8322043: HeapDumper should use parallel dump by default

Reviewed-by: yyang, sspitsyn, dholmes

src/hotspot/share/services/attachListener.cpp

@@ -254,7 +254,7 @@ static jint dump_heap(AttachOperation* op, outputStream* out) {
     // This helps reduces the amount of unreachable objects in the dump
     // and makes it easier to browse.
     HeapDumper dumper(live_objects_only /* request GC */);
-    dumper.dump(path, out, level, false, HeapDumper::default_num_of_dump_threads());
+    dumper.dump(path, out, level);
   }
   return JNI_OK;
 }

src/hotspot/share/services/heapDumper.cpp

@@ -2604,6 +2604,18 @@ int HeapDumper::dump(const char* path, outputStream* out, int compression, bool
     out->print_cr("Dumping heap to %s ...", path);
     timer()->start();
   }
+
+  if (_oome && num_dump_threads > 1) {
+    // Each additional parallel writer requires several MB of internal memory
+    // (DumpWriter buffer, DumperClassCacheTable, GZipCompressor buffers).
+    // For the OOM handling we may already be limited in memory.
+    // Lets ensure we have at least 20MB per thread.
+    julong max_threads = os::free_memory() / (20 * M);
+    if (num_dump_threads > max_threads) {
+      num_dump_threads = MAX2<uint>(1, (uint)max_threads);
+    }
+  }
+
   // create JFR event
   EventHeapDump event;
 

src/hotspot/share/services/heapDumper.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2005, 2023, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2005, 2024, 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
@@ -60,8 +60,8 @@ class HeapDumper : public StackObj {
   // dumps the heap to the specified file, returns 0 if success.
   // additional info is written to out if not null.
   // compression >= 0 creates a gzipped file with the given compression level.
-  // parallel_thread_num >= 0 indicates thread numbers of parallel object dump
-  int dump(const char* path, outputStream* out = nullptr, int compression = -1, bool overwrite = false, uint parallel_thread_num = 1);
+  // parallel_thread_num >= 0 indicates thread numbers of parallel object dump.
+  int dump(const char* path, outputStream* out = nullptr, int compression = -1, bool overwrite = false, uint parallel_thread_num = default_num_of_dump_threads());
 
   // returns error message (resource allocated), or null if no error
   char* error_as_C_string() const;