RFR: 8316131: runtime/cds/appcds/TestParallelGCWithCDS.java fails with JNI error

[David Holmes]

On Fri, 31 May 2024 22:37:56 GMT, Calvin Cheung <ccheung at openjdk.org> wrote:

> The `runtime/cds/appcds/TestParallelGCWithCDS.java` test currently expects specific output. When running with a small heap, JVM could fail at different places with different output.
> Updating the test if the expected output isn't there, checks for non-zero exit value and that the JVM hasn't crashed.
> 
> Tested with the following jtreg options on linux-x64:
> 
> -javaoptions:"-Xcomp -ea -esa -XX:CompileThreshold=100 -XX:+UnlockExperimentalVMOptions -server -XX:-TieredCompilation -XX:-UseCompressedOops -XX:+UnlockExperimentalVMOptions -XX:+UseJVMCICompiler -Xmx8m -Xms2m"
> -timeout:25

This will treat any "VM exit during initialization" error as a pass for this test. That seems rather broad but I'm not sure we can really do better unless all potential exit paths are uncovered. That said IIUC the original problem with no exception being thrown has been fixed, so perhaps it is only the new case you discovered that really needs to be allowed for?

test/hotspot/jtreg/runtime/cds/appcds/TestParallelGCWithCDS.java line 124:

> 122:                     } catch (RuntimeException ex) {
> 123:                         // If the test is run with a very small heap, the above expected
> 124:                         // output may not occurred as the JVM may fail in many locations

Nit: s/occurred/occur/

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

PR Review: https://git.openjdk.org/jdk/pull/19506#pullrequestreview-2092766244
PR Review Comment: https://git.openjdk.org/jdk/pull/19506#discussion_r1623769400