This is the mail archive of the
libc-alpha@sources.redhat.com
mailing list for the glibc project.
Re: Results with glibc-2.2.92
- From: Dan Kegel <dank at kegel dot com>
- To: Jeff Chua <jchua at fedex dot com>
- Cc: Ulrich Drepper <drepper at redhat dot com>, GNU libc <libc-alpha at sources dot redhat dot com>
- Date: Fri, 30 Aug 2002 07:55:25 -0700
- Subject: Re: Results with glibc-2.2.92
- References: <Pine.LNX.4.44.0208301439000.1363-100000@boston.corp.fedex.com>
Jeff Chua wrote:
> > > So, why does the gcc/glib not default to "strip" during installation?
> > > Most people don't bother with these files after install.
> >
> > The extra size doesn't add a bit to the runtime costs. But the debug
> > info is invaluable when you need it. If you then strip the binaries it
> > is too late.
>
> I ran into a problem trying to build a minimum filesystem using
> initrd/ramdisk and realized that the bin and lib are too big to fit into
> initrd ... there seems to be a limitation of <4MB for celeron and <8M for
> pentium to boot, so stripping these files make it easier to fit the
> necessary tools into the limited fs.
I run into this situation, too; what I do is strip all the files
on their way into the init boot image. The original binaries
remain unstripped. That way I can do gdbserver on the
stripped binary with gdb on the unstripped binary.
BTW normal binaries, if stripped with strip -g, are still
bloated; you have to do strip --remove-unneeded. (Is that a
bug in strip?)
Kernel modules should be stripped with -g, because they
don't work if stripped with --remove-unneeded.
- Dan