RFR: 8317294: Classloading throws exceptions over already pending exceptions

David Holmes dholmes at openjdk.org
Tue Oct 3 08:20:32 UTC 2023


On Fri, 29 Sep 2023 11:23:53 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

> See the bug for reproducer. I think we are just missing `return` after throwing the previous exception in `ClassFileParser::check_super_interface_access`. So we re-enter the method again with another exception in hand and fail the assert. This path is special, because we are looping, and we really need to exit after the first exception.
> 
> It is not easy to add a regression test that pokes into super-interfaces from some JDK internals.
> 
> Additional testing:
>  - [x] CTW reproducer no longer fails
>  - [x] Linux AArch64 `tier1 tier2 tier3` fastdebug

Yep this is an oversight from the initial module system implementation.

Thanks.

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

Marked as reviewed by dholmes (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/15984#pullrequestreview-1654610437


More information about the hotspot-runtime-dev mailing list