All builds of Eclipse OpenJ9 prior to 0.15 contain a bug where the loop versioner may fail to privatize a value that is pulled out of the loop by versioning - for example if there is a condition that is moved out of the loop that reads a field we may not privatize the value of that field in the modified copy of the loop allowing the test to see one value of the field and subsequently the loop to see a modified field value without retesting the condition moved out of the loop. This can lead to a variety of different issues but read out of array bounds is one major consequence of these problems.

All builds of Eclipse OpenJ9 prior to 0.15 contain a bug where the loop versioner may fail to privatize a value that is pulled out of the loop by versioning - for example if there is a condition that is moved out of the loop that reads a field we may not privatize the value of that field in the modified copy of the loop allowing the test to see one value of the field and subsequently the loop to see a modified field value without retesting the condition moved out of the loop. This can lead to a variety of different issues but read out of array bounds is one major consequence of these problems.

OpenJDK: Insufficient checks of suppressed exceptions in deserialization (Utilities, 8212328) OpenJDK: Unbounded memory allocation during deserialization in Collections (Utilities, 8213432) OpenJDK: Missing URL format validation (Networking, 8221518) libpng: use-after-free in png_image_free in png.c JDK: Failure to privatize a value pulled out of the loop by versioning

Security update for java-1_7_1-ibm