Merge pull request #253 from sysprog21/compound-literal

Consolidates initial compound literal improvements

Author: jserv

src/arm-codegen.c

@@ -136,9 +136,9 @@ void update_elf_offset(ph2_ir_t *ph2_ir)
 void cfg_flatten(void)
 {
     func_t *func = find_func("__syscall");
-    func->bbs->elf_offset = 44; /* offset of start + exit in codegen */
+    func->bbs->elf_offset = 48; /* offset of start + branch + exit in codegen */
 
-    elf_offset = 80; /* offset of start + exit + syscall in codegen */
+    elf_offset = 84; /* offset of start + branch + exit + syscall in codegen */
     GLOBAL_FUNC->bbs->elf_offset = elf_offset;
 
     for (ph2_ir_t *ph2_ir = GLOBAL_FUNC->bbs->ph2_ir_list.head; ph2_ir;
@@ -457,6 +457,8 @@ void code_generate(void)
     emit(__sub_r(__AL, __sp, __sp, __r8));
     emit(__mov_r(__AL, __r12, __sp));
     emit(__bl(__AL, GLOBAL_FUNC->bbs->elf_offset - elf_code->size));
+    /* After global init, jump to main preparation */
+    emit(__b(__AL, 56)); /* PC+8: skip exit (24) + syscall (36) + ret (4) - 8 */
 
     /* exit */
     emit(__movw(__AL, __r8, GLOBAL_FUNC->stack_size));
@@ -490,7 +492,7 @@ void code_generate(void)
     emit(__add_i(__AL, __r1, __r8, 4));
     emit(__bl(__AL, MAIN_BB->elf_offset - elf_code->size));
 
-    /* exit with main's return value */
+    /* exit with main's return value - r0 already has the return value */
     emit(__mov_i(__AL, __r7, 1));
     emit(__svc());
 

src/elf.c

@@ -21,11 +21,15 @@ void elf_write_str(strbuf_t *elf_array, char *vals)
      * If necessary, use elf_write_byte() to append the null character
      * after calling elf_write_str().
      */
+    if (!elf_array || !vals)
+        return;
     strbuf_puts(elf_array, vals);
 }
 
 void elf_write_byte(strbuf_t *elf_array, int val)
 {
+    if (!elf_array)
+        return;
     strbuf_putc(elf_array, val);
 }
 
@@ -36,19 +40,29 @@ char e_extract_byte(int v, int b)
 
 void elf_write_int(strbuf_t *elf_array, int val)
 {
+    if (!elf_array)
+        return;
     for (int i = 0; i < 4; i++)
         strbuf_putc(elf_array, e_extract_byte(val, i));
 }
 
 void elf_write_blk(strbuf_t *elf_array, void *blk, int sz)
 {
+    if (!elf_array || !blk || sz <= 0)
+        return;
     char *ptr = blk;
     for (int i = 0; i < sz; i++)
         strbuf_putc(elf_array, ptr[i]);
 }
 
 void elf_generate_header(void)
 {
+    /* Check for null pointers to prevent crashes */
+    if (!elf_code || !elf_data || !elf_symtab || !elf_strtab || !elf_header) {
+        error("ELF buffers not initialized");
+        return;
+    }
+
     elf32_hdr_t hdr;
     /*
      * The following table explains the meaning of each field in the
@@ -175,6 +189,12 @@ void elf_generate_header(void)
 
 void elf_generate_sections(void)
 {
+    /* Check for null pointers to prevent crashes */
+    if (!elf_symtab || !elf_strtab || !elf_section) {
+        error("ELF section buffers not initialized");
+        return;
+    }
+
     /* symtab section */
     for (int b = 0; b < elf_symtab->size; b++)
         elf_write_byte(elf_section, elf_symtab->elements[b]);
@@ -312,6 +332,12 @@ void elf_generate_sections(void)
 
 void elf_align(void)
 {
+    /* Check for null pointers to prevent crashes */
+    if (!elf_data || !elf_symtab || !elf_strtab) {
+        error("ELF buffers not initialized for alignment");
+        return;
+    }
+
     while (elf_data->size & 3)
         elf_write_byte(elf_data, 0);
 
@@ -324,6 +350,12 @@ void elf_align(void)
 
 void elf_add_symbol(char *symbol, int pc)
 {
+    /* Check for null pointers to prevent crashes */
+    if (!symbol || !elf_symtab || !elf_strtab) {
+        error("Invalid parameters for elf_add_symbol");
+        return;
+    }
+
     elf_write_int(elf_symtab, elf_strtab->size);
     elf_write_int(elf_symtab, pc);
     elf_write_int(elf_symtab, 0);
@@ -344,6 +376,11 @@ void elf_generate(char *outfile)
         outfile = "a.out";
 
     FILE *fp = fopen(outfile, "wb");
+    if (!fp) {
+        error("Unable to open output file for writing");
+        return;
+    }
+
     for (int i = 0; i < elf_header->size; i++)
         fputc(elf_header->elements[i], fp);
     for (int i = 0; i < elf_code->size; i++)

src/lexer.c

@@ -14,6 +14,12 @@
 #define NUM_DIRECTIVES 11
 #define NUM_KEYWORDS 16
 
+/* Token mapping structure for elegant initialization */
+typedef struct {
+    char *name;
+    token_t token;
+} token_mapping_t;
+
 /* Preprocessor directive hash table using existing shecc hashmap */
 hashmap_t *DIRECTIVE_MAP = NULL;
 /* C keywords hash table */
@@ -29,41 +35,25 @@ void lex_init_directives()
 
     DIRECTIVE_MAP = hashmap_create(16); /* Small capacity for directives */
 
-    /* Initialization using indexed for-loop */
+    /* Initialization using struct compound literals for elegance */
     directive_tokens_storage =
         arena_alloc(GENERAL_ARENA, NUM_DIRECTIVES * sizeof(token_t));
 
-    char *names[NUM_DIRECTIVES];
-    token_t token_values[NUM_DIRECTIVES];
-
-    /* Populate arrays using index-based assignments for compatibility */
-    names[0] = "#define";
-    token_values[0] = T_cppd_define;
-    names[1] = "#elif";
-    token_values[1] = T_cppd_elif;
-    names[2] = "#else";
-    token_values[2] = T_cppd_else;
-    names[3] = "#endif";
-    token_values[3] = T_cppd_endif;
-    names[4] = "#error";
-    token_values[4] = T_cppd_error;
-    names[5] = "#if";
-    token_values[5] = T_cppd_if;
-    names[6] = "#ifdef";
-    token_values[6] = T_cppd_ifdef;
-    names[7] = "#ifndef";
-    token_values[7] = T_cppd_ifndef;
-    names[8] = "#include";
-    token_values[8] = T_cppd_include;
-    names[9] = "#pragma";
-    token_values[9] = T_cppd_pragma;
-    names[10] = "#undef";
-    token_values[10] = T_cppd_undef;
+    /* Use array compound literal for directive mappings */
+    token_mapping_t directives[] = {
+        {"#define", T_cppd_define},   {"#elif", T_cppd_elif},
+        {"#else", T_cppd_else},       {"#endif", T_cppd_endif},
+        {"#error", T_cppd_error},     {"#if", T_cppd_if},
+        {"#ifdef", T_cppd_ifdef},     {"#ifndef", T_cppd_ifndef},
+        {"#include", T_cppd_include}, {"#pragma", T_cppd_pragma},
+        {"#undef", T_cppd_undef},
+    };
 
     /* hashmap insertion */
     for (int i = 0; i < NUM_DIRECTIVES; i++) {
-        directive_tokens_storage[i] = token_values[i];
-        hashmap_put(DIRECTIVE_MAP, names[i], &directive_tokens_storage[i]);
+        directive_tokens_storage[i] = directives[i].token;
+        hashmap_put(DIRECTIVE_MAP, directives[i].name,
+                    &directive_tokens_storage[i]);
     }
 }
 
@@ -74,51 +64,34 @@ void lex_init_keywords()
 
     KEYWORD_MAP = hashmap_create(32); /* Capacity for keywords */
 
-    /* Initialization using indexed for-loop */
+    /* Initialization using struct compound literals for elegance */
     keyword_tokens_storage =
         arena_alloc(GENERAL_ARENA, NUM_KEYWORDS * sizeof(token_t));
 
-    char *names[NUM_KEYWORDS];
-    token_t token_values[NUM_KEYWORDS];
-
-    /* Populate arrays using index-based assignments for compatibility */
-    names[0] = "if";
-    token_values[0] = T_if;
-    names[1] = "while";
-    token_values[1] = T_while;
-    names[2] = "for";
-    token_values[2] = T_for;
-    names[3] = "do";
-    token_values[3] = T_do;
-    names[4] = "else";
-    token_values[4] = T_else;
-    names[5] = "return";
-    token_values[5] = T_return;
-    names[6] = "typedef";
-    token_values[6] = T_typedef;
-    names[7] = "enum";
-    token_values[7] = T_enum;
-    names[8] = "struct";
-    token_values[8] = T_struct;
-    names[9] = "sizeof";
-    token_values[9] = T_sizeof;
-    names[10] = "switch";
-    token_values[10] = T_switch;
-    names[11] = "case";
-    token_values[11] = T_case;
-    names[12] = "break";
-    token_values[12] = T_break;
-    names[13] = "default";
-    token_values[13] = T_default;
-    names[14] = "continue";
-    token_values[14] = T_continue;
-    names[15] = "union";
-    token_values[15] = T_union;
+    /* Use array compound literal for keyword mappings */
+    token_mapping_t keywords[] = {
+        {"if", T_if},
+        {"while", T_while},
+        {"for", T_for},
+        {"do", T_do},
+        {"else", T_else},
+        {"return", T_return},
+        {"typedef", T_typedef},
+        {"enum", T_enum},
+        {"struct", T_struct},
+        {"sizeof", T_sizeof},
+        {"switch", T_switch},
+        {"case", T_case},
+        {"break", T_break},
+        {"default", T_default},
+        {"continue", T_continue},
+        {"union", T_union},
+    };
 
     /* hashmap insertion */
     for (int i = 0; i < NUM_KEYWORDS; i++) {
-        keyword_tokens_storage[i] = token_values[i];
-        hashmap_put(KEYWORD_MAP, names[i], &keyword_tokens_storage[i]);
+        keyword_tokens_storage[i] = keywords[i].token;
+        hashmap_put(KEYWORD_MAP, keywords[i].name, &keyword_tokens_storage[i]);
     }
 }