In Bouncy Castle for Java FIPS (BC-FJA) versions prior to 2.1.3, the native entropy source on Intel platforms used unbounded retries for the RDSEED and RDRAND CPU instructions when they reported failure via the carry flag. The JNI seeding routine would spin indefinitely re-issuing these instructions if the entropy source failed persistently due to hardware faults, DRBG exhaustion, or hypervisor limitations. This caused the calling thread to hang without interruption or timeout, leading to denial of service. The fix introduced bounded retry loops (200 attempts for RDSEED and 20 for RDRAND), pauses between attempts, and throwing an exception on exhaustion after clearing any partial buffer securely with memzero. Other Bouncy Castle components without native entropy sources or older FIPS series are unaffected.

Applications using the native entropy source in affected versions of Bouncy Castle for Java FIPS on Intel platforms may experience indefinite hangs during entropy generation if the CPU entropy instructions fail persistently. This results in denial of service, as the calling thread cannot be interrupted or timed out while spinning. There is no direct compromise of cryptographic material or data confidentiality indicated.

Upgrade to Bouncy Castle for Java FIPS version 2.1.3 or later, where retry loops for CPU entropy instructions are bounded and failures throw exceptions instead of hanging. No other mitigation is indicated. Patch status is not explicitly confirmed in vendor advisories; users should verify upgrade availability from official Bouncy Castle sources.