RFR: 8335358: [premain] Explore alternative ways to trigger the end of training run [v3]

[Mat Carter]

On Mon, 16 Sep 2024 16:30:04 GMT, Ashutosh Mehra <asmehra at openjdk.org> wrote:

>> Mat Carter has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   another missing include that impacts some build configurations
>
> src/hotspot/cpu/aarch64/interp_masm_aarch64.cpp line 1403:
> 
>> 1401: 
>> 1402: void InterpreterMacroAssembler::end_training_check() {
>> 1403:   if (!FLAG_IS_DEFAULT(AOTEndTrainingOnMethodEntry) && CDSPreimage == nullptr) {
> 
> CDSPreimage == nullptr may not be enough as it is would be null when AOTCache is not used at all. There is an API in CDSConfig that can be used here to check for training run: CDSConfig::is_dumping_preimage_static_archive().
> Also, I think it would be cleaner if these checks can be encapsulated in an API in CDSConfig.

added a simple is_dumping_preimage_static_archive_with_triggers

> src/hotspot/cpu/aarch64/templateInterpreterGenerator_aarch64.cpp line 1276:
> 
>> 1274: 
>> 1275:   // AOT training run support
>> 1276:   __ end_training_check();
> 
> Does it make sense to do these checks after `notify_method_entry()`? It could aid in debugging/testing that the training run indeed ended on this method entry.

good call, will swap order of notify_method_entry and end_training_check

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

PR Review Comment: https://git.openjdk.org/leyden/pull/21#discussion_r1761995190
PR Review Comment: https://git.openjdk.org/leyden/pull/21#discussion_r1761994473