[PATCH] drivers/applicom:Removed the unused variable byte_reset_it

[PATCH] drivers/applicom:Removed the unused variable byte_reset_it

[Zhu Jun]

Remove unused variable and redundant readb in ac_register_board

Signed-off-by: Zhu Jun <zhujun2@cmss.chinamobile.com>
---
 drivers/char/applicom.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/char/applicom.c b/drivers/char/applicom.c
index 9fed9706d9cd..ea25caf35346 100644
--- a/drivers/char/applicom.c
+++ b/drivers/char/applicom.c
@@ -127,8 +127,6 @@ static int dummy;	/* dev_id for request_irq() */
 static int ac_register_board(unsigned long physloc, void __iomem *loc, 
 		      unsigned char boardno)
 {
-	volatile unsigned char byte_reset_it;
-
 	if((readb(loc + CONF_END_TEST)     != 0x00) ||
 	   (readb(loc + CONF_END_TEST + 1) != 0x55) ||
 	   (readb(loc + CONF_END_TEST + 2) != 0xAA) ||
@@ -156,7 +154,6 @@ static int ac_register_board(unsigned long physloc, void __iomem *loc,
 	apbs[boardno].RamIO = loc;
 	init_waitqueue_head(&apbs[boardno].FlagSleepSend);
 	spin_lock_init(&apbs[boardno].mutex);
-	byte_reset_it = readb(loc + RAM_IT_TO_PC);
 
 	numboards++;
 	return boardno + 1;
-- 
2.17.1

Re: [PATCH] drivers/applicom:Removed the unused variable byte_reset_it

[Arnd Bergmann]

On Tue, Oct 8, 2024, at 07:28, Zhu Jun wrote:
> Remove unused variable and redundant readb in ac_register_board
>
> Signed-off-by: Zhu Jun <zhujun2@cmss.chinamobile.com>

This is probably intentional, as the readb() can both have side-effects
on the device and serialize the device against a previous instructions.

It would have been nice to have a comment in there, but since that
is 30 year old code, I doubt we get more information on it now.

Just don't remove those.

      Arnd

Re: [PATCH] drivers/applicom:Removed the unused variable byte_reset_it

[Greg KH]

On Tue, Oct 08, 2024 at 12:28:25AM -0700, Zhu Jun wrote:
> Remove unused variable and redundant readb in ac_register_board
> 
> Signed-off-by: Zhu Jun <zhujun2@cmss.chinamobile.com>
> ---
>  drivers/char/applicom.c | 3 ---
>  1 file changed, 3 deletions(-)
> 
> diff --git a/drivers/char/applicom.c b/drivers/char/applicom.c
> index 9fed9706d9cd..ea25caf35346 100644
> --- a/drivers/char/applicom.c
> +++ b/drivers/char/applicom.c
> @@ -127,8 +127,6 @@ static int dummy;	/* dev_id for request_irq() */
>  static int ac_register_board(unsigned long physloc, void __iomem *loc, 
>  		      unsigned char boardno)
>  {
> -	volatile unsigned char byte_reset_it;
> -
>  	if((readb(loc + CONF_END_TEST)     != 0x00) ||
>  	   (readb(loc + CONF_END_TEST + 1) != 0x55) ||
>  	   (readb(loc + CONF_END_TEST + 2) != 0xAA) ||
> @@ -156,7 +154,6 @@ static int ac_register_board(unsigned long physloc, void __iomem *loc,
>  	apbs[boardno].RamIO = loc;
>  	init_waitqueue_head(&apbs[boardno].FlagSleepSend);
>  	spin_lock_init(&apbs[boardno].mutex);
> -	byte_reset_it = readb(loc + RAM_IT_TO_PC);

As Arnd said, this is doing something, please be more familiar with
hardware types (i.e. reads are required to ensure a write happens),
before removing lines that feel they are not needed, unless you can test
the driver change yourself.

thanks,

greg k-h