RFR: 8369559: Identify owning method for MethodTrainingData and CompileTrainingData in AOT map output
María Arias de Reyna Domínguez
duke at openjdk.org
Tue Oct 14 07:27:05 UTC 2025
On Fri, 10 Oct 2025 11:27:30 GMT, María Arias de Reyna Domínguez <duke at openjdk.org> wrote:
> Provides part of fixes for: https://bugs.openjdk.org/browse/JDK-8363440
> Fixes: https://bugs.openjdk.org/browse/JDK-8369559
>
> Right now, MethodTrainingData and CompileTrainingData do not show to what Method they belong to:
>
>
> $ cat aot.map | grep MethodTrainingData
> 0x00000008019d54c0: @@ MethodTrainingData 96
> 0x00000008019dcec8: @@ MethodTrainingData 96
> [...]
>
> $ cat aot.map | grep CompileTrainingData
> [...]
> 0x000000080079c8a0: @@ CompileTrainingData 80
> 0x00000008007a7660: @@ CompileTrainingData 80
>
>
> Add the method name to those lines in the AOT map to add more context. Also, for CompileTrainingData, add the level of compilation.
>
> The output should look like the following:
>
>
> $ cat aot.map | grep CompileTrainingData
> 0x000000080079c8a0: @@ CompileTrainingData 80 3 int java.lang.Byte.hashCode()
> 0x00000008007a7660: @@ CompileTrainingData 80 3 java.lang.Object java.lang.ref.Reference.get()
> 0x00000008007a76b0: @@ CompileTrainingData 80 4 java.lang.Object java.lang.ref.Reference.get()
> [...]
>
> $ cat aot.map | grep MethodTrainingData
> 0x00000008007864c0: @@ MethodTrainingData 96 void java.lang.System.arraycopy(java.lang.Object, int, java.lang.Object, int, int)
> 0x00000008007884f0: @@ MethodTrainingData 96 boolean java.lang.Module.isNamed()
> 0x00000008007892f8: @@ MethodTrainingData 96 boolean java.lang.Module.implIsExportedOrOpen(java.lang.String, java.lang.Module, boolean)
> [...]
>
>
> Note the number before the method signature on CompileTrainingData that represents the level, important because there are two CompileTrainingData for the same method with different compilation levels.
>
> Some elements have a nullpointer instead of a valid Method, and those will be represented without method signature (as they appear currently, with no changes).
@adinn @iklam can you please review this one too? More log messages on the AOT Map file.
-------------
PR Comment: https://git.openjdk.org/jdk/pull/27740#issuecomment-3400457145
More information about the hotspot-runtime-dev
mailing list