initial mips32 (r2000ish mips32r6) assembler

1 files changed, 365 insertions, 0 deletions

diff --git a/masm/asm_mips32.c b/masm/asm_mips32.c
new file mode 100644
index 0000000..dcb81e5
--- /dev/null
+++ b/masm/asm_mips32.c
@@ -0,0 +1,365 @@
+#include <merror.h>
+#include <mips.h>
+#include <mips32.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <elf.h>
+#include <string.h>
+#include <stddef.h>
+
+#include "asm.h"
+#include "mlimits.h"
+#include "parse.h"
+#include "parse_mips32.h"
+
+extern char *current_file;
+
+#define SHDR_SYMTBL   0
+#define SHDR_STRTBL   1
+#define SHDR_SECTIONS 2
+
+static int parse_file(struct parser *parser)
+{
+	while (1) {
+		struct expr expr;
+		if (parser_next(parser, &expr)) {
+			break;
+		}
+
+                if (expr.type == EXPR_INS)
+                        if (sectbl_push(&parser->sec_tbl,
+                                        parser->sec_tbl.current, expr.ins))
+                                return M_ERROR;
+	}
+
+        for (uint32_t i = 0; i < parser->ref_tbl.count; i++) {
+                struct reference *ref = &parser->ref_tbl.references[i];
+                struct symbol *sym;
+                struct mips32_instruction *ins;
+
+                if (symtbl_find(&parser->sym_tbl, &sym, ref->name)) {
+                        ERROR("undefined symbol '%s'", ref->name);
+                        return M_ERROR;
+                }
+
+                ins = &ref->section->ins[ref->index].mips32;
+
+                switch (ref->type) {
+                case REF_OFFESET:
+                        ins->B_data.offset += sym->position -
+                                        (ref->section->start + ref->index);
+                        break;
+                case REF_TARGET:
+                        ins->J_data.target += sym->position;
+                        break;
+                }
+        };
+
+        return M_SUCCESS;
+}
+
+static int assemble_phdr(struct assembler *asm, Elf32_Phdr **res,
+                         uint32_t *res2)
+{
+        struct parser *parser = asm->parser;
+        Elf32_Phdr *phdr = malloc(sizeof(Elf32_Phdr) *
+                                  parser->sec_tbl.count);
+        size_t ins_sz = sizeof(struct mips32_instruction);
+
+        if (phdr == NULL) {
+                ERROR("cannot alloc");
+                return M_ERROR;;
+        }
+
+        for (uint32_t i = 0; i < parser->sec_tbl.count; i++) {
+                Elf32_Phdr *hdr = &phdr[i];
+                struct section *sec = &parser->sec_tbl.sections[i];
+
+                hdr->p_type = PT_LOAD;
+                hdr->p_flags = PF_X | PF_W | PF_R; // FIXME: this is bad
+                hdr->p_offset = sec->start * ins_sz;
+                hdr->p_vaddr = sec->start * ins_sz;
+                hdr->p_paddr = 0x00;
+                hdr->p_filesz = sec->count * ins_sz;
+                hdr->p_memsz = sec->count * ins_sz;
+                hdr->p_align = sec->alignment;
+        }
+
+        *res = phdr;
+        *res2 = parser->sec_tbl.count;
+        return M_SUCCESS;
+}
+
+static int assemble_symtbl(struct assembler *asm, Elf32_Sym **res,
+                           uint32_t *res2)
+{
+        Elf32_Sym *stbl = malloc(sizeof(Elf32_Sym) * asm->parser->sym_tbl
+                                 .count);
+
+        if (stbl == NULL)
+                return M_ERROR;
+
+        for (uint32_t i = 0; i < asm->parser->sym_tbl.count; i++) {
+                struct symbol *sym = &asm->parser->sym_tbl.symbols[i];
+                size_t str_off;
+
+                if (strtbl_write_str(&asm->str_tbl, sym->name, &str_off)) {
+                        free(stbl);
+                        return M_ERROR;
+                }
+
+                int viz = STB_LOCAL;
+                switch (sym->flag) {
+                case SYM_LOCAL:
+                        viz = STB_LOCAL;
+                        break;
+                case SYM_GLOBAL:
+                case SYM_EXTERNAL:
+                        viz = STB_GLOBAL;
+                        break;
+                }
+
+                stbl[i] = (Elf32_Sym) {
+                        .st_name = str_off,
+                        .st_value = sym->position,
+                        .st_size = 0,
+                        .st_info = (unsigned char)
+                                        ELF32_ST_INFO(SYMINFO_BT_SELF,
+                                                 SYMINFO_FLG_DIRECT),
+                        .st_other = (unsigned char)
+                                        ELF32_ST_VISIBILITY(viz),
+                        .st_shndx = 0, // FIXME: specify section
+                };
+        };
+
+        *res = stbl;
+        *res2 = asm->parser->sym_tbl.count;
+
+        return M_SUCCESS;
+}
+
+static int assemble_shdr(struct assembler *asm, Elf32_Shdr **res,
+                         uint32_t *res2)
+{
+        uint32_t entries = 2; // str table and sym tabel
+        entries += asm->parser->sec_tbl.count; // sections
+
+        Elf32_Shdr *shdr = malloc(sizeof(Elf32_Shdr) * entries);
+
+        size_t str_off;
+        if (strtbl_write_str(&asm->str_tbl, ".shsymtab", &str_off)) {
+                free(shdr);
+                return M_ERROR;
+        }
+
+        // symbol table
+        shdr[SHDR_SYMTBL] = (Elf32_Shdr) {
+                .sh_name = str_off,
+                .sh_type = SHT_SYMTAB,
+                .sh_flags = 0,
+                .sh_addr = 0,
+                .sh_offset = 0,
+                .sh_size = 0,
+                .sh_link = 1,
+                .sh_info = 0,
+                .sh_addralign = 1,
+                .sh_entsize = sizeof(Elf32_Sym),
+        };
+
+        if (strtbl_write_str(&asm->str_tbl, ".shstrtab", &str_off)) {
+                free(shdr);
+                return M_ERROR;
+        }
+
+        // string table
+        shdr[SHDR_STRTBL] = (Elf32_Shdr) {
+                .sh_name = str_off,
+                .sh_type = SHT_STRTAB,
+                .sh_flags = SHF_STRINGS,
+                .sh_addr = 0,
+                .sh_offset = 0,
+                .sh_size = 0,
+                .sh_link = 0,
+                .sh_info = 0,
+                .sh_addralign = 1,
+                .sh_entsize = 0,
+        };
+
+        // for each section
+        for (uint32_t i = 0; i < asm->parser->sec_tbl.count; i++) {
+                struct section *sec = &asm->parser->sec_tbl.sections[i];
+                char name[MAX_LEX_LENGTH+1] = ".";
+                strcat(name, sec->name);
+                if (strtbl_write_str(&asm->str_tbl, name, &str_off)) {
+                        free(shdr);
+                        return M_ERROR;
+                }
+                shdr[i+SHDR_SECTIONS] = (Elf32_Shdr) {
+                        .sh_name = str_off,
+                        .sh_type = SHT_PROGBITS,
+                        .sh_flags = SHF_WRITE | SHF_ALLOC | SHF_EXECINSTR,
+                        .sh_addr = 0,
+                        .sh_offset = 0,
+                        .sh_size = 0,
+                        .sh_link = 0,
+                        .sh_info = 0,
+                        .sh_addralign = sec->alignment,
+                        .sh_entsize = sizeof(struct mips32_instruction),
+                };
+        }
+
+        *res = shdr;
+        *res2 = entries;
+
+        return M_SUCCESS;
+}
+
+static int assemble_file(struct assembler *asm)
+{
+        Elf32_Phdr *phdr;
+        Elf32_Shdr *shdr;
+        Elf32_Sym *symtbl;
+        uint32_t phdr_len;
+        uint32_t shdr_len;
+        uint32_t symtbl_len;
+
+        if (assemble_symtbl(asm, &symtbl, &symtbl_len))
+                return M_ERROR;
+
+        if (assemble_phdr(asm, &phdr, &phdr_len)) {
+                free(symtbl);
+                return M_ERROR;
+        }
+
+        if (assemble_shdr(asm, &shdr, &shdr_len)) {
+                free(symtbl);
+                free(phdr);
+                return M_ERROR;
+        };
+
+        Elf32_Ehdr ehdr = {
+                .e_ident = {
+                        [EI_MAG0] = ELFMAG0,
+                        [EI_MAG1] = ELFMAG1,
+                        [EI_MAG2] = ELFMAG2,
+                        [EI_MAG3] = ELFMAG3,
+                        [EI_CLASS] = ELFCLASS32,
+                        [EI_DATA] = ELFDATA2LSB,
+                        [EI_VERSION] = EV_CURRENT,
+                        [EI_OSABI] = ELFOSABI_NONE,
+                        [EI_ABIVERSION] = 0x00,
+                        [EI_PAD] = 0x00,
+                },
+                .e_type = ET_REL,
+                .e_machine = EM_MIPS,
+                .e_version = EV_CURRENT,
+                .e_entry = 0x00,
+                .e_phoff = 0x00,
+                .e_shoff = 0x00,
+                .e_flags = EF_MIPS_ARCH_32R6,
+                .e_ehsize = sizeof(Elf32_Ehdr),
+                .e_phentsize = sizeof(Elf32_Phdr),
+                .e_phnum = phdr_len,
+                .e_shentsize = sizeof(Elf32_Shdr),
+                .e_shnum = shdr_len,
+                .e_shstrndx = SHDR_STRTBL,
+        };
+
+        uint32_t ptr = 0;
+
+        // we must now correct offets and sizes inside the ehdr, phdr,
+        // and shdr
+
+        ptr += sizeof(Elf32_Ehdr);
+
+        // phdr
+        ehdr.e_phoff = ptr;
+        ptr += phdr_len * sizeof(Elf32_Phdr);
+
+        // sections
+        for (uint32_t i = 0; i < asm->parser->sec_tbl.count; i++) {
+                phdr[i].p_offset = ptr;
+                phdr[i].p_vaddr = ptr;
+                shdr[i+SHDR_SECTIONS].sh_offset = ptr;
+                shdr[i+SHDR_SECTIONS].sh_size = phdr[i].p_filesz;
+                ptr += phdr[i].p_filesz;
+        }
+
+        // symtbl
+        shdr[SHDR_SYMTBL].sh_offset = ptr;
+        shdr[SHDR_SYMTBL].sh_size = symtbl_len * sizeof(Elf32_Sym);
+        ptr += symtbl_len * sizeof(Elf32_Sym);
+
+        // strtbl
+        shdr[SHDR_STRTBL].sh_offset = ptr;
+        shdr[SHDR_STRTBL].sh_size = asm->str_tbl.size;
+        ptr += asm->str_tbl.size;
+
+        // shdr
+        ehdr.e_shoff = ptr;
+
+        FILE *out = fopen("out.o", "w");
+
+        // ehdr
+        fwrite(&ehdr, sizeof(Elf32_Ehdr), 1, out);
+
+        // phdr
+        fwrite(phdr, sizeof(Elf32_Phdr), phdr_len, out);
+
+        // sections
+        for (uint32_t i = 0; i < asm->parser->sec_tbl.count; i++) {
+                struct section *sec = &asm->parser->sec_tbl.sections[i];
+                for (uint32_t j = 0; j < sec->count; j++) {
+                        struct mips32_instruction *ins = &sec->ins[j].mips32;
+                        fwrite(ins, sizeof(struct mips32_instruction),
+                               1, out);
+                }
+        }
+
+        // sym tbl
+        fwrite(symtbl, sizeof(Elf32_Sym), symtbl_len, out);
+
+        // str tbl
+        fwrite(asm->str_tbl.ptr, asm->str_tbl.size, 1, out);
+
+        // shdr
+        fwrite(shdr, sizeof(Elf32_Shdr), shdr_len, out);
+
+        fclose(out);
+
+        free(shdr);
+        free(phdr);
+        free(symtbl);
+
+        return M_SUCCESS;
+}
+
+int assemble_file_mips32(char *path)
+{
+	struct lexer lexer;
+	struct parser parser;
+	current_file = path;
+        int res = M_SUCCESS;
+
+	if (lexer_init(current_file, &lexer))
+		return M_ERROR;
+
+	if (mips32_parser_init(&lexer, &parser))
+		return M_ERROR;
+
+        if (res == M_SUCCESS)
+                res = parse_file(&parser);
+
+        struct assembler assembler;
+        assembler.parser = &parser;
+        strtbl_init(&assembler.str_tbl);
+
+        if (res == M_SUCCESS)
+                res = assemble_file(&assembler);
+
+        strtbl_free(&assembler.str_tbl);
+	lexer_free(&lexer);
+	parser_free(&parser);
+
+        return res;
+}