libcore.git - libcore

	Commit message (Collapse)	Author	Age	Files	Lines
*	SecureRandom should be thread safe	Brian Carlstrom	2010-12-01	1	-27/+28
\| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \|	Random is documented to be thread safe and our implementation appears to be. However, the SecureRandom subclass implementation was not. This was quite reproducible with the javax.net.ssl unit tests when run on a multicore device, but not seen on a uniprocessor. Details: Actual bug fix by adding synchronized to engine* methods luni/src/main/java/org/apache/harmony/security/provider/crypto/SHA1PRNG_SecureRandomImpl.java New testSecureRandomThreadSafety based on generateSeed errors in javax.net.ssl tests. luni/src/test/java/tests/security/SecureRandomTest.java Removing dalvik.annotation.* from assorted SecureRandom tests. luni/src/test/java/org/apache/harmony/security/tests/java/security/SecureRandom2Test.java luni/src/test/java/org/apache/harmony/security/tests/java/security/SecureRandomSpiTest.java luni/src/test/java/tests/java/security/SecureRandomTest.java luni/src/test/java/tests/security/SecureRandomTest.java Change-Id: I5e8ece4c0836b02277d7cac1b9b2f60b3c4a7755
*	Scrubbing tests marked @BrokenTest.	Jesse Wilson	2010-06-24	1	-0/+95
	This rearranges the security test support infrastructure. We no longer rely on many top-level classes defined in CipherHelper.java to provide test support. Instead these each have their own top level class in our test support package: support/src/test/java/tests/security. Similarly for abstract classes intended to be subclassed by cipher-specific tests. Other test methods that were duplicated in Harmony have been removed. We need to pay closer attention to Harmony failures because they are now our only source of coverage for some of these tests. Change-Id: I1a1ca8a046bc9b6a33d5fa3f55fecc0d39f72c16