RFR: 8318455: Fix the compiler/sharedstubs/SharedTrampolineTest.java and SharedStubToInterpTest.java

Aleksey Shipilev shade at openjdk.org
Fri Oct 20 10:51:37 UTC 2023


On Thu, 19 Oct 2023 20:23:18 GMT, Evgeny Nikitin <enikitin at openjdk.org> wrote:

> A continuation of #16258.
> 
> Said tests try to control the compilers used via arguments passed to subprocesses' they create. The [JDK-8314838](https://bugs.openjdk.org/browse/JDK-8314838) added support for additional forwarding of the JTReg javaoptions to the subprocesses, but broke the original arguments passing. They are now passed to the enclosing driver and simply ignored by ProcessTools.createTestJvm. As a result, proper compiler may be not chosen on subprocess start, etc. This PR fixes that.
> 
> Testing: both tests run in CI, with additional arguments (tests were ignored, as expected) and without (passed).

test/hotspot/jtreg/compiler/sharedstubs/SharedTrampolineTest.java line 53:

> 51:         String testClassName = SharedTrampolineTest.class.getName() + "$" + test;
> 52:         ArrayList<String> command = new ArrayList<String>();
> 53:         command.add("-XX:-TieredCompilation");  // C2-compiler only

It is not quite clear to me why we should not pass `args[0]` here, like we do in another test fix. 
Would this test ever make sense to run with different compiler, e.g. C1?

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

PR Review Comment: https://git.openjdk.org/jdk/pull/16277#discussion_r1366813006


More information about the hotspot-compiler-dev mailing list