feat: Allow linking with extern vars, fixes #58

examples/inp_extern.c

@@ -0,0 +1,11 @@
+void fprintf(void *, char *, ...);
+
+extern void *stderr;
+extern void *stdout;
+
+int main()
+{
+    fprintf(stdout, "This is stdout!\n");
+    fprintf(stderr, "This is stderr!\n");
+    return 0;
+}

lexer.c

@@ -123,6 +123,8 @@ typed_token *next_keyword_or_identifier(char **inp_ptr)
             return new_simp_tkn(TKN_CASE);
         else if (strcmp(val, "default") == 0)
             return new_simp_tkn(TKN_DEFAULT);
+        else if (strcmp(val, "extern") == 0)
+            return new_simp_tkn(TKN_EXTERN);
         else
             return new_tkn(TKN_ID, val, ident_tkn_debug);
     }

lexer.h

@@ -27,6 +27,7 @@ typed_token *tokenize(char *inp);
 #define TKN_SIZEOF 11
 #define TKN_TYPEDEF 12
 #define TKN_GOTO 13
+#define TKN_EXTERN 14
 
 // Single letter symbols
 #define TKN_L_PAREN 16

parser/extern.c

@@ -0,0 +1,53 @@
+#include "extern.h"
+#include <stdlib.h>
+#include <string.h>
+#include "../codegen/codegen.h"
+
+#include "parser.h"
+#include "var_decl.h"
+#include "type.h"
+
+void extern_debug(int depth, parser_node *node)
+{
+    node_extern *ex = (node_extern *)node->data;
+    printtabs(depth);
+    printf("Extern:\n");
+    ex->var_decl->debug(depth + 1, ex->var_decl);
+}
+
+apply_result *extern_apply(parser_node *node, context *ctx)
+{
+    node_extern *ex = (node_extern *)node->data;
+    node_var_decl *vd = ex->var_decl->data;
+    node_type *tp = vd->type->data;
+
+    add_text(ctx, "extern %s", vd->identity);
+    new_global_symbol(ctx, vd->identity, cc_asprintf("[%s]", vd->identity), tp->type);
+
+    return NULL;
+}
+
+parser_node *parse_extern(typed_token **tkns_ptr)
+{
+    typed_token *tkn = *tkns_ptr;
+    if (tkn->type_id == TKN_EXTERN)
+    {
+        tkn = tkn->next;
+        parser_node *decl = parse_var_decl(&tkn);
+        if (decl)
+        {
+            *tkns_ptr = tkn;
+
+            parser_node *node = (parser_node *)malloc(sizeof(parser_node));
+            node->data = (void *)malloc(sizeof(node_extern));
+            node->debug = extern_debug;
+            node->apply = extern_apply;
+            node_extern *ex = (node_extern *)node->data;
+            ex->var_decl = decl;
+
+            return node;
+        }
+    }
+
+    return NULL;
+}

parser/extern.h

@@ -0,0 +1,14 @@
+#ifndef EXTERN_H
+#define EXTERN_H
+
+#include "../lexer.h"
+#include "parser.h"
+
+parser_node *parse_extern(typed_token **tkns_ptr);
+
+typedef struct
+{
+    parser_node *var_decl;
+} node_extern;
+
+#endif

parser/program.c

@@ -12,12 +12,18 @@
 #include "../linked_list.h"
 #include "struct_def.h"
 #include "type.h"
+#include "extern.h"
 
 void program_debug(int depth, parser_node *node)
 {
     node_program *prog = (node_program *)node->data;
     printtabs(depth);
     printf("Program(\n");
+    for (int i = 0; i < prog->externs->total; i++)
+    {
+        parser_node *node = (parser_node *)get_vec(prog->externs, i);
+        node->debug(depth + 1, node);
+    }
     for (int i = 0; i < prog->num_functions; i++)
     {
         parser_node *node = prog->functions[i];
@@ -35,6 +41,11 @@ void program_debug(int depth, parser_node *node)
 apply_result *program_apply(parser_node *node, context *ctx)
 {
     node_program *prog = (node_program *)node->data;
+    for (int i = 0; i < prog->externs->total; i++)
+    {
+        parser_node *node = (parser_node *)get_vec(prog->externs, i);
+        node->apply(node, ctx);
+    }
     for (int i = 0; i < prog->num_struct_defs; i++)
     {
         prog->struct_defs[i]->apply(prog->struct_defs[i], ctx);
@@ -61,7 +72,7 @@ apply_result *program_apply(parser_node *node, context *ctx)
         int total = ctx->stack_size;
         // 16 byte stack alignment
         total = total + (16 - total % 16);
-        if(nfd->statements)
+        if (nfd->statements)
             add_data(ctx, "__%s_size: equ %u", ((node_func_def *)node->data)->identity, total);
     }
     add_text(ctx, "extern exit");
@@ -86,7 +97,8 @@ parser_node *parse_program(typed_token **tkn_ptr)
     int var_decl_count = 0;
     parser_node **funcs = (parser_node **)malloc(sizeof(parser_node *) * 128);
     parser_node **struct_defs = (parser_node **)malloc(sizeof(parser_node *) * 128);
-    vector* var_decls = initialize_vec(sizeof(parser_node*));
+    vector *externs = initialize_vec(sizeof(parser_node *));
+    vector *var_decls = initialize_vec(sizeof(parser_node *));
     typed_token *tkn = *tkn_ptr;
     while (tkn)
     {
@@ -97,7 +109,7 @@ parser_node *parse_program(typed_token **tkn_ptr)
             node->debug = program_debug;
             node->apply = program_apply;
             node_program *prog = (node_program *)node->data;
-
+            prog->externs = externs;
             prog->num_functions = func_count;
             prog->functions = funcs;
             prog->num_struct_defs = struct_def_count;
@@ -118,6 +130,10 @@ parser_node *parse_program(typed_token **tkn_ptr)
         {
             push_vec(var_decls, f);
         }
+        else if ((f = parse_extern(&tkn)))
+        {
+            push_vec(externs, f);
+        }
         else
         {
             return NULL;

parser/program.h

@@ -3,6 +3,7 @@
 
 #include "../lexer.h"
 #include "parser.h"
+#include "../vec.h"
 
 parser_node *parse_program(typed_token **tkns_ptr);
 
@@ -12,6 +13,7 @@ typedef struct
     parser_node **functions;
     int num_struct_defs;
     parser_node **struct_defs;
+    vector *externs;
 } node_program;
 
 #endif

scripts/test.py

@@ -25,6 +25,7 @@
     "./examples/inp_loop.c": [],
     "./examples/lots_of_variables.c": [],
     "./examples/inp_arg_macro.c": [],
+    "./examples/inp_extern.c": [],
 }
 C_PROGRAM_NAME = "./a.out"
 OUTPUT_FOLDER = "tests/output"

tests/output/inp_extern.c_asm_output.asm

@@ -0,0 +1,56 @@
+section .data
+__temp_str_0 db `This is stdout!\n`, 0
+__temp_str_1 db `This is stderr!\n`, 0
+__main_size: equ 80
+section .text
+extern stderr
+extern stdout
+extern fprintf
+global main
+main:
+push rbp
+mov rbp, rsp
+sub rsp, __main_size
+mov rax, rsp
+add rax, 0
+mov [rsp+0], rax
+mov rax, [stdout]
+mov [rsp+8], rax
+mov rax, __temp_str_0
+mov [rsp+16], rax
+mov rdi, [rsp+8]
+mov rsi, [rsp+16]
+mov rax, rsp
+add rax, 0
+mov [rsp+24], rax
+call fprintf
+mov rax, rsp
+add rax, 0
+mov [rsp+32], rax
+mov rax, [stderr]
+mov [rsp+40], rax
+mov rax, __temp_str_1
+mov [rsp+48], rax
+mov rdi, [rsp+40]
+mov rsi, [rsp+48]
+mov rax, rsp
+add rax, 0
+mov [rsp+56], rax
+call fprintf
+mov rax, 0
+mov rsp, rbp
+pop rbp
+ret
+mov rsp, rbp
+pop rbp
+ret
+extern exit
+global _start
+_start:
+; Pass argc and argv
+mov rdi, [rsp]
+mov rsi, rsp
+add rsi, 8
+call main
+mov rdi, rax
+call exit

tests/output/inp_extern.c_lex_output.txt

@@ -0,0 +1,48 @@
+TKN_VOID
+TKN_ID(fprintf)
+TKN_L_PAREN
+TKN_VOID
+TKN_STAR
+TKN_COMMA
+TKN_CHAR
+TKN_STAR
+TKN_COMMA
+TKN_DOTS
+TKN_R_PAREN
+TKN_SEMICOLON
+TKN_EXTERN
+TKN_VOID
+TKN_STAR
+TKN_ID(stderr)
+TKN_SEMICOLON
+TKN_EXTERN
+TKN_VOID
+TKN_STAR
+TKN_ID(stdout)
+TKN_SEMICOLON
+TKN_INT
+TKN_ID(main)
+TKN_L_PAREN
+TKN_R_PAREN
+TKN_L_BRACE
+TKN_ID(fprintf)
+TKN_L_PAREN
+TKN_ID(stdout)
+TKN_COMMA
+TKN_LIT_STR(This is stdout!
+)
+TKN_R_PAREN
+TKN_SEMICOLON
+TKN_ID(fprintf)
+TKN_L_PAREN
+TKN_ID(stderr)
+TKN_COMMA
+TKN_LIT_STR(This is stderr!
+)
+TKN_R_PAREN
+TKN_SEMICOLON
+TKN_RETURN
+TKN_LIT_INT(0)
+TKN_SEMICOLON
+TKN_R_BRACE
+TKN_EOF

tests/output/inp_extern.c_prep_output.txt

@@ -0,0 +1,48 @@
+TKN_VOID
+TKN_ID(fprintf)
+TKN_L_PAREN
+TKN_VOID
+TKN_STAR
+TKN_COMMA
+TKN_CHAR
+TKN_STAR
+TKN_COMMA
+TKN_DOTS
+TKN_R_PAREN
+TKN_SEMICOLON
+TKN_EXTERN
+TKN_VOID
+TKN_STAR
+TKN_ID(stderr)
+TKN_SEMICOLON
+TKN_EXTERN
+TKN_VOID
+TKN_STAR
+TKN_ID(stdout)
+TKN_SEMICOLON
+TKN_INT
+TKN_ID(main)
+TKN_L_PAREN
+TKN_R_PAREN
+TKN_L_BRACE
+TKN_ID(fprintf)
+TKN_L_PAREN
+TKN_ID(stdout)
+TKN_COMMA
+TKN_LIT_STR(This is stdout!
+)
+TKN_R_PAREN
+TKN_SEMICOLON
+TKN_ID(fprintf)
+TKN_L_PAREN
+TKN_ID(stderr)
+TKN_COMMA
+TKN_LIT_STR(This is stderr!
+)
+TKN_R_PAREN
+TKN_SEMICOLON
+TKN_RETURN
+TKN_LIT_INT(0)
+TKN_SEMICOLON
+TKN_R_BRACE
+TKN_EOF