linux-kernel - Re: [PATCH 2/3] Turn bitmap_set and bitmap

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: <87bmpzbiru.fsf@rasmusvillemoes.dk>
Date:   Wed, 07 Jun 2017 23:16:37 +0200
From:   Rasmus Villemoes <linux@...musvillemoes.dk>
To:     Matthew Wilcox <willy@...radead.org>
Cc:     linux-kernel@...r.kernel.org,
        Andrew Morton <akpm@...ux-foundation.org>,
        Martin Schwidefsky <schwidefsky@...ibm.com>,
        Matthew Wilcox <mawilcox@...rosoft.com>
Subject: Re: [PATCH 2/3] Turn bitmap_set and bitmap_clear into memset when possible

On Wed, Jun 07 2017, Matthew Wilcox <willy@...radead.org> wrote:

> From: Matthew Wilcox <mawilcox@...rosoft.com>
>
> Several callers have constant 'start' and an 'nbits' that is a multiple of
> 8, so we can turn them into calls to memset.  We don't need the entirety
> of 'start' and 'nbits' to  be constant, we just need to know whether
> they're divisible by 8.
>
> Signed-off-by: Matthew Wilcox <mawilcox@...rosoft.com>
> ---
>  include/linux/bitmap.h | 6 ++++++
>  1 file changed, 6 insertions(+)
>
> diff --git a/include/linux/bitmap.h b/include/linux/bitmap.h
> index 4e0f0c8167af..0b3e4452b054 100644
> --- a/include/linux/bitmap.h
> +++ b/include/linux/bitmap.h
> @@ -319,6 +319,9 @@ static __always_inline void bitmap_set(unsigned long *map, unsigned int start,
>  {
>  	if (__builtin_constant_p(nbits) && nbits == 1)
>  		__set_bit(start, map);
> +	else if (__builtin_constant_p(start & 7) && IS_ALIGNED(start, 8) &&
> +		 __builtin_constant_p(nbits & 7) && IS_ALIGNED(nbits, 8))
> +		memset(map + start / 8, 0xff, nbits / 8);
>  	else

Isn't the pointer arithmetic wrong here? I think you need to cast map to
(char*).

> 
>  		__bitmap_set(map, start, nbits);
>  }
> @@ -328,6 +331,9 @@ static __always_inline void bitmap_clear(unsigned long *map, unsigned int start,
>  {
>  	if (__builtin_constant_p(nbits) && nbits == 1)
>  		__clear_bit(start, map);
> +	else if (__builtin_constant_p(start & 7) && IS_ALIGNED(start, 8) &&
> +		 __builtin_constant_p(nbits & 7) && IS_ALIGNED(nbits, 8))
> +		memset(map + start / 8, 0, nbits / 8);
>  	else

Ditto.

Do you have an example of how the generated code changes, both in the
case of actual constants and a case where gcc can see that start and
nbits are byte-aligned without knowing their actual values?