KONNO Kazuhiro
KONNO Kazuhiro
The IL trees above are from the "scorching" compilation. The "hot" compilation in the same trace file does not change the inlineProfiledGuard into an inlineHierarchyGuard like the "scorching" compilation does....
Another "scorching" trace file is available in the artifact of https://openj9-jenkins.osuosl.org/job/Grinder_iteration_1/139/ as aqa-tests/TKG/output_16597331655943/jdk_lang_1/work/java/lang/String/UnicodeCasingTest/parseInt.txt.1.1426511.33813.20220805.211013.1426511 .
The SEGV on AArch64 in https://github.com/eclipse-openj9/openj9/issues/14115#issuecomment-1009956805 seems to have occurred in the following location, where `javaLangClass` in x21 is 0x100000, and the result of `offsetOfObjectVftField()` in x0 is 0: https://github.com/eclipse-openj9/openj9/blob/684717766630da76ef64c0fedb269e3965900ede/runtime/compiler/env/VMJ9.cpp#L8391
FYI @Akira1Saitoh
jenkins test sanity plinux jdk19
jenkins test sanity plinux jdk19
jenkins compile win jdk19
jenkins compile win jdk19
Windows build was successful.
jenkins test sanity plinux jdk19