On Mon, 9 Mar 2026 10:17:38 GMT, Volkan Yazici <[email protected]> wrote:

> Improve input validation checks and documentation of the 
> `java.lang.StringUTF16` intrinsics.
> 
> ### References:
> 
> - [JDK-8374582] #29164 — ditto for `java.lang.StringCoding`
> - [JDK-8367129] #28832 — ditto for `java.lang.StringLatin1`
> 
> [JDK-8367129]: https://bugs.openjdk.org/browse/JDK-8367129
> [JDK-8374582]: https://bugs.openjdk.org/browse/JDK-8374582

Thanks for your continued work on range checks @vy. Overall it look good. I've 
had a first look and left a couple of inline comments.

src/hotspot/share/opto/library_call.cpp line 846:

> 844: }
> 845: 
> 846: #define RETURN_ON_BAILOUT(region, ...) \

I was wondering if it wouldn't be cleaner to avoid creating a macro in this 
case and instead e.g. have 2 regular methods, one that "initializes the halt 
nodes" and the other that "finalizes the halt nodes" (it might be slightly 
better if you need to debug and slightly less risky with variable shadowing... 
no strong opinion about it though)

src/hotspot/share/opto/library_call.cpp line 862:

> 860:   if (stopped()) { \
> 861:     return true; \
> 862:   }

Since you use the `while(0)` idiom I think it would make sense to move this 
inside as well.

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

PR Review: https://git.openjdk.org/jdk/pull/30139#pullrequestreview-3916364543
PR Review Comment: https://git.openjdk.org/jdk/pull/30139#discussion_r2906440868
PR Review Comment: https://git.openjdk.org/jdk/pull/30139#discussion_r2906283852

Reply via email to