linux-kernel - Re: [PATCH 13/15] arm64: fix mrs_s/msr

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAKwvOd=zNQqyta42hm86rXquCd67OoO=azT7NWMF94=6xVS_4g@mail.gmail.com>
Date:   Fri, 3 Nov 2017 11:15:55 -0700
From:   Nick Desaulniers <ndesaulniers@...gle.com>
To:     Mark Rutland <mark.rutland@....com>
Cc:     Sami Tolvanen <samitolvanen@...gle.com>,
        Matthias Kaehlcke <mka@...omium.org>,
        Greg Hackmann <ghackmann@...gle.com>,
        LKML <linux-kernel@...r.kernel.org>,
        linux-arm-kernel@...ts.infradead.org,
        Kees Cook <keescook@...omium.org>
Subject: Re: [PATCH 13/15] arm64: fix mrs_s/msr_s macros for clang LTO

On Fri, Nov 3, 2017 at 11:06 AM, Mark Rutland <mark.rutland@....com> wrote:
> On Fri, Nov 03, 2017 at 10:53:08AM -0700, Nick Desaulniers wrote:
>> These mrs_s and msr_s macros in particular were preventing us from
>> linking arm64 with Clang's integrated assembler, regardless of LTO.
>> Those macros ran into: https://bugs.llvm.org/show_bug.cgi?id=19749.
>> So while I appreciate how clever they are, they prevent us from
>> assembling with Clang so I would like to see them go.
>
> They're necessary to work with some currently-supported toolchains
> (which don't support the s*_*_c*_c*_* syntax), so they're not going to
> go completely.
>
> If you could suggest something that clang might prefer, which doesn't
> adversely affect GCC, I'm all ears.

I wasn't clear in my point; I recognize what they do and agree that
they are needed.

More specifically, my problem is the use of .macro assembly
directives, since Clang "considers it a feature" not to let you invoke
a .macro defined in one inline asm block from another asm inline
block.

So I would like to see *the use of .macro in separate inline asm
blocks within mrs_s and msr_s c-preprocessor macros go.*