[PATCH] i2c: Not all adapters have a parent

[PATCH] i2c: Not all adapters have a parent

[Jean Delvare]

The code in acpi_i2c_register_devices() assumes that all i2c adapters
have a parent. This is not necessarily the case, for example the
i2c-stub driver instantiate a virtual i2c adapter without a parent.
Check for this to avoid a NULL pointer deference.

Signed-off-by: Jean Delvare <jdelvare-l3A5Bk7waGM@public.gmane.org>
Cc: Wolfram Sang <wsa-z923LK4zBo2bacvFa/9K2g@public.gmane.org>
Cc: Mika Westerberg <mika.westerberg-VuQAYsv1563Yd54FQh9/CA@public.gmane.org>
Cc: "Rafael J. Wysocki" <rafael.j.wysocki-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/i2c/i2c-core.c |    3 +++
 1 file changed, 3 insertions(+)

--- linux-3.12-rc4.orig/drivers/i2c/i2c-core.c	2013-09-24 00:41:09.000000000 +0200
+++ linux-3.12-rc4/drivers/i2c/i2c-core.c	2013-10-10 07:46:12.244886047 +0200
@@ -1134,6 +1134,9 @@ static void acpi_i2c_register_devices(st
 	acpi_handle handle;
 	acpi_status status;
 
+	if (!adap->dev.parent)
+		return;
+
 	handle = ACPI_HANDLE(adap->dev.parent);
 	if (!handle)
 		return;


-- 
Jean Delvare
Suse L3 Support

Re: [PATCH] i2c: Not all adapters have a parent

[Mika Westerberg]

On Thu, Oct 10, 2013 at 08:04:06AM +0200, Jean Delvare wrote:
> The code in acpi_i2c_register_devices() assumes that all i2c adapters
> have a parent. This is not necessarily the case, for example the
> i2c-stub driver instantiate a virtual i2c adapter without a parent.
> Check for this to avoid a NULL pointer deference.
> 
> Signed-off-by: Jean Delvare <jdelvare-l3A5Bk7waGM@public.gmane.org>
> Cc: Wolfram Sang <wsa-z923LK4zBo2bacvFa/9K2g@public.gmane.org>
> Cc: Mika Westerberg <mika.westerberg-VuQAYsv1563Yd54FQh9/CA@public.gmane.org>
> Cc: "Rafael J. Wysocki" <rafael.j.wysocki-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

Oh, I didn't know that. Thanks for the information.

Acked-by: Mika Westerberg <mika.westerberg-VuQAYsv1563Yd54FQh9/CA@public.gmane.org>

Probably should go to stable as well.

Re: [PATCH] i2c: Not all adapters have a parent

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 456 bytes --]

On Thu, Oct 10, 2013 at 08:04:06AM +0200, Jean Delvare wrote:
> The code in acpi_i2c_register_devices() assumes that all i2c adapters
> have a parent. This is not necessarily the case, for example the
> i2c-stub driver instantiate a virtual i2c adapter without a parent.
> Check for this to avoid a NULL pointer deference.
> 
> Signed-off-by: Jean Delvare <jdelvare-l3A5Bk7waGM@public.gmane.org>

Applied to for-current with stable tag, thanks!


[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH] i2c: Not all adapters have a parent

[Jean Delvare]

Hi Wolfram, Mika,

Le Thursday 10 October 2013 à 09:35 +0200, Wolfram Sang a écrit :
> On Thu, Oct 10, 2013 at 08:04:06AM +0200, Jean Delvare wrote:
> > The code in acpi_i2c_register_devices() assumes that all i2c adapters
> > have a parent. This is not necessarily the case, for example the
> > i2c-stub driver instantiate a virtual i2c adapter without a parent.
> > Check for this to avoid a NULL pointer deference.
> > 
> > Signed-off-by: Jean Delvare <jdelvare-l3A5Bk7waGM@public.gmane.org>
> 
> Applied to for-current with stable tag, thanks!

I don't hit the bug on 3.11. Looking at the code, before it was moved in
3.12-rc1, acpi_i2c_register_devices() was only called explicitly from
i2c-designware. So the fix doesn't actually need to go to stable.

-- 
Jean Delvare
Suse L3 Support