IndexOf method of String file fails using Eclipse OpenJ9 VM 11.0.22.0

[gpezzini]

Java -version output

openjdk version "11.0.22" 2024-01-16
IBM Semeru Runtime Open Edition 11.0.22.0 (build 11.0.22+7)
Eclipse OpenJ9 VM 11.0.22.0 (build openj9-0.43.0, JRE 11 Windows 11 amd64-64-Bit Compressed References 20240131_835 (JIT enabled, AOT enabled)
OpenJ9 - 2c3d78b
OMR - ea8124dbc
JCL - 7876cac747 based on jdk-11.0.22+7)

Summary of problem

The attached simple class demonstrates that the 'indexOf' method fails after reading a string stored in an xml property file.

After execution with the above5ra indicated jsk the output is:
The string: Field1ç°^§|Field2 does NOT contains ç°^§|

Attention:

1. using JDK
   openjdk version "11.0.20" 2023-07-18
   IBM Semeru Runtime Open Edition 11.0.20.0 (build 11.0.20+8)
   Eclipse OpenJ9 VM 11.0.20.0 (build openj9-0.40.0, JRE 11 Windows 11 amd64-64-Bit Compressed References 20230810_700 (JIT enabled, AOT enabled)
   OpenJ9 - d12d10c
   OMR - e80bff83b
   JCL - f53b132192 based on jdk-11.0.20+8)

and older jdk version, it works.

The output is:
The string: Field1ç°^§|Field2 contains ç°^§|

2. Also using jdk
   openjdk version "17.0.10" 2024-01-16
   IBM Semeru Runtime Open Edition 17.0.10.0 (build 17.0.10+7)
   Eclipse OpenJ9 VM 17.0.10.0 (build openj9-0.43.0, JRE 17 Windows 11 amd64-64-Bit Compressed References 20240116_624 (JIT enabled, AOT enabled)
   OpenJ9 - 2c3d78b
   OMR - ea8124dbc
   JCL - 2aad089841f based on jdk-17.0.10+7)

The output is:
The string: Field1ç°^§|Field2 contains ç°^§|

So it works with 17 too

TestIndexOf.java.txt

[gpezzini]

I further simplified the example class
The same results described above can also be reproduced with the class attached here.
In practice, the problem can be reproduced even without switching between writing and reading an xml property file.
TestIndexOf2.java.txt

[pshipton]

Running with -XX:-CompactStrings is a workaround for the bug.

[patric42]

It's even worse: any String longer than 1 character including anything needing UTF-8 multibyte encoding will not be found. So "ä!", "!ä" or "wörd" won't work. Yikes!

[gpezzini]

Thanks and pardon me: we are only talking about the .indexOf() method.
I personally haven't tested other methods like .split() or similar that require a scan of the string.
I hope the fix covers all the scenarios involved.

[pshipton]

The fix is released into the upcoming 0.44 release, expected by early May.

[pshipton]

An open source pre-release 0.44 Windows build with the fix.
https://openj9-artifactory.osuosl.org/artifactory/ci-openj9/Build_JDK11_x86-64_windows_Release/134/OpenJ9-JDK11-x86-64_windows-20240419-025627.tar.gz

Let me know if you want a direct link to another platform.

The full builds:
https://openj9-jenkins.osuosl.org/job/Pipeline-Release-0.44.0-Build11/12/
https://openj9-jenkins.osuosl.org/job/Pipeline-Release-0.44.0-Build11x/10/

[patric42]

Linux amd64 and arm64 build please?

[pshipton]

https://openj9-artifactory.osuosl.org/artifactory/ci-openj9/Build_JDK11_x86-64_linux_Release/123/OpenJ9-JDK11-x86-64_linux-20240419-015834.tar.gz

https://openj9-artifactory.osuosl.org/artifactory/ci-openj9/Build_JDK11_aarch64_linux_Release/117/OpenJ9-JDK11-aarch64_linux-20240419-014102.tar.gz

https://openj9-artifactory.osuosl.org/artifactory/ci-openj9/Build_JDK11_aarch64_mac_Release/100/OpenJ9-JDK11-aarch64_mac-20240419-111829.tar.gz

[patric42]

thank you very much :-)