Suspicious code in GraphKit::record_profiled_receiver_for_speculation

[Tobias Hartmann]

Hi Radek,

On 06.07.2016 23:14, Radosław Smogura wrote:
> Hi all,
> 
> I have found following code, in JDK9 dev, which confused me a little bit:
> 
> Node* GraphKit::record_profiled_receiver_for_speculation(Node* n) {
>   if (!UseTypeSpeculation) {
>     return n;
>   }
>   ciKlass* exact_kls = profile_has_unique_klass();
>   bool maybe_null = true;
>   if (java_bc() == Bytecodes::_checkcast ||
>       java_bc() == Bytecodes::_instanceof ||
>       java_bc() == Bytecodes::_aastore) {
>     ciProfileData* data = method()->method_data()->bci_to_data(bci());
>     bool maybe_null = data == NULL ? true : data->as_BitData()->null_seen();
>   }
>   return record_profile_for_speculation(n, exact_kls, maybe_null);
>   return n;
> }
> 
> There are two subsequent returns and shadowing maybe_null. At a glance it looks like that closing bracket should be put one line down, between 
> `return record_profile_for_speculation` and `return n`?

Thanks for reporting this issue! The "return n" was introduced by JDK-8031755 [1] and I think it can be removed. I filed JDK-8160942 [2] to investigate.

Best regards,
Tobias

[1] http://hg.openjdk.java.net/jdk9/hs-comp/hotspot/rev/37023a7f1e1b#l7.99
[2] https://bugs.openjdk.java.net/browse/JDK-8160942

> 
> Kind regards,
> Radek
>