On 10/1/15 7:20 AM, Pierre Labastie wrote:
I am not sure, but it seems that the recent thread about this problem
pointed to cc1plus (the c++ front end in libexec),
see
http://lists.linuxfromscratch.org/pipermail/blfs-support/2015-September/077082.html
It may be that GMP is at fault, but if so, it has been statically
linked into cc1plus. So, after recompiling gmp, it would be necessary
to recompile gcc.
Pierre
First, thanks for helping out a newb. I appreciate your patience and
substantive responses. Second, I truly appreciate your and Douglas
Reno's responses. Both were correct. The thread
http://lists.linuxfromscratch.org/pipermail/blfs-support/2015-September/077082.html
covers the issue in detail, but I will summarize for posterity:
There appears to be a dependency on the Virtualbox Host's CPU in the GMP
code. Mike Shell points to the correct test for the issue in
http://lists.linuxfromscratch.org/pipermail/blfs-support/2015-September/077091.html,
this is the hack that I used:
echo '#include <limits>' > dummy.c && g++ -c dummy.c
Which, on a broken toolchain, results in:
In file included from dummy.c:1:0:
/usr/include/c++/5.2.0/limits:1598:7: internal compiler error: Illegal
instruction
min() _GLIBCXX_USE_NOEXCEPT { return __FLT_MIN__; }
^
0xa527df crash_signal
../../gcc-5.2.0/gcc/toplev.c:383
0x7fa191d9e64f ???
/sources/7.8-rc1/glibc-2.22/signal/../sysdeps/unix/sysv/linux/x86_64/sigaction.c:0
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <http://gcc.gnu.org/bugs.html> for instructions.
The solution that I used was pointed out by William Harrington in
http://lists.linuxfromscratch.org/pipermail/blfs-support/2015-September/077094.html.
In order to apply the solution, I reverted my mac-arch vm to just after
6.16 (pre GCC, ideally, you would go back to 6.14, but I only had
snapshots for just after 6.10 where the tool chain is adjusted and just
after 6.16 before GCC). This brings the LFS machine to a state, just
prior to the big honking build of GCC. I then redid the GMP build
described in 6.14, but as follows (ala William Harrington's solution):
== 6.14. GMP-6.0.0a
tar xvf gmp-6.0.0a.tar.xz
cd gmp-6.0.0
# added the next two steps to avoid cpu issues while moving between
# the mac host and the dell host - basically using a generic configuration
# rather than an optimized version... or somesuch
cp -v config{fsf,}.guess
# ‘configfsf.guess’ -> ‘config.guess’
cp -v config{fsf,}.sub
# ‘configfsf.sub’ -> ‘config.sub’
./configure --prefix=/usr \
--enable-cxx \
--disable-static \
--docdir=/usr/share/doc/gmp-6.0.0a
make && make html
make check 2>&1 | tee gmp-check-log
awk '/tests passed/{total+=$2} ; END{print total}' gmp-check-log
# should return 188
make install && make install-html
cd ..
rm -fr ./gmp-6.0.0
==
I then rebuilt GCC following the steps in the book and exported the VM
from my mac host to the dell host. I brought the dell VM up and retried
the offending code - success!
Bruce noted in
http://lists.linuxfromscratch.org/pipermail/blfs-support/2015-September/077108.html
that he would be updating the book, but I wanted to summarize as this
was a related, but not identical thread.
Thanks all.
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
Do not top post on this list.
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
http://en.wikipedia.org/wiki/Posting_style
