[sanitizer] Move out stack trace pointer from header StackDepot

Trace pointers accessed very rarely and don't need to
be in hot data.

Depends on D111613.

Reviewed By: dvyukov

Differential Revision: https://reviews.llvm.org/D111614

compiler-rt/lib/sanitizer_common/sanitizer_chained_origin_depot.cpp

@@ -38,9 +38,9 @@ struct ChainedOriginDepotNode {
 
   static bool is_valid(const args_type &args);
 
-  void store(const args_type &args, hash_type other_hash);
+  void store(u32 id, const args_type &args, hash_type other_hash);
 
-  args_type load() const;
+  args_type load(u32 id) const;
 
   struct Handle {
     const ChainedOriginDepotNode *node_ = nullptr;
@@ -106,13 +106,13 @@ ChainedOriginDepotNode::hash_type ChainedOriginDepotNode::hash(
 
 bool ChainedOriginDepotNode::is_valid(const args_type &args) { return true; }
 
-void ChainedOriginDepotNode::store(const args_type &args,
+void ChainedOriginDepotNode::store(u32 id, const args_type &args,
                                    hash_type other_hash) {
   here_id = args.here_id;
   prev_id = args.prev_id;
 }
 
-ChainedOriginDepotNode::args_type ChainedOriginDepotNode::load() const {
+ChainedOriginDepotNode::args_type ChainedOriginDepotNode::load(u32 id) const {
   args_type ret = {here_id, prev_id};
   return ret;
 }

compiler-rt/lib/sanitizer_common/sanitizer_stackdepot.cpp

@@ -24,7 +24,6 @@ static PersistentAllocator<uptr> traceAllocator;
 struct StackDepotNode {
   using hash_type = u64;
   hash_type stack_hash;
-  uptr *stack_trace;
   u32 link;
   atomic_uint32_t tag_and_use_count;  // tag : 12 high bits; use_count : 20;
 
@@ -47,22 +46,8 @@ struct StackDepotNode {
   static bool is_valid(const args_type &args) {
     return args.size > 0 && args.trace;
   }
-  void store(const args_type &args, hash_type hash) {
-    CHECK_EQ(args.tag & (~kUseCountMask >> kUseCountBits), args.tag);
-    atomic_store(&tag_and_use_count, args.tag << kUseCountBits,
-                 memory_order_relaxed);
-    stack_hash = hash;
-    stack_trace = traceAllocator.alloc(args.size + 1);
-    *stack_trace = args.size;
-    internal_memcpy(stack_trace + 1, args.trace, args.size * sizeof(uptr));
-  }
-  args_type load() const {
-    if (!stack_trace)
-      return {};
-    u32 tag =
-        atomic_load(&tag_and_use_count, memory_order_relaxed) >> kUseCountBits;
-    return args_type(stack_trace + 1, *stack_trace, tag);
-  }
+  void store(u32 id, const args_type &args, hash_type hash);
+  args_type load(u32 id) const;
   static StackDepotHandle get_handle(u32 id);
 
   typedef StackDepotHandle handle_type;
@@ -85,6 +70,30 @@ void StackDepotHandle::inc_use_count_unsafe() {
 typedef StackDepotBase<StackDepotNode, 1, StackDepotNode::kTabSizeLog>
     StackDepot;
 static StackDepot theDepot;
+// Keep rarely accessed stack traces out of frequently access nodes to improve
+// caching efficiency.
+static TwoLevelMap<uptr *, StackDepot::kNodesSize1, StackDepot::kNodesSize2>
+    tracePtrs;
+
+void StackDepotNode::store(u32 id, const args_type &args, hash_type hash) {
+  CHECK_EQ(args.tag & (~kUseCountMask >> kUseCountBits), args.tag);
+  atomic_store(&tag_and_use_count, args.tag << kUseCountBits,
+               memory_order_relaxed);
+  stack_hash = hash;
+  uptr *stack_trace = traceAllocator.alloc(args.size + 1);
+  *stack_trace = args.size;
+  internal_memcpy(stack_trace + 1, args.trace, args.size * sizeof(uptr));
+  tracePtrs[id] = stack_trace;
+}
+
+StackDepotNode::args_type StackDepotNode::load(u32 id) const {
+  const uptr *stack_trace = tracePtrs[id];
+  if (!stack_trace)
+    return {};
+  u32 tag =
+      atomic_load(&tag_and_use_count, memory_order_relaxed) >> kUseCountBits;
+  return args_type(stack_trace + 1, *stack_trace, tag);
+}
 
 StackDepotStats StackDepotGetStats() { return theDepot.GetStats(); }
 

compiler-rt/lib/sanitizer_common/sanitizer_stackdepotbase.h

@@ -135,7 +135,7 @@ u32 StackDepotBase<Node, kReservedBits, kTabSizeLog>::Put(args_type args,
   CHECK_EQ(s & kUnlockMask, s);
   CHECK_EQ(s & (((u32)-1) >> kReservedBits), s);
   Node &new_node = nodes[s];
-  new_node.store(args, h);
+  new_node.store(s, args, h);
   new_node.link = s2;
   unlock(p, s);
   if (inserted) *inserted = true;
@@ -151,7 +151,7 @@ StackDepotBase<Node, kReservedBits, kTabSizeLog>::Get(u32 id) {
   if (!nodes.contains(id))
     return args_type();
   const Node &node = nodes[id];
-  return node.load();
+  return node.load(id);
 }
 
 template <class Node, int kReservedBits, int kTabSizeLog>
@@ -178,7 +178,7 @@ void StackDepotBase<Node, kReservedBits, kTabSizeLog>::PrintAll() {
     for (; s;) {
       const Node &node = nodes[s];
       Printf("Stack for id %u:\n", s);
-      node.load().Print();
+      node.load(s).Print();
       s = node.link;
     }
   }