add sema

Makefile

@@ -1,5 +1,5 @@
 
-larpcc: main.c lex.c preproc.c collection.c parse.c ast_print.c resolve.c
+larpcc: main.c lex.c preproc.c collection.c parse.c ast_print.c resolve.c sema.c
 	gcc $^ -o $@ -std=c23 -pedantic -Wall -Wextra -g -fsanitize=address -Werror=switch
 
 

examples/hello.c

@@ -1,6 +1,6 @@
 #include "not_stdio.h"
 
-int main()
+int main(void)
 {
     puts("hello world");
 }

lex.c

@@ -86,6 +86,8 @@ const char* tokty_str(TokTy ty)
             return "KwChar";
         case Tok_KwConst:
             return "KwConst";
+        case Tok_KwVoid:
+            return "KwVoid";
         case Tok_EndOfMacroInvocation:
             return "EndOfMacroInvocation";
     }

main.c

@@ -2,6 +2,7 @@
 #include "collection.h"
 #include "parse.h"
 #include "resolve.h"
+#include "sema.h"
 #include <assert.h>
 #include <stddef.h>
 #include <stdio.h>
@@ -71,16 +72,18 @@ int main(int argc, char* argv[])
         TokStream tokstream;
         tokstream_init(&tokstream, preproc, &tok_arena);
 
+        Sema* sema = sema_new();
+
         Arena ast_arena;
         arena_init(&ast_arena, 4096);
 
         printf("--- ast ---\n");
 
-        Parser* parser = parser_new(&tokstream, &ast_arena);
+        Parser* parser = parser_new(&tokstream, sema, &ast_arena);
 
-        Arena sym_arena;
+        // Arena sym_arena;
-        arena_init(&sym_arena, 4096);
+        // arena_init(&sym_arena, 4096);
-        Resolver* resolver = resolver_new(&sym_arena);
+        // Resolver* resolver = resolver_new(&sym_arena);
 
         AstStmt* stmt;
         while (!parser_done(parser)) {
@@ -90,16 +93,18 @@ int main(int argc, char* argv[])
 
             printf("\n");
 
-            resolver_resolve_top_stmt(resolver, stmt);
+            // resolver_resolve_top_stmt(resolver, stmt);
         }
 
-        resolver_free(resolver);
+        // resolver_free(resolver);
 
         parser_free(parser);
-
         preproc_free(preproc);
         arena_deinit(&tok_arena);
+
         arena_deinit(&ast_arena);
-        arena_deinit(&sym_arena);
+        // arena_deinit(&sym_arena);
+
+        sema_free(sema);
     }
 }

parse.c

@@ -2,6 +2,7 @@
 #include "ast.h"
 #include "collection.h"
 #include "resolve.h"
+#include "sema.h"
 #include <assert.h>
 #include <stddef.h>
 #include <stdio.h>
@@ -115,6 +116,7 @@ static void kwmatcher_init(KwMatcher* m, TokStream* tokstream)
         { Tok_KwInt, "int" },
         { Tok_KwChar, "char" },
         { Tok_KwConst, "const" },
+        { Tok_KwVoid, "void" },
     };
 
     for (size_t i = 0; i < sizeof(kws) / sizeof(kws[0]); ++i) {
@@ -150,6 +152,7 @@ typedef enum : unsigned int {
 
 struct Parser {
     KwMatcher kwmatcher;
+    Sema* sema;
     Arena* arena;
     Arena sym_arena;
 
@@ -160,12 +163,13 @@ struct Parser {
     ParserFlags flags;
 };
 
-Parser* parser_new(TokStream* tokstream, Arena* arena)
+Parser* parser_new(TokStream* tokstream, Sema* sema, Arena* arena)
 {
     Parser* p = malloc(sizeof(Parser));
 
     *p = (Parser) {
         .kwmatcher = { },
+        .sema = sema,
         .arena = arena,
         .sym_arena = { },
 
@@ -194,20 +198,21 @@ static void step(Parser* p)
     p->curr = kwmatcher_next(&p->kwmatcher);
 }
 
-#define DEFINE_NEW(TYPE, FN_NAME)                                              \
+#define DEFINE_NEW(TYPE, FN_NAME, POST)                                        \
     [[maybe_unused]] static TYPE* new_##FN_NAME(Parser* p, TYPE init)          \
     {                                                                          \
         TYPE* val = arena_alloc(p->arena, sizeof(TYPE));                       \
         *val = init;                                                           \
+        POST;                                                                  \
         return val;                                                            \
     }
 
-DEFINE_NEW(AstStmt, stmt)
+DEFINE_NEW(AstStmt, stmt, { sema_visit_stmt(p->sema, val); })
-DEFINE_NEW(AstDecl, decl)
+DEFINE_NEW(AstDecl, decl, { })
-DEFINE_NEW(AstDeclar, declar)
+DEFINE_NEW(AstDeclar, declar, { })
-DEFINE_NEW(AstExpr, expr)
+DEFINE_NEW(AstExpr, expr, { sema_visit_expr(p->sema, val); })
 
-DEFINE_NEW(AstStmtBlock, stmt_block)
+DEFINE_NEW(AstStmtBlock, stmt_block, { })
 
 static AstStmtBlock* parse_stmt_block(Parser* p);
 static AstStmt* parse_stmt(Parser* p);
@@ -260,8 +265,8 @@ static bool is_defty_ident(Parser* p, Tok tok)
 {
     if (tok.ty != Tok_Ident)
         return false;
-    Sym* sym = syms_get(p->syms, tok.text);
+    OldSym* sym = syms_get(p->syms, tok.text);
-    return sym && sym->ty == Sym_Typedef;
+    return sym && sym->ty == OldSym_Typedef;
 }
 
 static AstStmt* parse_stmt(Parser* p)
@@ -322,7 +327,11 @@ static AstDecl* parse_decl(Parser* p, bool is_param)
     bool multi_allowed = !is_param;
 
     if (!is_param && first->ty == AstDeclar_Function && p->curr.ty == '{') {
+
+        sema_function_enter(p->sema, declars[0]);
         function_body = parse_stmt_block(p);
+        sema_function_leave(p->sema);
+
         multi_allowed = false;
     }
 
@@ -432,7 +441,9 @@ static AstDeclar* parse_declar_l3(Parser* p)
 
         ARRAY_VAR(AstDecl*, params);
 
-        if (p->curr.ty != Tok_Eof && p->curr.ty != ')') {
+        if (p->curr.ty == Tok_KwVoid) {
+            step(p);
+        } else if (p->curr.ty != Tok_Eof && p->curr.ty != ')') {
             AstDecl* d = parse_decl(p, true);
             ARENA_ARRAY_VAR_PUSH(p->arena, params, &d);
 
@@ -509,8 +520,8 @@ static AstSpec parse_decl_spec(Parser* p)
     while (true) {
         switch (p->curr.ty) {
             case Tok_Ident: {
-                Sym* sym = syms_get(p->syms, p->curr.text);
+                OldSym* sym = syms_get(p->syms, p->curr.text);
-                if (!sym || sym->ty != Sym_Typedef)
+                if (!sym || sym->ty != OldSym_Typedef)
                     break;
                 spec |= AstSpec_DefTy;
                 step(p);

parse.h

@@ -3,6 +3,7 @@
 
 #include "ast.h"
 #include "collection.h"
+#include "sema.h"
 #include <stddef.h>
 
 char* read_textfile(const char* filename);
@@ -47,6 +48,7 @@ typedef enum {
     Tok_Char,
     Tok_Whitespace,
     Tok_HeaderName,
+
     Tok_Eol = '\n', // NOTE: has value 10
     Tok_Semicolon = ';',
     Tok_LParen = '(',
@@ -57,12 +59,14 @@ typedef enum {
     Tok_Asterisk = '*',
     Tok_Hashtag = '#',
 
-    Tok_KwInt,
+    Tok_KwInt = 256,
     Tok_KwChar,
     Tok_KwConst,
+    Tok_KwVoid,
 
     // used internally in preprocessor
     Tok_EndOfMacroInvocation,
+
 } TokTy;
 
 const char* tokty_str(TokTy ty);
@@ -105,7 +109,7 @@ Tok tokstream_next(TokStream* s);
 
 typedef struct Parser Parser;
 
-Parser* parser_new(TokStream* tokstream, Arena* arena);
+Parser* parser_new(TokStream* tokstream, Sema* sema, Arena* arena);
 void parser_free(Parser* p);
 AstStmt* parser_parse_stmt(Parser* p);
 bool parser_done(Parser* p);

resolve.c

@@ -17,7 +17,7 @@ Syms* syms_new()
     };
 
     ARRAY_FIELD_PUSH(s, maps, &(Hashmap) { });
-    hashmap_init(&s->maps[0], sizeof(Sym*));
+    hashmap_init(&s->maps[0], sizeof(OldSym*));
 
     return s;
 }
@@ -33,7 +33,7 @@ void syms_free(Syms* s)
 void syms_push_scope(Syms* s)
 {
     ARRAY_FIELD_PUSH(s, maps, &(Hashmap) { });
-    hashmap_init(&s->maps[s->maps_size - 1], sizeof(Sym*));
+    hashmap_init(&s->maps[s->maps_size - 1], sizeof(OldSym*));
 }
 
 void syms_pop_scope(Syms* s)
@@ -43,60 +43,21 @@ void syms_pop_scope(Syms* s)
     s->maps_size -= 1;
 }
 
-void syms_define(Syms* s, const char* ident, Sym* sym)
+void syms_define(Syms* s, const char* ident, OldSym* sym)
 {
     Hashmap* m = &s->maps[s->maps_size - 1];
     hashmap_set(m, ident, &sym);
 }
 
-Sym* syms_get(Syms* s, const char* ident)
+OldSym* syms_get(Syms* s, const char* ident)
 {
-    Sym** res;
+    OldSym** res;
     for (size_t i = 0; i < s->maps_size; ++i)
         if ((res = hashmap_get(&s->maps[s->maps_size - i - 1], ident)))
             return *res;
     return nullptr;
 }
 
-static void print_sym(Sym* s)
-{
-    switch (s->ty) {
-        case Sym_Typedef:
-            printf("Typedef");
-            break;
-        case Sym_Declarator:
-            printf("Declarator");
-            break;
-    }
-}
-
-struct SymPrinter {
-    int depth;
-};
-
-static void print_sym_hashmap(void* self, const char* key, void* val)
-{
-    struct SymPrinter* p = self;
-    Sym* sym = *(Sym**)val;
-
-    printf("  ");
-    for (int i = 0; i < p->depth; ++i)
-        printf("  ");
-    printf("'%s' -> ", key);
-    print_sym(sym);
-    printf("\n");
-}
-
-void syms_print(Syms* s)
-{
-    for (size_t i = 0; i < s->maps_size; ++i) {
-        struct SymPrinter p = {
-            .depth = i,
-        };
-        hashmap_for_each(&s->maps[i], print_sym_hashmap, &p);
-    }
-}
-
 typedef enum : unsigned int {
     ResolverFlags_InTypedef,
 } ResolverFlags;
@@ -130,7 +91,7 @@ Resolver* resolver_new(Arena* arena)
         .flags = 0,
     };
 
-    keymap_init(&r->resol_map, sizeof(Sym*));
+    keymap_init(&r->resol_map, sizeof(OldSym*));
 
     return r;
 }
@@ -226,9 +187,9 @@ static AstVisitorResult visit_declar(AstVisitor* v, AstDeclar* node)
     if (node->ty == AstDeclar_Ident) {
         assert(r->current_decl_stmt);
 
-        Sym* sym = arena_alloc(r->arena, sizeof(Sym));
+        OldSym* sym = arena_alloc(r->arena, sizeof(OldSym));
-        *sym = (Sym) {
+        *sym = (OldSym) {
-            .ty = Sym_Declarator,
+            .ty = OldSym_Declarator,
             .declarator = {
                 .stmt = r->current_decl_stmt,
                 .declar = node,
@@ -247,15 +208,12 @@ static AstVisitorResult visit_expr(AstVisitor* v, AstExpr* node)
 
     if (node->ty == AstExpr_Ident) {
 
-        Sym* sym = syms_get(r->ord_syms, node->ident);
+        OldSym* sym = syms_get(r->ord_syms, node->ident);
         if (!sym) {
             fprintf(stderr,
                 "error: use of undeclared '%s' on line %d\n",
                 node->ident,
                 node->loc.line);
-
-            syms_print(r->ord_syms);
-
             exit(1);
         }
     }

resolve.h

@@ -5,19 +5,19 @@
 #include "collection.h"
 
 typedef enum {
-    Sym_Typedef,
+    OldSym_Typedef,
-    Sym_Declarator,
+    OldSym_Declarator,
-} SymTy;
+} OldSymTy;
 
 typedef struct {
-    SymTy ty;
+    OldSymTy ty;
     union {
         struct {
             AstStmt* stmt;
             AstDeclar* declar;
         } declarator;
     };
-} Sym;
+} OldSym;
 
 typedef struct Syms Syms;
 
@@ -27,8 +27,8 @@ void syms_free(Syms* s);
 void syms_push_scope(Syms* s);
 void syms_pop_scope(Syms* s);
 
-void syms_define(Syms* s, const char* ident, Sym* sym);
+void syms_define(Syms* s, const char* ident, OldSym* sym);
-Sym* syms_get(Syms* s, const char* ident);
+OldSym* syms_get(Syms* s, const char* ident);
 
 typedef struct Resolver Resolver;
 

sema.c

@@ -0,0 +1,106 @@
+#include "sema.h"
+#include "ast.h"
+#include "collection.h"
+#include <assert.h>
+#include <stdlib.h>
+
+typedef struct {
+    Hashmap ord_syms;
+    Hashmap tag_syms;
+} Rib;
+
+static void rib_init(Rib* r)
+{
+    *r = (Rib) {
+        .ord_syms = { },
+        .tag_syms = { },
+    };
+    hashmap_init(&r->ord_syms, sizeof(Sym*));
+    hashmap_init(&r->tag_syms, sizeof(Sym*));
+}
+
+static void rib_deinit(Rib* r)
+{
+    hashmap_deinit(&r->ord_syms);
+    hashmap_deinit(&r->tag_syms);
+}
+
+struct Sema {
+    Arena sym_arena;
+    ARRAY_FIELD(Rib, ribs);
+    Hashmap labels;
+};
+
+Sema* sema_new(void)
+{
+    Sema* s = malloc(sizeof(Sema));
+    *s = (Sema) {
+        .sym_arena = { },
+        ARRAY_FIELD_INIT(ribs),
+        .labels = { },
+    };
+
+    arena_init(&s->sym_arena, 1024);
+
+    ARRAY_FIELD_PUSH(s, ribs, (&(Rib) { }));
+    rib_init(&s->ribs[0]);
+
+    hashmap_init(&s->labels, sizeof(LabelSym*));
+
+    return s;
+}
+
+void sema_free(Sema* s)
+{
+    arena_deinit(&s->sym_arena);
+    for (size_t i = 0; i < s->ribs_size; ++i)
+        rib_deinit(&s->ribs[i]);
+    if (s->ribs)
+        free(s->ribs);
+    hashmap_deinit(&s->labels);
+    free(s);
+}
+
+void sema_scope_enter(Sema* s)
+{
+    ARRAY_FIELD_PUSH(s, ribs, (&(Rib) { }));
+    rib_init(&s->ribs[s->ribs_size - 1]);
+}
+
+void sema_scope_leave(Sema* s)
+{
+    assert(s->ribs_size > 0);
+    rib_deinit(&s->ribs[s->ribs_size - 1]);
+    s->ribs_size -= 1;
+}
+
+void sema_function_enter(Sema* s, AstDeclar* declar)
+{
+    assert(declar->ty == AstDeclar_Function);
+
+    // clear label map
+    hashmap_deinit(&s->labels);
+    hashmap_init(&s->labels, sizeof(LabelSym*));
+
+    sema_scope_enter(s);
+
+    for (size_t i = 0; i < declar->function.params_size; ++i) { }
+}
+
+void sema_function_leave(Sema* s)
+{
+    sema_scope_leave(s);
+}
+
+void sema_visit_function_declar(Sema* s, AstDeclar* declar)
+{
+}
+
+void sema_visit_stmt(Sema* s, AstStmt* stmt)
+{
+    if (stmt->ty == AstStmt_Decl) { }
+}
+
+void sema_visit_expr(Sema* s, AstExpr* expr)
+{
+}

sema.h

@@ -0,0 +1,31 @@
+#ifndef SEMA_H
+#define SEMA_H
+
+#include "ast.h"
+typedef enum {
+    Sym_Declar,
+} SymTy;
+
+typedef struct {
+    SymTy ty;
+} Sym;
+
+typedef struct {
+    int _;
+} LabelSym;
+
+typedef struct Sema Sema;
+
+Sema* sema_new(void);
+void sema_free(Sema* s);
+
+void sema_scope_enter(Sema* s);
+void sema_scope_leave(Sema* s);
+
+void sema_function_enter(Sema* s, AstDeclar* declar);
+void sema_function_leave(Sema* s);
+
+void sema_visit_stmt(Sema* s, AstStmt* stmt);
+void sema_visit_expr(Sema* s, AstExpr* expr);
+
+#endif