RFR: 8317951: Refactor loading of zip library to help resolve JDK-8315220 [v4]
David Holmes
dholmes at openjdk.org
Mon Oct 30 05:20:32 UTC 2023
On Thu, 26 Oct 2023 10:01:55 GMT, Markus Grönlund <mgronlun at openjdk.org> wrote:
>> Greetings,
>>
>> This refactorization resolves the problem associated with [JDK-8315220](https://bugs.openjdk.org/browse/JDK-8315220) and, subsequently, [JDK-8315892](https://bugs.openjdk.org/browse/JDK-8315892).
>>
>> Testing: jdk_jfr, tier1-9, stress testing
>>
>> Thanks
>> Markus
>
> Markus Grönlund has updated the pull request incrementally with one additional commit since the last revision:
>
> do not exit vm on failure to resolve InitParams and Compress
Okay I finally understand. We used to initialize the entry points in two stages: one eager that would do VM exit if `dl_load` failed. The other two lazily with no vm-exit, or any error reporting. With the new code everything is initialized at once, but you extended the error handling to the individual `dll_lookup`s and so apply the vm-exit or not vm-exit to them based on earlier behaviour.
I can only assume that it is optional to have the Zip `InitParams`/`Fully` API's available so failing to find them is not fatal. A comment to that effect in `store_function_pointers` would clarify the initially strange seeming initialization logic. A future RFE could see if these truly are optional APIs, and if not simplify the code to treat everything consistently.
Thanks.
-------------
Marked as reviewed by dholmes (Reviewer).
PR Review: https://git.openjdk.org/jdk/pull/16149#pullrequestreview-1703280289
More information about the hotspot-runtime-dev
mailing list