Re: [lworld] RFR: 8380434: [lworld] C2: Wrong execution caused by treating a non-buffered InlineType as an oop in PredictedCallGenerator

[Marc Chevalier]

On Mon, 30 Mar 2026 14:14:18 GMT, Tobias Hartmann <thartm...@openjdk.org> wrote:

>> src/hotspot/share/opto/callGenerator.cpp line 1126:
>> 
>>> 1124:     Node* n = slow_map->in(i);
>>> 1125:     if (m != n) {
>>> 1126: #ifdef assert
>> 
>> Suggestion:
>> 
>> #ifdef ASSERT
>
> Interesting, that this builds 😄

"is `assert` defined? No." It builds for the same reason `#ifdef ASSERT` builds 
in product mode.

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

PR Review Comment: 
https://git.openjdk.org/valhalla/pull/2256#discussion_r3010150462