feat: add defrag support for small strings

src/core/compact_object.cc

@@ -218,7 +218,7 @@ struct TL {
 
 thread_local TL tl;
 
-constexpr bool kUseSmallStrings = false;
+constexpr bool kUseSmallStrings = true;
 
 /// TODO: Ascii encoding becomes slow for large blobs. We should factor it out into a separate
 /// file and implement with SIMD instructions.
@@ -877,8 +877,7 @@ bool CompactObj::DefragIfNeeded(float ratio) {
       }
       return false;
     case SMALL_TAG:
-      // TODO - support this later
-      return false;
+      return u_.small_str.DefragIfNeeded(ratio);
     case INT_TAG:
       // this is not relevant in this case
       return false;

src/core/small_string.cc

@@ -4,6 +4,7 @@
 
 #include "core/small_string.h"
 
+#include <mimalloc.h>
 #include <xxhash.h>
 
 #include <memory>
@@ -155,6 +156,7 @@ void SmallString::Get(std::string* dest) const {
 }
 
 unsigned SmallString::GetV(string_view dest[2]) const {
+  DCHECK_GT(size_, kPrefLen);
   if (size_ <= kPrefLen) {
     dest[0] = string_view{prefix_, size_};
     return 1;
@@ -166,4 +168,23 @@ unsigned SmallString::GetV(string_view dest[2]) const {
   return 2;
 }
 
+bool SmallString::DefragIfNeeded(float ratio) {
+  DCHECK_GT(size_, kPrefLen);
+  if (size_ <= kPrefLen) {
+    return false;
+  }
+
+  uint8_t* cur_real_ptr = tl.seg_alloc->Translate(small_ptr_);
+  if (!mi_heap_page_is_underutilized(tl.seg_alloc->heap(), cur_real_ptr, ratio))
+    return false;
+
+  auto [sp, rp] = tl.seg_alloc->Allocate(size_ - kPrefLen);
+
+  memcpy(rp, cur_real_ptr, size_ - kPrefLen);
+  tl.seg_alloc->Free(small_ptr_);
+  small_ptr_ = sp;
+
+  return true;
+}
+
 }  // namespace dfly

src/core/small_string.h

@@ -17,8 +17,7 @@ class SmallString {
   static constexpr unsigned kPrefLen = 10;
 
  public:
-
-  static void InitThreadLocal(void * heap);
+  static void InitThreadLocal(void* heap);
   static size_t UsedThreadLocal();
 
   void Reset() {
@@ -48,6 +47,8 @@ class SmallString {
   // With current implementation, it will return 2 slices for a non-empty string.
   unsigned GetV(std::string_view dest[2]) const;
 
+  bool DefragIfNeeded(float ratio);
+
  private:
   // prefix of the string that is broken down into 2 parts.
   char prefix_[kPrefLen];