[7u-60] backport request: 6571600 JNI use results in UnsatisfiedLinkError looking for libmawt.so (original) (raw)

david buck david.buck at oracle.com
Tue Jan 7 16:30:18 PST 2014

Hi!

I am sorry about not following the correct procedure. The bug has now been tagged 7u60-critical-approved by the release team. Between that and the 2 7u60-specific code reviews, I believe I have met the conditions for phase 2 approval. May I please push this fix into 7u60?

Cheers, -Buck

(2014/01/07 23:34), Seán Coffey wrote:

Hey Buck,

Such fixes should be pushed to the jdk7u-dev forest first [1]. It gives everyone an option to test the fix via the early access bundles. I'd suggest that steps should be taken to get this fix into jdk7u-dev also. It looks like you've already marked the bug with 7u60-critical-request. Thanks, Sean. [1] http://openjdk.java.net/projects/jdk7u/criticalcpufixes.html On 07/01/14 01:24, david buck wrote: Hi!

Please disregard this request. Due to the urgency of this change, it has been approved for inclusion into the April CPU update. It will be merged into 7u-dev at the same time the security fixes are. Best Regards, -Buck (2013/12/25 21:29), david buck wrote: Hi!

Sergey Bylokhov and Artem Ananiev have reviewed this fix for 7u60 and have approved. Now all we need is the RM team to approve (7u60-critical-approved tag added in JBS). Cheers, -Buck (2013/12/25 5:11), david buck wrote: Hi!

I am resending this as a phase 2 request for inclusion into 7u60. This is a critical issue impacting multiple users and is an extremely low-risk change. I have tested jdk7 builds with this fix against our reproducer and all of the standard (default jprt) tests indicate no issues. This change has already passed review for inclusion into 5 / 6 / 9. Best Regards, -Buck (2013/12/24 6:02), david buck wrote: Hi!

This is a request to backport the flowing fix to JDK7. [ JDK-6571600 JNI use results in UnsatisfiedLinkError looking for libmawt.so ] https://bugs.openjdk.java.net/browse/JDK-657160 The root cause of this issue is a name conflict with JNIOnLoad provided by awtLoadLibrary.c and code loaded manually by the user (usually their own JNI libraries). The fix is to use a different name for calling dladdr that is much less likely to conflict with any names used by customer code. The fix is exactly the same in JDK7 as it is in both JDK8 and JDK9. I have of course tested the fix on JDK7 as well. The JDK9 fix can be found here: http://hg.openjdk.java.net/jdk9/client/jdk/rev/35dfaa9954c2 As the forest for JDK8u-dev is not up yet, I am submitting individual requests to backport to both JDK8 and JDK7 in parallel. Cheers, -Buck

More information about the jdk7u-dev mailing list