[lworld] RFR: 8372113: [lworld] Fix various issues with TypeAryPtr and TypeAryKlassPtr

[Quan Anh Mai]

On Fri, 21 Nov 2025 14:48:48 GMT, Tobias Hartmann <thartmann at openjdk.org> wrote:

>> Hi,
>> 
>> This PR fixes various issues with `TypeAryPtr` and `TypeAryKlassPtr`. It starts with me trying to tighten the properties (`flat`, `null_free`, etc) of these classes, then fixing all the revealed issues until there are no crashes or wrong results left.
>> 
>> Please take a look and leave your reviews, thanks a lot.
>
> Thanks for working on this!
> 
> Unfortunately, it triggers a lot of failures in testing. Here's a few, I can provide more details next week.
> 
> 
> compiler/c2/TestMergeStores.java#id0
> -ea -esa -XX:CompileThreshold=100 -XX:+UnlockExperimentalVMOptions -server -XX:+TieredCompilation
> 
> #  Error: assert(klass_is_exact()) failed
> 
> Current CompileTask:
> C2:6719 1466    b  4       compiler.c2.TestMergeStores::test600a (165 bytes)
> 
> Stack: [0x00007f992c11b000,0x00007f992c21b000],  sp=0x00007f992c216b40,  free space=1006k
> Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
> V  [libjvm.so+0xbc272a]  Compile::flatten_alias_type(TypePtr const*) const+0xeba  (type.hpp:1467)
> V  [libjvm.so+0xbc2f5c]  Compile::find_alias_type(TypePtr const*, bool, ciField*, bool) [clone .part.0]+0x8c  (compile.cpp:1677)
> V  [libjvm.so+0x1590623]  LibraryCallKit::inline_unsafe_access(bool, BasicType, LibraryCallKit::AccessKind, bool, bool)+0x9e3  (compile.hpp:1005)
> V  [libjvm.so+0x15c211a]  LibraryIntrinsic::generate(JVMState*)+0x22a  (library_call.cpp:130)
> V  [libjvm.so+0xdc5f4f]  Parse::do_call()+0xebf  (doCall.cpp:771)
> V  [libjvm.so+0x190ded8]  Parse::do_one_bytecode()+0x458  (parse2.cpp:3526)
> V  [libjvm.so+0x18f4f9c]  Parse::do_one_block()+0x24c  (parse1.cpp:1713)
> V  [libjvm.so+0x18f63a0]  Parse::do_all_blocks()+0x130  (parse1.cpp:770)
> V  [libjvm.so+0x18f9f01]  Parse::Parse(JVMState*, ciMethod*, float)+0xea1  (parse1.cpp:674)
> V  [libjvm.so+0x9e1205]  ParseGenerator::generate(JVMState*)+0x135  (callGenerator.cpp:99)
> V  [libjvm.so+0xbe7ee3]  Compile::Compile(ciEnv*, ciMethod*, int, Options, DirectiveSet*)+0x15e3  (compile.cpp:824)
> V  [libjvm.so+0x9ddd30]  C2Compiler::compile_method(ciEnv*, ciMethod*, int, bool, DirectiveSet*)+0x4c0  (c2compiler.cpp:149)
> V  [libjvm.so+0xbf7f40]  CompileBroker::invoke_compiler_on_method(CompileTask*)+0x780  (compileBroker.cpp:2345)
> V  [libjvm.so+0xbf97a0]  CompileBroker::compiler_thread_loop()+0x530  (compileBroker.cpp:1989)
> V  [libjvm.so+0x119c49b]  JavaThread::thread_main_inner()+0x13b  (javaThread.cpp:777)
> V  [libjvm.so+0x1c7c0b6]  Thread::call_run()+0xb6  (thread.cpp:242)
> V  [libjvm.so+0x18a2a78]  thread_native_entry(Thread*)+0x118  (os_linux.cpp:879)
> 
> 
> 
> compiler/codegen/TestRedundantLea.java#StoreNParallel
> -XX:UseAVX=0 -XX:UseSSE=3
> 
> Multiple IR failures
> 
> 
> 
> compiler/valhalla/inlinetypes/TestArrays.java
> -Xcomp -XX:-TieredCompilation -DIgnoreCompilerControls=true
> 
> 
> #  assert(_base >= KlassPtr && _base <= AryKlassPtr) failed: Not a klass po...

@TobiHartmann Thanks a lot, I have fixed the failure you mentioned here. It also seems to reveal an issue with C2 alias analysis.

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

PR Comment: https://git.openjdk.org/valhalla/pull/1755#issuecomment-3567731934