On 01/16/16 17:54, Daniel Schepler wrote:
On Sat, Jan 16, 2016 at 8:36 AM, Tim Tassonis <[email protected]> wrote:
As "we" initially define our own target as
LFS_TGT=$(uname -m)-lfs-linux-gnu
wouldn't it be nicer to set this target here accordingly, as:
../gcc-5.3.0/configure \
--prefix=/usr \
--target=$LFS_TGT \
--enable-languages=c,c++ \
--disable-multilib \
--disable-bootstrap \
--with-system-zlib
, resulting in a gcc with files containing x86_64-lfs-linux-gnu instead of
x86_64-unknown-linux-gnu ?
Or am I missing something that makes this a bad idea?
First of all, unless you're building a cross-compiler, you shouldn't
be specifying --target; you should use --build=$LFS_TGT instead if you
really want to adjust the full compiler names. (Same goes for
binutils. It would also work to specify all three explicitly:
--build=$LFS_TGT --host=$LFS_TGT --target=$LFS_TGT .)
Thanks, I'll try with --build, then.
Also, according to this from 4.4, LFS_TGT is meant to be just for the
intermediate tools and the final compiler build is intentionally left
to use the default name:
The LFS_TGT variable sets a non-default, but compatible machine
description for use when building our cross compiler and linker and
when cross compiling our temporary toolchain. More information is
contained in Section 5.2, “Toolchain Technical Notes”.
The intention to finally use the default "unknown" vendor however is not
really explained. I assumed, as we are the builder of the whole system,
this would make us the "vendor"?
Kind regards
Tim
--
http://lists.linuxfromscratch.org/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
