eclipse-omr/omr

In Eclipse OMR versions up to 0.11, the arraycmp SIMD implementation for Z and P does not check if the number of bytes to compare is zero.

In the Eclipse OMR port library component since release 0.2.0, an API function to return the textual names of all supported processor features was not accounting for the separator inserted between processor features. If the output buffer supplied to this function was incorrectly sized, failing to account for the separator when determining when a write to the buffer was safe could lead to a buffer overflow. This issue is fixed in Eclipse OMR version 0.8.0.

In the Eclipse OMR compiler component, since release 0.7.0, an optimization enabled for Eclipse OpenJ9 consumers of OMR on Z processors incorrectly handles NUL (0x00) characters during the Latin-compatible charset (UTF-8, ISO8859-1, ASCII, etc) to IBM-1047/037 translation sequence. This can cause the output byte array to be truncated, discarding the first NUL byte and all subsequent characters, and thereby exposing a possible buffer over-read problem. This issue is fixed in Eclipse OMR version 0.8.0.

In Eclipse OMR, from the initial contribution to version 0.4.0, some OMR internal port library and utilities consumers of z/OS atoe functions do not check their return values for NULL memory pointers or for memory allocation failures. This can lead to NULL pointer dereference crashes. Beginning in version 0.5.0, internal OMR consumers of atoe functions handle NULL return values and memory allocation failures correctly.

In Eclipse OMR versions 0.2.0 to 0.4.0, some of the z/OS atoe print functions use a constant length buffer for string conversion. If the input format string and arguments are larger than the buffer size then buffer overflow occurs. Beginning in version 0.5.0, the conversion buffers are sized correctly and checked appropriately to prevent buffer overflows.

In Eclipse OpenJ9 release versions prior to 0.44.0 and after 0.13.0, when running with JVM option -Xgc:concurrentScavenge, the sequence generated for System.arrayCopy on the IBM Z platform with hardware and software support for guarded storage [1], could allow access to a buffer with an incorrect length value when executing an arraycopy sequence while the Concurrent Scavenge Garbage Collection cycle is active and the source and destination memory regions for arraycopy overlap. This allows read and write to addresses beyond the end of the array range.

In Eclipse Openj9 before version 0.35.0, interface calls can be inlined without a runtime type check. Malicious bytecode could make use of this inlining to access or modify memory via an incompatible type.