linux-kernel - Re: Linux 5.19-rc8

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]

Date:   Mon, 25 Jul 2022 13:40:58 -0700
From:   Linus Torvalds <torvalds@...ux-foundation.org>
To:     Yury Norov <yury.norov@...il.com>
Cc:     Guenter Roeck <linux@...ck-us.net>,
        Dennis Zhou <dennis@...nel.org>,
        Russell King - ARM Linux <linux@...linux.org.uk>,
        Catalin Marinas <catalin.marinas@....com>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: Linux 5.19-rc8

On Mon, Jul 25, 2022 at 1:35 PM Yury Norov <yury.norov@...il.com> wrote:
>
> Here we have 2 ugly options - having pairs of almost identical
> functions, or passing dummy variables. I decided that copy-pasting is
> worse than abusing branch predictor.

The thing is, we have solutions for this - just use a single inline
function, and make it *see* the constant.

That way the compiler can DTRT, either generating two copies (with the
constant elided) or depending on the branch predictor.

That's particularly true in a case like this, when there is only *one*
case for the normal situation (ie little-endian). Because let's face
it, big-endian is completely dead and legacy model.

              Linus