On Fri, 16 Jan 2026 03:42:18 GMT, Chris Plummer <[email protected]> wrote:
>> Yasumasa Suenaga has updated the pull request incrementally with two >> additional commits since the last revision: >> >> - Update copyright year in ConstantPool.java >> - Update ClhsdbDumpheap to verify HPROF_GC_ROOT_JAVA_FRAME > > I filed https://bugs.openjdk.org/browse/JDK-8306591 about 3 years ago, which > noted that hotspot produces the following records, but SA does not: > > HPROF_GC_ROOT_STICKY_CLASS > HPROF_GC_ROOT_JAVA_FRAME > HPROF_GC_ROOT_NATIVE_STACK > > I suspected that this could lead to incomplete heap dumps. It seems with this > PR you are fixing the lack of HPROF_GC_ROOT_JAVA_FRAME records in SA. @plummercj Thanks, I haven't been aware of JDK-8306591. I checked HotSpot again, heapDumper.cpp would dump `HPROF_GC_ROOT_STICKY_CLASS` and `HPROF_GC_ROOT_JAVA_FRAME`, but `HPROF_GC_ROOT_NATIVE_STACK` wouldn't be dumped I think. Should I close this and will start to work as JDK-8306591? or can I continue to work this? IMHO we can proceed this first because `HPROF_GC_ROOT_JAVA_FRAME` is very important to analyze memory leak, and it is very rare case that system classes (defined by null class loader) grabs large memory. I think it is ease to understand the commit if we make fix to align heapdump related code in SA to HotSpot in this PR, and add other unimplemented sub records like `HPROF_GC_ROOT_STICKY_CLASS` in another commits(s). ------------- PR Comment: https://git.openjdk.org/jdk/pull/29058#issuecomment-3758761803
