Added object file format spec

This is a WIP.

Makefile

@@ -10,7 +10,7 @@ OBJECTS = $(SOURCES:.c=.o)
 DEPENDENCIES = $(SOURCES:.c=.d)
 TARGET?=oas
 OUTPUTS=oas oas-asan oas-msan oas-afl
-RUNARGUMENTS?=-tokens tests/input/valid.asm
+RUNARGUMENTS?=ast tests/input/valid.asm
 
 all: $(TARGET)
 	

doc/object_format.md

@@ -0,0 +1,55 @@
+# Linker file format
+
+```C
+
+struct object_file {
+    uint64_t    magic;   // ".oo-bin"
+    uint64_t    version; // 1
+    uint64_t    architecture; // AMD64(0)
+    uint64_t    offsets_offset;
+
+    struct offsets {
+        uint64_t strings;
+        uint64_t sections;
+        uint64_t symbols;
+        uint64_t relocations;
+    } offsets;
+
+    struct string_table {
+        uint64_t size;
+        uint8_t data[static size];
+    } strings;
+
+    struct section_table {
+        uint32_t count;
+        struct section_entry {
+            uint32_t name;
+            uint64_t offset;
+            uint64_t size_on_disk;
+            uint64_t size_in_memory;
+            uint64_t flags;
+        } sections[static count];
+    } sections;
+
+    struct symbol_table {
+        uint32_t count;
+        struct symbol_entry {
+            uint32_t name;
+            uint8_t  kind;  // IMPORT(0) | EXPORT(1) | LOCAL(2)
+            uint32_t section;
+            uint64_t offset;
+        } symbols[static count];
+    } symbols;
+
+    struct relocation_table {
+        uint32_t count;
+        struct relocation_entry {
+            uint32_t section;
+            uint64_t offset;
+            uint8_t  size;
+            uint32_t symbol;
+            uint8_t  kind;  // ABSOLUTE(0) | RELATIVE(1)
+        } relocations[static count];
+    } relocations;
+};
+```

doc/parser_grammar.txt

@@ -1,24 +1,20 @@
-/* string literals are lexer identifier tokens with that particular value */
 <program>   ::= <statement>*
-<statement> ::= ( <label> | <directive> | <instruction> ) <newline>
+<statement> ::= <label> | <directive> | <instruction>
 
 <label> ::= <identifier> <colon>
 
-<directive> ::= <dot> <section>
+<directive> ::= <dot> <section_directive>
 
-<section>   ::= "section" <identifier>
+<section_directive> ::= "section" <identifier>
 
 <instruction> ::= <identifier> <operands>
 
-<operands> ::= <operand> ( <comma> <operands> )*
+<operands> ::= <operand> ( <comma> <operand> )*
 
 <operand>  ::= <register> | <immediate> | <memory>
 
-<register> ::= <register_base> | <register_extra>
-<register_base> ::= "rax" | "rbx" | "rcx" | "rdx" | "rsi" | "rdi" | "rbp" | "rsp"
-<register_extra> ::= "r8" | "r9" | "r10" | "r11" | "r12" | "r13" | "r14" | "r15" 
-
 <immediate> ::= <number> | <label_reference>
+
 <number> ::= <octal> | <binary> | <decimal> | <hexadecimal>
 
 <label_reference> ::= <identifier>
@@ -27,10 +23,17 @@
 
 <memory_expression> ::= <label_reference> | <register_expression>
 
-<register_expression> ::= <register> ( <plus> <register> <asterisk> <number> )? ( <plus_or_minus> <number> )?
+<register_expression> ::= <register> <register_index>? <register_offset>?
 
-<register_displacement> ::= <plus> <register> <asterisk> <number>
+<register_index> ::= <plus> <register> <asterisk> <number>
 
 <register_offset> ::= <plus_or_minus> <number>
 
 <plus_or_minus> ::= <plus> | <minus>
+
+
+/* These are lexer identifiers with the correct string value */
+<section> ::= "section"
+
+<register> ::= "rax" | "rbx" | "rcx" | "rdx" | "rsi" | "rdi" | "rbp" | "rsp" |
+"r8" | "r9" | "r10" | "r11" | "r12" | "r13" | "r14" | "r15"

src/ast.c

@@ -0,0 +1,187 @@
+#include "ast.h"
+#include "error.h"
+#include <assert.h>
+#include <string.h>
+
+error_t *err_node_children_cap = &(error_t){
+    .message = "Failed to increase ast node children, max capacity reached"};
+
+error_t *ast_node_alloc(ast_node_t **output) {
+    *output = nullptr;
+
+    ast_node_t *node = calloc(1, sizeof(ast_node_t));
+    if (node == nullptr)
+        return err_allocation_failed;
+
+    *output = node;
+    return nullptr;
+}
+
+void ast_node_free_value(ast_node_t *node) {
+    // TODO: decide how value ownership will work and clean it up here
+}
+
+void ast_node_free(ast_node_t *node) {
+    if (node == nullptr)
+        return;
+    if (node->children) {
+        for (size_t i = 0; i < node->len; ++i)
+            ast_node_free(node->children[i]);
+        free(node->children);
+    }
+
+    ast_node_free_value(node);
+
+    memset(node, 0, sizeof(ast_node_t));
+    free(node);
+}
+
+/**
+ * @pre node->children must be nullptr
+ */
+error_t *ast_node_alloc_children(ast_node_t *node) {
+    node->children = calloc(node_default_children_cap, sizeof(ast_node_t *));
+    if (node->children == nullptr)
+        return err_allocation_failed;
+
+    node->cap = node_default_children_cap;
+    return nullptr;
+}
+
+error_t *ast_node_grow_cap(ast_node_t *node) {
+    if (node->cap >= node_max_children_cap) {
+        return err_node_children_cap;
+    }
+
+    size_t new_cap = node->cap * 2;
+    if (new_cap > node_max_children_cap) {
+        new_cap = node_max_children_cap;
+    }
+
+    ast_node_t **new_children =
+        realloc(node->children, new_cap * sizeof(ast_node_t *));
+    if (new_children == nullptr) {
+        return err_allocation_failed;
+    }
+
+    node->children = new_children;
+    node->cap = new_cap;
+
+    return nullptr;
+}
+
+error_t *ast_node_add_child(ast_node_t *node, ast_node_t *child) {
+    error_t *err = nullptr;
+    if (node->children == nullptr)
+        err = ast_node_alloc_children(node);
+    else if (node->len >= node->cap)
+        err = ast_node_grow_cap(node);
+    if (err)
+        return err;
+
+    node->children[node->len] = child;
+    node->len += 1;
+
+    return nullptr;
+}
+
+const char *ast_node_id_to_cstr(node_id_t id) {
+    switch (id) {
+    case NODE_INVALID:
+        return "NODE_INVALID";
+    case NODE_PROGRAM:
+        return "NODE_PROGRAM";
+    case NODE_STATEMENT:
+        return "NODE_STATEMENT";
+    case NODE_LABEL:
+        return "NODE_LABEL";
+    case NODE_DIRECTIVE:
+        return "NODE_DIRECTIVE";
+    case NODE_INSTRUCTION:
+        return "NODE_INSTRUCTION";
+    case NODE_OPERANDS:
+        return "NODE_OPERANDS";
+    case NODE_OPERAND:
+        return "NODE_OPERAND";
+    case NODE_IMMEDIATE:
+        return "NODE_IMMEDIATE";
+    case NODE_MEMORY:
+        return "NODE_MEMORY";
+    case NODE_NUMBER:
+        return "NODE_NUMBER";
+    case NODE_LABEL_REFERENCE:
+        return "NODE_LABEL_REFERENCE";
+    case NODE_MEMORY_EXPRESSION:
+        return "NODE_MEMORY_EXPRESSION";
+    case NODE_REGISTER_EXPRESSION:
+        return "NODE_REGISTER_EXPRESSION";
+    case NODE_REGISTER_INDEX:
+        return "NODE_REGISTER_INDEX";
+    case NODE_REGISTER_OFFSET:
+        return "NODE_REGISTER_OFFSET";
+    case NODE_PLUS_OR_MINUS:
+        return "NODE_PLUS_OR_MINUS";
+    case NODE_SECTION_DIRECTIVE:
+        return "NODE_SECTION_DIRECTIVE";
+    case NODE_REGISTER:
+        return "NODE_REGISTER";
+    case NODE_SECTION:
+        return "NODE_SECTION";
+    case NODE_IDENTIFIER:
+        return "NODE_IDENTIFIER";
+    case NODE_DECIMAL:
+        return "NODE_DECIMAL";
+    case NODE_HEXADECIMAL:
+        return "NODE_HEXADECIMAL";
+    case NODE_OCTAL:
+        return "NODE_OCTAL";
+    case NODE_BINARY:
+        return "NODE_BINARY";
+    case NODE_CHAR:
+        return "NODE_CHAR";
+    case NODE_STRING:
+        return "NODE_STRING";
+    case NODE_COLON:
+        return "NODE_COLON";
+    case NODE_COMMA:
+        return "NODE_COMMA";
+    case NODE_LBRACKET:
+        return "NODE_LBRACKET";
+    case NODE_RBRACKET:
+        return "NODE_RBRACKET";
+    case NODE_PLUS:
+        return "NODE_PLUS";
+    case NODE_MINUS:
+        return "NODE_MINUS";
+    case NODE_ASTERISK:
+        return "NODE_ASTERISK";
+    case NODE_DOT:
+        return "NODE_DOT";
+    }
+    assert(!"Unreachable, weird node id" && id);
+    __builtin_unreachable();
+}
+
+static void ast_node_print_internal(ast_node_t *node, int indent) {
+    if (node == NULL) {
+        return;
+    }
+
+    for (int i = 0; i < indent; i++) {
+        printf("  ");
+    }
+    printf("%s", ast_node_id_to_cstr(node->id));
+
+    if (node->token_entry && node->token_entry->token.value) {
+        printf(" \"%s\"", node->token_entry->token.value);
+    }
+    printf("\n");
+
+    for (size_t i = 0; i < node->len; i++) {
+        ast_node_print_internal(node->children[i], indent + 1);
+    }
+}
+
+void ast_node_print(ast_node_t *node) {
+    ast_node_print_internal(node, 0);
+}

src/ast.h

@@ -0,0 +1,120 @@
+#ifndef INCLUDE_SRC_AST_H_
+#define INCLUDE_SRC_AST_H_
+
+#include "error.h"
+#include "lexer.h"
+#include "tokenlist.h"
+#include <stddef.h>
+#include <stdint.h>
+
+typedef enum node_id {
+    NODE_INVALID,
+
+    NODE_PROGRAM,
+    NODE_STATEMENT,
+    NODE_LABEL,
+    NODE_DIRECTIVE,
+    NODE_INSTRUCTION,
+    NODE_OPERANDS,
+    NODE_OPERAND,
+    NODE_IMMEDIATE,
+    NODE_MEMORY,
+    NODE_NUMBER,
+    NODE_LABEL_REFERENCE,
+    NODE_MEMORY_EXPRESSION,
+    NODE_REGISTER_EXPRESSION,
+    NODE_REGISTER_INDEX,
+    NODE_REGISTER_OFFSET,
+    NODE_PLUS_OR_MINUS,
+    NODE_SECTION_DIRECTIVE,
+
+    // Validated primitives
+    NODE_REGISTER,
+    NODE_SECTION,
+
+    // Primitive nodes
+    NODE_IDENTIFIER,
+    NODE_DECIMAL,
+    NODE_HEXADECIMAL,
+    NODE_OCTAL,
+    NODE_BINARY,
+    NODE_CHAR,
+    NODE_STRING,
+    NODE_COLON,
+    NODE_COMMA,
+    NODE_LBRACKET,
+    NODE_RBRACKET,
+    NODE_PLUS,
+    NODE_MINUS,
+    NODE_ASTERISK,
+    NODE_DOT,
+} node_id_t;
+
+typedef struct ast_node ast_node_t;
+
+constexpr size_t node_default_children_cap = 8;
+/* 65K ought to be enough for anybody */
+constexpr size_t node_max_children_cap = 1 << 16;
+
+struct ast_node {
+    node_id_t id;
+    tokenlist_entry_t *token_entry;
+    size_t len;
+    size_t cap;
+    ast_node_t **children;
+
+    union {
+        struct {
+            uint64_t value;
+            int size;
+        } integer;
+        char *name;
+    } value;
+};
+
+/**
+ * @brief Allocates a new AST node
+ *
+ * Creates and initializes a new AST node with default (zero) values.
+ *
+ * @param[out] output Pointer to store the allocated node
+ * @return error_t* nullptr on success, allocation error on failure
+ */
+error_t *ast_node_alloc(ast_node_t **node);
+
+/**
+ * @brief Frees an AST node and all its children recursively
+ *
+ * Recursively frees all children of the node, then frees the node itself.
+ * If node is nullptr, the function returns without doing anything.
+ *
+ * @param node The node to free
+ */
+void ast_node_free(ast_node_t *node);
+
+/**
+ * @brief Adds a child node to a parent node
+ *
+ * Adds the specified child node to the parent's children array.
+ * If this is the first child, the function allocates the children array.
+ * If the children array is full, the function increases its capacity.
+ *
+ * @param node The parent node to add the child to
+ * @param child The child node to add
+ * @return error_t* nullptr on success, allocation error on failure,
+ *                  or err_node_children_cap if maximum capacity is reached
+ */
+error_t *ast_node_add_child(ast_node_t *node, ast_node_t *child);
+
+/**
+ * @brief Prints an AST starting from the given node
+ *
+ * Prints a representation of the AST with indentation to show structure.
+ * Each node's type is shown, and if a node has an associated token value,
+ * that value is printed in quotes.
+ *
+ * @param node The root node of the AST to print
+ */
+void ast_node_print(ast_node_t *node);
+
+#endif // INCLUDE_SRC_AST_H_

src/error.c

@@ -10,6 +10,9 @@ error_t *const err_errorf_length = &(error_t){
     .message =
         "Formatting of another error failed to determine the error length"};
 
+error_t *err_allocation_failed =
+    &(error_t){.message = "Memory allocation failed"};
+
 error_t *errorf(const char *fmt, ...) {
     error_t *err = calloc(1, sizeof(error_t));
     if (err == nullptr)

src/error.h

@@ -18,4 +18,7 @@ static inline void error_free(error_t *err) {
     free(err);
 }
 
+/* Some global errors */
+extern error_t *err_allocation_failed;
+
 #endif // INCLUDE_SRC_ERROR_H_

src/lexer.c

@@ -20,9 +20,6 @@ error_t *err_eof =
 
 error_t *err_unknown_read = &(error_t){.message = "Unknown read error"};
 
-error_t *err_allocation_failed =
-    &(error_t){.message = "Memory allocation failed"};
-
 typedef bool (*char_predicate_t)(char);
 
 const char *lexer_token_id_to_cstr(lexer_token_id_t id) {
@@ -186,7 +183,7 @@ error_t *lexer_consume_n(lexer_t *lex, const size_t len,
                          char buffer[static len], const size_t n) {
     if (lex->buffer_count < n)
         return err_buffer_underrun;
-    if (len > n)
+    if (n > len)
         return err_consume_excessive_length;
 
     memcpy(buffer, lex->buffer, n);

src/main.c

@@ -1,62 +1,108 @@
 #include "error.h"
 #include "lexer.h"
+#include "parser/parser.h"
+#include "tokenlist.h"
 
 #include <limits.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
-bool print_token(lexer_token_t *token) {
+typedef enum mode { MODE_AST, MODE_TEXT, MODE_TOKENS } mode_t;
-    lexer_token_print(token);
+
-    return true;
+void print_tokens(tokenlist_t *list) {
+    for (auto entry = list->head; entry; entry = entry->next) {
+        auto token = &entry->token;
+        lexer_token_print(token);
+    }
 }
 
-bool print_value(lexer_token_t *token) {
+void print_text(tokenlist_t *list) {
-    if (token->id == TOKEN_ERROR) {
+    for (auto entry = list->head; entry; entry = entry->next) {
-        printf("%s\n", token->value);
+        auto token = &entry->token;
-        for (size_t i = 0; i < token->character_number; ++i)
+        if (token->id == TOKEN_ERROR) {
-            printf(" ");
+            printf("%s\n", token->value);
-        printf("^-- %s\n", token->explanation);
+            for (size_t i = 0; i < token->character_number; ++i)
-    } else {
+                printf(" ");
-        printf("%s", token->value);
+            printf("^-- %s\n", token->explanation);
+            return;
+        } else {
+            printf("%s", token->value);
+        }
     }
-    return token->id != TOKEN_ERROR;
+}
+
+void print_ast(tokenlist_t *list) {
+    parse_result_t result = parse(list->head);
+    if (result.err) {
+        puts(result.err->message);
+        error_free(result.err);
+        return;
+    }
+    ast_node_print(result.node);
+
+    if (result.next != nullptr) {
+        puts("First unparsed token:");
+        lexer_token_print(&result.next->token);
+    }
+
+    ast_node_free(result.node);
+}
+
+int get_execution_mode(int argc, char *argv[]) {
+    if (argc != 3 || (strcmp(argv[1], "tokens") != 0 &&
+                      strcmp(argv[1], "text") != 0 && strcmp(argv[1], "ast"))) {
+        puts("Usage: oas [tokens|text|ast] <filename>");
+        exit(1);
+    }
+
+    if (strcmp(argv[1], "tokens") == 0)
+        return MODE_TOKENS;
+    if (strcmp(argv[1], "text") == 0)
+        return MODE_TEXT;
+    return MODE_AST;
 }
 
 int main(int argc, char *argv[]) {
-    if (argc != 3 ||
+    mode_t mode = get_execution_mode(argc, argv);
-        (strcmp(argv[1], "-tokens") != 0 && strcmp(argv[1], "-text") != 0)) {
-        puts("Usage: oas -tokens <filename>");
-        puts("Usage: oas -text <filename>");
-        return 1;
-    }
-
-    bool (*print_fn)(lexer_token_t *);
     char *filename = argv[2];
-    if (strcmp(argv[1], "-tokens") == 0) {
+
-        print_fn = print_token;
+    lexer_t *lex = &(lexer_t){};
-    } else {
+    error_t *err = lexer_open(lex, filename);
-        print_fn = print_value;
+    if (err)
+        goto cleanup_error;
+
+    tokenlist_t *list;
+    err = tokenlist_alloc(&list);
+    if (err)
+        goto cleanup_lexer;
+
+    err = tokenlist_fill(list, lex);
+    if (err)
+        goto cleanup_tokens;
+
+    switch (mode) {
+    case MODE_TOKENS:
+        print_tokens(list);
+        break;
+    case MODE_TEXT:
+        print_text(list);
+        break;
+    case MODE_AST:
+        print_ast(list);
+        break;
     }
 
-    lexer_t lex = {0};
+    tokenlist_free(list);
-    lexer_token_t token;
-    error_t *err = lexer_open(&lex, filename);
-    if (err) {
-        puts(err->message);
-        error_free(err);
-        return 1;
-    }
-
-    bool keep_going = true;
-    while (keep_going && (err = lexer_next(&lex, &token)) == nullptr) {
-        keep_going = print_fn(&token);
-        free(token.value);
-    }
-
-    if (err && err != err_eof) {
-        puts(err->message);
-    }
     error_free(err);
     return 0;
+
+cleanup_tokens:
+    tokenlist_free(list);
+cleanup_lexer:
+    lexer_close(lex);
+cleanup_error:
+    puts(err->message);
+    error_free(err);
+    return 1;
 }

src/parser/combinators.c

@@ -0,0 +1,126 @@
+#include "combinators.h"
+
+// Parse a list of the given parser delimited by the given token id. Does not
+// store the delimiters in the parent node
+parse_result_t parse_list(tokenlist_entry_t *current, node_id_t id,
+                          bool allow_none, lexer_token_id_t delimiter_id,
+                          parser_t parser) {
+    ast_node_t *many;
+    error_t *err = ast_node_alloc(&many);
+    parse_result_t result;
+    if (err)
+        return parse_error(err);
+    many->id = id;
+
+    while (current) {
+        // Skip beyond the delimiter on all but the first iteration
+        if (many->len > 0) {
+            if (current->token.id != delimiter_id)
+                break;
+            current = tokenlist_next(current);
+            if (current == nullptr) {
+                // FIXME: this isn't quite right, we can't consume the delimiter
+                // if the next element will fail to parse but it's late and I
+                // must think this through tomorrow
+                break;
+            }
+        }
+
+        result = parser(current);
+        if (result.err == err_parse_no_match)
+            break;
+        if (result.err) {
+            ast_node_free(many);
+            return result;
+        }
+        err = ast_node_add_child(many, result.node);
+        if (err) {
+            ast_node_free(many);
+            ast_node_free(result.node);
+            return parse_error(err);
+        }
+        current = result.next;
+    }
+
+    if (!allow_none && many->len == 0) {
+        ast_node_free(many);
+        return parse_no_match();
+    }
+    return parse_success(many, current);
+}
+
+parse_result_t parse_any(tokenlist_entry_t *current, parser_t parsers[]) {
+    parser_t parser;
+    while ((parser = *parsers++)) {
+        parse_result_t result = parser(current);
+        if (result.err == nullptr)
+            return result;
+    }
+    return parse_no_match();
+}
+
+// parse as many of the giver parsers objects in a row as possible,
+// potentially allowing none wraps the found objects in a new ast node with
+// the given note id
+parse_result_t parse_many(tokenlist_entry_t *current, node_id_t id,
+                          bool allow_none, parser_t parser) {
+    ast_node_t *many;
+    error_t *err = ast_node_alloc(&many);
+    parse_result_t result;
+    if (err)
+        return parse_error(err);
+    many->id = id;
+
+    while (current) {
+        result = parser(current);
+        if (result.err == err_parse_no_match)
+            break;
+        if (result.err) {
+            ast_node_free(many);
+            return result;
+        }
+        err = ast_node_add_child(many, result.node);
+        if (err) {
+            ast_node_free(many);
+            ast_node_free(result.node);
+            return parse_error(err);
+        }
+        current = result.next;
+    }
+
+    if (!allow_none && many->len == 0) {
+        ast_node_free(many);
+        return parse_no_match();
+    }
+    return parse_success(many, current);
+}
+
+// Parse all tries to parse all parsers consecutively and if it succeeds it
+// wraps the parsed nodes in a new parent node.
+parse_result_t parse_consecutive(tokenlist_entry_t *current, node_id_t id,
+                                 parser_t parsers[]) {
+    ast_node_t *all;
+    error_t *err = ast_node_alloc(&all);
+    parse_result_t result;
+    if (err)
+        return parse_error(err);
+
+    all->id = id;
+
+    parser_t parser;
+    while ((parser = *parsers++) && current) {
+        result = parser(current);
+        if (result.err) {
+            ast_node_free(all);
+            return result;
+        }
+        err = ast_node_add_child(all, result.node);
+        if (err) {
+            ast_node_free(result.node);
+            ast_node_free(all);
+            return parse_error(err);
+        }
+        current = result.next;
+    }
+    return parse_success(all, current);
+}

src/parser/combinators.h

@@ -0,0 +1,25 @@
+#ifndef INCLUDE_PARSER_COMBINATORS_H_
+#define INCLUDE_PARSER_COMBINATORS_H_
+
+#include "util.h"
+
+typedef parse_result_t (*parser_t)(tokenlist_entry_t *);
+
+parse_result_t parse_any(tokenlist_entry_t *current, parser_t parsers[]);
+
+// parse as many of the giver parsers objects in a row as possible, potentially
+// allowing none wraps the found objects in a new ast node with the given note
+// id
+parse_result_t parse_many(tokenlist_entry_t *current, node_id_t id,
+                          bool allow_none, parser_t parser);
+
+parse_result_t parse_list(tokenlist_entry_t *current, node_id_t id,
+                          bool allow_none, lexer_token_id_t delimiter_id,
+                          parser_t parser);
+
+// Parse all tries to parse all parsers consecutively and if it succeeds it
+// wraps the parsed nodes in a new parent node.
+parse_result_t parse_consecutive(tokenlist_entry_t *current, node_id_t id,
+                                 parser_t parsers[]);
+
+#endif // INCLUDE_PARSER_COMBINATORS_H_

src/parser/parser.c

@@ -0,0 +1,140 @@
+#include "parser.h"
+#include "../ast.h"
+#include "../lexer.h"
+#include "../tokenlist.h"
+#include "combinators.h"
+#include "primitives.h"
+#include "util.h"
+
+parse_result_t parse_number(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_octal, parse_decimal, parse_hexadecimal,
+                          parse_binary, nullptr};
+    parse_result_t result = parse_any(current, parsers);
+    return parse_result_wrap(NODE_NUMBER, result);
+}
+
+parse_result_t parse_plus_or_minus(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_plus, parse_minus, nullptr};
+    return parse_any(current, parsers);
+}
+
+parse_result_t parse_register_index(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_plus, parse_register, parse_asterisk,
+                          parse_number, nullptr};
+    return parse_consecutive(current, NODE_REGISTER_INDEX, parsers);
+}
+
+parse_result_t parse_register_offset(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_plus_or_minus, parse_number, nullptr};
+    return parse_consecutive(current, NODE_REGISTER_OFFSET, parsers);
+}
+
+parse_result_t parse_register_expression(tokenlist_entry_t *current) {
+    parse_result_t result;
+
+    ast_node_t *expr;
+    error_t *err = ast_node_alloc(&expr);
+    if (err)
+        return parse_error(err);
+    expr->id = NODE_REGISTER_EXPRESSION;
+
+    // <register>
+    result = parse_register(current);
+    if (result.err) {
+        ast_node_free(expr);
+        return result;
+    }
+    err = ast_node_add_child(expr, result.node);
+    if (err) {
+        ast_node_free(result.node);
+        ast_node_free(expr);
+        return parse_error(err);
+    }
+    current = result.next;
+
+    // <register_index>?
+    result = parse_register_index(current);
+    if (result.err) {
+        error_free(result.err);
+    } else {
+        err = ast_node_add_child(expr, result.node);
+        if (err) {
+            ast_node_free(result.node);
+            ast_node_free(expr);
+            return parse_error(err);
+        }
+        current = result.next;
+    }
+
+    // <register_offset>?
+    result = parse_register_offset(current);
+    if (result.err) {
+        error_free(result.err);
+    } else {
+        err = ast_node_add_child(expr, result.node);
+        if (err) {
+            ast_node_free(result.node);
+            ast_node_free(expr);
+            return parse_error(err);
+        }
+        current = result.next;
+    }
+    return parse_success(expr, current);
+}
+
+parse_result_t parse_immediate(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_number, parse_identifier, nullptr};
+    parse_result_t result = parse_any(current, parsers);
+    return parse_result_wrap(NODE_IMMEDIATE, result);
+}
+
+parse_result_t parse_memory_expression(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_register_expression, parse_identifier, nullptr};
+    return parse_any(current, parsers);
+}
+
+parse_result_t parse_memory(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_lbracket, parse_memory_expression,
+                          parse_rbracket, nullptr};
+    return parse_consecutive(current, NODE_MEMORY, parsers);
+}
+
+parse_result_t parse_operand(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_register, parse_memory, parse_immediate,
+                          nullptr};
+    return parse_any(current, parsers);
+}
+
+parse_result_t parse_operands(tokenlist_entry_t *current) {
+    return parse_list(current, NODE_OPERANDS, true, TOKEN_COMMA, parse_operand);
+}
+
+parse_result_t parse_label(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_identifier, parse_colon, nullptr};
+    return parse_consecutive(current, NODE_LABEL, parsers);
+}
+
+parse_result_t parse_section_directive(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_section, parse_identifier, nullptr};
+    return parse_consecutive(current, NODE_SECTION_DIRECTIVE, parsers);
+}
+
+parse_result_t parse_directive(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_dot, parse_section_directive, nullptr};
+    return parse_consecutive(current, NODE_DIRECTIVE, parsers);
+}
+
+parse_result_t parse_instruction(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_identifier, parse_operands, nullptr};
+    return parse_consecutive(current, NODE_INSTRUCTION, parsers);
+}
+
+parse_result_t parse_statement(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_label, parse_directive, parse_instruction,
+                          nullptr};
+    return parse_any(current, parsers);
+}
+
+parse_result_t parse(tokenlist_entry_t *current) {
+    return parse_many(current, NODE_PROGRAM, true, parse_statement);
+}

src/parser/parser.h

@@ -0,0 +1,9 @@
+#ifndef INCLUDE_PARSER_PARSER_H_
+#define INCLUDE_PARSER_PARSER_H_
+
+#include "../tokenlist.h"
+#include "util.h"
+
+parse_result_t parse(tokenlist_entry_t *current);
+
+#endif // INCLUDE_PARSER_PARSER_H_

src/parser/primitives.c

@@ -0,0 +1,103 @@
+#include "primitives.h"
+#include "../ast.h"
+#include <string.h>
+
+parse_result_t parse_identifier(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_IDENTIFIER, NODE_IDENTIFIER, nullptr);
+}
+
+parse_result_t parse_decimal(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_DECIMAL, NODE_DECIMAL, nullptr);
+}
+
+parse_result_t parse_hexadecimal(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_HEXADECIMAL, NODE_HEXADECIMAL, nullptr);
+}
+
+parse_result_t parse_binary(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_BINARY, NODE_BINARY, nullptr);
+}
+
+parse_result_t parse_octal(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_OCTAL, NODE_OCTAL, nullptr);
+}
+
+parse_result_t parse_string(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_STRING, NODE_STRING, nullptr);
+}
+
+parse_result_t parse_char(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_CHAR, NODE_CHAR, nullptr);
+}
+
+parse_result_t parse_colon(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_COLON, NODE_COLON, nullptr);
+}
+
+parse_result_t parse_comma(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_COMMA, NODE_COMMA, nullptr);
+}
+
+parse_result_t parse_lbracket(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_LBRACKET, NODE_LBRACKET, nullptr);
+}
+
+parse_result_t parse_rbracket(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_RBRACKET, NODE_RBRACKET, nullptr);
+}
+
+parse_result_t parse_plus(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_PLUS, NODE_PLUS, nullptr);
+}
+
+parse_result_t parse_minus(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_MINUS, NODE_MINUS, nullptr);
+}
+
+parse_result_t parse_asterisk(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_ASTERISK, NODE_ASTERISK, nullptr);
+}
+
+parse_result_t parse_dot(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_DOT, NODE_DOT, nullptr);
+}
+
+parse_result_t parse_label_reference(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_IDENTIFIER, NODE_LABEL_REFERENCE,
+                       nullptr);
+}
+
+const char *registers[] = {
+    // 64-bit registers
+    "rax", "rcx", "rdx", "rbx", "rsp", "rbp", "rsi", "rdi", "r8", "r9", "r10",
+    "r11", "r12", "r13", "r14", "r15",
+    // 32-bit registers
+    "eax", "ecx", "edx", "ebx", "esp", "ebp", "esi", "edi", "r8d", "r9d",
+    "r10d", "r11d", "r12d", "r13d", "r14d", "r15d",
+    // 16-bit registers
+    "ax", "cx", "dx", "bx", "sp", "bp", "si", "di", "r8w", "r9w", "r10w",
+    "r11w", "r12w", "r13w", "r14w", "r15w",
+    // 8-bit low registers
+    "al", "cl", "dl", "bl", "spl", "bpl", "sil", "dil", "r8b", "r9b", "r10b",
+    "r11b", "r12b", "r13b", "r14b", "r15b", nullptr};
+
+bool is_register_token(lexer_token_t *token) {
+    for (size_t i = 0; registers[i] != nullptr; ++i)
+        if (strcmp(token->value, registers[i]) == 0)
+            return true;
+    return false;
+}
+
+parse_result_t parse_register(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_IDENTIFIER, NODE_REGISTER,
+                       is_register_token);
+}
+
+bool is_section_token(lexer_token_t *token) {
+    return strcmp(token->value, "section") == 0;
+}
+
+parse_result_t parse_section(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_IDENTIFIER, NODE_SECTION,
+                       is_section_token);
+}

src/parser/primitives.h

@@ -0,0 +1,30 @@
+#ifndef INCLUDE_PARSER_PRIMITIVES_H_
+#define INCLUDE_PARSER_PRIMITIVES_H_
+
+#include "util.h"
+
+parse_result_t parse_identifier(tokenlist_entry_t *current);
+parse_result_t parse_decimal(tokenlist_entry_t *current);
+parse_result_t parse_hexadecimal(tokenlist_entry_t *current);
+parse_result_t parse_binary(tokenlist_entry_t *current);
+parse_result_t parse_octal(tokenlist_entry_t *current);
+parse_result_t parse_string(tokenlist_entry_t *current);
+parse_result_t parse_char(tokenlist_entry_t *current);
+parse_result_t parse_colon(tokenlist_entry_t *current);
+parse_result_t parse_comma(tokenlist_entry_t *current);
+parse_result_t parse_lbracket(tokenlist_entry_t *current);
+parse_result_t parse_rbracket(tokenlist_entry_t *current);
+parse_result_t parse_plus(tokenlist_entry_t *current);
+parse_result_t parse_minus(tokenlist_entry_t *current);
+parse_result_t parse_asterisk(tokenlist_entry_t *current);
+parse_result_t parse_dot(tokenlist_entry_t *current);
+parse_result_t parse_label_reference(tokenlist_entry_t *current);
+
+/* These are "primitives" with a different name and some extra validation on top
+ * for example, register is just an identifier but it only matches a limited set
+ * of values
+ */
+parse_result_t parse_register(tokenlist_entry_t *current);
+parse_result_t parse_section(tokenlist_entry_t *current);
+
+#endif // INCLUDE_PARSER_PRIMITIVES_H_

src/parser/util.c

@@ -0,0 +1,56 @@
+#include "util.h"
+#include "../tokenlist.h"
+
+error_t *err_parse_no_match =
+    &(error_t){.message = "parsing failed to find the correct token sequence"};
+
+parse_result_t parse_error(error_t *err) {
+    return (parse_result_t){.err = err};
+}
+
+parse_result_t parse_no_match() {
+    return parse_error(err_parse_no_match);
+}
+
+parse_result_t parse_success(ast_node_t *ast, tokenlist_entry_t *next) {
+    next = tokenlist_skip_trivia(next);
+    return (parse_result_t){.node = ast, .next = next};
+}
+
+parse_result_t parse_token(tokenlist_entry_t *current,
+                           lexer_token_id_t token_id, node_id_t ast_id,
+                           token_validator_t is_valid) {
+    if (current->token.id != token_id ||
+        (is_valid && !is_valid(&current->token)))
+        return parse_no_match();
+
+    ast_node_t *node;
+    error_t *err = ast_node_alloc(&node);
+    if (err)
+        return parse_error(err);
+    node->id = ast_id;
+    node->token_entry = current;
+
+    return parse_success(node, current->next);
+}
+
+parse_result_t parse_result_wrap(node_id_t id, parse_result_t result) {
+    if (result.err)
+        return result;
+
+    ast_node_t *node;
+    error_t *err = ast_node_alloc(&node);
+    if (err) {
+        ast_node_free(result.node);
+        return parse_error(err);
+    }
+    node->id = id;
+
+    err = ast_node_add_child(node, result.node);
+    if (err) {
+        ast_node_free(result.node);
+        return parse_error(err);
+    }
+
+    return parse_success(node, result.next);
+}

src/parser/util.h

@@ -0,0 +1,26 @@
+#ifndef INCLUDE_PARSER_UTIL_H_
+#define INCLUDE_PARSER_UTIL_H_
+
+#include "../ast.h"
+#include "../error.h"
+#include "../tokenlist.h"
+
+typedef struct parse_result {
+    error_t *err;
+    tokenlist_entry_t *next;
+    ast_node_t *node;
+} parse_result_t;
+
+typedef bool (*token_validator_t)(lexer_token_t *);
+
+parse_result_t parse_error(error_t *err);
+parse_result_t parse_no_match();
+parse_result_t parse_success(ast_node_t *ast, tokenlist_entry_t *next);
+parse_result_t parse_token(tokenlist_entry_t *current,
+                           lexer_token_id_t token_id, node_id_t ast_id,
+                           token_validator_t is_valid);
+parse_result_t parse_result_wrap(node_id_t id, parse_result_t result);
+
+extern error_t *err_parse_no_match;
+
+#endif // INCLUDE_PARSER_UTIL_H_

src/tokenlist.c

@@ -0,0 +1,106 @@
+#include "tokenlist.h"
+#include "error.h"
+#include "lexer.h"
+#include <stdlib.h>
+
+error_t *tokenlist_alloc(tokenlist_t **output) {
+    *output = nullptr;
+
+    tokenlist_t *list = calloc(1, sizeof(tokenlist_t));
+    if (list == nullptr)
+        return err_allocation_failed;
+
+    list->head = nullptr;
+    list->tail = nullptr;
+
+    *output = list;
+    return nullptr;
+}
+
+error_t *tokenlist_entry_alloc(tokenlist_entry_t **output) {
+    *output = nullptr;
+
+    tokenlist_entry_t *entry = calloc(1, sizeof(tokenlist_entry_t));
+    if (entry == nullptr)
+        return err_allocation_failed;
+
+    entry->next = nullptr;
+    entry->prev = nullptr;
+
+    *output = entry;
+    return nullptr;
+}
+
+void tokenlist_append(tokenlist_t *list, tokenlist_entry_t *entry) {
+    if (list->head == nullptr) {
+        list->head = entry;
+        list->tail = entry;
+        entry->next = nullptr;
+        entry->prev = nullptr;
+    } else {
+        entry->prev = list->tail;
+        entry->next = nullptr;
+        list->tail->next = entry;
+        list->tail = entry;
+    }
+}
+
+void tokenlist_entry_free(tokenlist_entry_t *entry) {
+    lexer_token_cleanup(&entry->token);
+    free(entry);
+}
+
+void tokenlist_free(tokenlist_t *list) {
+    if (list == nullptr)
+        return;
+
+    tokenlist_entry_t *current = list->head;
+    while (current) {
+        tokenlist_entry_t *next = current->next;
+        tokenlist_entry_free(current);
+        current = next;
+    }
+
+    free(list);
+}
+
+error_t *tokenlist_fill(tokenlist_t *list, lexer_t *lex) {
+    error_t *err = nullptr;
+    lexer_token_t token = {};
+    while ((err = lexer_next(lex, &token)) == nullptr) {
+        tokenlist_entry_t *entry;
+        err = tokenlist_entry_alloc(&entry);
+        if (err) {
+            lexer_token_cleanup(&token);
+            return err;
+        }
+        entry->token = token;
+        tokenlist_append(list, entry);
+    }
+    if (err != err_eof)
+        return err;
+    return nullptr;
+}
+
+bool is_trivia(tokenlist_entry_t *trivia) {
+    switch (trivia->token.id) {
+    case TOKEN_WHITESPACE:
+    case TOKEN_COMMENT:
+    case TOKEN_NEWLINE:
+        return true;
+    default:
+        return false;
+    }
+}
+
+tokenlist_entry_t *tokenlist_skip_trivia(tokenlist_entry_t *current) {
+    while (current && is_trivia(current))
+        current = current->next;
+    return current;
+}
+
+tokenlist_entry_t *tokenlist_next(tokenlist_entry_t *current) {
+    if (!current)
+        return nullptr;
+    return tokenlist_skip_trivia(current->next);
+}

src/tokenlist.h

@@ -0,0 +1,40 @@
+#ifndef INCLUDE_SRC_TOKENLIST_H_
+#define INCLUDE_SRC_TOKENLIST_H_
+#include "lexer.h"
+
+typedef struct tokenlist_entry tokenlist_entry_t;
+
+struct tokenlist_entry {
+    lexer_token_t token;
+    tokenlist_entry_t *next;
+    tokenlist_entry_t *prev;
+};
+
+typedef struct tokenlist {
+    tokenlist_entry_t *head;
+    tokenlist_entry_t *tail;
+} tokenlist_t;
+
+/**
+ * @brief Allocate a new doubly linked list of lexer tokens
+ */
+error_t *tokenlist_alloc(tokenlist_t **list);
+
+/**
+ * Consume all tokens from the lexer and add them to the list
+ */
+error_t *tokenlist_fill(tokenlist_t *list, lexer_t *lex);
+
+void tokenlist_free(tokenlist_t *list);
+
+/**
+ * Return the first token entry that isn't whitespace, newline or comment
+ */
+tokenlist_entry_t *tokenlist_skip_trivia(tokenlist_entry_t *current);
+
+/**
+ * Return the next token entry that isn't whitespace, newline or comment
+ */
+tokenlist_entry_t *tokenlist_next(tokenlist_entry_t *current);
+
+#endif // INCLUDE_SRC_TOKENLIST_H_

tests/input/valid.asm

@@ -1,5 +1,17 @@
+.section text
+
+; Small valid code snippet that should contain all different AST nodes
+
 _start:
-    mov eax, 555            ; move 555 into eax
+    mov eax, ebx
+    lea eax, [eax + ebx * 4 + 8]
+    lea eax, [eax + 8]
+    lea eax, [eax + ebx * 8]
+    lea eax, [esp - 24]
+    lea eax, [eax + ebx * 4 - 8]
+    lea eax, [_start]
+    mov eax, _start
+    mov eax, 555
     push 0o777
     xor eax, 0xDEADBEEF
     and ecx, 0o770

validate.sh

@@ -10,7 +10,7 @@ scan-build -o reports/static-analysis/ -plist-html --status-bugs make all
 # Run the sanitizer builds and valgrind
 make clean sanitize all
 
-ARGUMENTS=("-tokens" "-text")
+ARGUMENTS=("tokens" "text" "ast")
 while IFS= read -r INPUT_FILE; do
     for ARGS in ${ARGUMENTS[@]}; do
         ./oas-asan $ARGS $INPUT_FILE > /dev/null