linux-kernel - RE: First kernel patch (optimization)

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: <063D6719AE5E284EB5DD2968C1650D6D1CB989FD@AcuExch.aculab.com>
Date:	Thu, 17 Sep 2015 08:45:57 +0000
From:	David Laight <David.Laight@...LAB.COM>
To:	'Jaime Arrocha' <jarr@...ercoder.com>,
	'Austin S Hemmelgarn' <ahferroin7@...il.com>,
	Steve Calfee <stevecalfee@...il.com>,
	Eric Curtin <ericcurtin17@...il.com>
CC:	Valentina Manea <valentina.manea.m@...il.com>,
	"shuah.kh@...sung.com" <shuah.kh@...sung.com>,
	USB list <linux-usb@...r.kernel.org>,
	"Kernel development list" <linux-kernel@...r.kernel.org>
Subject: RE: First kernel patch (optimization)

From: Jaime Arrocha
> Sent: 17 September 2015 02:50
..
> One interesting observation I found was that in O0 and O2, it does make
> a call to strlen while in O1 it calculates
> the length of the string using:
> 

You want an 'xor %rcx,%rcx' here.
> repnz scas    %es:(%rdi),%al
> not                %rcx
> sub               $0x2,%rcx
> 
> Why does it do that? Is the code above faster? If yes, why not do it in
> O2 too?

Because 'repnz scasb' is slow, especially on some cpu types.
It may win for -Os on 32 bit systems.
Pentium 4 netburst have about 40 clocks setup for all the 'rep' instructions,
later cpus are better but you might still be talking double figures.
On 64 bit cpu there are much faster ways of detecting a zero byte in a
64 bit word by using shifts and masks - so the function call can be a win.

	David