On Tue, 23 Jun 2026 13:50:05 GMT, Roland Westrelin <[email protected]> wrote:
>> The problem, here, is that, when a `Proj` is created for a late >> inlining call and transformed, it is replaced by a `Proj` for a >> dominating call. As a consequence, the code tries to clear an already >> cleared flag. >> >> Beyond the fix for the crash here, 8375694 (that added that code) >> expects a single return projection but there can be several with the >> scalarized calling convention. I filed 8387131 to investigate if the >> change from 8375694 needs to be reworked. >> >> --------- >> - [x] I confirm that I make this contribution in accordance with the >> [OpenJDK Interim AI Policy](https://openjdk.org/legal/ai). > > Roland Westrelin has updated the pull request incrementally with one > additional commit since the last revision: > > more src/hotspot/share/opto/callGenerator.cpp line 187: > 185: kit.set_edges_for_java_call(call, false, _separate_io_proj); > 186: Node* ret = kit.set_results_for_java_call(call, _separate_io_proj); > 187: if (is_late_inline() && !call->is_boxing_method() && ret->is_Proj() && > ret->in(0) == call) { Why do you need `!call->is_boxing_method()` here? ------------- PR Review Comment: https://git.openjdk.org/valhalla/pull/2575#discussion_r3463007694
