[ruby/json] Allocate the initial generator buffer on the stack

Ref: https://github.com/ruby/json/issues/655
Followup: https://github.com/ruby/json/issues/657

Assuming the generator might be used for fairly small documents
we can start with a reasonable buffer size of the stack, and if
we outgrow it, we can spill on the heap.

In a way this is optimizing for micro-benchmarks, but there are
valid use case for fiarly small JSON document in actual real world
scenarios, so trashing the GC less in such case make sense.

Before:

```
ruby 3.3.4 (2024-07-09 revision https://github.com/ruby/json/commit/be1089c8ec) +YJIT [arm64-darwin23]
Warming up --------------------------------------
                  Oj   518.700k i/100ms
          JSON reuse   483.370k i/100ms
Calculating -------------------------------------
                  Oj      5.722M (± 1.8%) i/s  (174.76 ns/i) -     29.047M in   5.077823s
          JSON reuse      5.278M (± 1.5%) i/s  (189.46 ns/i) -     26.585M in   5.038172s

Comparison:
                  Oj:  5722283.8 i/s
          JSON reuse:  5278061.7 i/s - 1.08x  slower
```

After:

```
ruby 3.3.4 (2024-07-09 revision https://github.com/ruby/json/commit/be1089c8ec) +YJIT [arm64-darwin23]
Warming up --------------------------------------
                  Oj   517.837k i/100ms
          JSON reuse   548.871k i/100ms
Calculating -------------------------------------
                  Oj      5.693M (± 1.6%) i/s  (175.65 ns/i) -     28.481M in   5.004056s
          JSON reuse      5.855M (± 1.2%) i/s  (170.80 ns/i) -     29.639M in   5.063004s

Comparison:
                  Oj:  5692985.6 i/s
          JSON reuse:  5854857.9 i/s - 1.03x  faster
```

https://github.com/ruby/json/commit/fe607f4806

5 files changed, 30 insertions, 7 deletions

diff --git a/ext/json/fbuffer/fbuffer.h b/ext/json/fbuffer/fbuffer.h
index dc09dde9cb..55fc0bba9d 100644
--- a/ext/json/fbuffer/fbuffer.h
+++ b/ext/json/fbuffer/fbuffer.h
@@ -4,13 +4,20 @@
 #include "ruby.h"
 #include "ruby/encoding.h"
 
+enum fbuffer_type {
+    HEAP = 0,
+    STACK = 1,
+};
+
 typedef struct FBufferStruct {
+    enum fbuffer_type type;
     unsigned long initial_length;
-    char *ptr;
     unsigned long len;
     unsigned long capa;
+    char *ptr;
 } FBuffer;
 
+#define FBUFFER_STACK_SIZE 512
 #define FBUFFER_INITIAL_LENGTH_DEFAULT 1024
 
 #define FBUFFER_PTR(fb) ((fb)->ptr)
@@ -35,14 +42,21 @@ static VALUE fbuffer_to_s(FBuffer *fb);
 #define RB_UNLIKELY(expr) expr
 #endif
 
-static void fbuffer_init(FBuffer *fb, unsigned long initial_length)
+static void fbuffer_stack_init(FBuffer *fb, unsigned long initial_length, char *stack_buffer, long stack_buffer_size)
 {
     fb->initial_length = (initial_length > 0) ? initial_length : FBUFFER_INITIAL_LENGTH_DEFAULT;
+    if (stack_buffer) {
+        fb->type = STACK;
+        fb->ptr = stack_buffer;
+        fb->capa = stack_buffer_size;
+    }
 }
 
 static void fbuffer_free(FBuffer *fb)
 {
-    if (fb->ptr) ruby_xfree(fb->ptr);
+    if (fb->ptr && fb->type == HEAP) {
+        ruby_xfree(fb->ptr);
+    }
 }
 
 #ifndef JSON_GENERATOR
@@ -65,7 +79,14 @@ static inline void fbuffer_inc_capa(FBuffer *fb, unsigned long requested)
         for (required = fb->capa; requested > required - fb->len; required <<= 1);
 
         if (required > fb->capa) {
-            REALLOC_N(fb->ptr, char, required);
+            if (fb->type == STACK) {
+                const char *old_buffer = fb->ptr;
+                fb->ptr = ALLOC_N(char, required);
+                fb->type = HEAP;
+                MEMCPY(fb->ptr, old_buffer, char, fb->len);
+            } else {
+                REALLOC_N(fb->ptr, char, required);
+            }
             fb->capa = required;
         }
     }
diff --git a/ext/json/generator/generator.c b/ext/json/generator/generator.c
index 4e1c0e314e..fc002ff737 100644
--- a/ext/json/generator/generator.c
+++ b/ext/json/generator/generator.c
@@ -954,8 +954,9 @@ static VALUE cState_partial_generate(VALUE self, VALUE obj, void (*func)(FBuffer
 {
     GET_STATE(self);
 
+    char stack_buffer[FBUFFER_STACK_SIZE];
     FBuffer buffer = {0};
-    fbuffer_init(&buffer, state->buffer_initial_length);
+    fbuffer_stack_init(&buffer, state->buffer_initial_length, stack_buffer, FBUFFER_STACK_SIZE);
 
     struct generate_json_data data = {
         .buffer = &buffer,
diff --git a/ext/json/generator/generator.h b/ext/json/generator/generator.h
index 05e8f0d809..b6898c955f 100644
--- a/ext/json/generator/generator.h
+++ b/ext/json/generator/generator.h
@@ -54,6 +54,7 @@ typedef struct JSON_Generator_StateStruct {
     JSON_Generator_State *state;              \
     GET_STATE_TO(self, state)
 
+
 static VALUE mHash_to_json(int argc, VALUE *argv, VALUE self);
 static VALUE mArray_to_json(int argc, VALUE *argv, VALUE self);
 #ifdef RUBY_INTEGER_UNIFICATION
diff --git a/ext/json/parser/parser.c b/ext/json/parser/parser.c
index 368d539bcd..a6d8ff2cc5 100644
--- a/ext/json/parser/parser.c
+++ b/ext/json/parser/parser.c
@@ -2159,7 +2159,7 @@ static VALUE cJSON_parser_s_allocate(VALUE klass)
 {
     JSON_Parser *json;
     VALUE obj = TypedData_Make_Struct(klass, JSON_Parser, &JSON_Parser_type, json);
-    fbuffer_init(&json->fbuffer, 0);
+    fbuffer_stack_init(&json->fbuffer, 0, NULL, 0);
     return obj;
 }
 
diff --git a/ext/json/parser/parser.rl b/ext/json/parser/parser.rl
index c627014593..499cc693a3 100644
--- a/ext/json/parser/parser.rl
+++ b/ext/json/parser/parser.rl
@@ -919,7 +919,7 @@ static VALUE cJSON_parser_s_allocate(VALUE klass)
 {
     JSON_Parser *json;
     VALUE obj = TypedData_Make_Struct(klass, JSON_Parser, &JSON_Parser_type, json);
-    fbuffer_init(&json->fbuffer, 0);
+    fbuffer_stack_init(&json->fbuffer, 0, NULL, 0);
     return obj;
 }