On Sat, 21 Nov 2026 04:07:36 GMT, Quan Anh Mai <[email protected]> wrote:

> The result has been computed for A, not C. The key here is that the 
> computation of C can contradict the computation of an unrelated class A.

Got it. Tricky case. Alternatively, it could fail to observe `D` and report `C` 
as having no subclasses, but then it has to be consistent with 
`ciInstanceKlass::has_subklass()`. 

> Previously, I experimented with failing the ciEnv when encountering CHA 
> inconsistency, but I gave up because ciEnv does not expect to fail from a 
> call to ciEnv::get_instance_klass.

But `ciEnv::get_instance_klass()` is allowed to return unloaded 
`ciInstanceKlass`es, isn't it? Or did you try to proactively fail the whole 
compilation from there (setting `ciEnv::_failure_reason`) and that caused 
problems?

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

PR Comment: https://git.openjdk.org/jdk/pull/33753#issuecomment-7057934090

Reply via email to