RFR: 8348570: CTW: Expose the code hidden by uncommon traps [v4]
Aleksey Shipilev
shade at openjdk.org
Thu Feb 6 20:23:17 UTC 2025
On Wed, 5 Feb 2025 15:23:07 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:
>> We have been looking at some related compiler behaviors, and realized that in the absence of profiling data, C2 routinely uncommon-traps a lot of code that is presumed to be never executed. This apparently is a norm in CTW tests: CTW runners never execute code, and so only the most basic java.base classes are having any profile. This seems to limit the scope of CTW testing.
>>
>> I think we need to run CTW in the mode that exposes more code to the compiler optimizations.
>>
>> Case in point: [JDK-8348572](https://bugs.openjdk.org/browse/JDK-8348572), which reliably fails with more aggressive compilation mode.
>>
>> Additional testing:
>> - [x] Linux x86-64 server fastdebug, `applications/ctw/modules`
>> - [x] Linux AArch64 server fastdebug, `applications/ctw/modules`
>
> Aleksey Shipilev has updated the pull request with a new target base due to a merge or a rebase. The incremental webrev excludes the unrelated changes brought in by the merge/rebase. The pull request contains four additional commits since the last revision:
>
> - Merge branch 'master' into JDK-8348570-ctw-uncommon-traps
> - Merge branch 'master' into JDK-8348570-ctw-uncommon-traps
> - Also do markMethodProfiled for extra scope
> - Fix
About half of my CTW pipeline passed without problems. So it does not look there are major breakages with it. I'll integrate and see what happens in weekly testing. Thanks!
-------------
PR Comment: https://git.openjdk.org/jdk/pull/23296#issuecomment-2640911603
More information about the hotspot-compiler-dev
mailing list