mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: "Richard B. Johnson" <root@chaos.analogic.com>
To: Jeff Garzik <jgarzik@pobox.com>
Cc: Linus Torvalds <torvalds@transmeta.com>,
	Rusty Trivial Russell <rusty@rustcorp.com.au>,
	linux-kernel@vger.kernel.org
Subject: Re: [TRIVIAL] kstrdup
Date: Fri, 18 Apr 2003 14:00:08 -0400 (EDT)	[thread overview]
Message-ID: <Pine.LNX.4.53.0304181323400.22493@chaos> (raw)
In-Reply-To: <3EA02E55.80103@pobox.com>

On Fri, 18 Apr 2003, Jeff Garzik wrote:

> Linus Torvalds wrote:
> > On Fri, 18 Apr 2003, Jeff Garzik wrote:
> >
> >>You should save the strlen result to a temp var, and then s/strcpy/memcpy/
> >
> >
> > No, you should just not do this. I don't see the point.
>
>
> strcpy has a test for each byte of its contents, and memcpy doesn't.
> Why search 's' for NULL twice?
>
> 	Jeff

Because it doesn't. strcpy() is usually implimented by getting
the string-length, using the same code sequence as strlen(), then
using the same code sequence as memcpy(), but copying the null-byte
as well. The check for the null-byte is done in the length routine.

If you do a memcpy(a, b, strlen(b));, then you are making two
procedure calls and dirtying the cache twice..

A typical Intel procedure, stripped of the push/pops to save
registers is here....

		cld
		movl	SRC(%esp), %edi
		movl	%edi, %esi	# Source = ESI and EDI
		orl	$-1, %ecx
		xorl	%eax,%eax	# Zero
		repnz	scasb		# Look for the null-byte
		negl	%ecx		# ecx = string length
		incl	%ecx		# copy the null
		movl	DST(%esp), %edi	# Destination buffer EDI
		shrl	$1, %ecx	# Make word count
		pushf			# Save result
		shrl	$1, %ecx	# Make longword count
		rep	movsl		# copy longwords
		adcl	%ecx, %ecx	# Possible word count
		rep	movsw		# Copy possible last word
		popf			# Result from first shift
		adcl	%ecx, %ecx	# Possible byte-count
		rep	movsb		# Copy last byte

There are no jumps and the locality of action is good. This could
be improved in the corner cases of long, aligned buffers.

A lot of persons who are unfamiliar with tools other than 'C' think
that strcpy() is made like this:

	while(*dsp++ = *src++)
                   ;

... ahaaa .. Wouderful short portable code! Wrong, awful code, possibly
portable.


.L2:
	movl 8(%ebp),%edx	# Memory reference (destination)
	movl 12(%ebp),%ecx	# Memory reference (source)
	movb (%ecx),%al		# Get a byte
	movb %al,(%edx)		# Put it into destination buffer
	incl 12(%ebp)		# Memory reference (bump the source)
	incl 8(%ebp)		# Memory reference (bump the destination)
	testb %al,%al		# Check for zero on every byte
	jne .L4			# Why a jump to here? (should be jne .L2)
	jmp .L3
	.align 4
.L4:
	jmp .L2			# Now we go back to the beginning.
	.align 4
.L3:
.L1:
	movl %ebp,%esp
	popl %ebp
	ret


Now; this code is 'shorter', but it is very slow compared to the
smarter strcpy() as previously shown.


Cheers,
Dick Johnson
Penguin : Linux version 2.4.20 on an i686 machine (797.90 BogoMips).
Why is the government concerned about the lunatic fringe? Think about it.


  parent reply	other threads:[~2003-04-18 17:45 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2003-04-18  7:58 Rusty Trivial Russell
2003-04-18  8:10 ` Jeff Garzik
2003-04-18 16:20   ` Linus Torvalds
2003-04-18 16:56     ` Jeff Garzik
2003-04-18 18:00       ` Linus Torvalds
2003-04-18 18:44         ` Jeff Garzik
2003-04-19  4:52         ` Rusty Russell
2003-04-18 18:00       ` Richard B. Johnson [this message]
2003-04-18 18:40         ` Jeff Garzik
2003-04-18 19:24           ` H. Peter Anvin
2003-04-18 23:37             ` Alan Cox
2003-04-18 19:29           ` Richard B. Johnson
2003-04-19 11:45             ` Kai Henningsen
2003-04-19 20:16             ` Ruth Ivimey-Cook
2003-04-19  4:14   ` Rusty Russell
2003-04-19  4:48     ` Jeff Garzik
2003-04-19  8:28       ` Rusty Russell
2003-04-19 12:27       ` Alan Cox
2003-04-19 14:44         ` Bernd Eckenfels
2003-04-20  8:05         ` Rusty Russell

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=Pine.LNX.4.53.0304181323400.22493@chaos \
    --to=root@chaos.analogic.com \
    --cc=jgarzik@pobox.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=rusty@rustcorp.com.au \
    --cc=torvalds@transmeta.com \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®