robert at linuxfromscratch.org
Wed Nov 21 17:05:04 PST 2007
The march=i486 is for Glibc. -march/mtune=i586, and -march/mtune=i686 are all
prone to produce extra test suite failures. I'm not sure if it's a problem
with the test suite, or the library. I don't think Glibc will have more or
less features based on the march=.
There was recent discussion on the lfs-dev mailing list about
the --enable-kernel option. It sounds like if you use --enable-kernel=2.6.0
then all newer features are still enabled, but there's a libc version if the
running kernel doesn't have the new feature. If you
use --enable-kernel=2.6.23 then the libc versions of supported kernel
features are not added to the library, and there would be failures if you
ever run an older kernel. I haven't checked this out, but if this is correct
then I'm partial to keep using --enable-kernel=2.6.0 so the library is more
robust, at the expense of some performance. Another option is to
remove --enable-kernel and use the default 2.0.
On Wednesday November 21 2007 05:43:25 pm marty at goodoldmarty.com wrote:
> building the initial C library in embryo build,
> echo "CFLAGS += -march=i486 -mtune=i486" > configparms
> This works fine to get things going, however will this library treat us
> like an antique at this point, dropping features we may want on a more
> advanced CPU, or does it matter?
> Marty B
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 189 bytes
Desc: not available
More information about the hlfs-dev