From 967a203414e528de7237fbfeb5c562010fd285d9 Mon Sep 17 00:00:00 2001
From: Yaossg <yaossg@outlook.com>
Date: Fri, 29 Nov 2024 11:16:58 +0800
Subject: [PATCH] string table and zero

---
 boot.c | 31 ++++++++++++++++++++++++++++---
 1 file changed, 28 insertions(+), 3 deletions(-)

diff --git a/boot.c b/boot.c
index 1b51e0e..91f6486 100644
--- a/boot.c
+++ b/boot.c
@@ -145,6 +145,24 @@ int parse_string() {
     return string_lut_size++;
 }
 
+void rewind_string(int new_data) {
+    string_offset = string_lut[token_data];
+    token_data = new_data;
+    --string_lut_size;
+}
+
+void dedup_string() {
+    int last_string = string_lut_size - 1;
+    char* latest = string_table + string_lut[last_string];
+    for (int i = 0; i < last_string; i++) {
+        char* candidate = string_table + string_lut[i];
+        if (streq(candidate, latest)) {
+            rewind_string(i);
+            return;
+        }
+    }
+}
+
 char id_table[65536];
 int id_offset;
 int id_lut[4096];
@@ -361,6 +379,7 @@ void next_token() {
     } else if (ch == '"') {
         token_type = TOKEN_STRING;
         token_data = parse_string();
+        dedup_string();
     } else if (ch == '.') {
         int ch2 = getchar();
         if (ch2 == '.') {
@@ -488,6 +507,7 @@ void reset_reg() {
         indirection[i] = 0;
         overflow[i] = 0;
     }
+    reg_type[REG_ZERO] = TYPE_INT;
 }
 
 const char* reg_name(int reg) {
@@ -764,17 +784,21 @@ void _asm_i(const char* op, int rd, const char* prefix1, const char* prefix2, in
     }
 }
 
+int is_not_reusable(int rs1, int expected_type) {
+    return indirection[rs1] || reg_type[rs1] != expected_type || rs1 == REG_ZERO;
+}
+
 int asm_r(int type, const char* op, int rs1) {
     int rd = rs1;
-    if (indirection[rs1] || reg_type[rs1] != type) rd = next_reg(type);
+    if (is_not_reusable(rs1, type)) rd = next_reg(type);
     _asm_r(op, rd, rs1);
     return rd;
 }
 
 int asm_rr(int type, const char* op, int rs1, int rs2) {
     int rd = rs1;
-    if (indirection[rs1] || reg_type[rs1] != type) rd = rs2;
-    if (indirection[rs2] || reg_type[rs2] != type) rd = next_reg(type);
+    if (is_not_reusable(rs1, type)) rd = rs2;
+    if (is_not_reusable(rs2, type)) rd = next_reg(type);
     _asm_rr(op, rd, rs1, rs2);
     return rd;
 }
@@ -1042,6 +1066,7 @@ int parse_primary_expr() {
     if (token_type == TOKEN_EOF) {
         exit(1);
     } else if (token_type == TOKEN_NUMBER) {
+        if (token_data == 0) return REG_ZERO;
         int reg = next_reg(TYPE_INT);
         _asm_i("li", reg, "", "", token_data);
         return reg;