[11u] RFR 8253923: C2 doesn't always run loop opts for compilations that include loops

Vladimir Kozlov vladimir.kozlov at oracle.com
Fri Mar 5 17:28:19 UTC 2021


Looks good.

Thanks,
Vladimir K

On 3/4/21 11:55 PM, Roland Westrelin wrote:
> 
> Original bug:
>    https://bugs.openjdk.java.net/browse/JDK-8253923
>    https://github.com/openjdk/jdk/commit/a6c23b77
>   
> Original patch does not apply cleanly to 11u because code around a
> couple changes is different. Patch is otherwise identical. See below for
> hunks that failed to apply.
>   
> 11u webrev:
>    https://cr.openjdk.java.net/~roland/8253923.11u/webrev.00/
>   
> Testing: x86_64 build, tier1, some complile the world testing
> 
> Roland.
> 
> --- callGenerator.cpp
> +++ callGenerator.cpp
> @@ -456,7 +456,6 @@ void LateInlineCallGenerator::do_late_inline() {
>         result = (result_size == 1) ? kit.pop() : kit.pop_pair();
>       }
>   
> -    C->set_has_loops(C->has_loops() || _inline_cg->method()->has_loops());
>       C->env()->notice_inlined_method(_inline_cg->method());
>       C->set_inlining_progress(true);
>       C->set_do_cleanup(kit.stopped()); // path is dead; needs cleanup
> 
> --- compile.cpp
> +++ compile.cpp
> @@ -1022,6 +1022,7 @@ void Compile::Init(int aliaslevel) {
>   #ifdef ASSERT
>     _type_verify_symmetry = true;
>     _phase_optimize_finished = false;
> +  _exception_backedge = false;
>   #endif
>   }
>   
> 


More information about the jdk-updates-dev mailing list