[jdk8u-dev] RFR: 8301143: [TESTBUG] jfr/event/sampling/TestNative was backported to JDK8u without proper native wrapper

[Zdenek Zambersky]

On Tue, 24 Jan 2023 16:12:14 GMT, Jiří Vaněk <jvanek at openjdk.org> wrote:

> This test was brouht in by 8223147: JFR Backport and never passed. Despite not being wrong nor having jfr broken in jdk8. jdk8 testsuite simply lacks native support casses.
> 
> This is adding the shell wrapper which is compiling and properly setting up the native longSleep  in native shared object.
> 
> This comit was tested only on linux.
> After this commit whole jdk_jfr group passes

Yy, shell wrapper will be necessary to make this work on JDK8.  (This wrapper is inspired by [1], I think.)
I have 2 remarks:

MacOS testing
-  do we want to enable this test on MacOS? (if so, it should probably be tested there)

Linux multilib
- on linux there is also multilib scenario, which is a bit tricky
- basically when building native test libs for 32-bit JDK running on 64-bit system, compiler flag is needed to build 32-bit test libs
- for hotspot tests this case is now detected in test_env.sh file (See this PR for details [2])
- question is: should this be done in test or should there be something like test_env.sh for jdk tests as well?

[1] https://github.com/openjdk/jdk8u-dev/blob/a5bd9018d5671f2399272b680e83239154443b11/jdk/test/jdk/tools/launcher/JliLaunchTest.sh
[2] https://github.com/openjdk/jdk8u-dev/pull/173

-------------

PR: https://git.openjdk.org/jdk8u-dev/pull/235