[PATCH] include/asm-generic/pgtable.h: Check once and return for pgprot_modify

[PATCH] include/asm-generic/pgtable.h: Check once and return for pgprot_modify

[Chen Gang]

pgprot_val() likes an empty operation, and all three checking statements
are related with oldprot, so only check once is enough, the reason is:

 - if both "oldprot == pgprot_noncached(oldprot)" and "oldprot ==
   pgprot_writecombine(oldprot)" are satisfied.

 - So "pgprot_noncached(oldprot) == pgprot_writecombine(oldprot)", too.

 - So as the result, it is still equal to check once (for pgprot_device
   can get the same result).

Then can return directly.

Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
---
 include/asm-generic/pgtable.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/include/asm-generic/pgtable.h b/include/asm-generic/pgtable.h
index 069536a..cd9e7c4 100644
--- a/include/asm-generic/pgtable.h
+++ b/include/asm-generic/pgtable.h
@@ -306,11 +306,11 @@ static inline int pmd_same(pmd_t pmd_a, pmd_t pmd_b)
 static inline pgprot_t pgprot_modify(pgprot_t oldprot, pgprot_t newprot)
 {
 	if (pgprot_val(oldprot) == pgprot_val(pgprot_noncached(oldprot)))
-		newprot = pgprot_noncached(newprot);
+		return pgprot_noncached(newprot);
 	if (pgprot_val(oldprot) == pgprot_val(pgprot_writecombine(oldprot)))
-		newprot = pgprot_writecombine(newprot);
+		return pgprot_writecombine(newprot);
 	if (pgprot_val(oldprot) == pgprot_val(pgprot_device(oldprot)))
-		newprot = pgprot_device(newprot);
+		return pgprot_device(newprot);
 	return newprot;
 }
 #endif
-- 
1.9.3

Re: [PATCH] include/asm-generic/pgtable.h: Check once and return for pgprot_modify

[Chen Gang]

Oh, after check again, it looks my original comments are not precise:
they check oldprot, but set newprot.

After check pgprot_noncached, pgprot_writecombine, and pgprot_device
again, for me, if still in order: 1st noncached, 2nd writecombine, 3rd
device, we can still treate them exclusive with each other:

 - For pgprot_device, Only amd64 redefine it, which is exclusive with
   the others (or the precisely same as one of the others).

 - For pgprot_writecombine, except metag and x86, all the other archs
   are exclusive with (or same as) pgprot_noncached.

   for metag arch, pgprot_writecombine is the part of pgprot_noncached.

   for x86 arch, it has its own pgprot_modify.


Thanks.

On 12/13/15 17:12, Chen Gang wrote:
> 
> pgprot_val() likes an empty operation, and all three checking statements
> are related with oldprot, so only check once is enough, the reason is:
> 
>  - if both "oldprot == pgprot_noncached(oldprot)" and "oldprot ==
>    pgprot_writecombine(oldprot)" are satisfied.
> 
>  - So "pgprot_noncached(oldprot) == pgprot_writecombine(oldprot)", too.
> 
>  - So as the result, it is still equal to check once (for pgprot_device
>    can get the same result).
> 
> Then can return directly.
> 
> Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
> ---
>  include/asm-generic/pgtable.h | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/include/asm-generic/pgtable.h b/include/asm-generic/pgtable.h
> index 069536a..cd9e7c4 100644
> --- a/include/asm-generic/pgtable.h
> +++ b/include/asm-generic/pgtable.h
> @@ -306,11 +306,11 @@ static inline int pmd_same(pmd_t pmd_a, pmd_t pmd_b)
>  static inline pgprot_t pgprot_modify(pgprot_t oldprot, pgprot_t newprot)
>  {
>  	if (pgprot_val(oldprot) == pgprot_val(pgprot_noncached(oldprot)))
> -		newprot = pgprot_noncached(newprot);
> +		return pgprot_noncached(newprot);
>  	if (pgprot_val(oldprot) == pgprot_val(pgprot_writecombine(oldprot)))
> -		newprot = pgprot_writecombine(newprot);
> +		return pgprot_writecombine(newprot);
>  	if (pgprot_val(oldprot) == pgprot_val(pgprot_device(oldprot)))
> -		newprot = pgprot_device(newprot);
> +		return pgprot_device(newprot);
>  	return newprot;
>  }
>  #endif
> 

-- 
Chen Gang (陈刚)

Open, share, and attitude like air, water, and life which God blessed