RFR: 8258956: Memory Leak in StringCoding on ThreadLocal resultCached StringCoding.Result

[Alan Bateman]

On Wed, 13 Jan 2021 18:41:05 GMT, Naoto Sato <naoto at openjdk.org> wrote:

> Please review the fix to this issue. Unused thread local StringCoding.Result is now wrapped with SoftReference, which can be GC'ed as needed. I confirmed it worked as expected with a test case (https://github.com/openjdk/jdk/blob/8b4e1e1cf8f5d753ed901406f73d67b21557fddb/test/jdk/java/lang/StringCoding/ResultCachedGCTest.java), but decided not to include it in this PR. This is because I purposefully chose the size of the byte array and max heap size which is fragile, and could become noise in a regression test run.

Marked as reviewed by alanb (Reviewer).

-------------

PR: https://git.openjdk.java.net/jdk/pull/2064