On Sat, 14 Feb 2026 05:02:32 GMT, Yasumasa Suenaga <[email protected]> wrote:
>> Coleen Phillimore has updated the pull request incrementally with one >> additional commit since the last revision: >> >> Remove the doc. > > Actually `printmdo` is not so needed for troubleshooting POV, but it is > useful for learning HotSpot and JIT compiler behavior IMHO (I gave a talk > about MDO in relation to JVMCI in past ( @ Java Day Tokyo 2017)). Replay data > might be much better as you wrote in CSR, but it is difficult to extract MDO > from that, and also it is difficult to check realtime data (e.g. num of > branches). > > I do not have big objections for removing `printmdo`, but I think it is > better if we can keep it for learning HotSpot if possible. I'm not familiar > with Valhalla, but I want to know whether it needs big changes to support > `printmdo`. @YaSuenag for your usage, I think a jcmd would be a better place to support this, rather than a Java copy of this code that can and will become out of date. I'll file an RFE for that, and could you elaborate in that RFE? ------------- PR Comment: https://git.openjdk.org/jdk/pull/29693#issuecomment-3914758515
