All of lore.kernel.org
 help / color / mirror / Atom feed
From: David Laight <david.laight.linux@gmail.com>
To: David Desobry <david.desobry@formalgen.com>
Cc: tglx@kernel.org, mingo@redhat.com, bp@alien8.de,
	dave.hansen@linux.intel.com, x86@kernel.org, hpa@zytor.com,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH v4] x86/lib: Optimize num_digits() and fix INT_MIN overflow
Date: Wed, 21 Jan 2026 11:36:43 +0000	[thread overview]
Message-ID: <20260121113643.22718a3d@pumpkin> (raw)
In-Reply-To: <20260121093911.27042-1-david.desobry@formalgen.com>

On Wed, 21 Jan 2026 10:39:11 +0100
David Desobry <david.desobry@formalgen.com> wrote:

> The current implementation of num_digits() uses a loop with repeated
> multiplication, which is inefficient. Furthermore, it takes a signed
> integer which leads to undefined behavior when negating INT_MIN.
> 
> As this function is only used for unsigned magnitudes in the kernel code,
> convert the interface to take an unsigned int. This naturally resolves
> the INT_MIN overflow issue.
> 
> Replace the loop with a branchless sequence using inline assembly. By
> using the 'sbb' instruction against the carry flag after comparisons,
> we eliminate all conditional branches. This provides constant-time
> performance and avoids CPU branch misprediction penalties.

Don't do this version, it isn't worth the effort.

	David

> 
> Update the function comment to reflect that signs are no longer handled.
> 
> Signed-off-by: David Desobry <david.desobry@formalgen.com>
> ---
>  v4:
>  - Switched to branchless inline assembly. 
>  - Changed function signature to unsigned int.
>  - Updated prototype in arch/x86/include/asm/misc.h. 
> 
>  arch/x86/include/asm/misc.h |  2 +-
>  arch/x86/lib/misc.c         | 28 +++++++++++++---------------
>  2 files changed, 14 insertions(+), 16 deletions(-)
> 
> diff --git a/arch/x86/include/asm/misc.h b/arch/x86/include/asm/misc.h
> index bb049cca3729..48b6bd7c08b9 100644
> --- a/arch/x86/include/asm/misc.h
> +++ b/arch/x86/include/asm/misc.h
> @@ -2,6 +2,6 @@
>  #ifndef _ASM_X86_MISC_H
>  #define _ASM_X86_MISC_H
>  
> -int num_digits(int val);
> +int num_digits(unsigned int val);
>  
>  #endif /* _ASM_X86_MISC_H */
> diff --git a/arch/x86/lib/misc.c b/arch/x86/lib/misc.c
> index 40b81c338ae5..9623795b059f 100644
> --- a/arch/x86/lib/misc.c
> +++ b/arch/x86/lib/misc.c
> @@ -2,23 +2,21 @@
>  #include <asm/misc.h>
>  
>  /*
> - * Count the digits of @val including a possible sign.
> - *
> - * (Typed on and submitted from hpa's mobile phone.)
> + * Count the decimal digits of an unsigned integer.
>   */
> -int num_digits(int val)
> +int num_digits(unsigned int x)
>  {
> -	long long m = 10;
> -	int d = 1;
> +	int n = 0;
>  
> -	if (val < 0) {
> -		d++;
> -		val = -val;
> -	}
> +	asm("cmp %2,%1; sbb $-2,%0" : "+r" (n) : "r" (x), "g" (10));
> +	asm("cmp %2,%1; sbb $-1,%0" : "+r" (n) : "r" (x), "g" (100));
> +	asm("cmp %2,%1; sbb $-1,%0" : "+r" (n) : "r" (x), "g" (1000));
> +	asm("cmp %2,%1; sbb $-1,%0" : "+r" (n) : "r" (x), "g" (10000));
> +	asm("cmp %2,%1; sbb $-1,%0" : "+r" (n) : "r" (x), "g" (100000));
> +	asm("cmp %2,%1; sbb $-1,%0" : "+r" (n) : "r" (x), "g" (1000000));
> +	asm("cmp %2,%1; sbb $-1,%0" : "+r" (n) : "r" (x), "g" (10000000));
> +	asm("cmp %2,%1; sbb $-1,%0" : "+r" (n) : "r" (x), "g" (100000000));
> +	asm("cmp %2,%1; sbb $-1,%0" : "+r" (n) : "r" (x), "g" (1000000000));
>  
> -	while (val >= m) {
> -		m *= 10;
> -		d++;
> -	}
> -	return d;
> +	return n;
>  }


  reply	other threads:[~2026-01-21 11:36 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-01-20 17:47 [PATCH v2] x86/lib: Optimize num_digits() and fix INT_MIN overflow David Desobry
2026-01-20 18:03 ` H. Peter Anvin
2026-01-20 21:46 ` David Laight
2026-01-20 23:19   ` [PATCH v3] " David Desobry
2026-01-20 23:32   ` [PATCH v2] " David Desobry
2026-01-20 23:49     ` H. Peter Anvin
2026-01-21  0:04       ` David Desobry
2026-01-21  0:17         ` H. Peter Anvin
2026-01-21  9:39           ` [PATCH v4] " David Desobry
2026-01-21 11:36             ` David Laight [this message]
2026-01-21 15:46               ` H. Peter Anvin
2026-01-21  9:54         ` [PATCH v2] " David Laight
2026-01-21 10:32           ` [PATCH v5] x86/lib: Rename num_digits() to num_digits_u32() and optimize David Desobry
2026-01-21 10:40           ` [PATCH v2] x86/lib: Optimize num_digits() and fix INT_MIN overflow David Desobry
2026-01-21 11:51       ` Maciej W. Rozycki
2026-01-23  7:06         ` H. Peter Anvin
2026-01-23 10:55           ` Maciej W. Rozycki

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260121113643.22718a3d@pumpkin \
    --to=david.laight.linux@gmail.com \
    --cc=bp@alien8.de \
    --cc=dave.hansen@linux.intel.com \
    --cc=david.desobry@formalgen.com \
    --cc=hpa@zytor.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=tglx@kernel.org \
    --cc=x86@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.