Compare commits

...

10 Commits

Author SHA1 Message Date
Aryadev Chavali
f4e0be8977 struct Label -> label_t 2024-12-24 23:21:01 +00:00
Aryadev Chavali
9fb9413c45 node_t -> ast_node_t
Clearer to understand the intent of node_t as a direct result of
parsing source code.
2024-12-24 23:18:52 +00:00
Aryadev Chavali
fbf2202cfd Remove asm_write_init and asm_write_exit
Just inline into asm_translate_nodes
2024-12-24 23:15:04 +00:00
Aryadev Chavali
3c39df6c95 Implement SIZE_FMT and vec_append_fmt
SIZE_FMT gets the size of a cstr with the given sprintf formatted
string and arguments using `snprintf`.

vec_append_fmt formats a cstr with the given sprintf arguments then
appends it to the given vector.

This simplifies a little bit of the wordy implementation of the
assembler.
2024-12-24 23:08:52 +00:00
Aryadev Chavali
70cfeac995 Move system calling code to the end 2024-12-24 22:57:28 +00:00
Aryadev Chavali
493b98c524 Change vec_append signature to ensure ptr is constant 2024-12-24 22:54:57 +00:00
Aryadev Chavali
2c1c89c048 Separate out node type as an enumeration. 2024-12-19 08:48:44 +00:00
Aryadev Chavali
b03ef6564a Refactor obc for new vector based assembler and better config parser
Simpler implementation for configuration parsing with default values.
Now using new assembler functions.
2024-12-19 08:46:19 +00:00
Aryadev Chavali
7686c1043c Refactor assembler to use vec_t instead of buffer
Clearer, precise and no double pointers.
2024-12-19 08:45:47 +00:00
Aryadev Chavali
ffc8068a74 Introduce and implement vec_t
A vec_t is a dynamic array, separate to a buffer.  Should make it
easier to distinguish purposes using these two data types.
2024-12-19 08:44:53 +00:00
8 changed files with 193 additions and 160 deletions

View File

@@ -9,79 +9,10 @@
#include <string.h>
// Structure representing a loop operative as indices into an ASM label array.
struct Label
typedef struct
{
u64 cur, next;
};
/* First, the functions to call external utilities (badly written, I know). They
* presume a filled asm_buffer. */
void asm_write(buffer_t **asm_buffer)
{
FILE *fp = fopen(asm_buffer[0]->name, "w");
fwrite(asm_buffer[0]->data, asm_buffer[0]->size, 1, fp);
fclose(fp);
}
int asm_assemble(const char *srcname, const char *objname)
{
#ifdef DEBUG
char *format_str = "yasm -f elf64 -g dwarf2 -o %s %s";
#else
char *format_str = "yasm -f elf64 -o %s %s";
#endif
char command[snprintf(NULL, 0, format_str, objname, srcname) + 1];
sprintf(command, format_str, objname, srcname);
return system(command);
}
int asm_link(const char *objname, const char *outname)
{
char *format_str = "ld -o %s %s";
char command[snprintf(NULL, 0, format_str, outname, objname) + 1];
sprintf(command, format_str, outname, objname);
return system(command);
}
int asm_compile(buffer_t **asm_buffer, const char *objname, const char *outname)
{
asm_write(asm_buffer);
asm_assemble(asm_buffer[0]->name, objname);
return asm_link(objname, outname);
}
void asm_setup_buffer(buffer_t **asm_buffer, const char *outname)
{
*asm_buffer = buffer_init_str(outname, NULL, 1024);
asm_buffer[0]->size = 0;
}
// Write the initial boilerplate of the assembly file
void asm_write_init(buffer_t **asm_buffer)
{
const char format_string[] = "section .bss\n"
" memory resb %d\n"
"section .text\n"
"global _start\n"
"_start:\n"
" mov r9, memory\n";
char initial_assembly[snprintf(NULL, 0, format_string, MEMORY_DEFAULT) + 1];
snprintf(initial_assembly, sizeof(initial_assembly), format_string,
MEMORY_DEFAULT);
buffer_append_bytes(asm_buffer, (u8 *)initial_assembly,
sizeof(initial_assembly) - 1);
}
// Write the exit code for the assembly file
void asm_write_exit(buffer_t **asm_buffer)
{
buffer_append_bytes(asm_buffer,
(u8 *)" mov rax, 60\n"
" mov rdi, 0\n"
" syscall\n",
37);
}
} label_t;
// A table which translates brainfuck operations into generic assembly code (I
// love C's array indexing)
@@ -110,87 +41,94 @@ static const struct
// Linear search in the ASM label array for a label with the current AST
// relative index
i64 get_abs_label(u64, struct Label *, u64);
i64 get_abs_label(u64, label_t *, u64);
// Translate an AST relative jump to a specific ASM label, storing the current
// and next items in given pointers.
void ast_ref_to_asm_label(u64, struct Label *, u64, i64 *, i64 *);
void ast_ref_to_asm_label(u64, label_t *, u64, i64 *, i64 *);
void asm_translate_nodes(buffer_t **asm_buffer, struct PResult nodes,
void asm_translate_nodes(vec_t *asm_buffer, struct PResult nodes,
const char *src_name)
{
asm_write_init(asm_buffer);
vec_append_fmt(asm_buffer,
"section .bss\n"
" memory resb %d\n"
"section .text\n"
"global _start\n"
"_start:\n"
" mov r9, memory\n",
MEMORY_DEFAULT);
// First pass: Setup the ASM label array
struct Label labels[nodes.labels ? nodes.labels * 2 : 1];
label_t labels[nodes.labels ? nodes.labels * 2 : 1];
if (nodes.labels)
{
u64 label_ptr = 0;
for (size_t i = 0; i < nodes.size; ++i)
if (nodes.nodes[i].type == LIN || nodes.nodes[i].type == LOUT)
labels[label_ptr++] =
(struct Label){.cur = i, .next = nodes.nodes[i].loop_ref};
(label_t){.cur = i, .next = nodes.nodes[i].loop_ref};
}
// Second pass: Translating to assembly
for (size_t i = 0; i < nodes.size; ++i)
{
node_t node = nodes.nodes[i];
if (node.type == LIN || node.type == LOUT)
ast_node_t node = nodes.nodes[i];
// Something we can compile with only the information at hand
if (!(node.type == LIN || node.type == LOUT))
{
// Translate the node-relative addresses to assembly-label-relative
// addresses
i64 cur_asm_label = 0, next_asm_label = 0;
ast_ref_to_asm_label(i, labels, nodes.labels * 2, &cur_asm_label,
&next_asm_label);
if (cur_asm_label == -1 || next_asm_label == -1)
{
print_error(
src_name, node.row, node.col,
"[ASSEMBLY ERROR]: Could not find label for current jump!\n");
exit(1);
}
// I love tables so goddamn much
vec_append(asm_buffer, table[node.type].str, table[node.type].len);
continue;
}
// Format labels
char current_label[128], next_label[128];
// NOTE: Must be a label
sprintf(current_label, ".L%lu", cur_asm_label);
sprintf(next_label, ".L%lu", next_asm_label);
// Translate the node-relative addresses to assembly-label-relative
// addresses
i64 cur_asm_label = 0, next_asm_label = 0;
ast_ref_to_asm_label(i, labels, nodes.labels * 2, &cur_asm_label,
&next_asm_label);
if (cur_asm_label == -1 || next_asm_label == -1)
{
print_error(src_name, node.row, node.col,
"[ASSEMBLY ERROR]: Could not find label for current jump!\n");
exit(1);
}
// Setup format string for assembly jump code
char *format_string = NULL;
if (node.type == LIN)
{
format_string = "%s:\n"
" cmp byte [r9], 0\n"
" je %s\n";
}
else
{
format_string = "%s:\n"
" cmp byte [r9], 0\n"
" jne %s\n";
}
char formatted_string[snprintf(NULL, 0, format_string, current_label,
next_label) +
1];
snprintf(formatted_string, sizeof(formatted_string), format_string,
current_label, next_label);
buffer_append_bytes(asm_buffer, (u8 *)formatted_string,
sizeof(formatted_string) - 1);
// Format labels
char current_label[128], next_label[128];
sprintf(current_label, ".L%lu", cur_asm_label);
sprintf(next_label, ".L%lu", next_asm_label);
// Setup format string for assembly jump code
char *format_string = NULL;
if (node.type == LIN)
{
format_string = "%s:\n"
" cmp byte [r9], 0\n"
" je %s\n";
}
else
{
// I love tables so goddamn much
buffer_append_bytes(asm_buffer, (u8 *)table[node.type].str,
table[node.type].len);
format_string = "%s:\n"
" cmp byte [r9], 0\n"
" jne %s\n";
}
vec_append_fmt(asm_buffer, format_string, current_label, next_label);
}
asm_write_exit(asm_buffer);
vec_append(asm_buffer,
" mov rax, 60\n"
" mov rdi, 0\n"
" syscall\n",
37);
}
/* Implementations for throwaway functions */
i64 get_abs_label(u64 ref, struct Label *labels, u64 size)
i64 get_abs_label(u64 ref, label_t *labels, u64 size)
{
for (u64 i = 0; i < size; ++i)
if (labels[i].cur == ref)
@@ -198,7 +136,7 @@ i64 get_abs_label(u64 ref, struct Label *labels, u64 size)
return -1;
}
void ast_ref_to_asm_label(u64 ref, struct Label *labels, u64 size, i64 *cur,
void ast_ref_to_asm_label(u64 ref, label_t *labels, u64 size, i64 *cur,
i64 *next)
{
*cur = get_abs_label(ref, labels, size);
@@ -206,3 +144,40 @@ void ast_ref_to_asm_label(u64 ref, struct Label *labels, u64 size, i64 *cur,
return;
*next = get_abs_label(labels[*cur].next, labels, size);
}
/* The system calling code - not exactly pretty, but it is what it is. */
void asm_write(const char *asm_name, vec_t *asm_buffer)
{
FILE *fp = fopen(asm_name, "w");
fwrite(asm_buffer->data, asm_buffer->size, 1, fp);
fclose(fp);
}
int asm_assemble(const char *asm_name, const char *objname)
{
#ifdef DEBUG
char *format_str = "yasm -f elf64 -g dwarf2 -o %s %s";
#else
char *format_str = "yasm -f elf64 -o %s %s";
#endif
char command[SIZE_FMT(format_str, objname, asm_name) + 1];
sprintf(command, format_str, objname, asm_name);
return system(command);
}
int asm_link(const char *objname, const char *outname)
{
char *format_str = "ld -o %s %s";
char command[SIZE_FMT(format_str, outname, objname) + 1];
sprintf(command, format_str, outname, objname);
return system(command);
}
int asm_compile(vec_t *asm_buffer, const char *asm_name, const char *objname,
const char *outname)
{
asm_write(asm_name, asm_buffer);
asm_assemble(asm_name, objname);
return asm_link(objname, outname);
}

View File

@@ -4,14 +4,13 @@
#include "./lib.h"
#include "./parser.h"
void asm_setup_buffer(buffer_t **asm_buffer, const char *outname);
void asm_translate_nodes(buffer_t **asm_buffer, struct PResult nodes,
void asm_translate_nodes(vec_t *asm_buffer, struct PResult nodes,
const char *src_name);
void asm_write(buffer_t **asm_buffer);
int asm_assemble(const char *srcname, const char *objname);
void asm_write(const char *asm_name, vec_t *asm_buffer);
int asm_assemble(const char *asm_name, const char *objname);
int asm_link(const char *objname, const char *outname);
int asm_compile(buffer_t **asm_buffer, const char *objname,
int asm_compile(vec_t *asm_buffer, const char *asm_name, const char *objname,
const char *outname);
#endif

38
lib.c
View File

@@ -8,6 +8,12 @@
#include "./lib.h"
bool usable_character(char c)
{
return c == '>' || c == '<' || c == '+' || c == '-' || c == '-' || c == '.' ||
c == ',' || c == '[' || c == ']';
}
buffer_t *buffer_init_str(const char *name, const char *str, size_t str_size)
{
buffer_t *buf = malloc(sizeof(*buf) + str_size + 1);
@@ -61,8 +67,34 @@ char *fread_all(FILE *fp)
return buffer.data;
}
bool usable_character(char c)
void vec_append(vec_t *vec, const void *const ptr, u64 size)
{
return c == '>' || c == '<' || c == '+' || c == '-' || c == '-' || c == '.' ||
c == ',' || c == '[' || c == ']';
vec_ensure_free(vec, size);
memcpy(vec->data + vec->size, ptr, size);
vec->size += size;
}
void vec_ensure(vec_t *vec, u64 abs_size)
{
if (vec->capacity < abs_size)
{
vec->capacity = MAX(vec->capacity * 2, abs_size);
vec->data = realloc(vec->data, vec->capacity);
}
else if (!vec->data)
{
vec->data = calloc(1, vec->capacity);
}
}
void vec_ensure_free(vec_t *vec, u64 rel_size)
{
vec_ensure(vec, vec->size + rel_size);
}
void vec_free(vec_t *vec)
{
if (vec->data)
free(vec->data);
memset(vec, 0, sizeof(*vec));
}

22
lib.h
View File

@@ -26,6 +26,8 @@ char *fread_all(FILE *fp);
void print_error(const char *handle, size_t row, size_t column,
const char *reason);
#define SIZE_FMT(FMT, ...) (snprintf(NULL, 0, (FMT), __VA_ARGS__))
typedef struct Buffer
{
const char *name;
@@ -34,4 +36,24 @@ typedef struct Buffer
} buffer_t;
buffer_t *buffer_init_str(const char *name, const char *str, size_t str_size);
typedef struct
{
u64 size, capacity;
u8 *data;
} vec_t;
void vec_append(vec_t *vec, const void *const ptr, u64 size);
void vec_ensure(vec_t *vec, u64 abs_size);
void vec_ensure_free(vec_t *vec, u64 rel_size);
void vec_free(vec_t *vec);
#define vec_append_fmt(VEC, FMT, ...) \
do \
{ \
char f[SIZE_FMT(FMT, __VA_ARGS__) + 1]; \
snprintf(f, sizeof(f), (FMT), __VA_ARGS__); \
vec_append((VEC), f, sizeof(f) - 1); \
} while (0)
#endif

21
obc.c
View File

@@ -59,6 +59,7 @@ int parse_config(struct Config *config, int argc, char *argv[])
}
if (!config->prog_name)
return 0;
return 1;
}
@@ -70,8 +71,9 @@ int main(int argc, char *argv[])
return 1;
}
char *file_data = NULL;
buffer_t *buffer = NULL, *asm_buffer = NULL;
char *file_data = NULL;
buffer_t *buffer = NULL;
vec_t asm_buffer = {0};
struct PResult res = {0};
struct Config config = {0};
@@ -81,6 +83,12 @@ int main(int argc, char *argv[])
usage(argv[0], stderr);
goto end;
}
if (!config.asm_name)
config.asm_name = "a.asm";
if (!config.obj_name)
config.obj_name = "a.o";
if (!config.exec_name)
config.exec_name = "a.out";
ret = 0;
FILE *handle = fopen(config.prog_name, "r");
@@ -102,15 +110,10 @@ int main(int argc, char *argv[])
goto end;
}
asm_setup_buffer(&asm_buffer, config.asm_name ? config.asm_name : "a.asm");
asm_translate_nodes(&asm_buffer, res, config.prog_name);
asm_write(&asm_buffer);
asm_assemble(asm_buffer->name, config.obj_name ? config.obj_name : "a.o");
asm_link(config.obj_name ? config.obj_name : "a.o",
config.exec_name ? config.exec_name : "a.out");
asm_compile(&asm_buffer, config.asm_name, config.obj_name, config.exec_name);
end:
if (asm_buffer)
free(asm_buffer);
vec_free(&asm_buffer);
if (res.nodes)
free(res.nodes);
if (buffer)

4
obi.c
View File

@@ -25,11 +25,11 @@ typedef struct
#endif
} machine_t;
void interpret(machine_t *cpu, node_t *ast, size_t num)
void interpret(machine_t *cpu, ast_node_t *ast, size_t num)
{
for (size_t i = 0; i < num; ++i)
{
node_t node = ast[i];
ast_node_t node = ast[i];
switch (node.type)
{
case NEXT:

View File

@@ -8,7 +8,7 @@
#include "./parser.h"
char *ast_to_str(node_t *ast, size_t size)
char *ast_to_str(ast_node_t *ast, size_t size)
{
char *out = calloc(size + 1, 1);
if (!out)
@@ -54,7 +54,7 @@ char *ast_to_str(node_t *ast, size_t size)
struct PResult parse_buffer(buffer_t *buffer)
{
node_t *nodes = NULL;
ast_node_t *nodes = NULL;
size_t usable = 0, loops = 0, loop_ends = 0;
// First pass: Compute |nodes|
@@ -161,11 +161,11 @@ struct PResult parse_buffer(buffer_t *buffer)
size_t stackptr = 0;
if (loops)
{
node_t *stack[loops];
ast_node_t *stack[loops];
memset(stack, 0, loops);
for (size_t i = 0; i < usable; ++i)
{
node_t *node = nodes + i;
ast_node_t *node = nodes + i;
if (node->type == LIN)
stack[stackptr++] = node;
else if (node->type == LOUT)
@@ -180,7 +180,7 @@ struct PResult parse_buffer(buffer_t *buffer)
if (stackptr > 0)
{
node_t *node = nodes + usable - 1;
ast_node_t *node = nodes + usable - 1;
print_error(buffer->name, node->row, node->col,
"ERROR: Unbalanced square brackets!");
goto error;

View File

@@ -3,31 +3,33 @@
#include "./lib.h"
typedef enum
{
NEXT = 0,
PREV,
INC,
DEC,
OUT,
READ,
LIN,
LOUT
} node_type_t;
typedef struct
{
size_t col, row;
enum
{
NEXT = 0,
PREV,
INC,
DEC,
OUT,
READ,
LIN,
LOUT
} type;
node_type_t type;
int loop_ref;
} node_t;
} ast_node_t;
struct PResult
{
node_t *nodes;
ast_node_t *nodes;
size_t size;
size_t labels;
};
char *ast_to_str(node_t *ast, size_t size);
char *ast_to_str(ast_node_t *ast, size_t size);
struct PResult parse_buffer(buffer_t *buffer);
#endif