-
由 Jiri Benc 提交于
Commit 5f70bde2 ("selftests: fix build behaviour on targets' failures") added a logic to track failure of builds of individual targets. However, it does exactly the opposite of what a distro kernel needs: we create a RPM package with a selected set of selftests and we need the build to fail if build of any of the targets fail. Both use cases are valid. A distribution kernel is in control of what is included in the kernel and what is being built; any error needs to be flagged and acted upon. A CI system that tries to build as many tests as possible on the best effort basis is not really interested in a failure here and there. Support both use cases by introducing a FORCE_TARGETS variable. It is switched off by default to make life for CI systems easier, distributions can easily switch it on while building their packages. Reported-by: NYauheni Kaliuta <yauheni.kaliuta@redhat.com> Signed-off-by: NJiri Benc <jbenc@redhat.com> Reviewed-by: NCristian Marussi <cristian.marussi@arm.com> Tested-by: NCristian Marussi <cristian.marussi@arm.com> Signed-off-by: NShuah Khan <skhan@linuxfoundation.org>
9d235a55