On 04/24/2014 09:38 PM, Bruce Dubbs wrote: > Pierre Labastie wrote: >> Le 24/04/2014 17:28, Frans de Boer a écrit : >>> Ok, followed the advises from ticket #3552, now binutils chapter 6 >>> reports failures: >>> >>> Running /sources-bss/binutils-2.24/ld/testsuite/ld-plugin/lto.exp ... >>> FAIL: PR ld/12758 >>> FAIL: PR ld/12760 >>> FAIL: LTO 3 symbol >>> FAIL: PR ld/13183 >>> FAIL: LTO 3a >>> FAIL: LTO 11 >>> Running /sources-bss/binutils-2.24/ld/testsuite/ld-plugin/plugin.exp ... >>> >>> Concerning LTO, thus induced by gcc-4.9.0. >>> Chapter 5 is completed without any errors, added --disable-werror to the >>> binutils configure...Seems that others having no problem, so what could >>> be wrong? >>> >>> Frans. >> I have exactly the same failures. > > Looking at a full build, I have: > > 077-binutils-2.24:FAIL: PR ld/12758 > 077-binutils-2.24:FAIL: PR ld/12760 > 077-binutils-2.24:FAIL: LTO 3 symbol > 077-binutils-2.24:FAIL: PR ld/13183 > 077-binutils-2.24:FAIL: LTO 3a > 077-binutils-2.24:FAIL: LTO 11
These are new to me. Using gcc-4.8.2 did not yield these results. > > 093-coreutils-8.22:FAIL: tests/misc/nohup.sh > 093-coreutils-8.22:# FAIL: 1 > 093-coreutils-8.22:FAIL: tests/misc/nohup > 093-coreutils-8.22:# FAIL: 1 These are older and can be avoided by echo "exit 0" > tests/misc/nohub.sh > > 106-perl-5.18.2:FAILED at test 104 > 106-perl-5.18.2:FAILED at test 84 > These are also new to me. > 131-systemd-212:FAIL: test-strv > 131-systemd-212:FAIL: test-bus-creds > 131-systemd-212:FAIL: test-journal-flush > 131-systemd-212:# FAIL: 3 > I am not sure of the above. > 133-util-linux-2.24.1: last: last ipv6 ... FAILED (last/ipv6) > 133-util-linux-2.24.1: last: last ... FAILED (last/last) > 133-util-linux-2.24.1: 2 tests of 127 FAILED > Errors are known as well as the reason too. > -- Bruce > I use bash scripts with set +h and set -e, so any error is terminating execution. Some scripts have "set +e; make check/tests ....; set -e" to catch known errors with no known (grave) severity. So new errors or lack of errors are not being seen, unless I run the scripts manually and check every output. It's clear that gcc-4.9.0 does introduce some new failures in tests. Maybe because the maintainers of those other packages are behind? It seems that new software - or software depended on new software - should be tested manually, and if all is working out, enable the "set +e" to generate the whole TC and/or BSS part in one go. Frans. -- http://linuxfromscratch.org/mailman/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/lfs/faq.html Unsubscribe: See the above information page