diff --git a/sdk_container/src/third_party/portage-stable/eclass/toolchain.eclass b/sdk_container/src/third_party/portage-stable/eclass/toolchain.eclass index 12a1ec2413..274789559d 100644 --- a/sdk_container/src/third_party/portage-stable/eclass/toolchain.eclass +++ b/sdk_container/src/third_party/portage-stable/eclass/toolchain.eclass @@ -1596,7 +1596,7 @@ toolchain_src_configure() { # If they've explicitly opt-ed in, do hardfloat, # otherwise let the gcc default kick in. case ${CTARGET//_/-} in - *-hardfloat-*|*eabihf) + *-hardfloat-*|*eabihf*) confgcc+=( --with-float=hard ) ;; esac @@ -2174,10 +2174,13 @@ gcc_do_filter_flags() { fi if ver_test -lt 15.1 ; then - filter-flags -fdiagnostics-explain-harder -fdiagnostics-details filter-flags -fdiagnostics-set-output=text:experimental-nesting=yes fi + if ver_test -lt 16.1 ; then + filter-flags -fdiagnostics-details + fi + # Ada: PR116226 # D: PR117002 (bug #940750) filter-flags -Warray-bounds @@ -2447,6 +2450,11 @@ toolchain_src_test() { # From opensuse's spec file: "asan needs a whole shadow address space" ulimit -v unlimited + # Use the magic value of 1 which avoids invoking coredump handlers + # like systemd-coredumpd which can be really slow when we have many + # intentional or expected crashes. + prlimit -c=1 -p $$ + # 'asan' wants to be preloaded first, so does 'sandbox'. # To make asan tests work, we disable sandbox for all of test suite. # The 'backtrace' tests also do not like the presence of 'libsandbox.so'. @@ -2575,6 +2583,7 @@ toolchain_src_test() { # the failures are tolerable or not, so we bail out. eerror "No reference test data at ${manifest}!" eerror "GCC's tests require a baseline to compare with for any reasonable interpretation of results." + eerror "See https://wiki.gentoo.org/wiki/Project:Toolchain/sys-devel/gcc#Test_suite for details." if [[ -n ${GCC_TESTS_IGNORE_NO_BASELINE} ]] ; then eerror "GCC_TESTS_IGNORE_NO_BASELINE is set, ignoring test result and creating a new baseline..."