8371964: C2 compilation asserts with "Unexpected load/store size"

Reviewed-by: chagedorn, epeter

Author: merykitty

src/hotspot/share/opto/vectornode.cpp

@@ -22,6 +22,8 @@
  */
 
 #include "memory/allocation.inline.hpp"
+#include "opto/c2_globals.hpp"
+#include "opto/compile.hpp"
 #include "opto/connode.hpp"
 #include "opto/convertnode.hpp"
 #include "opto/mulnode.hpp"
@@ -1145,7 +1147,14 @@ Node* LoadVectorMaskedNode::Ideal(PhaseGVN* phase, bool can_reshape) {
     if (ty && ty->is_con()) {
       BasicType mask_bt = Matcher::vector_element_basic_type(in(3));
       int load_sz = type2aelembytes(mask_bt) * ty->get_con();
-      assert(load_sz <= MaxVectorSize, "Unexpected load size");
+      if (load_sz > MaxVectorSize) {
+        // After loop opts, cast nodes are aggressively removed, if the input is then transformed
+        // into a constant that is outside the range of the removed cast, we may encounter it here.
+        // This should be a dead node then.
+        assert(Compile::current()->post_loop_opts_phase(), "Unexpected load size");
+        return phase->C->top();
+      }
+
       if (load_sz == MaxVectorSize) {
         Node* ctr = in(MemNode::Control);
         Node* mem = in(MemNode::Memory);
@@ -1164,7 +1173,14 @@ Node* StoreVectorMaskedNode::Ideal(PhaseGVN* phase, bool can_reshape) {
     if (ty && ty->is_con()) {
       BasicType mask_bt = Matcher::vector_element_basic_type(in(4));
       int load_sz = type2aelembytes(mask_bt) * ty->get_con();
-      assert(load_sz <= MaxVectorSize, "Unexpected store size");
+      if (load_sz > MaxVectorSize) {
+        // After loop opts, cast nodes are aggressively removed, if the input is then transformed
+        // into a constant that is outside the range of the removed cast, we may encounter it here.
+        // This should be a dead node then.
+        assert(Compile::current()->post_loop_opts_phase(), "Unexpected store size");
+        return phase->C->top();
+      }
+
       if (load_sz == MaxVectorSize) {
         Node* ctr = in(MemNode::Control);
         Node* mem = in(MemNode::Memory);

test/hotspot/jtreg/compiler/arraycopy/TestArrayCopyDisjoint.java

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020, 2022, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2020, 2025, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -26,7 +26,7 @@
 
 /**
  * @test
- * @bug 8251871 8285301
+ * @bug 8251871 8285301 8371964
  * @summary Optimize arrayCopy using AVX-512 masked instructions.
  *
  * @run main/othervm/timeout=600 -XX:-TieredCompilation -Xbatch -XX:+IgnoreUnrecognizedVMOptions
@@ -52,6 +52,9 @@
  *      compiler.arraycopy.TestArrayCopyDisjoint
  * @run main/othervm/timeout=600 -XX:-TieredCompilation -Xbatch -XX:+UnlockExperimentalVMOptions -XX:+AlwaysAtomicAccesses
  *      compiler.arraycopy.TestArrayCopyDisjoint
+ * @run main/othervm/timeout=600 -XX:-TieredCompilation -Xbatch -XX:+IgnoreUnrecognizedVMOptions
+ *      -XX:+UnlockDiagnosticVMOptions -XX:UseAVX=3 -XX:MaxVectorSize=32 -XX:ArrayOperationPartialInlineSize=32 -XX:+StressIGVN
+ *      compiler.arraycopy.TestArrayCopyDisjoint
  *
  */