DMARC-Filter: OpenDMARC Filter v1.4.2 delorie.com 61S2VRXJ1990096
Authentication-Results: delorie.com; dmarc=pass (p=none dis=none) header.from=cygwin.com
Authentication-Results: delorie.com; spf=pass smtp.mailfrom=cygwin.com
DKIM-Filter: OpenDKIM Filter v2.11.0 delorie.com 61S2VRXJ1990096
Authentication-Results: delorie.com;
	dkim=pass (1024-bit key, unprotected) header.d=cygwin.com header.i=@cygwin.com header.a=rsa-sha256 header.s=default header.b=qAMzxjuN
X-Recipient: archive-cygwin AT delorie DOT com
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 4378E4B9DB64
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=cygwin.com;
	s=default; t=1772245885;
	bh=gdN3TnC0xYAkUo9JVFdVTAhW+NBzgfLNlGyHHjz9kVw=;
	h=Date:To:Cc:In-Reply-To:References:Subject:List-Id:
	 List-Unsubscribe:List-Archive:List-Post:List-Help:List-Subscribe:
	 From:Reply-To:From;
	b=qAMzxjuNXJc0EI098N7ou0YynmEzOoTNNheQgUdp3dobKT3rXLSJFNY9EupSbpVlH
	 JgEpfEs0EVb52Cj3CSvejUpTjxxqYC9ZNc6l0DQvLEfoxXLnXyiAB9dOgnENiKuxKp
	 AfjScNNJre2Q6RgeShUidH0VOjxBk+uVofIMoiXI=
X-Original-To: cygwin AT cygwin DOT com
Delivered-To: cygwin AT cygwin DOT com
DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 0C79D4BA23C2
ARC-Filter: OpenARC Filter v1.0.0 sourceware.org 0C79D4BA23C2
ARC-Seal: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1772245836; cv=none;
 b=XRHAagK+P4sgHGvMx+gSOtOEFN6cv8d8GjmXxY3sCBVPW4+AvKIbY8YkqKl/KUg1khUO1tf6DSQqy2aftyezvZtLEDygo4ib/NV/03+ivN8VfBCHlOPnFVjacubj4kpck/bqYTiZZRYzmyCh4T5jZkDm66XCi5uzPadUlAaCoBQ=
ARC-Message-Signature: i=1; a=rsa-sha256; d=sourceware.org; s=key;
 t=1772245836; c=relaxed/simple;
 bh=amXZ3Xkn5iIs0mP3mph98AMeMtqVMklBMOJLZK4dQ88=;
 h=DKIM-Signature:Date:From:To:Message-ID:Subject:MIME-Version;
 b=pjmrUi04TwWSjJypDrW65IRqJ38y7EdjUk80U66uUljejkkWSzv/9AuRCpwXczyH6orprBB4N4FUjsKDRY7dmEpXkaDxocD/ZQzn/7HuT9V/tQ+B0EEBGqGgk1B+f5dlEokEn/eEl0MOV90xgOvVucN61ipAHcNLCDJnizkmQ24=
ARC-Authentication-Results: i=1; server2.sourceware.org
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 0C79D4BA23C2
Date: Fri, 27 Feb 2026 18:30:30 -0800 (PST)
To: General Cygwin discussions and problem reports <cygwin AT cygwin DOT com>
Cc: Brian Inglis <Brian DOT Inglis AT SystematicSW DOT ab DOT ca>
Message-ID: <103536920.1558501.1772245830440@connect.xfinity.com>
In-Reply-To: <45c133f7-8285-4cb3-9701-2642cb76ab37@SystematicSW.ab.ca>
References: <547312365 DOT 1464244 DOT 1771958282029 AT connect DOT xfinity DOT com>
 <aZ7PrbisVR1R4A7v AT dimstar DOT local DOT net>
 <1670201592 DOT 1489273 DOT 1772043520008 AT connect DOT xfinity DOT com>
 <e91d8b5b-2690-4271-aa74-e6226440e33d AT SystematicSW DOT ab DOT ca>
 <1044918836 DOT 1507810 DOT 1772086967212 AT connect DOT xfinity DOT com>
 <1579472684 DOT 1508349 DOT 1772092747339 AT connect DOT xfinity DOT com>
 <aaABFf5iEowV1l7I AT xps13>
 <1148572549 DOT 1808180 DOT 1772097444036 AT mail DOT yahoo DOT com>
 <1901597260 DOT 1508573 DOT 1772100378936 AT connect DOT xfinity DOT com>
 <0C965DD0-856E-41FF-B5A4-15E472292A32 AT unified-streaming DOT com>
 <483908609 DOT 1508714 DOT 1772103775739 AT connect DOT xfinity DOT com>
 <2346fd41-2500-0db6-5849-6788174b5a1d AT cs DOT umass DOT edu>
 <1462848037 DOT 1521935 DOT 1772136952077 AT connect DOT xfinity DOT com>
 <399745a1-429a-ebb4-0f67-c32f6282caa6 AT cs DOT umass DOT edu>
 <1093316506 DOT 1533154 DOT 1772157883568 AT connect DOT xfinity DOT com>
 <3e0de899-a7dd-8fea-7743-10e6b05cc6b6 AT cs DOT umass DOT edu>
 <1990836634 DOT 1545853 DOT 1772216419837 AT connect DOT xfinity DOT com>
 <45c133f7-8285-4cb3-9701-2642cb76ab37 AT SystematicSW DOT ab DOT ca>
Subject: Re: Memmove causing program crashes, giving SIGTRAP in GDB(?)
MIME-Version: 1.0
X-Priority: 3
Importance: Normal
X-Mailer: Open-Xchange Mailer v7.10.6-Rev83
X-Originating-IP: ::ffff:50.47.202.14
X-Originating-Port: 36348
X-Originating-Client: open-xchange-appsuite
X-CMAE-Envelope: MS4xfI/UU4H+SaPHknYXzHUK2BoZF4RB0MXCfODbxdfEs9f3ctJvJ8W/tJHdVD8KU9Eo9f0uyz1KUsQnU2avaNSnsFpdxb3apBMe2q+Fbq4iBY4QVhcX1o4o
 4H+Szh9uoTg2x9ljlp+SmZVkfRyQG92g58IFE8NcXJIAFhB5d4mfONGshSEpo/CKbFUkTVgXOWoleg03yEKI5Eob2biwlTf7k47WC+gCRPYuw3k+0pg8P/g2
 CzDPegWTIm97vJ3e9NxRASkcm/TzCw1xoP7zb9j21Gc=
X-BeenThere: cygwin AT cygwin DOT com
X-Mailman-Version: 2.1.30
Precedence: list
List-Id: General Cygwin discussions and problem reports <cygwin.cygwin.com>
List-Unsubscribe: <https://cygwin.com/mailman/options/cygwin>,
 <mailto:cygwin-request AT cygwin DOT com?subject=unsubscribe>
List-Archive: <https://cygwin.com/pipermail/cygwin/>
List-Post: <mailto:cygwin AT cygwin DOT com>
List-Help: <mailto:cygwin-request AT cygwin DOT com?subject=help>
List-Subscribe: <https://cygwin.com/mailman/listinfo/cygwin>,
 <mailto:cygwin-request AT cygwin DOT com?subject=subscribe>
From: KENNON J CONRAD via Cygwin <cygwin AT cygwin DOT com>
Reply-To: KENNON J CONRAD <kennonconrad AT comcast DOT net>
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Errors-To: cygwin-bounces~archive-cygwin=delorie DOT com AT cygwin DOT com
Sender: "Cygwin" <cygwin-bounces~archive-cygwin=delorie DOT com AT cygwin DOT com>

Hi Brian,

I just wanted to add that the stash and store idea you suggest that is also used in memmove has a very nice impact
on the assembly code.

With the old code that does this for the last 0 to 7 words:
        while (candidate_ptr > score_ptr) {
          *candidate_ptr = *(candidate_ptr - 1);
          candidate_ptr--;
        }

the assembly code shows this from the point where the move starts:
.L24:
	movdqu	-16(%rax), %xmm1
	subq	$16, %rax
	movups	%xmm1, 2(%rax)
	cmpq	%rdx, %rax
	jnb	.L24
	movq	%r10, %rax
	subq	%r9, %rax
	subq	$16, %rax
	notq	%rax
	andq	$-16, %rax
	addq	%r10, %rax
	cmpq	%rax, %r9
	jnb	.L28
	movq	%rax, %rcx
	movq	%rax, %rdx
	movq	%r9, 48(%rsp)
	subq	%r9, %rcx
	subq	$1, %rcx
	shrq	%rcx
	leaq	2(%rcx,%rcx), %r8
	negq	%rcx
	subq	%r8, %rdx
	leaq	(%rax,%rcx,2), %rcx
	call	memmove
	movq	48(%rsp), %r9
	jmp	.L28

But with stash and store:
        *(uint64_t *)&candidates_index[new_score_rank + 1] = first_four;
        *(uint64_t *)&candidates_index[new_score_rank + 5] = next_four;

the assembly code from the point where the move start is this:
.L24:
	movdqu	-16(%r9), %xmm1
	subq	$16, %r9
	movups	%xmm1, 2(%r9)
	cmpq	%rax, %r9
	jnb	.L24
	movups	%xmm0, 2(%rdi,%rdx)
	jmp	.L26

There are a couple of extra assembly instructions to stash into xmm0 before the move, but this is a big reduction in
assembly code size for the backward memory move.  Not as fast as memmove if the DF wasn't getting corrupted, but much
better than the old code plus it completely avoids the risk of DF corruption during rep movsq in memmove for backward
move sizes >= 8!  I like it because there is no need to worry about whether rep movsb or rep movsw could also be
vulnerable to DF corruption.

Best Regards,

Kennon

> On 02/27/2026 11:49 AM PST Brian Inglis via Cygwin <cygwin AT cygwin DOT com> wrote:
> 
>  
> Hi Kennon,
> 
> Some perf reports and analysis imply that backward moves (with overlap?) are no 
> faster than straight rep movsb on some CPUs, so it may be better to just 
> simplify to that, unless you want to stash the final element(s) to be moved out 
> of the way in register(s), and use multiple registers in unrolled wide moves for 
> the aligned portion?
>

-- 
Problem reports:      https://cygwin.com/problems.html
FAQ:                  https://cygwin.com/faq/
Documentation:        https://cygwin.com/docs.html
Unsubscribe info:     https://cygwin.com/ml/#unsubscribe-simple