Implement single byte OP_WRITE_RAW instruction

Shopify · Oct 16, 2020 · e85d5eb · e85d5eb
1 parent c95ba50
commit e85d5eb
Show file tree

Hide file tree

Showing 4 changed files with 60 additions and 14 deletions.
diff --git a/ext/liquid_c/block.c b/ext/liquid_c/block.c
@@ -279,10 +279,15 @@ static VALUE block_body_remove_blank_strings(VALUE self)
 
     while (*ip != OP_LEAVE) {
         if (*ip == OP_WRITE_RAW) {
-            if (ip[1] || ip[2] || ip[3]) { // if (size != 0)
+            if (ip[1]) { // if (size != 0)
                 ip[0] = OP_JUMP_FWD; // effectively a no-op
                 body->render_score--;
             }
+        } else if (*ip == OP_WRITE_RAW_W) {
+            if (ip[1] || ip[2] || ip[3]) { // if (size != 0)
+                ip[0] = OP_JUMP_FWD_W; // effectively a no-op
+                body->render_score--;
+            }
         }
         liquid_vm_next_instruction((const uint8_t **)&ip, (const size_t **)&const_ptr);
     }
@@ -320,10 +325,18 @@ static VALUE block_body_nodelist(VALUE self)
         switch (*ip) {
             case OP_LEAVE:
                 goto loop_break;
+            case OP_WRITE_RAW_W:
             case OP_WRITE_RAW:
             {
-                const char *text = (const char *)&ip[4];
-                size_t size = bytes_to_uint24(&ip[1]);
+                const char *text;
+                size_t size;
+                if (*ip == OP_WRITE_RAW_W) {
+                    size = bytes_to_uint24(&ip[1]);
+                    text = (const char *)&ip[4];
+                } else {
+                    size = ip[1];
+                    text = (const char *)&ip[2];
+                }
                 VALUE string = rb_enc_str_new(text, size, utf8_encoding);
                 rb_ary_push(nodelist, string);
                 break;

diff --git a/ext/liquid_c/vm.c b/ext/liquid_c/vm.c
@@ -356,22 +356,38 @@ static VALUE vm_render_until_error(VALUE uncast_args)
 
             // Rendering instructions
 
+            case OP_WRITE_RAW_W:
             case OP_WRITE_RAW:
             {
-                const char *text = (const char *)&ip[3];
-                size_t size = bytes_to_uint24(ip);
+                const char *text;
+                size_t size;
+                if (ip[-1] == OP_WRITE_RAW_W) {
+                    size = bytes_to_uint24(ip);
+                    text = (const char *)&ip[3];
+                    ip += 3 + size;
+                } else {
+                    size = *ip;
+                    text = (const char *)&ip[1];
+                    ip += 1 + size;
+                }
                 rb_str_cat(output, text, size);
                 resource_limits_increment_write_score(vm->resource_limits, output);
-                ip += 3 + size;
                 break;
             }
-            case OP_JUMP_FWD:
+            case OP_JUMP_FWD_W:
             {
                 size_t size = bytes_to_uint24(ip);
                 ip += 3 + size;
                 break;
             }
 
+            case OP_JUMP_FWD:
+            {
+                uint8_t size = *ip;
+                ip += 1 + size;
+                break;
+            }
+
             case OP_WRITE_NODE:
                 rb_funcall(cLiquidBlockBody, id_render_node, 3, args->context, output, (VALUE)*const_ptr++);
                 if (RARRAY_LEN(vm->interrupts)) {
@@ -468,14 +484,22 @@ void liquid_vm_next_instruction(const uint8_t **ip_ptr, const size_t **const_ptr
             (*const_ptr_ptr)++;
             break;
 
-        case OP_WRITE_RAW:
-        case OP_JUMP_FWD:
+        case OP_WRITE_RAW_W:
+        case OP_JUMP_FWD_W:
         {
             size_t size = bytes_to_uint24(ip);
             ip += 3 + size;
             break;
         }
 
+        case OP_WRITE_RAW:
+        case OP_JUMP_FWD:
+        {
+            uint8_t size = *ip;
+            ip += 1 + size;
+            break;
+        }
+
         default:
             rb_bug("invalid opcode: %u", ip[-1]);
     }

diff --git a/ext/liquid_c/vm_assembler.c b/ext/liquid_c/vm_assembler.c
@@ -22,11 +22,18 @@ void vm_assembler_gc_mark(vm_assembler_t *code)
 
 void vm_assembler_add_write_raw(vm_assembler_t *code, const char *string, size_t size)
 {
-    uint8_t instructions[4];
-    instructions[0] = OP_WRITE_RAW;
-    uint24_to_bytes((unsigned int)size, &instructions[1]);
+    if (size > UINT8_MAX) {
+        uint8_t instructions[4];
+        instructions[0] = OP_WRITE_RAW_W;
+        uint24_to_bytes((unsigned int)size, &instructions[1]);
+
+        c_buffer_write(&code->instructions, &instructions, sizeof(instructions));
+    } else {
+        uint8_t instructions[2] = { OP_WRITE_RAW, size };
+
+        c_buffer_write(&code->instructions, &instructions, sizeof(instructions));
+    }
 
-    c_buffer_write(&code->instructions, &instructions, sizeof(instructions));
     c_buffer_write(&code->instructions, (char *)string, size);
 }
 

diff --git a/ext/liquid_c/vm_assembler.h b/ext/liquid_c/vm_assembler.h
@@ -8,7 +8,7 @@
 
 enum opcode {
     OP_LEAVE = 0,
-    OP_WRITE_RAW = 1,
+    OP_WRITE_RAW_W = 1,
     OP_WRITE_NODE = 2,
     OP_POP_WRITE_VARIABLE,
     OP_PUSH_CONST,
@@ -26,6 +26,8 @@ enum opcode {
     OP_HASH_NEW, // rb_hash_new & rb_hash_bulk_insert
     OP_FILTER,
     OP_RENDER_VARIABLE_RESCUE, // setup state to rescue variable rendering
+    OP_WRITE_RAW,
+    OP_JUMP_FWD_W,
     OP_JUMP_FWD,
 };