drivers/mtd/devices/doc2000.c - address of 'eccbuf' will always evaluate as 'true' , warnings

drivers/mtd/devices/doc2000.c - address of 'eccbuf' will always evaluate as 'true' , warnings

[Gabriel C]

Hi,

I noticed this warnings on current git with gcc 4.2.1

....

drivers/mtd/devices/doc2000.c: In function 'doc_read':
drivers/mtd/devices/doc2000.c:635: warning: the address of 'eccbuf' will always evaluate as 'true'
drivers/mtd/devices/doc2000.c: In function 'doc_write':
drivers/mtd/devices/doc2000.c:899: warning: the address of 'eccbuf' will always evaluate as 'true'

....

This patch fixes it but maybe there is a better way to do it.


Signed-off-by: Gabriel Craciunescu <nix.or.die@googlemail.com>
---


diff --git a/drivers/mtd/devices/doc2000.c b/drivers/mtd/devices/doc2000.c
index c73e96b..ce8fbf5 100644
--- a/drivers/mtd/devices/doc2000.c
+++ b/drivers/mtd/devices/doc2000.c
@@ -632,7 +632,7 @@ static int doc_read(struct mtd_info *mtd, loff_t from, size_t len,
 			len = ((from | 0x1ff) + 1) - from;
 
 		/* The ECC will not be calculated correctly if less than 512 is read */
-		if (len != 0x200 && eccbuf)
+		if (len != 0x200 && eccbuf != NULL)
 			printk(KERN_WARNING
 			       "ECC needs a full sector read (adr: %lx size %lx)\n",
 			       (long) from, (long) len);
@@ -896,7 +896,7 @@ static int doc_write(struct mtd_info *mtd, loff_t to, size_t len,
 		/* Let the caller know we completed it */
 		*retlen += len;
 
-		if (eccbuf) {
+		if (eccbuf != NULL) {
 			unsigned char x[8];
 			size_t dummy;
 			int ret;

Re: drivers/mtd/devices/doc2000.c - address of 'eccbuf' will always evaluate as 'true' , warnings

[Jörn Engel]

On Sun, 22 July 2007 14:13:26 +0200, Gabriel C wrote:
> 
> I noticed this warnings on current git with gcc 4.2.1
> 
> ....
> 
> drivers/mtd/devices/doc2000.c: In function 'doc_read':
> drivers/mtd/devices/doc2000.c:635: warning: the address of 'eccbuf' will always evaluate as 'true'
> drivers/mtd/devices/doc2000.c: In function 'doc_write':
> drivers/mtd/devices/doc2000.c:899: warning: the address of 'eccbuf' will always evaluate as 'true'
> 
> ....
> 
> This patch fixes it but maybe there is a better way to do it.
> 
> 
> Signed-off-by: Gabriel Craciunescu <nix.or.die@googlemail.com>
> ---
> 
> 
> diff --git a/drivers/mtd/devices/doc2000.c b/drivers/mtd/devices/doc2000.c
> index c73e96b..ce8fbf5 100644
> --- a/drivers/mtd/devices/doc2000.c
> +++ b/drivers/mtd/devices/doc2000.c
> @@ -632,7 +632,7 @@ static int doc_read(struct mtd_info *mtd, loff_t from, size_t len,
>  			len = ((from | 0x1ff) + 1) - from;
>  
>  		/* The ECC will not be calculated correctly if less than 512 is read */
> -		if (len != 0x200 && eccbuf)
> +		if (len != 0x200 && eccbuf != NULL)
>  			printk(KERN_WARNING
>  			       "ECC needs a full sector read (adr: %lx size %lx)\n",
>  			       (long) from, (long) len);
> @@ -896,7 +896,7 @@ static int doc_write(struct mtd_info *mtd, loff_t to, size_t len,
>  		/* Let the caller know we completed it */
>  		*retlen += len;
>  
> -		if (eccbuf) {
> +		if (eccbuf != NULL) {
>  			unsigned char x[8];
>  			size_t dummy;
>  			int ret;

The patches doesn't fix anything.  If you take a look at git history for
the file, you may find this:
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=7f8a894066b826a4baea49c2a3adbba0a56a192f

And if I read the patch correctly, Thomas merely forgot to remove these
two checks on eccbuf.  So the correct patch would remove '&& eccbuf'
from one conditional and nuke the other completely.

Care to send such a patch?

Jörn

-- 
ticks = jiffies;
while (ticks == jiffies);
ticks = jiffies;
-- /usr/src/linux/init/main.c

Re: drivers/mtd/devices/doc2000.c - address of 'eccbuf' will always evaluate as 'true' , warnings

[Gabriel C]

Jörn Engel wrote:
> On Sun, 22 July 2007 14:13:26 +0200, Gabriel C wrote:
>> I noticed this warnings on current git with gcc 4.2.1
>>
>> ....
>>
>> drivers/mtd/devices/doc2000.c: In function 'doc_read':
>> drivers/mtd/devices/doc2000.c:635: warning: the address of 'eccbuf' will always evaluate as 'true'
>> drivers/mtd/devices/doc2000.c: In function 'doc_write':
>> drivers/mtd/devices/doc2000.c:899: warning: the address of 'eccbuf' will always evaluate as 'true'
>>
>> ....
>>
>> This patch fixes it but maybe there is a better way to do it.
>>
>>
>> Signed-off-by: Gabriel Craciunescu <nix.or.die@googlemail.com>
>> ---
>>
>>
>> diff --git a/drivers/mtd/devices/doc2000.c b/drivers/mtd/devices/doc2000.c
>> index c73e96b..ce8fbf5 100644
>> --- a/drivers/mtd/devices/doc2000.c
>> +++ b/drivers/mtd/devices/doc2000.c
>> @@ -632,7 +632,7 @@ static int doc_read(struct mtd_info *mtd, loff_t from, size_t len,
>>  			len = ((from | 0x1ff) + 1) - from;
>>  
>>  		/* The ECC will not be calculated correctly if less than 512 is read */
>> -		if (len != 0x200 && eccbuf)
>> +		if (len != 0x200 && eccbuf != NULL)
>>  			printk(KERN_WARNING
>>  			       "ECC needs a full sector read (adr: %lx size %lx)\n",
>>  			       (long) from, (long) len);
>> @@ -896,7 +896,7 @@ static int doc_write(struct mtd_info *mtd, loff_t to, size_t len,
>>  		/* Let the caller know we completed it */
>>  		*retlen += len;
>>  
>> -		if (eccbuf) {
>> +		if (eccbuf != NULL) {
>>  			unsigned char x[8];
>>  			size_t dummy;
>>  			int ret;
> 
> The patches doesn't fix anything.  If you take a look at git history for
> the file, you may find this:
> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=7f8a894066b826a4baea49c2a3adbba0a56a192f
> 
> And if I read the patch correctly, Thomas merely forgot to remove these
> two checks on eccbuf.  So the correct patch would remove '&& eccbuf'
> from one conditional and nuke the other completely.
> 
> Care to send such a patch?

Yes. I noticed doc2001plus.c has an leftover too.

...

drivers/mtd/devices/doc2001plus.c:751: warning: the address of 'eccbuf' will always evaluate as 'true'

...


> 
> Jörn
> 


Gabriel

diff --git a/drivers/mtd/devices/doc2000.c b/drivers/mtd/devices/doc2000.c
index c73e96b..d0e6ee8 100644
--- a/drivers/mtd/devices/doc2000.c
+++ b/drivers/mtd/devices/doc2000.c
@@ -632,7 +632,7 @@ static int doc_read(struct mtd_info *mtd, loff_t from, size_t len,
 			len = ((from | 0x1ff) + 1) - from;
 
 		/* The ECC will not be calculated correctly if less than 512 is read */
-		if (len != 0x200 && eccbuf)
+		if (len != 0x200)
 			printk(KERN_WARNING
 			       "ECC needs a full sector read (adr: %lx size %lx)\n",
 			       (long) from, (long) len);
@@ -896,25 +896,6 @@ static int doc_write(struct mtd_info *mtd, loff_t to, size_t len,
 		/* Let the caller know we completed it */
 		*retlen += len;
 
-		if (eccbuf) {
-			unsigned char x[8];
-			size_t dummy;
-			int ret;
-
-			/* Write the ECC data to flash */
-			for (di=0; di<6; di++)
-				x[di] = eccbuf[di];
-
-			x[6]=0x55;
-			x[7]=0x55;
-
-			ret = doc_write_oob_nolock(mtd, to, 8, &dummy, x);
-			if (ret) {
-				mutex_unlock(&this->lock);
-				return ret;
-			}
-		}
-
 		to += len;
 		left -= len;
 		buf += len;
diff --git a/drivers/mtd/devices/doc2001plus.c b/drivers/mtd/devices/doc2001plus.c
index 2b30b58..83be346 100644
--- a/drivers/mtd/devices/doc2001plus.c
+++ b/drivers/mtd/devices/doc2001plus.c
@@ -748,7 +748,7 @@ static int doc_write(struct mtd_info *mtd, loff_t to, size_t len,
 	WriteDOC(DoC_GetDataOffset(mtd, &fto), docptr, Mplus_FlashCmd);
 
 	/* On interleaved devices the flags for 2nd half 512 are before data */
-	if (eccbuf && before)
+	if (before)
 		fto -= 2;
 
 	/* issue the Serial Data In command to initial the Page Program process */

Re: drivers/mtd/devices/doc2000.c - address of 'eccbuf' will always evaluate as 'true' , warnings

[Jörn Engel]

On Sun, 22 July 2007 22:29:29 +0200, Gabriel C wrote:
> 
> @@ -896,25 +896,6 @@ static int doc_write(struct mtd_info *mtd, loff_t to, size_t len,
>  		/* Let the caller know we completed it */
>  		*retlen += len;
>  
> -		if (eccbuf) {
> -			unsigned char x[8];
> -			size_t dummy;
> -			int ret;
> -
> -			/* Write the ECC data to flash */
> -			for (di=0; di<6; di++)
> -				x[di] = eccbuf[di];
> -
> -			x[6]=0x55;
> -			x[7]=0x55;
> -
> -			ret = doc_write_oob_nolock(mtd, to, 8, &dummy, x);
> -			if (ret) {
> -				mutex_unlock(&this->lock);
> -				return ret;
> -			}
> -		}
> -
>  		to += len;
>  		left -= len;
>  		buf += len;

You should remove the condition, but not the conditional code.

Jörn

-- 
A quarrel is quickly settled when deserted by one party; there is
no battle unless there be two.
-- Seneca

Re: drivers/mtd/devices/doc2000.c - address of 'eccbuf' will always evaluate as 'true' , warnings

[Gabriel C]

Jörn Engel wrote:
> On Sun, 22 July 2007 22:29:29 +0200, Gabriel C wrote:
>> @@ -896,25 +896,6 @@ static int doc_write(struct mtd_info *mtd, loff_t to, size_t len,
>>  		/* Let the caller know we completed it */
>>  		*retlen += len;
>>  
>> -		if (eccbuf) {
>> -			unsigned char x[8];
>> -			size_t dummy;
>> -			int ret;
>> -
>> -			/* Write the ECC data to flash */
>> -			for (di=0; di<6; di++)
>> -				x[di] = eccbuf[di];
>> -
>> -			x[6]=0x55;
>> -			x[7]=0x55;
>> -
>> -			ret = doc_write_oob_nolock(mtd, to, 8, &dummy, x);
>> -			if (ret) {
>> -				mutex_unlock(&this->lock);
>> -				return ret;
>> -			}
>> -		}
>> -
>>  		to += len;
>>  		left -= len;
>>  		buf += len;
> 
> You should remove the condition, but not the conditional code.

Sorry , I was away from keyboard.

Then I can't understand what this code is doing :/ 

So let Thomas or someone know this code fix it.

> 
> Jörn
> 

Regards,

Gabriel