(&pos->list);
+		kfree(pos);
+	}
+
 	done = true;
 	iounmap(irqc);
 	return 0;
@@ -122,7 +142,12 @@  static struct notifier_block regulator_quirk_nb = {
 
 static int __init rcar_gen2_regulator_quirk(void)
 {
-	u32 mon;
+	struct regulator_quirk *quirk, *pos, *tmp;
+	struct of_phandle_args *argsa, *argsb;
+	const struct of_device_id *id;
+	struct device_node *np;
+	u32 mon, addr;
+	int ret;
 
 	if (!of_machine_is_compatible("renesas,koelsch") &&
 	    !of_machine_is_compatible("renesas,lager") &&
@@ -130,22 +155,76 @@  static int __init rcar_gen2_regulator_quirk(void)
 	    !of_machine_is_compatible("renesas,gose"))
 		return -ENODEV;
 
+	for_each_matching_node_and_match(np, rcar_gen2_quirk_match, &id) {
+		if (!of_device_is_available(np))
+			break;
+
+		quirk = kzalloc(sizeof(*quirk), GFP_KERNEL);
+		if (!quirk) {
+			ret = -ENOMEM;
+			goto err_mem;
+		}
+
+		argsa = &quirk->irq_args;
+		memcpy(&quirk->i2c_msg, id->data, sizeof(quirk->i2c_msg));
+
+		ret = of_property_read_u32(np, "reg", &addr);
+		if (ret)
+			return ret;
+
+		quirk->id = id;
+		quirk->i2c_msg.addr = addr;
+
+		ret = of_irq_parse_one(np, 0, argsa);
+		if (ret)
+			return ret;
+
+		list_for_each_entry(pos, &quirk_list, list) {
+			argsb = &pos->irq_args;
+
+			if (argsa->args_count != argsb->args_count)
+				continue;
+
+			ret = memcmp(argsa->args, argsb->args,
+				     argsa->args_count *
+				     sizeof(argsa->args[0]));
+			if (!ret) {
+				pos->shared = true;
+				quirk->shared = true;
+			}
+		}
+
+		list_add_tail(&quirk->list, &quirk_list);
+	}
+
 	irqc = ioremap(IRQC_BASE, PAGE_SIZE);
-	if (!irqc)
-		return -ENOMEM;
+	if (!irqc) {
+		ret = -ENOMEM;
+		goto err_mem;
+	}
 
 	mon = ioread32(irqc + IRQC_MONITOR);
 	if (mon & REGULATOR_IRQ_MASK) {
 		pr_debug("%s: IRQ2 is not asserted, not installing quirk\n",
 			 __func__);
-		iounmap(irqc);
-		return 0;
+		ret = 0;
+		goto err_free;
 	}
 
 	pr_info("IRQ2 is asserted, installing da9063/da9210 regulator quirk\n");
 
 	bus_register_notifier(&i2c_bus_type, &regulator_quirk_nb);
 	return 0;
+
+err_free:
+	iounmap(irqc);
+err_mem:
+	list_for_each_entry_safe(pos, tmp, &quirk_list, list) {
+		list_del(&pos->list);
+		kfree(pos);
+	}
+
+	return ret;
 }
 
 arch_initcall(rcar_gen2_regulator_quirk);