diskusage and 'gcc -s'
archaic at comcast.net
Mon Oct 14 20:14:02 PDT 2002
Well, I just finished an LFS-3.3 using the chapter separtation like
LFS-4.0. I wanted to keep track of the highest point in diskusage during
the scripted build. Not knowing a pretty way to do it I just ran:
df -h |grep hda2 >>diskusage && sleep 3
continuously about a billion times. :) It worked, however dirty it is,
and I came up with 437MB as the high point (building the kernel). The
way I build is to use mount --bind and I delete the source prior to
building the next package to keep diskusage down and prevent re-using a
dirty source tree. If I put the tarballs on the LFS partition itself it
adds ~104 MB and if i assume LFS-4.0 is 30% larger, then 104 + 568 =
672. That's quite a savings now that we use export CC='gcc -s' and
perhaps the book should reflect the lower requirements?
Also, if we don't export CXX='g++ -s' then GCC builds a lot of stuff not
stripped. Groff uses CCC=g++ so for that we could also export or just:
make CCC='g++ -s' && make install. IIRC, someone already mentioned the
Bzip install in chapter 5 needed the -s manually added. Also, Bzip2 in
chapter 6 seems to ignore the fact that CC was exported. I got around
that by adding CC= to both the make -f Makefile-libbz2_so and make
commands. I'm rebuilding as we speak with a global export of CC, CXX,
and CCC to see how it goes, along with the manual passing of CC to
bzip2. I'll post the results as soon as I'm finished.
"I hold it, that a little rebellion, now and then, is a good thing, and
as necessary in the political world as storms in the physical."
- Thomas Jefferson, Letter to James Madison, January 30, 1787
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe lfs-dev' in the subject header of the message
More information about the lfs-dev