Re: uClibc vs. eglibc size on ARM

[Johannes Stezenbach <js at sig21 dot net>]

On Mon, Oct 22, 2012 at 12:09:43PM -0700, Khem Raj wrote:
> On Mon, Oct 22, 2012 at 11:18 AM, Johannes Stezenbach <js@sig21.net> wrote:
> > On Mon, Oct 22, 2012 at 09:36:24AM -0700, Khem Raj wrote:
> >> On Mon, Oct 22, 2012 at 6:22 AM, Johannes Stezenbach <js@sig21.net> wrote:
> >> >
> >> > The minimal eglibc config for eglibc-1.16 (libm needed by gcc,
> >> > the other options to un-break eglibc build):
> >> >
> >> > OPTION_EGLIBC_INET = y
> >> > OPTION_EGLIBC_LIBM = y
> >> > OPTION_EGLIBC_NSSWITCH = y
> >> > OPTION_EGLIBC_RTLD_DEBUG = y
> >> > OPTION_POSIX_C_LANG_WIDE_CHAR = y
> >> > OPTION_POSIX_WIDE_CHAR_DEVICE_IO = y
> >>
> >> check option-groups.config file in eglibc's top build dir. I wonder if
> >> others options were turned of or not.
> >
> > I started with all options off and then enabled them one by one
> > until it built, looking at the failures to figure out which
> > option might be responsible.   OPTION_EGLIBC_RTLD_DEBUG was
> > the first one, which depends on OPTION_POSIX_WIDE_CHAR_DEVICE_IO
> > which in turn depends on OPTION_POSIX_C_LANG_WIDE_CHAR.
> > OPTION_EGLIBC_NSSWITCH and OPTION_EGLIBC_INET were next.
> > OPTION_EGLIBC_LIBM is needed by gcc (eglibc itself builz
> > without it).
> >
> 
> and it did not change size of libc compared to when you have all options
> on ? is that what you observed ?

Apprently my first mail was too long to read...
It contained numbers for full eglibc vs. minimal eglibc vs. uClibc,
for public record because it would have saved me from doing
the experiments myself if I had been able to find those numbers on the net.


Johannes

--
For unsubscribe information see http://sourceware.org/lists.html#faq