RFR: [XS] 8234625: hs test serviceability/sa/ClhsdbCDSCore.java fails on macOS 10.15

Thu Nov 28 08:55:29 UTC 2019

Hi Matthias,

First of all, thanks for tackling this issue which pops up with MacOS 10.15.

In the test, however, I think we should not/don’t need to query for the explicit MacOS version. Just do the writability check and if it fails, throw the SkippedException.

As for the message text, I’d prefer: “Directory \“” + coresDir +”\” is not writable”. I think that’s both, explicit and concise. @Igor, what do you think?

And there’s one last thing what I was thinking about: Is it really necessary that /cores is writable for the user? Maybe the system can still write a core into /cores? I’m trying to verify this now on my Catalina MacBook… will let you know.

Cheers
Christoph

From: Baesken, Matthias <matthias.baesken at sap.com>
Sent: Donnerstag, 28. November 2019 09:33
To: Igor Ignatyev <igor.ignatyev at oracle.com>
Cc: hotspot-dev at openjdk.java.net; Langer, Christoph <christoph.langer at sap.com>
Subject: RE: RFR: [XS] 8234625: hs test serviceability/sa/ClhsdbCDSCore.java fails on macOS 10.15

  *   so I'd suggest to change it to something like '/cores is not writable'.
Hi  Igor, done :

http://cr.openjdk.java.net/~mbaesken/webrevs/8234625.2/

May I add you as reviewer now ?

Best regards, Matthias

From: Igor Ignatyev <igor.ignatyev at oracle.com<mailto:igor.ignatyev at oracle.com>>
Sent: Mittwoch, 27. November 2019 19:15
To: Baesken, Matthias <matthias.baesken at sap.com<mailto:matthias.baesken at sap.com>>
Cc: hotspot-dev at openjdk.java.net<mailto:hotspot-dev at openjdk.java.net>; Langer, Christoph <christoph.langer at sap.com<mailto:christoph.langer at sap.com>>
Subject: Re: RFR: [XS] 8234625: hs test serviceability/sa/ClhsdbCDSCore.java fails on macOS 10.15

SkippedException's message is used by jtreg as a reason why a test was skipped, so I'd suggest to change it to something like '/cores is not writable'.

I'm also not sure if we really need to check version string at L#123 now, but it's fine if you decide to keep it.

Thanks.
-- Igor

On Nov 27, 2019, at 6:30 AM, Baesken, Matthias <matthias.baesken at sap.com<mailto:matthias.baesken at sap.com>> wrote:

>  I'd recommend you to use jtreg.SkippedException to signal that the test can't be run or to introduce new @requires property

Hello, I  now  changed  the test  to throw a  SkippedException  :

http://cr.openjdk.java.net/~mbaesken/webrevs/8234625.1/

Best regards, Matthias

From: Igor Ignatyev <igor.ignatyev at oracle.com<mailto:igor.ignatyev at oracle.com>>
Sent: Dienstag, 26. November 2019 18:35
To: Baesken, Matthias <matthias.baesken at sap.com<mailto:matthias.baesken at sap.com>>
Cc: hotspot-dev at openjdk.java.net<mailto:hotspot-dev at openjdk.java.net>
Subject: Re: RFR: [XS] 8234625: hs test serviceability/sa/ClhsdbCDSCore.java fails on macOS 10.15

for now, we have only test/jtreg-ext/requires/VMProps.java which sets @requires properties; although 'env.core.accessible' and similar properties aren't vm-related properties per-se so VMProps isn't the best place to set them. however VMProps already sets a few non vm properties, and given introduction of another class to set these properties is a bit of hassle, I think it's fine to just add `env.core.accessible` to VMProps (and later rename VMProps to be more appropriate for all kinds of properties).

there is one thing which you should be aware than add any new requires properties, jtreg runs VMProps once for *every* execution even if none of "target" tests use @requires, in other words, all test executions pay price of setting all properties, therefore it should be costly.

-- Igor

On Nov 26, 2019, at 2:07 AM, Baesken, Matthias <matthias.baesken at sap.com<mailto:matthias.baesken at sap.com>> wrote:

  *   or to introduce new @requires property, say `env.core.accessible`, which is true iif core dumping is enabled and dumped cores can be accessed from the test code.

Hello Igor,  that sounds interesting.

Who would set the say env.core.accessible property  accordingly (so that in our environment on 10.15 it would be false) ?

Best regards, Matthias

From: Igor Ignatyev <igor.ignatyev at oracle.com<mailto:igor.ignatyev at oracle.com>>
Sent: Montag, 25. November 2019 21:10
To: Baesken, Matthias <matthias.baesken at sap.com<mailto:matthias.baesken at sap.com>>
Cc: hotspot-dev at openjdk.java.net<mailto:hotspot-dev at openjdk.java.net>
Subject: Re: RFR: [XS] 8234625: hs test serviceability/sa/ClhsdbCDSCore.java fails on macOS 10.15

Hi Matthias,

your solution will hide the fact that the coverage from this test will be missed on macos 10.15+. I'd recommend you to use jtreg.SkippedException to signal that the test can't be run or to introduce new @requires property, say `env.core.accessible`, which is true iif core dumping is enabled and dumped cores can be accessed from the test code.

-- Igor

On Nov 25, 2019, at 1:58 AM, Baesken, Matthias <matthias.baesken at sap.com<mailto:matthias.baesken at sap.com>> wrote:

Hello,   the test
serviceability/sa/ClhsdbCDSCore.java
fails on macOS 10.15 .
exception :
java.lang.Error: cores is not a directory or does not have write permissions
                at ClhsdbCDSCore.main(ClhsdbCDSCore.java:115)
                at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
                at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
                at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
                at java.base/java.lang.reflect.Method.invoke(Method.java:564)
                at com.sun.javatest.regtest.agent.MainWrapper$MainThread.run(MainWrapper.java:127)
                at java.base/java.lang.Thread.run(Thread.java:833)

Looks like the test checks that directory /cores is writable :
File coresDir = new File("/cores");
if (!coresDir.isDirectory() || !coresDir.canWrite()) { ... // fail
However on macOS 10.15 /cores is not writable any more (at least for most users, including our test user).
So the test fails.

My change adjusts the test, so that it gives a clearer error message,  and  returns gracefully in case  we are running on macOS 10.15 and notice a non-writeable /cures directory .

Bug/webrev :

https://bugs.openjdk.java.net/browse/JDK-8234625

http://cr.openjdk.java.net/~mbaesken/webrevs/8234625.0/

Best regards, Matthias