This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug bootstrap/47016] New: bootstrap on darwin needs much more disk space than expected to complete
- From: "Denis.Excoffier at airbus dot com" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Sun, 19 Dec 2010 20:31:35 +0000
- Subject: [Bug bootstrap/47016] New: bootstrap on darwin needs much more disk space than expected to complete
- Auto-submitted: auto-generated
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47016
Summary: bootstrap on darwin needs much more disk space than
expected to complete
Product: gcc
Version: 4.5.2
Status: UNCONFIRMED
Severity: minor
Priority: P3
Component: bootstrap
AssignedTo: unassigned@gcc.gnu.org
ReportedBy: Denis.Excoffier@airbus.com
Created attachment 22825
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22825
created during stage2-gcc, produces a huge temporary during compare-debug
I have bootstraped GCC 4.5.2 on darwin (ppc 9.8.0) successfully, with all
default options, in
particular "bootstrap-debug".
Nevertheless i noticed that the disk space used is 2Gb more than expected. This
is due to the
- contrib/compare-debug script
- that uses, in case of Darwin: ld -S -r -no_uuid xxx.o -o xxx.o.stripped
(instead of a simple strip)
- which produces:
- for the host-darwin.o created during stage2 (with -gtoggle)
- and also for the host-darwin.o created during stage3 (without -gtoggle)
a huge host-darwin.o.stripped containing 1Gb with many many zeroes
(all the other xxx.o in stage2-gcc or stage3-gcc are ok)
As a coincidence, both host-darwin.o.stripped are exactly equal, therefore no
failure in comparison of stage2 and stage3, and consequently no failure in
bootstrap. Of
course if you don't have such extra disk space, something will surely fail. I
could suggest, in case this is not a local problem occurring only to my
configuration, to add gcc/host-darwin.o in compare_exclusions (in the main
./configure).
Attached stage2-gcc/host-darwin.o. I can also add stage2/host-darwin.o.stripped
(ie `od -c`on it) if
needed.
I cannot say whether the host-darwin.o created for GCC 4.5.1 had the same
problem.
I suppose this could be also reported to Apple, but in case someone meets that
very same
problem, a bypass is given: increase free disk space.