JDK-8255915 : jdk/incubator/vector/AddTest.java timed out
  • Type: Bug
  • Component: hotspot
  • Sub-Component: compiler
  • Affected Version: 16,17
  • Priority: P4
  • Status: Resolved
  • Resolution: Fixed
  • OS: os_x
  • CPU: x86_64
  • Submitted: 2020-11-05
  • Updated: 2021-05-04
  • Resolved: 2021-04-27
The Version table provides details related to the release that this issue/RFE will be addressed.

Unresolved : Release in which this issue/RFE will be addressed.
Resolved: Release in which this issue/RFE has been resolved.
Fixed : Release in which this issue/RFE has been fixed. The release containing this fix may be available for download as an Early Access Release or a General Availability Release.

To download the current JDK release, click here.
JDK 17
17 b20Fixed
Related Reports
Relates :  
JDK-8223347 - Integration of Vector API (Incubator)
Description
The following test timed out in the JDK16 CI:

jdk/incubator/vector/AddTest.java

Here's a snippet from the log file:

#section:main
----------messages:(7/276)----------
command: main AddTest
reason: Assumed action based on file name: run main AddTest 
Mode: othervm [test needs --add-modules]
Additional options from @modules: --add-modules jdk.incubator.vector
Timeout information:
--- Timeout information end.
elapsed time (seconds): 1227.578
----------configuration:(3/48)----------
Boot Layer
  add modules: jdk.incubator.vector

----------System.out:(1/27)----------
Timeout refired 1200 times
----------System.err:(1/55)----------
WARNING: Using incubator modules: jdk.incubator.vector
----------rerun:(49/6050)*----------

I'm starting this bug off in hotspot/compiler because that's the
cat/subcat for: JDK-8223347 Integration of Vector API (Incubator)
Comments
Changeset: 0a88f0a1 Author: Paul Sandoz <psandoz@openjdk.org> Date: 2021-04-27 18:58:37 +0000 URL: https://git.openjdk.java.net/jdk/commit/0a88f0a1
27-04-2021
I can reproduce. The test runs very slowly under C1, and is made worse with assertions enabled and -Xcomp, specifically for execution of the following: static void workloadIndexMapped() { for (int i = 0; i < a.length; i += SPECIES.length()) { FloatVector av = FloatVector.fromArray(SPECIES, a, i, IDENTITY_INDEX_MAPPING, 0); FloatVector bv = FloatVector.fromArray(SPECIES, b, i, IDENTITY_INDEX_MAPPING, 0); av.add(bv).intoArray(c, i, IDENTITY_INDEX_MAPPING, 0); } } ... for (int i = 0; i < 30_0000; i++) { workloadIndexMapped(); } It indicates there is some limitation with C1, possibly with method handle functionality. I don't know if it's worth the time investigating. At this point i would recommend marking the test as C2 only, and on advice from HotSpot engineers i can log a bug against C1 (linked to this issue).
25-01-2021
Hi Paul, assigning to you. I hope that's fine with you. Can you take a look, please?
25-01-2021
Here's a log file snippet from the jdk-17+7-398-tier5 sighting: #section:main ----------messages:(7/276)---------- command: main AddTest reason: Assumed action based on file name: run main AddTest Mode: othervm [test needs --add-modules] Additional options from @modules: --add-modules jdk.incubator.vector Timeout information: --- Timeout information end. elapsed time (seconds): 1264.672 ----------configuration:(3/48)---------- Boot Layer add modules: jdk.incubator.vector ----------System.out:(1/27)---------- Timeout refired 1200 times ----------System.err:(1/55)---------- WARNING: Using incubator modules: jdk.incubator.vector ----------rerun:(49/6000)*---------- <snip> result: Error. Program `/scratch/mesos/jib-master/install/jdk-17+7-398/macosx-x64-debug.jdk/jdk-17/fastdebug/bin/java' timed out (timeout set to 1200000ms, elapsed time including timeout handling was 1264586ms). The JVM args for the test task are: -XX:+CreateCoredumpOnCrash -ea -esa -XX:CompileThreshold=100 -XX:+UnlockExperimentalVMOptions -Xcomp -server -XX:+TieredCompilation -XX:TieredStopAtLevel=1 The test machine is a Macmini5_3 running Mac_OS_X_10.13.6.
23-01-2021
Timeout was observed once. Reason may be machine dependent or related to a temporary issue.
17-11-2020
Thanks for your comments. It's valid to use the Vector API with C1 even though it won't be fast, so it should be testable. It would be possible to run the tests with increased timeout when C2 is disabled, but closing it as suggested by Paul is my favorite choice.
17-11-2020
Consider adding to test: @requires vm.compiler2.enabled
16-11-2020
This test is quite simple. Timing out after 20 minutes suggests some transient HotSpot issue given the flags it was run with -Xcomp -XX:+TieredCompilation -XX:TieredStopAtLevel=1 (FWIW the Vector API is not intrinsified on C1.) My suggestion is to close it down and reopen if it persists.
16-11-2020
I haven't seen timeouts in this test. It's very differen from the VectorConversion tests which we are investigating in JDK-8255959. Maybe it's possible to profile the test on the particular machine?
11-11-2020
Hi [~mdoerr], Request your help with this task. Similar to JDK-8255959 for now assigning this one also to you. Please un-assign/reassign if required. Thank you.
09-11-2020