Toolchain upgrades

Robert Connolly robert at linuxfromscratch.org
Tue Jul 4 13:02:36 PDT 2006


I don't know I'm losing my mind, or if I entered the twilight zone. This is a 
draft I wrote but never sent.

Also, have any of you ever seen this before:

$ uptime
 16:00:28 up 13 days, 20:51,  0 users,  load average: 0.05, 0.10, 0.08

... "0 users".

robert

On July 3, 2006 09:43 pm, Robert Connolly wrote:
> Oops, sent send without realizing it.
>
> On July 3, 2006 09:41 pm, Robert Connolly wrote:
> > I have attached experimental differences to build HLFS with gcc-4.1.1 and
> > uclibc-20060613. Comments are welcome.
> >
> > gcc-4.1.1 has a demented check for "__stack_chk_fail in target GNU C
> > library". Setting gcc_cv_libc_provides_ssp=yes is pretty much the only
> > way to use uClibc's ssp with gcc-4.1.1; and it doesn't hurt with Glibc
> > either, especially when installing to /tools.
> >
> > I like the idea of tossing in a 'make bootstrap' somewhere, either in
> > chapter 5 or 6. Many host systems may be using gcc-3.2 or gcc-3.3, and a
> > proper bootstrap would help assure the gcc-4.1.1 compiler is compiled
> > correctly. Doing this in chapter 5 is more traditional, but in chapter 6
> > a profiledbootstrap is more convenient.
> >
> > Also, I noticed binutils-2.16.93 (and binutils-2.17.50.0.2) are more
> > sensitive to CFLAGS than before.
> >
> > ----
> >
> > robert



More information about the hlfs-dev mailing list