unit-test.py: Problems with analyzes

[ewalkusx]

All BMC downstream repos should generate UT coverage report using upstream script, but there are problems with analyzes. There is no coverage report if the previous analyzes fail.

1. clang-tidy
   With clang 17 there are compiler errors for nested tuple construction with flag -std=c++23. This is resolved in clang 18, but we are still waiting for clang update: https://gerrit.openbmc.org/c/openbmc/openbmc-build-scripts/+/69586
2. sanitizers:
   linker error with -Db_lundef=false and -Db_sanitize=address,undefined

I can propose 2 solutions:
1) changing the order of the analyzes https://gerrit.openbmc.org/c/openbmc/openbmc-build-scripts/+/72247
2) adding coverage report to 'TEST_ONLY' option

--- a/scripts/unit-test.py
+++ b/scripts/unit-test.py
@@ -921,6 +921,7 @@ class Meson(BuildSystem):
             # is supported by meson's build types. Configure it manually.
             meson_flags.append("-Ddebug=true")
             meson_flags.append("-Doptimization=g")
+            meson_flags.append("-Db_coverage=true")
         else:
             meson_flags.append("--buildtype=debugoptimized")
         if OptionKey("tests") in meson_options:
@@ -1186,6 +1187,8 @@ class Package(object):
         system.test()
         if not TEST_ONLY:
             system.analyze()
+        else:
+            check_call_cmd("ninja", "-C", "build", "coverage-html")
 
     def test(self):
         for system in self.build_systems():

Any other ideas? 🙂

[amboar]

Perhaps add a coverage() method to the BuildSystem class in https://github.com/openbmc/openbmc-build-scripts/blob/master/scripts/unit-test.py#L526 ? That way it can be run independent of other analysis.

[ewalkusx]

https://gerrit.openbmc.org/c/openbmc/openbmc-build-scripts/+/72727
Add an optional parameter ANALYSES to enable the ability to run only the listed analyses in the given order.

I think this solution has more potential.