Merge pull request #39 from cyuria/merge-passes

feat: merge assembly passes into one

.github/workflows/test.yml

@@ -52,7 +52,7 @@ jobs:
     steps:
     - uses: actions/checkout@v4
 
-    - run: sudo apt install qemu-user lld
+    - run: sudo apt update && sudo apt install qemu-user lld
 
     - uses: ./.github/setup_cc_env
       with: 

h/generation.h

@@ -6,5 +6,4 @@
 void parse_file(FILE *, FILE *);
 int parse_line(char *, struct sectionpos);
 
-int symbol_forward_declare(char *);
 int parse_label(char *, struct sectionpos);

h/symbols.h

@@ -26,6 +26,7 @@ extern struct symbolmap symbols[SYMBOLMAP_ENTRIES];
 
 struct symbol *get_symbol(const char *);
 struct symbol *create_symbol(const char *, enum symbol_types);
+struct symbol *get_or_create_symbol(const char *, enum symbol_types);
 
 struct elf64sym create_symtab_entry(const char *);
 

src/bytecode.c

@@ -7,6 +7,7 @@
 #include "debug.h"
 #include "elf/output.h"
 #include "form/generic.h"
+#include "symbols.h"
 #include "xmalloc.h"
 
 static struct instruction *instructions = NULL;
@@ -69,6 +70,12 @@ int write_instruction(struct instruction i)
 	       "Generating bytecode for %s instruction (offset: %zu)",
 	       i.formation.name, i.position.offset);
 	set_section(i.position.section);
+
+	if (i.args.sym)
+		if (i.args.sym->section == SECTION_NULL)
+			logger(ERROR, error_unknown, "Symbol %s not found",
+			       i.args.sym->name);
+
 	struct bytecode bytecode =
 		i.formation.form_handler(i.formation.name, i.formation.idata,
 					 i.args, calc_fileoffset(i.position));

src/directives.c

@@ -164,10 +164,10 @@ int parse_section(const char *str)
 }
 int parse_global(const char *str)
 {
-	struct symbol *sym = get_symbol(str);
+	struct symbol *sym = get_or_create_symbol(str, SYMBOL_LABEL);
 	if (!sym) {
-		logger(ERROR, error_unknown, "Uknown symbol (%s) encountered",
-		       sym->name);
+		logger(ERROR, error_internal, "Uknown symbol %s encountered",
+		       str);
 		return 1;
 	}
 	sym->binding = 0x10;

src/generation.c

@@ -69,8 +69,6 @@ void parse_file(FILE *ifp, FILE *ofp)
 
 	linenumber = 0;
 
-	while ((nread = getl(&line, &linesize, ifp)) != (size_t)-1)
-		symbol_forward_declare(line);
 	fseek(ifp, 0L, SEEK_SET);
 
 	while ((nread = getl(&line, &linesize, ifp)) != (size_t)-1) {
@@ -104,7 +102,7 @@ void parse_file(FILE *ifp, FILE *ofp)
 	free_symbols();
 }
 
-static int parse_line_trimmed(char *, struct sectionpos);
+static inline int parse_line_trimmed(char *, struct sectionpos);
 int parse_line(char *line, struct sectionpos position)
 {
 	char *trimmed_line = trim_whitespace(line);
@@ -113,63 +111,40 @@ int parse_line(char *line, struct sectionpos position)
 	return result;
 }
 
-static int parse_line_trimmed(char *line, struct sectionpos position)
+static inline int parse_line_trimmed(char *line, struct sectionpos position)
 {
 	logger(DEBUG, no_error, " |-> \"%s\"", line);
 
-	if (*line == '\0')
+	switch (*line) {
+	case '\0':
+	case ';':
 		return 0;
-
-	if (*line == '.' || *line == '[')
+	case '.':
+	case '[':
 		return parse_directive(line);
-	if (*line == ';')
-		return 0;
-	if (*line == '/' && *(line + 1) == '/')
-		return 0;
+	case '/':
+		if (line[1] == '/')
+			return 0;
+	}
+
 	if (strchr(line, ':'))
 		return parse_label(line, position);
 
 	return parse_asm(line, position);
 }
 
-int symbol_forward_declare(char *line)
-{
-	char *colon = strchr(line, ':');
-	if (colon == NULL)
-		return 0;
-	*colon = '\0';
-	char *name = trim_whitespace(line);
-	struct symbol *sym = create_symbol(name, SYMBOL_LABEL);
-	free(name);
-	return !sym;
-}
-
 int parse_label(char *line, struct sectionpos position)
 {
 	char *end = strchr(line, ':');
 	/* check for invalid label definition */
 	if (end == NULL)
 		return 0;
 
-	logger(DEBUG, no_error, "Setting position of label (%s)", line);
+	logger(DEBUG, no_error, "Creating label (%s)", line);
 
 	*(end++) = '\0';
 	char *name = trim_whitespace(line);
-
-	struct symbol *label = get_symbol(name);
-	if (!label) {
-		logger(ERROR, error_internal, "Unknown label encountered (%s)",
-		       name);
-		free(name);
-		return 1;
-	}
-	if (label->type != SYMBOL_LABEL) {
-		logger(ERROR, error_internal,
-		       "%s is defined but is not a label", name);
-		free(name);
-		return 1;
-	}
-
+	struct symbol *label = get_or_create_symbol(name, SYMBOL_LABEL);
 	free(name);
 
 	const struct sectionpos fpos = get_outputpos();

src/main.c

@@ -13,6 +13,16 @@ FILE *inputfile = NULL;
 FILE *outputtempfile = NULL;
 FILE *outputfile = NULL;
 
+static int open(FILE **f, const char *filename, const char *flags)
+{
+#ifdef __STDC_LIB_EXT1__
+	return fopen_s(f, filename, flags));
+#else
+	*f = fopen(filename, flags);
+	return !*f;
+#endif
+}
+
 void closefiles(void)
 {
 	if (inputfile)
@@ -35,13 +45,8 @@ void open_files(void)
 	}
 
 	logger(DEBUG, no_error, "Opening %s", *cmdargs.inputfile->filename);
-#ifdef __STDC_LIB_EXT1__
-	if (fopen_s(&inputfile, *cmdargs.inputfile->filename, "r"))
-#else
-	inputfile = fopen(*cmdargs.inputfile->filename, "r");
-	if (!inputfile)
-#endif
-	{
+	if (open(&inputfile, *cmdargs.inputfile->filename, "r")) {
+		perror("Error: ");
 		logger(ERROR, error_system, "Unable to open input file");
 		exit(EXIT_FAILURE);
 	}
@@ -53,15 +58,9 @@ void open_files(void)
 	}
 
 	logger(DEBUG, no_error, "Opening %s", *cmdargs.outputfile->filename);
-#ifdef __STDC_LIB_EXT1__
-	if (fopen_s(&outputfile, *cmdargs.outputfile->filename, "wb"))
-#else
-	outputfile = fopen(*cmdargs.outputfile->filename, "wb");
-	if (!outputfile)
-#endif
-	{
-		logger(ERROR, error_system, "Unable to open output file");
+	if (open(&outputfile, *cmdargs.outputfile->filename, "wb")) {
 		perror("Error: ");
+		logger(ERROR, error_system, "Unable to open output file");
 		exit(EXIT_FAILURE);
 	}
 

src/parse.c

@@ -14,11 +14,7 @@
 #include "symbols.h"
 #include "xmalloc.h"
 
-const struct args empty_args = { .rd = 0,
-				 .rs1 = 0,
-				 .rs2 = 0,
-				 .imm = 0,
-				 .sym = NULL };
+const struct args empty_args = { .sym = NULL };
 
 static char *trim_arg(char *s)
 {
@@ -82,15 +78,6 @@ static uint32_t expect_imm(char *arg)
 	return (uint32_t)imm;
 }
 
-static struct symbol *expect_sym(char *arg)
-{
-	struct symbol *sym = get_symbol(arg);
-	if (!sym)
-		logger(ERROR, error_instruction_other,
-		       "Expected symbol but got %s", arg);
-	return sym;
-}
-
 static int expect_one_arg(char *first)
 {
 	if (strtok(NULL, ",")) {
@@ -220,6 +207,7 @@ struct args parse_rtype(char *argstr)
 		.rd = expect_reg(first),
 		.rs1 = expect_reg(second),
 		.rs2 = expect_reg(third),
+		.sym = NULL,
 	};
 
 	free(first);
@@ -248,6 +236,7 @@ struct args parse_itype(char *argstr)
 		.rd = expect_reg(first),
 		.rs1 = expect_reg(second),
 		.imm = expect_imm(third),
+		.sym = NULL,
 	};
 
 	free(first);
@@ -273,6 +262,7 @@ struct args parse_ltype(char *argstr)
 
 	struct args args = {
 		.rd = expect_reg(first),
+		.sym = NULL,
 	};
 
 	expect_offreg(second, &args.imm, &args.rs1);
@@ -299,6 +289,7 @@ struct args parse_stype(char *argstr)
 
 	struct args args = {
 		.rs2 = expect_reg(second),
+		.sym = NULL,
 	};
 
 	expect_offreg(first, &args.imm, &args.rs1);
@@ -326,6 +317,7 @@ struct args parse_utype(char *argstr)
 	struct args args = {
 		.rd = expect_reg(first),
 		.imm = expect_imm(second),
+		.sym = NULL,
 	};
 
 	free(first);
@@ -351,7 +343,7 @@ struct args parse_btype(char *argstr)
 	struct args args = {
 		.rs1 = expect_reg(first),
 		.rs2 = expect_reg(second),
-		.sym = expect_sym(third),
+		.sym = get_or_create_symbol(third, SYMBOL_LABEL),
 	};
 
 	free(first);
@@ -377,7 +369,7 @@ struct args parse_bztype(char *argstr)
 
 	struct args args = {
 		.rs1 = expect_reg(first),
-		.sym = expect_sym(second),
+		.sym = get_or_create_symbol(second, SYMBOL_LABEL),
 	};
 
 	free(first);
@@ -403,6 +395,7 @@ struct args parse_pseudo(char *argstr)
 	struct args args = {
 		.rd = expect_reg(first),
 		.rs1 = expect_reg(second),
+		.sym = NULL,
 	};
 
 	free(first);
@@ -444,6 +437,7 @@ struct args parse_fence(char *argstr)
 			.rd = 0x0,
 			.rs1 = 0x0,
 			.imm = 0xFF,
+			.sym = NULL,
 		};
 
 	if (strtok(NULL, ","))
@@ -466,6 +460,7 @@ struct args parse_fence(char *argstr)
 		.rd = 0x0,
 		.rs1 = 0x0,
 		.imm = (predecessor << 4) | successor,
+		.sym = NULL,
 	};
 }
 
@@ -499,7 +494,7 @@ struct args parse_jal(char *argstr)
 			       "Expected a second argument");
 	}
 
-	args.sym = expect_sym(sym);
+	args.sym = get_or_create_symbol(sym, SYMBOL_LABEL);
 	free(sym);
 
 	logger(DEBUG, no_error, "Registers parsed, x%d, %s", args.rd,
@@ -572,7 +567,7 @@ struct args parse_la(char *argstr)
 
 	struct args args = {
 		.rd = expect_reg(first),
-		.sym = expect_sym(second),
+		.sym = get_or_create_symbol(second, SYMBOL_LABEL),
 	};
 
 	free(first);
@@ -618,11 +613,9 @@ struct args parse_j(char *argstr)
 		return empty_args;
 
 	struct args args = {
-		.sym = expect_sym(first),
+		.sym = get_or_create_symbol(first, SYMBOL_LABEL),
 	};
 
-	free(first);
-
 	logger(DEBUG, no_error, "Symbol parsed %s", args.sym->name);
 
 	return args;

src/symbols.c

@@ -32,11 +32,28 @@ struct symbol *get_symbol(const char *name)
 	return NULL;
 }
 
+struct symbol *get_or_create_symbol(const char *name, enum symbol_types type)
+{
+	struct symbol *sym = get_symbol(name);
+	if (sym)
+		return sym;
+	return create_symbol(name, type);
+}
+
 struct symbol *create_symbol(const char *name, enum symbol_types type)
 {
 	const size_t hash = hash_str(name);
+
 	const size_t index = symbols[hash].count;
 
+	for (size_t i = 0; i < symbols[hash].count; i++) {
+		if (!strcmp(name, symbols[hash].data[i].name)) {
+			logger(ERROR, error_invalid_syntax,
+			       "Duplicate symbol %s encountered");
+			return NULL;
+		}
+	}
+
 	symbols[hash].count++;
 	symbols[hash].data =
 		xrealloc(symbols[hash].data,