[PATCH] saa7146: fix sparse warnings

[PATCH] saa7146: fix sparse warnings

[Harvey Harrison]

Went along with the existing file style.
drivers/media/common/saa7146_core.c:309:6: warning: Using plain integer as NULL pointer
drivers/media/common/saa7146_core.c:311:8: warning: Using plain integer as NULL pointer
drivers/media/common/saa7146_core.c:319:7: warning: Using plain integer as NULL pointer
drivers/media/common/saa7146_core.c:319:28: warning: Using plain integer as NULL pointer
drivers/media/common/saa7146_core.c:325:7: warning: Using plain integer as NULL pointer
drivers/media/common/saa7146_core.c:325:28: warning: Using plain integer as NULL pointer
drivers/media/common/saa7146_fops.c:275:12: warning: Using plain integer as NULL pointer

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
---
 drivers/media/common/saa7146_core.c |    8 ++++----
 drivers/media/common/saa7146_fops.c |    2 +-
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/media/common/saa7146_core.c b/drivers/media/common/saa7146_core.c
index 168a8d3..8c755a1 100644
--- a/drivers/media/common/saa7146_core.c
+++ b/drivers/media/common/saa7146_core.c
@@ -306,9 +306,9 @@ static irqreturn_t interrupt_hw(int irq, void *dev_id)
 		return IRQ_NONE;
 	}
 
-	if( 0 != (dev->ext)) {
+	if( NULL != (dev->ext)) {
 		if( 0 != (dev->ext->irq_mask & isr )) {
-			if( 0 != dev->ext->irq_func ) {
+			if( NULL != dev->ext->irq_func ) {
 				dev->ext->irq_func(dev, &isr);
 			}
 			isr &= ~dev->ext->irq_mask;
@@ -316,13 +316,13 @@ static irqreturn_t interrupt_hw(int irq, void *dev_id)
 	}
 	if (0 != (isr & (MASK_27))) {
 		DEB_INT(("irq: RPS0 (0x%08x).\n",isr));
-		if( 0 != dev->vv_data && 0 != dev->vv_callback) {
+		if( NULL != dev->vv_data && NULL != dev->vv_callback) {
 			dev->vv_callback(dev,isr);
 		}
 		isr &= ~MASK_27;
 	}
 	if (0 != (isr & (MASK_28))) {
-		if( 0 != dev->vv_data && 0 != dev->vv_callback) {
+		if( NULL != dev->vv_data && NULL != dev->vv_callback) {
 			dev->vv_callback(dev,isr);
 		}
 		isr &= ~MASK_28;
diff --git a/drivers/media/common/saa7146_fops.c b/drivers/media/common/saa7146_fops.c
index f0703d8..a73db79 100644
--- a/drivers/media/common/saa7146_fops.c
+++ b/drivers/media/common/saa7146_fops.c
@@ -272,7 +272,7 @@ static int fops_open(struct inode *inode, struct file *file)
 
 	result = 0;
 out:
-	if( fh != 0 && result != 0 ) {
+	if( fh != NULL && result != 0 ) {
 		kfree(fh);
 		file->private_data = NULL;
 	}
-- 
1.5.4.2.200.g99e75

Re: [PATCH] saa7146: fix sparse warnings

[Al Viro]

On Thu, Feb 21, 2008 at 09:52:36AM -0800, Harvey Harrison wrote:

Could you please use more descriptive names?  NULL noise removal
is not the same as shadowing or endianness annotations or endianness
fixes or __user/__iomem annotations/fixes, etc.

Re: [PATCH] saa7146: fix sparse warnings

[Harvey Harrison]

On Thu, 2008-02-21 at 21:19 +0000, Al Viro wrote:
> On Thu, Feb 21, 2008 at 09:52:36AM -0800, Harvey Harrison wrote:
> 
> Could you please use more descriptive names?  NULL noise removal
> is not the same as shadowing or endianness annotations or endianness
> fixes or __user/__iomem annotations/fixes, etc.

Point taken, will do so going forward.

For now, I'm just doing the trivial ones to make the more interesting
warnings stand out a little better.

Harvey

Re: [PATCH] saa7146: fix sparse warnings

[Dmitri Vorobiev]

Harvey Harrison пишет:
> Went along with the existing file style.
> drivers/media/common/saa7146_core.c:309:6: warning: Using plain integer as NULL pointer
> drivers/media/common/saa7146_core.c:311:8: warning: Using plain integer as NULL pointer
> drivers/media/common/saa7146_core.c:319:7: warning: Using plain integer as NULL pointer
> drivers/media/common/saa7146_core.c:319:28: warning: Using plain integer as NULL pointer
> drivers/media/common/saa7146_core.c:325:7: warning: Using plain integer as NULL pointer
> drivers/media/common/saa7146_core.c:325:28: warning: Using plain integer as NULL pointer
> drivers/media/common/saa7146_fops.c:275:12: warning: Using plain integer as NULL pointer
> 
> Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
> ---
>  drivers/media/common/saa7146_core.c |    8 ++++----
>  drivers/media/common/saa7146_fops.c |    2 +-
>  2 files changed, 5 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/media/common/saa7146_core.c b/drivers/media/common/saa7146_core.c
> index 168a8d3..8c755a1 100644
> --- a/drivers/media/common/saa7146_core.c
> +++ b/drivers/media/common/saa7146_core.c
> @@ -306,9 +306,9 @@ static irqreturn_t interrupt_hw(int irq, void *dev_id)
>  		return IRQ_NONE;
>  	}
>  
> -	if( 0 != (dev->ext)) {
> +	if( NULL != (dev->ext)) {

At the risk of looking an idiot, I'm taking a liberty to ask what is
the point in explicit comparison to zero in conditional operators? Is
it not a fundamental C idiom to write

if (a) {
}

instead of

if (a != 0) {
}

and, similarly, to write

if (!a) {
}

instead of

if (a == 0) {
}

?

Thanks,

Dmitri

Re: [PATCH] saa7146: fix sparse warnings

[Harvey Harrison]

On Sun, 2008-03-02 at 20:34 +0300, Dmitri Vorobiev wrote:
> Harvey Harrison пишет:
> >  
> > -	if( 0 != (dev->ext)) {
> > +	if( NULL != (dev->ext)) {
> 
> At the risk of looking an idiot, I'm taking a liberty to ask what is
> the point in explicit comparison to zero in conditional operators? Is
> it not a fundamental C idiom to write

<snip>

Yes, that's how I would have written it, but I tried to keep with the
prevailing style in that file.  I suppose I could see an argument for
consistency if you had a long series of if() statements to keep a
similar style.

if (foo == value1)

if (bar == value2)

if (baz == NULL)

I'll leave the discussion of putting the constant first in the comparison
for someone else to comment on.

Harvey

Re: [PATCH] saa7146: fix sparse warnings

[Dmitri Vorobiev]

Harvey Harrison пишет:
> On Sun, 2008-03-02 at 20:34 +0300, Dmitri Vorobiev wrote:
>> Harvey Harrison пишет:
>>>  
>>> -	if( 0 != (dev->ext)) {
>>> +	if( NULL != (dev->ext)) {
>> At the risk of looking an idiot, I'm taking a liberty to ask what is
>> the point in explicit comparison to zero in conditional operators? Is
>> it not a fundamental C idiom to write
> 
> <snip>
> 
> Yes, that's how I would have written it, but I tried to keep with the
> prevailing style in that file.  I suppose I could see an argument for
> consistency if you had a long series of if() statements to keep a
> similar style.

To me it looks that the original style in this file can be sacrificed
in favor of

1) satisfying the coding style rules of the kernel;
2) keeping with informal, but commonly known idioms of C language.

So I thought that while you're at it, you could also comb through
this file and make the checkpatch.pl script happy with it.

But this is just my very personal opinion :)

Dmitri

Re: [PATCH] saa7146: fix sparse warnings

[Andrew Morton]

On Sun, 02 Mar 2008 10:06:44 -0800 Harvey Harrison <harvey.harrison@gmail.com> wrote:

> On Sun, 2008-03-02 at 20:34 +0300, Dmitri Vorobiev wrote:
> > Harvey Harrison пишет:
> > >  
> > > -	if( 0 != (dev->ext)) {
> > > +	if( NULL != (dev->ext)) {
> > 
> > At the risk of looking an idiot, I'm taking a liberty to ask what is
> > the point in explicit comparison to zero in conditional operators? Is
> > it not a fundamental C idiom to write
> 
> <snip>
> 
> Yes, that's how I would have written it, but I tried to keep with the
> prevailing style in that file.

I'm not a bit fan of the match-the-existing-style approach.

You'll find that files which started out with a non-standard style already
contain a mixup of styles, because later changes were often made in
standard-style.

And given that we're churning the code anyway, we might as well fix it up now. 
Yes, that'll make the code look partially-weird rather than wholly-weird,
but I doubt if that will harm anyone much.