Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Thu, 30 Nov 2017 11:08:30 -0500
From: Rich Felker <dalias@...c.org>
To: musl@...ts.openwall.com
Subject: Re: [PATCH] arm: respect both __ARM_ARCH_6KZ__ and
 __ARM_ARCH_6ZK__ macros

On Wed, Nov 29, 2017 at 11:20:48PM -0800, Andre McCurdy wrote:
> On Wed, Nov 29, 2017 at 10:01 PM, Rich Felker <dalias@...c.org> wrote:
> > On Wed, Nov 29, 2017 at 09:57:55PM -0800, Andre McCurdy wrote:
> >> 6KZ is the correct form, 6ZK is a gcc specific historical typo.
> >> Respect both for the widest compatibility with clang and older
> >> versions of gcc.
> >
> > Probably ok, but is it needed?
> 
> As far as I know, __ARM_ARCH_6ZK__ is gcc specific and clang only ever
> defines __ARM_ARCH_6KZ__. Older versions of gcc only define
> __ARM_ARCH_6ZK__ and newer versions of gcc define both
> __ARM_ARCH_6KZ__ and __ARM_ARCH_6ZK__.
> 
> > Ideally we would just use
> > __ARM_ARCH>=N, but old gcc lacked __ARM_ARCH, thus necessitating all
> > the awful cases.
> 
> In atomic_arch.h, replacing the current ARMv6 tests with __ARM_ARCH >=
> 6 would be OK (if we could rely on __ARM_ARCH).
> 
> For pthread_arch.h, there are ARMv6 cores which lack the c13
> coprocessor thread and process ID registers so __ARM_ARCH >= 6
> wouldn't be fine grained enough.

I see. That's the core issue here -- that it's needed to choose the
right (or at least optimal) thread pointer access method for v6 vs
v6kz.

Rich

Powered by blists - more mailing lists

Your e-mail address:

Powered by Openwall GNU/*/Linux - Powered by OpenVZ