Tests can be put into two categories, native (tests that run on the build system) and non-native (tests run in QEMU). For any two targets (even of completely different architectures), the native tests will be running the same binary code (because they will be compiled for and run on the build machine), and thus will have the same result. So when building and running tests for multiple targets on a build machine, the native tests on need be run once. This can decrease the runtime of a multi-target test run significantly (like hours). This patch series makes it possible to run only run the non-native tests (skipping the native tests) by partitioning the set of tests into native and non-native based on whether QEMU is used by the test. The first patch is not necessary, but makes things look cleaner.
Phcoder, could you confirm that the logic above is sound? Glenn Glenn Washburn (2): conf/Makefile.common: Order alphabetically variables tests: Add check-native and check-nonnative make targets Makefile.am | 9 +++ Makefile.util.def | 164 +++++++++++++++++++++---------------------- conf/Makefile.common | 22 +++--- gentpl.py | 6 +- 4 files changed, 106 insertions(+), 95 deletions(-) -- 2.27.0 _______________________________________________ Grub-devel mailing list Grub-devel@gnu.org https://lists.gnu.org/mailman/listinfo/grub-devel
