linux-kernel - RE: static_branch/jump

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: <0a2e658a936a4b7788b23c57ff37119b@AcuMS.aculab.com>
Date:   Sat, 10 Apr 2021 12:44:55 +0000
From:   David Laight <David.Laight@...LAB.COM>
To:     'David Malcolm' <dmalcolm@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>
CC:     Ard Biesheuvel <ardb@...nel.org>,
        "linux-toolchains@...r.kernel.org" <linux-toolchains@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Josh Poimboeuf <jpoimboe@...hat.com>,
        "Jason Baron" <jbaron@...mai.com>,
        "Steven Rostedt (VMware)" <rostedt@...dmis.org>
Subject: RE: static_branch/jump_label vs branch merging

From: David Malcolm
> Sent: 09 April 2021 14:49
...
> With the caveat that my knowledge of GCC's middle-end is mostly about
> implementing warnings, rather than optimization, I did some
> experimentation, with gcc trunk on x86_64 FWIW.
> 
> Given:
> 
> int __attribute__((pure)) foo(void);
> 
> int t(void)
> {
>   int a;
           = 0;
>   if (foo())
>     a++;
>   if (foo())
>     a++;
>   if (foo())
>     a++;
>   return a;
> }
> 
> At -O1 and above this is optimized to a single call to foo, returning 0
> or 3 accordingly.

Interesting horrid idea.
The code generated for the above should be:
	call	foo
	jz	label
So objtool could find the relocation entries for 'foo'
and use that information to replace the call instruction
(and maybe the jz) with a suitable alternate instruction
sequence.

Although that might end up with a game of 'whack-a-mole'
on the perverse instruction sequences the compiler
generates.

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)