RFR: 8365229: ARM32: c2i_no_clinit_check_entry assert failed after JDK-8364269

Ashutosh Mehra asmehra at openjdk.org
Tue Aug 12 15:09:15 UTC 2025


On Tue, 12 Aug 2025 14:56:19 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

> If we propagate -1 there, we then need to handle it around the accessors that compute the actual address as well. This can be done, but it would be more intrusive.

Yes, I agree. And there is only one place, if I am not wrong, that computes the actual address from the offset - `AdapterHandlerLibrary::lookup_aot_cache`. That's not very intrusive IMHO.

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

PR Comment: https://git.openjdk.org/jdk/pull/26746#issuecomment-3179752562


More information about the hotspot-compiler-dev mailing list