linux-kernel - Re: [PATCH] ktime: Simplify ktime

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <1495794020.29207.26.camel@perches.com>
Date:   Fri, 26 May 2017 03:20:20 -0700
From:   Joe Perches <joe@...ches.com>
To:     Mariusz Skamra <mariuszx.skamra@...el.com>,
        linux-kernel@...r.kernel.org
Cc:     tglx@...utronix.de
Subject: Re: [PATCH] ktime: Simplify ktime_* comparison functions

On Fri, 2017-05-26 at 12:09 +0200, Mariusz Skamra wrote:
> This simplifies ktime_compare, ktime_after and ktime_before to be defines.
> ktime_compare verified on x86_64 machine, gives a code 4 asm instruction
> less compared to the previous solution. Time measured is up to 2 times less.
> Signed-off-by: Mariusz Skamra <mariuszx.skamra@...el.com>
> Acked-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@...el.com>
> ---
>  include/linux/ktime.h | 19 +++----------------
>  1 file changed, 3 insertions(+), 16 deletions(-)
> 
> diff --git a/include/linux/ktime.h b/include/linux/ktime.h
> index 0c8bd45..3f3dfcd 100644
> --- a/include/linux/ktime.h
> +++ b/include/linux/ktime.h
> @@ -106,14 +106,7 @@ static inline ktime_t timeval_to_ktime(struct timeval tv)
>   *   cmp1 == cmp2: return 0
>   *   cmp1  > cmp2: return >0
>   */
> -static inline int ktime_compare(const ktime_t cmp1, const ktime_t cmp2)
> -{
> -	if (cmp1 < cmp2)
> -		return -1;
> -	if (cmp1 > cmp2)
> -		return 1;
> -	return 0;
> -}
> +#define ktime_compare(cmp1, cmp2)	ktime_sub(cmp1, cmp2)

Why not just change ktime_compare to

static inline int ktime_compare(const ktime_t cmp1, const ktime_t cmp2)
{
	return ktime_sub(cmp1, cmp2);
}

> 
>  /**
>   * ktime_after - Compare if a ktime_t value is bigger than another one.
> @@ -122,10 +115,7 @@ static inline int ktime_compare(const ktime_t cmp1, const ktime_t cmp2)
>   *
>   * Return: true if cmp1 happened after cmp2.
>   */
> -static inline bool ktime_after(const ktime_t cmp1, const ktime_t cmp2)
> -{
> -	return ktime_compare(cmp1, cmp2) > 0;
> -}
> +#define ktime_after(cmp1, cmp2)		(ktime_compare(cmp1, cmp2) > 0)

The ktime_after and ktime_before changes shouldn't
matter in generated objects.

I suggest these conversions from static inline to
#define are not changes for the better.
 
>  /**
>   * ktime_before - Compare if a ktime_t value is smaller than another one.
> @@ -134,10 +124,7 @@ static inline bool ktime_after(const ktime_t cmp1, const ktime_t cmp2)
>   *
>   * Return: true if cmp1 happened before cmp2.
>   */
> -static inline bool ktime_before(const ktime_t cmp1, const ktime_t cmp2)
> -{
> -	return ktime_compare(cmp1, cmp2) < 0;
> -}
> +#define ktime_before(cmp1, cmp2)	(ktime_compare(cmp1, cmp2) < 0)
>  
>  #if BITS_PER_LONG < 64
>  extern s64 __ktime_divns(const ktime_t kt, s64 div);