Closed connglli closed 1 year ago
@BradleyWood : please add this codegen crash to your list of issues to investigate.
I reproduced the crash.
The reduced test case uses FuzzerUtils, but I commented it out and returned 0 and still reproduced a crash.
_addressRegister is null
https://github.com/eclipse/omr/blob/master/compiler/x/amd64/codegen/OMRMemoryReference.cpp#L618-L629
For some reason on initialisation it was determined that the memory reference does not need an address register at:
However, the opposite conclusion is made at binary encoding, but no register was allocated.
This is not a regression in 0.33. Reproducible on JDK8 back to at least 0.29.
This will not be fixed for 0.35. Moving to 0.36.
Moving to 0.38.
No change in status. Moving to 0.40.
In this test case, we have a small integer array which becomes stack allocated, and is indexed at ax$1 - Integer.MIN_VALUE. I have identified 3 separate issues related to this crash.
I will open a fix shortly.
Java -version output
Summary of problem
The following
Test.java
, which is reduced by us, crashes OpenJ9's JIT compilerDiagnostic files
By issuing
the following crash log is given:
Please also check openj9-bug-81.tar.gz for all the logs (jitdump, snap, etc.), the test (Test.java, Test.class), and the unreduced test (Test.java.orig).
Notice
The given
Test.java
(which is reduced by us) is always reproducible for us. If it is not reproducible for you, please useTest.java.orig
in the above link.