RFR(S): 8209164: [TESTBUG] Apply jtreg skipped status to cds tests
ioi.lam at oracle.com
Thu Aug 9 23:26:25 UTC 2018
Also, some test cases executes many JVM processes. E.g.,
VerifierTest.java. On Windows, sometimes you get a mapping failure once
for several JVM launches, so VerifierTest.java is bound to always exit
with the SkippedException.
I worry that would reduce test coverage on Windows.
For such tests, maybe you should catch the SkippedException and move on
to the next sub-case?
On 8/9/18 4:22 PM, Ioi Lam wrote:
> Hi Misha,
> I think you should change checkCommonExecExceptions to a void
> function, since it now always throws or returns false.
> That way, you can catch places that try to read the return value, and
> fix these as appropriate. E.g., this in the Result class
> hasMappingFailure =
> hasAbnormalExit = (!hasMappingFailure) &&
> (output.getExitValue() != 0);
> hasNormalExit = (!hasMappingFailure) &&
> (output.getExitValue() == 0);
> I think the Result.hasMappingFailure field can now be removed.
> - Ioi
> On 8/9/18 1:50 PM, Mikhailo Seledtsov wrote:
>> Please review this straight forward change. Now that JTReg harness
>> supports the "skipped status",
>> this change updates CDS and AppCDS tests to use this mechanism where
>> JBS: https://bugs.openjdk.java.net/browse/JDK-8209164
>> Webrev: http://cr.openjdk.java.net/~mseledtsov/8209164.01/index.html
>> 1. Locally: exercised tests under runtime/SharedArchiveFile
>> and runtime/appcds on Linux-x64
>> No new failures
>> 2. Run runtime/SharedArchiveFile and runtime/appcds via
>> distributed test system
>> Run tier1,tier2
>> In progress
>> Thank you,
More information about the hotspot-runtime-dev