RFR: 8228604: StackMapFrames are missing from redefined class bytes of retransformed classes

Serguei Spitsyn sspitsyn at openjdk.org
Wed Feb 1 05:01:50 UTC 2023


On Wed, 1 Feb 2023 04:32:34 GMT, David Holmes <dholmes at openjdk.org> wrote:

> > My understanding is that eliding the StackMapTable attribute when there is no split-verifier is an optimization which can cause problems for agents/tools.
> 
> Sure we can keep the attribute, but I have to question why these agents/tools are using something that is only present for the purposes of verification and is not required to be present. Seems like we are encouraging buggy agent/tool behaviour.

I see your point. Tools may need this attribute for similar purpose as the split-verifier. Your question is valid anyway.
In our case (testing purpose), we want to make sure we can reconstitute this attribute correctly.

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

PR: https://git.openjdk.org/jdk/pull/12155


More information about the serviceability-dev mailing list