[PATCH] input: pxa27x_keypad: fix NULL pointer dereference

A NULL pointer dereference exception occurs in the driver probe function when
device tree is used.  The pdata pointer will be NULL in this case, but the code
dereferences it in all cases.  When device tree is used, a platform data
structure is allocated and initialized, and in all cases this pointer is copied
to the driver's private data, so the variable being tested should be accessed
through the driver's private data structure.

Signed-off-by: Mike Dunn <redacted>
---
 drivers/input/keyboard/pxa27x_keypad.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/input/keyboard/pxa27x_keypad.c b/drivers/input/keyboard/pxa27x_keypad.c
index 134c3b4..3b2a614 100644
--- a/drivers/input/keyboard/pxa27x_keypad.c
+++ b/drivers/input/keyboard/pxa27x_keypad.c

@@ -795,8 +795,10 @@ static int pxa27x_keypad_probe(struct platform_device *pdev)
 		goto failed_put_clk;
 	}
 
-	if ((pdata->enable_rotary0 && keypad->rotary_rel_code[0] != -1) ||
-	    (pdata->enable_rotary1 && keypad->rotary_rel_code[1] != -1)) {
+	if ((keypad->pdata->enable_rotary0 &&
+	     keypad->rotary_rel_code[0] != -1) ||
+	    (keypad->pdata->enable_rotary1 &&
+	     keypad->rotary_rel_code[1] != -1)) {
 		input_dev->evbit[0] |= BIT_MASK(EV_REL);
 	}
 

-- 
1.8.1.5

Dear Mike Dunn,

Nice find. Acked-by: Marek Vasut [off-list ref]

Best regards,
Marek Vasut

On Mon, Sep 16, 2013 at 06:49:53PM +0200, Marek Vasut wrote:

Dear Mike Dunn,

quoted

A NULL pointer dereference exception occurs in the driver probe function
when device tree is used.  The pdata pointer will be NULL in this case,
but the code dereferences it in all cases.  When device tree is used, a
platform data structure is allocated and initialized, and in all cases
this pointer is copied to the driver's private data, so the variable being
tested should be accessed through the driver's private data structure.

Signed-off-by: Mike Dunn <redacted>
---
 drivers/input/keyboard/pxa27x_keypad.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/input/keyboard/pxa27x_keypad.c

b/drivers/input/keyboard/pxa27x_keypad.c index 134c3b4..3b2a614 100644

--- a/drivers/input/keyboard/pxa27x_keypad.c
+++ b/drivers/input/keyboard/pxa27x_keypad.c

@@ -795,8 +795,10 @@ static int pxa27x_keypad_probe(struct platform_device

*pdev) goto failed_put_clk;
 	}

-	if ((pdata->enable_rotary0 && keypad->rotary_rel_code[0] != -1) ||
-	    (pdata->enable_rotary1 && keypad->rotary_rel_code[1] != -1)) {
+	if ((keypad->pdata->enable_rotary0 &&
+	     keypad->rotary_rel_code[0] != -1) ||
+	    (keypad->pdata->enable_rotary1 &&
+	     keypad->rotary_rel_code[1] != -1)) {
 		input_dev->evbit[0] |= BIT_MASK(EV_REL);
 	}

Nice find. Acked-by: Marek Vasut [off-list ref]

Excellent booby trap. I would prefer if we explicitly did

	pdata = keypad->pdata;

after calling the parse DT fucntion with a nice comment, because we
somebody might want to rearrange the code and accidentially revert the
checks to the original state.

Thanks.

-- 
Dmitry

On 09/16/2013 10:06 AM, Dmitry Torokhov wrote:

On Mon, Sep 16, 2013 at 06:49:53PM +0200, Marek Vasut wrote:

quoted

Dear Mike Dunn,

quoted

A NULL pointer dereference exception occurs in the driver probe function
when device tree is used.  The pdata pointer will be NULL in this case,
but the code dereferences it in all cases.  When device tree is used, a
platform data structure is allocated and initialized, and in all cases
this pointer is copied to the driver's private data, so the variable being
tested should be accessed through the driver's private data structure.

Signed-off-by: Mike Dunn <redacted>
---
 drivers/input/keyboard/pxa27x_keypad.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/input/keyboard/pxa27x_keypad.c

b/drivers/input/keyboard/pxa27x_keypad.c index 134c3b4..3b2a614 100644

--- a/drivers/input/keyboard/pxa27x_keypad.c
+++ b/drivers/input/keyboard/pxa27x_keypad.c

@@ -795,8 +795,10 @@ static int pxa27x_keypad_probe(struct platform_device

*pdev) goto failed_put_clk;
 	}

-	if ((pdata->enable_rotary0 && keypad->rotary_rel_code[0] != -1) ||
-	    (pdata->enable_rotary1 && keypad->rotary_rel_code[1] != -1)) {
+	if ((keypad->pdata->enable_rotary0 &&
+	     keypad->rotary_rel_code[0] != -1) ||
+	    (keypad->pdata->enable_rotary1 &&
+	     keypad->rotary_rel_code[1] != -1)) {
 		input_dev->evbit[0] |= BIT_MASK(EV_REL);
 	}

Nice find. Acked-by: Marek Vasut [off-list ref]

Excellent booby trap. I would prefer if we explicitly did

	pdata = keypad->pdata;

after calling the parse DT fucntion with a nice comment, because we
somebody might want to rearrange the code and accidentially revert the
checks to the original state.

Yes, that would have been better.  Is someone picking this up?  I'm not familir
with the input subsystem maintainer (sorry).  If this will be upstreamed in
someone's tree, I'll be glad to resubmit with this change.  Or, if you prefer,
please feel free to shepherd this Dmitry.

Sorry for the delay.

Thanks,
Mike

On Tue, Sep 17, 2013 at 09:05:54PM -0700, Mike Dunn wrote:

On 09/16/2013 10:06 AM, Dmitry Torokhov wrote:

quoted

On Mon, Sep 16, 2013 at 06:49:53PM +0200, Marek Vasut wrote:

quoted

Dear Mike Dunn,

quoted

A NULL pointer dereference exception occurs in the driver probe function
when device tree is used.  The pdata pointer will be NULL in this case,
but the code dereferences it in all cases.  When device tree is used, a
platform data structure is allocated and initialized, and in all cases
this pointer is copied to the driver's private data, so the variable being
tested should be accessed through the driver's private data structure.

Signed-off-by: Mike Dunn <redacted>
---
 drivers/input/keyboard/pxa27x_keypad.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/input/keyboard/pxa27x_keypad.c

b/drivers/input/keyboard/pxa27x_keypad.c index 134c3b4..3b2a614 100644

--- a/drivers/input/keyboard/pxa27x_keypad.c
+++ b/drivers/input/keyboard/pxa27x_keypad.c

@@ -795,8 +795,10 @@ static int pxa27x_keypad_probe(struct platform_device

*pdev) goto failed_put_clk;
 	}

-	if ((pdata->enable_rotary0 && keypad->rotary_rel_code[0] != -1) ||
-	    (pdata->enable_rotary1 && keypad->rotary_rel_code[1] != -1)) {
+	if ((keypad->pdata->enable_rotary0 &&
+	     keypad->rotary_rel_code[0] != -1) ||
+	    (keypad->pdata->enable_rotary1 &&
+	     keypad->rotary_rel_code[1] != -1)) {
 		input_dev->evbit[0] |= BIT_MASK(EV_REL);
 	}

Nice find. Acked-by: Marek Vasut [off-list ref]

Excellent booby trap. I would prefer if we explicitly did

	pdata = keypad->pdata;

after calling the parse DT fucntion with a nice comment, because we
somebody might want to rearrange the code and accidentially revert the
checks to the original state.

Yes, that would have been better.  Is someone picking this up?  I'm not familir
with the input subsystem maintainer (sorry).

That would be yours truly.

If this will be upstreamed in
someone's tree, I'll be glad to resubmit with this change.

If you could resubmit that would be great - I do not have the hardware
and I prefer applying patches that were tested, if possible.

Thanks.

-- 
Dmitry

On 09/17/2013 09:24 PM, Dmitry Torokhov wrote:

On Tue, Sep 17, 2013 at 09:05:54PM -0700, Mike Dunn wrote:

quoted

On 09/16/2013 10:06 AM, Dmitry Torokhov wrote:

quoted

On Mon, Sep 16, 2013 at 06:49:53PM +0200, Marek Vasut wrote:

quoted

Dear Mike Dunn,

quoted

A NULL pointer dereference exception occurs in the driver probe function
when device tree is used.  The pdata pointer will be NULL in this case,
but the code dereferences it in all cases.  When device tree is used, a
platform data structure is allocated and initialized, and in all cases
this pointer is copied to the driver's private data, so the variable being
tested should be accessed through the driver's private data structure.

Signed-off-by: Mike Dunn <redacted>
---
 drivers/input/keyboard/pxa27x_keypad.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/input/keyboard/pxa27x_keypad.c

b/drivers/input/keyboard/pxa27x_keypad.c index 134c3b4..3b2a614 100644

--- a/drivers/input/keyboard/pxa27x_keypad.c
+++ b/drivers/input/keyboard/pxa27x_keypad.c

@@ -795,8 +795,10 @@ static int pxa27x_keypad_probe(struct platform_device

*pdev) goto failed_put_clk;
 	}

-	if ((pdata->enable_rotary0 && keypad->rotary_rel_code[0] != -1) ||
-	    (pdata->enable_rotary1 && keypad->rotary_rel_code[1] != -1)) {
+	if ((keypad->pdata->enable_rotary0 &&
+	     keypad->rotary_rel_code[0] != -1) ||
+	    (keypad->pdata->enable_rotary1 &&
+	     keypad->rotary_rel_code[1] != -1)) {
 		input_dev->evbit[0] |= BIT_MASK(EV_REL);
 	}

Nice find. Acked-by: Marek Vasut [off-list ref]

Excellent booby trap. I would prefer if we explicitly did

	pdata = keypad->pdata;

after calling the parse DT fucntion with a nice comment, because we
somebody might want to rearrange the code and accidentially revert the
checks to the original state.

Yes, that would have been better.  Is someone picking this up?  I'm not familir
with the input subsystem maintainer (sorry).

That would be yours truly.

quoted

If this will be upstreamed in
someone's tree, I'll be glad to resubmit with this change.

If you could resubmit that would be great - I do not have the hardware
and I prefer applying patches that were tested, if possible.

OK, will do.  Yes, I'm testing this on a palm treo 680 with matrix keys and one
direct key.

Thanks,
Mike