Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 19 Mar 2013 18:08:50 -0400
From:      George Neville-Neil <gnn@neville-neil.com>
To:        Warner Losh <imp@bsdimp.com>
Cc:        freebsd-arm@freebsd.org, freebsd-toolchain@freebsd.org, Andrew Turner <andrew@fubar.geek.nz>
Subject:   Re: RFC: Make clang the default compiler on ARM
Message-ID:  <471C5F38-2401-4308-871F-32D6A395D209@neville-neil.com>
In-Reply-To: <773A0A0D-4208-4F5B-AA73-416512709A4A@bsdimp.com>
References:  <20130318220747.29d3fd29@bender> <773A0A0D-4208-4F5B-AA73-416512709A4A@bsdimp.com>

next in thread | previous in thread | raw e-mail | index | archive | help

On Mar 18, 2013, at 11:33 , Warner Losh <imp@bsdimp.com> wrote:

>=20
> On Mar 18, 2013, at 3:07 AM, Andrew Turner wrote:
>=20
>> I would like to make clang the default compiler on ARM using the =
patch
>> at [1]. This only affects little-endian ARM as there is no support =
for
>> big-endian ARM in clang.
>>=20
>> This will help me with my work to update the FreeBSD ARM ABI as I am
>> planning on moving to a hard-float ABI on armv6 as all SoCs we
>> are likely to support have at least a VFPv3-D16 floating point
>> coprocessor. The version of gcc we have in our tree does not support
>> the hard float ABI with VFP and support was added after gcc moved to
>> the GPLv3 so we are usable to make use of it.
>>=20
>> The only issue I know about is an incompatibility between gcc and =
clang
>> with ARM EABI however this has been fixed upstream and I plan on
>> bringing it into our tree before making clang the default. If anyone
>> knows of any other issues please let me know now so I can look into
>> them.
>>=20
>> If nobody brings up any issues or has any objections to me doing this
>> I will make clang the default earlier than 25 March (UTC+0).
>>=20
>> Andrew
>>=20
>> [1] http://people.freebsd.org/~andrew/clang/arm_clang_default.diff
>=20
> This scares me, but we're at a great time in our schedule for it.
>=20

I think we ought to do this, in particular since LLVM is the future of =
compiler tech.

Sooner we break it the sooner we can fix it :-)

Best,
George





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?471C5F38-2401-4308-871F-32D6A395D209>