[PATCH 12/14] scsi: use printk_once

[PATCH 12/14] scsi: use printk_once

[Marcin Slusarz]

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com>
Cc: linux-scsi@vger.kernel.org
---
 drivers/scsi/53c700.c             |    6 +-----
 drivers/scsi/cxgb3i/cxgb3i_init.c |    7 +------
 2 files changed, 2 insertions(+), 11 deletions(-)

diff --git a/drivers/scsi/53c700.c b/drivers/scsi/53c700.c
index f5a9add..c5f4fa7 100644
--- a/drivers/scsi/53c700.c
+++ b/drivers/scsi/53c700.c
@@ -291,7 +291,6 @@ NCR_700_detect(struct scsi_host_template *tpnt,
 	__u8 *memory;
 	__u32 *script;
 	struct Scsi_Host *host;
-	static int banner = 0;
 	int j;
 
 	if(tpnt->sdev_attrs == NULL)
@@ -385,10 +384,7 @@ NCR_700_detect(struct scsi_host_template *tpnt,
 	else
 		hostdata->rev = (NCR_700_readb(host, CTEST7_REG)>>4) & 0x0f;
 	hostdata->fast = (NCR_700_readb(host, CTEST9_REG) == 0);
-	if (banner == 0) {
-		printk(KERN_NOTICE "53c700: Version " NCR_700_VERSION " By James.Bottomley@HansenPartnership.com\n");
-		banner = 1;
-	}
+	printk_once(KERN_NOTICE "53c700: Version " NCR_700_VERSION " By James.Bottomley@HansenPartnership.com\n");
 	printk(KERN_NOTICE "scsi%d: %s rev %d %s\n", host->host_no,
 	       hostdata->chip710 ? "53c710" :
 	       (hostdata->fast ? "53c700-66" : "53c700"),
diff --git a/drivers/scsi/cxgb3i/cxgb3i_init.c b/drivers/scsi/cxgb3i/cxgb3i_init.c
index 042d9bc..8c9285f 100644
--- a/drivers/scsi/cxgb3i/cxgb3i_init.c
+++ b/drivers/scsi/cxgb3i/cxgb3i_init.c
@@ -43,12 +43,7 @@ static struct cxgb3_client t3c_client = {
  */
 static void open_s3_dev(struct t3cdev *t3dev)
 {
-	static int vers_printed;
-
-	if (!vers_printed) {
-		printk(KERN_INFO "%s", version);
-		vers_printed = 1;
-	}
+	printk_once(KERN_INFO "%s", version);
 
 	cxgb3i_ddp_init(t3dev);
 	cxgb3i_sdev_add(t3dev, &t3c_client);
-- 
1.6.3.3

Re: [PATCH 12/14] scsi: use printk_once

[James Bottomley]

On Sun, 2009-08-09 at 21:54 +0200, Marcin Slusarz wrote:
> Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com>
> Cc: linux-scsi@vger.kernel.org
> ---
>  drivers/scsi/53c700.c             |    6 +-----
>  drivers/scsi/cxgb3i/cxgb3i_init.c |    7 +------
>  2 files changed, 2 insertions(+), 11 deletions(-)
> 
> diff --git a/drivers/scsi/53c700.c b/drivers/scsi/53c700.c
> index f5a9add..c5f4fa7 100644
> --- a/drivers/scsi/53c700.c
> +++ b/drivers/scsi/53c700.c
> @@ -291,7 +291,6 @@ NCR_700_detect(struct scsi_host_template *tpnt,
>  	__u8 *memory;
>  	__u32 *script;
>  	struct Scsi_Host *host;
> -	static int banner = 0;
>  	int j;
>  
>  	if(tpnt->sdev_attrs == NULL)
> @@ -385,10 +384,7 @@ NCR_700_detect(struct scsi_host_template *tpnt,
>  	else
>  		hostdata->rev = (NCR_700_readb(host, CTEST7_REG)>>4) & 0x0f;
>  	hostdata->fast = (NCR_700_readb(host, CTEST9_REG) == 0);
> -	if (banner == 0) {
> -		printk(KERN_NOTICE "53c700: Version " NCR_700_VERSION " By James.Bottomley@HansenPartnership.com\n");
> -		banner = 1;
> -	}
> +	printk_once(KERN_NOTICE "53c700: Version " NCR_700_VERSION " By James.Bottomley@HansenPartnership.com\n");
>  	printk(KERN_NOTICE "scsi%d: %s rev %d %s\n", host->host_no,
>  	       hostdata->chip710 ? "53c710" :
>  	       (hostdata->fast ? "53c700-66" : "53c700"),
> diff --git a/drivers/scsi/cxgb3i/cxgb3i_init.c b/drivers/scsi/cxgb3i/cxgb3i_init.c
> index 042d9bc..8c9285f 100644
> --- a/drivers/scsi/cxgb3i/cxgb3i_init.c
> +++ b/drivers/scsi/cxgb3i/cxgb3i_init.c
> @@ -43,12 +43,7 @@ static struct cxgb3_client t3c_client = {
>   */
>  static void open_s3_dev(struct t3cdev *t3dev)
>  {
> -	static int vers_printed;
> -
> -	if (!vers_printed) {
> -		printk(KERN_INFO "%s", version);
> -		vers_printed = 1;
> -	}
> +	printk_once(KERN_INFO "%s", version);

There's really no point to this beyond code churn.  Both patterns are
correct uses so conversion buys us nothing.

James

Re: [PATCH 12/14] scsi: use printk_once

[Roland Dreier]

 > > -	static int vers_printed;
 > > -
 > > -	if (!vers_printed) {
 > > -		printk(KERN_INFO "%s", version);
 > > -		vers_printed = 1;
 > > -	}
 > > +	printk_once(KERN_INFO "%s", version);

 > There's really no point to this beyond code churn.  Both patterns are
 > correct uses so conversion buys us nothing.

It's a pretty small gain but I do think 6 lines -> 1 line does have value.

 - R.

Re: [PATCH 12/14] scsi: use printk_once

[Joe Perches]

On Tue, 2009-08-11 at 11:40 -0500, James Bottomley wrote:
> On Sun, 2009-08-09 at 21:54 +0200, Marcin Slusarz wrote:
> > --- a/drivers/scsi/cxgb3i/cxgb3i_init.c
> > +++ b/drivers/scsi/cxgb3i/cxgb3i_init.c
> > @@ -43,12 +43,7 @@ static struct cxgb3_client t3c_client = {
> >   */
> >  static void open_s3_dev(struct t3cdev *t3dev)
> >  {
> > -	static int vers_printed;
> > -
> > -	if (!vers_printed) {
> > -		printk(KERN_INFO "%s", version);
> > -		vers_printed = 1;
> > -	}
> > +	printk_once(KERN_INFO "%s", version);
> 
> There's really no point to this beyond code churn.  Both patterns are
> correct uses so conversion buys us nothing.

Standards are for hacks.  Stasis is the best policy.
Using patterns isn't worth the electron movement.
Code shortening and readability?  Humbug.

Re: [PATCH 12/14] scsi: use printk_once

[James Bottomley]

On Tue, 2009-08-11 at 09:46 -0700, Roland Dreier wrote:
> > > -	static int vers_printed;
>  > > -
>  > > -	if (!vers_printed) {
>  > > -		printk(KERN_INFO "%s", version);
>  > > -		vers_printed = 1;
>  > > -	}
>  > > +	printk_once(KERN_INFO "%s", version);
> 
>  > There's really no point to this beyond code churn.  Both patterns are
>  > correct uses so conversion buys us nothing.
> 
> It's a pretty small gain but I do think 6 lines -> 1 line does have value.

Not at the expense of churning the code base of marginal drivers.

I'm not entirely convinced this is a good pattern to begin with.  If you
add too many patterns you're effectively developing an overly complex
development language and rule set.  Even if I go with this on the basis
of readability (as you say, one line for six), then I still don't think
we should churn the code base to convert correct open coding to it; we
should just use it ongoing.

James