The vulnerability activates when the decoding routine processes specific invalid inputs, such as 128 "A" characters ending with a single padding character. Because Java allocates a 98-byte output array prior to input validation, an integer calculation error yields a negative length value during the final padding phase. This miscalculation causes an AVX-512 masked store instruction to execute an unmasked 64-byte write, spilling 62 bytes past the memory boundary before an exception is thrown. Unlike modern SIMD implementations in browsers and runtimes that handle whitespace gracefully, Java's fast-path decoder still halts on spaces or line breaks.
Key points
Claims that the OpenJDK Base64 decoding vulnerability permits remote code execution remai…: Claims that the OpenJDK Base64 decoding vulnerability permits remote code execution remain unverified.
Flawed length calculations cause an AVX-512 masked store instruction to perform an uninte…: Flawed length calculations cause an AVX-512 masked store instruction to perform an unintended unmasked write.
Java's fast-path Base64 implementation fails to process spaces or line breaks without hal…: Java's fast-path Base64 implementation fails to process spaces or line breaks without halting execution.
Sources checked by our analysis pipeline: 1
18 November 2026 · 4 min
Let us start with an example that triggers the issue.
. The string is not valid base64, and a correct decoder rejects it. OpenJDK writes past the end of the array anyway. The bug is CVE-2026-114938. It has been in the JDK since 17.0.6. The bad write is in hand-written assembly.
There are claims that it leads to a major security vulnerability. I have not checked. But I care about the issue because I am indirectly involved.
Base64 turns three bytes into four ASCII characters. Email attachments are base64, and a string often ends with =
. Java sizes the output before it checks the input. One trailing =
counts as one byte of padding, so the allocated length is
3 × ((129 + 3) / 4) − 1 = 98.
So decode
allocates 98 bytes, then calls the decoding routine.
On an x86-64 machine with AVX-512 VBMI, the routine takes the first 128 characters as two blocks…
The programming language's maintenance strategy has been updated to deliver both security patches and select backported bug resolutions going forward. Because of this expanded focus on stability, the development team has officially rebranded the support period as the legacy maintenance phase. This ongoing support remains fully guaranteed through the previously established timeline ending in March 2027.
The core language maintainers have delivered critical maintenance updates addressing identified security vulnerabilities and routine software flaws. Published by k0kubun, this distribution arrives alongside coordinated maintenance releases across multiple supported release lines to ensure ecosystem stability. Developers relying on these platforms can now access updated packages through standard project distribution channels.
Maintainers have delivered vital security patches and general bugfixes through community contributions from numerous developers and testers. This deployment addresses ongoing stability concerns for systems running modern branches of the language. It arrived as part of a synchronized maintenance cycle impacting several supported release lines on the same day.