https://bugzilla.redhat.com/show_bug.cgi?id=1906980 Ben Beasley <code@xxxxxxxxxxxxxxxxxx> changed: What |Removed |Added ---------------------------------------------------------------------------- Flags|needinfo?(code@musicinmybra | |in.net) | --- Comment #12 from Ben Beasley <code@xxxxxxxxxxxxxxxxxx> --- I’ve never seen this before: fedora-review spends a while building, then spits this out: > WARNING: Package highway-devel-0.11.1-1.fc35 not built > WARNING: Package highway-doc-0.11.1-1.fc35 not built > ERROR: 'No srpm found for highway' (logs in /home/reviewer/.cache/fedora-review.log) Eventually I figured out that this was just a consequence of the SRPM in your last comment not matching the spec file. Once I got the correct SRPM from the Koji scratch-build, I got a similar error to before, although it was later in the build this time: > [ 92%] Built target arithmetic_test > [ 95%] Linking CXX executable tests/hwy_test > /usr/bin/cmake -E cmake_link_script CMakeFiles/hwy_test.dir/link.txt --verbose=1 > /usr/bin/g++ -O2 -flto=auto -ffat-lto-objects -fexceptions -g -grecord-gcc-switches -pipe -Wall -Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -fstack-protector-strong -specs=/usr/lib/rpm/redhat/redhat-annobin-cc1 -m64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -O2 -g -DNDEBUG -Wl,-z,relro -Wl,--as-needed -Wl,-z,now -specs=/usr/lib/rpm/redhat/redhat-hardened-ld -fPIE -pie CMakeFiles/hwy_test.dir/hwy/tests/hwy_test.cc.o -o tests/hwy_test libhwy.a /usr/lib64/libgtest.so /usr/lib64/libgtest_main.so /usr/lib64/libgtest.so -lpthread > /usr/bin/cmake -D TEST_TARGET=hwy_test -D TEST_EXECUTABLE=/builddir/build/BUILD/highway-0.11.1/x86_64-redhat-linux-gnu/tests/hwy_test -D TEST_EXECUTOR= -D TEST_WORKING_DIR=/builddir/build/BUILD/highway-0.11.1/x86_64-redhat-linux-gnu -D TEST_EXTRA_ARGS= -D TEST_PROPERTIES= -D TEST_PREFIX= -D TEST_SUFFIX= -D NO_PRETTY_TYPES=FALSE -D NO_PRETTY_VALUES=FALSE -D TEST_LIST=hwy_test_TESTS -D CTEST_FILE=/builddir/build/BUILD/highway-0.11.1/x86_64-redhat-linux-gnu/hwy_test[1]_tests.cmake -D TEST_DISCOVERY_TIMEOUT=60 -D TEST_XML_OUTPUT_DIR= -P /usr/share/cmake/Modules/GoogleTestAddTests.cmake > CMake Error at /usr/share/cmake/Modules/GoogleTestAddTests.cmake:77 (message): > Error running test executable. > Path: '/builddir/build/BUILD/highway-0.11.1/x86_64-redhat-linux-gnu/tests/hwy_test' > Result: Illegal instruction > Output: > > Call Stack (most recent call first): > /usr/share/cmake/Modules/GoogleTestAddTests.cmake:173 (gtest_discover_tests_impl) > gmake[2]: Leaving directory '/builddir/build/BUILD/highway-0.11.1/x86_64-redhat-linux-gnu' > gmake[2]: *** [CMakeFiles/hwy_test.dir/build.make:111: tests/hwy_test] Error 1 > gmake[2]: *** Deleting file 'tests/hwy_test' > gmake[1]: *** [CMakeFiles/Makefile2:239: CMakeFiles/hwy_test.dir/all] Error 2 > gmake[1]: *** Waiting for unfinished jobs.... It looks like this is the offending section in GoogleTestAddTests.cmake: > # Run test executable to get list of available tests > if(NOT EXISTS "${_TEST_EXECUTABLE}") > message(FATAL_ERROR > "Specified test executable does not exist.\n" > " Path: '${_TEST_EXECUTABLE}'" > ) > endif() > execute_process( > COMMAND ${_TEST_EXECUTOR} "${_TEST_EXECUTABLE}" --gtest_list_tests > WORKING_DIRECTORY "${_TEST_WORKING_DIR}" > TIMEOUT ${_TEST_DISCOVERY_TIMEOUT} > OUTPUT_VARIABLE output > RESULT_VARIABLE result > ) > if(NOT ${result} EQUAL 0) > string(REPLACE "\n" "\n " output "${output}") > message(FATAL_ERROR > "Error running test executable.\n" > " Path: '${_TEST_EXECUTABLE}'\n" > " Result: ${result}\n" > " Output:\n" > " ${output}\n" > ) > endif() So the test executables are getting run with --gtest_list_tests to discover the tests they implement, and the code that sets up the test suites has been compiled such that even this requires–well, here I haven’t waded through all of the macro soup, but something I don’t have, anyway. I think if it required all supported ISA extensions to run it would not be working on Koji. Note that the baseline target is scalar: > ./hwy_list_targets || ( exit 0 ) > Compiled HWY_TARGETS: AVX3 AVX2 SSE4 Scalar > HWY_BASELINE_TARGETS: Scalar …but that this was also the case in your older Koji scratch build, https://koji.fedoraproject.org/koji/taskinfo?taskID=65535443. So I’m pretty sure that adding -DHWY_COMPILE_ALL_ATTAINABLE is not doing anything at all here on x86_64, since scalar is the only baseline target. (This makes sense as baseline targets are supposed to be only those enabled in the compilation environment, and you are not adding anything beyond the default Fedora flags.) is causing it to additionally compile targets below the best baseline for testing, but it does not keep the test executable from requiring extensions. When I have a chance, I’m going to read the CMake and C sources a little more, and perhaps try a couple of experiments, to try to understand better what’s really happening here. -- You are receiving this mail because: You are on the CC list for the bug. You are always notified about changes to this product and component _______________________________________________ package-review mailing list -- package-review@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to package-review-leave@xxxxxxxxxxxxxxxxxxxxxxx Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/package-review@xxxxxxxxxxxxxxxxxxxxxxx Do not reply to spam on the list, report it: https://pagure.io/fedora-infrastructure
