[PATCH] input: signedness bug in cma3000_thread_irq()

[PATCH] input: signedness bug in cma3000_thread_irq()

[Xi Wang]

The error check (intr_status < 0) didn't work because intr_status is
a u8.  Change its type to signed int.

Signed-off-by: Xi Wang <xi.wang@gmail.com>
---
 drivers/input/misc/cma3000_d0x.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/input/misc/cma3000_d0x.c b/drivers/input/misc/cma3000_d0x.c
index 80793f1..06517e6 100644
--- a/drivers/input/misc/cma3000_d0x.c
+++ b/drivers/input/misc/cma3000_d0x.c
@@ -115,8 +115,8 @@ static void decode_mg(struct cma3000_accl_data *data, int *datax,
 static irqreturn_t cma3000_thread_irq(int irq, void *dev_id)
 {
 	struct cma3000_accl_data *data = dev_id;
-	int datax, datay, dataz;
-	u8 ctrl, mode, range, intr_status;
+	int datax, datay, dataz, intr_status;
+	u8 ctrl, mode, range;
 
 	intr_status = CMA3000_READ(data, CMA3000_INTSTATUS, "interrupt status");
 	if (intr_status < 0)
-- 
1.7.5.4

Re: [PATCH] input: signedness bug in cma3000_thread_irq()

[Dmitry Torokhov]

On Mon, Dec 12, 2011 at 01:49:26AM -0500, Xi Wang wrote:
> The error check (intr_status < 0) didn't work because intr_status is
> a u8.  Change its type to signed int.
> 

Applied, thank you.

-- 
Dmitry

Re: [PATCH] input: signedness bug in cma3000_thread_irq()

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 511 bytes --]

On Mon, 12 Dec 2011 01:49:26 EST, Xi Wang said:
> The error check (intr_status < 0) didn't work because intr_status is
> a u8.  Change its type to signed int.
>
> Signed-off-by: Xi Wang <xi.wang@gmail.com>
> ---
>  drivers/input/misc/cma3000_d0x.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/input/misc/cma3000_d0x.c b/drivers/input/misc/cma3000_d0x.c

That will fix the signed/unsigned issue.  However, was there a reason
the field was an 8-bit-wide to begin with?

[-- Attachment #2: Type: application/pgp-signature, Size: 227 bytes --]

Re: [PATCH] input: signedness bug in cma3000_thread_irq()

[Xi Wang]

On Dec 14, 2011, at 9:27 PM, Valdis.Kletnieks@vt.edu wrote:
> That will fix the signed/unsigned issue.  However, was there a reason
> the field was an 8-bit-wide to begin with?

Which field?  The ->read() function pointer?

- xi

Re: [PATCH] input: signedness bug in cma3000_thread_irq()

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 354 bytes --]

On Wed, 14 Dec 2011 21:35:22 EST, Xi Wang said:
> On Dec 14, 2011, at 9:27 PM, Valdis.Kletnieks@vt.edu wrote:
> > That will fix the signed/unsigned issue.  However, was there a reason
> > the field was an 8-bit-wide to begin with?
>
> Which field?  The ->read() function pointer?

No, the variable that used to be a u8 - was it 8 bits wide for a reason?

[-- Attachment #2: Type: application/pgp-signature, Size: 227 bytes --]

Re: [PATCH] input: signedness bug in cma3000_thread_irq()

[Xi Wang]

On Dec 14, 2011, at 10:21 PM, Valdis.Kletnieks@vt.edu wrote:
> No, the variable that used to be a u8 - was it 8 bits wide for a reason?

The return value of CMA3000_READ?  I don't see any particular reason.  

- xi

Re: [PATCH] input: signedness bug in cma3000_thread_irq()

[Dmitry Torokhov]

On Thu, Dec 15, 2011 at 05:17:02PM -0500, Xi Wang wrote:
> On Dec 14, 2011, at 10:21 PM, Valdis.Kletnieks@vt.edu wrote:
> > No, the variable that used to be a u8 - was it 8 bits wide for a reason?
> 
> The return value of CMA3000_READ?  I don't see any particular reason.  
> 

The data returned by CMA3000_READ() is a byte, so that's the reason (and
I am guessing here) that the author used u8 to store the result,
forgetting that the range was extended to report error conditions.

-- 
Dmitry