parsing some special arm instructions

macho-go/pkg/ios/macho/arm.c

@@ -0,0 +1,168 @@
+#include <stdio.h>
+#include <stdint.h>
+
+#define true 1
+#define false 0
+
+
+// this file encodes and decodes arm instructions
+// only support for ARMv8 or (aarch64) because
+// it is the default ISA in Apple devices
+// https://yurichev.com/mirrors/ARMv8-A_Architecture_Reference_Manual_(Issue_A.a).pdf
+
+// using bit field in little endian is crazy
+// but clang does not support switching endianess (fuck u llvm)
+// gcc does, but default on Apple compiler is clang
+// TODO: default to using big endian for bit field parsing
+
+// this file assumes everything is LITTLE ENDIAN
+// the struct is assigned as bit field, and parsed using get_bits
+// to encode the instruction, can use the bit field and load each bits
+// through shifting, by default it has 0 padded so should work
+
+// mask to delete last 12 bits
+uint32_t ZEROS_12_LOWER = ~0xFFF;
+
+uint32_t get_bits(uint32_t value, uint32_t from, uint32_t to) {
+  // should assert compiler error
+  if (to < from) return false;
+  if (from == to) {
+    // bit at position is set
+    return (value & (1 << from)) != 0;
+  }
+
+  int32_t mask = ~(1U << (to - from + 1));
+  return (value & mask) >> from;
+}
+
+int is_bit_set(uint32_t value, uint32_t at) {
+  return (value & (1 << at)) != 0;
+}
+
+struct add {
+  uint32_t sf : 1;
+  uint32_t op : 1;
+  uint32_t s : 1;
+  uint32_t sig : 5;
+  uint32_t imm : 12;
+  uint32_t rn : 5;
+  uint32_t rd : 5;
+};
+
+struct add to_add(uint32_t inst) {
+  struct add parsed;
+  parsed.sf = is_bit_set(inst, 31);
+  parsed.op = is_bit_set(inst, 30);
+  parsed.s = is_bit_set(inst, 29);
+  parsed.sig = get_bits(inst, 24, 28);
+  parsed.imm = get_bits(inst, 10, 21);
+  parsed.rn = get_bits(inst, 5, 9);
+  parsed.rd = get_bits(inst, 0, 4);
+  return parsed;
+}
+
+void from_add(struct add parsed, uint32_t *inst) {
+  *inst = 0;
+  *inst |= parsed.sf << 31;
+  *inst |= parsed.op << 30;
+  *inst |= parsed.s << 29;
+  *inst |= parsed.sig << 24;
+  *inst |= parsed.imm << 10;
+  *inst |= parsed.rn << 5;
+  *inst |= parsed.rd;
+}
+
+int add_imm_set(uint32_t *inst, uint32_t offset) {
+  struct add parsed = to_add(*inst);
+  if (parsed.op != 0 || parsed.sig != 0b10001) {
+    return false;
+  }
+  parsed.imm = offset; // auto truncate?
+  from_add(parsed, inst);
+  return true;
+}
+
+uint32_t add_imm_get(uint32_t inst) {
+  struct add parsed = to_add(inst);
+  return parsed.imm;
+}
+
+int is_add(uint32_t inst) {
+  struct add parsed = to_add(inst);
+  if (parsed.op != 0 || parsed.sig != 0b10001) {
+    return false;
+  }
+  return true;
+}
+
+struct adrp {
+  uint32_t op : 1;
+  uint32_t immlo : 2;
+  uint32_t sig : 5;
+  uint32_t immhi : 19;
+  uint32_t rd : 5;
+};
+
+struct adrp to_adrp(uint32_t inst) {
+  struct adrp parsed;
+  parsed.op = is_bit_set(inst, 31);
+  parsed.immlo = get_bits(inst, 29, 30);
+  parsed.sig = get_bits(inst, 24, 28);
+  parsed.immhi = get_bits(inst, 5, 23);
+  parsed.rd = get_bits(inst, 0, 4);
+  return parsed;
+}
+
+void from_adrp(struct adrp parsed, uint32_t *inst) {
+  *inst = 0;
+  *inst |= parsed.op << 31;
+  *inst |= parsed.immlo << 29;
+  *inst |= parsed.sig << 24;
+  *inst |= parsed.immhi << 5;
+  *inst |= parsed.rd;
+}
+
+int is_adrp(uint32_t inst) {
+  struct adrp parsed = to_adrp(inst);
+  if (parsed.op != 1 || parsed.sig != 0b10000) {
+    return false;
+  }
+  return true;
+}
+
+// change adrp imm to something else
+int adrp_imm_set(uint32_t *inst, uint32_t offset) {
+  struct adrp parsed = to_adrp(*inst);
+  if (parsed.op != 1 || parsed.sig != 0b10000) {
+    return false;
+  }
+
+  // uint32_t imm = 0;
+
+  // imm = parsed.immhi << 14;
+  // imm |= parsed.immlo << 12;
+  // printf("old adrp is %x\n", imm);
+  // printf("  immlo %x\n", parsed.immlo);
+  // printf("  immhi %x\n", parsed.immhi);
+
+  // adrp: register = (base masked lower 12-bit) + imm
+  parsed.immlo = get_bits(offset >> 12, 0, 1);
+  parsed.immhi = offset >> 14;
+
+  // imm = parsed.immhi << 14;
+  // imm |= parsed.immlo << 12;
+  // printf("new adrp is %x\n", imm);
+  // printf("  immlo %x\n", parsed.immlo);
+  // printf("  immhi %x\n", parsed.immhi);
+
+  from_adrp(parsed, inst);
+  return true;
+}
+
+uint32_t adrp_imm_get(uint32_t inst) {
+  struct adrp parsed = to_adrp(inst);
+  uint32_t imm = 0;
+  imm = parsed.immhi << 14;
+  imm |= parsed.immlo << 12;
+  return imm;
+}

macho-go/pkg/ios/macho/arm.h

@@ -0,0 +1,7 @@
+int is_add(uint32_t inst);
+int add_imm_set(uint32_t *inst, uint32_t offset);
+uint32_t add_imm_get(uint32_t inst);
+
+int is_adrp(uint32_t inst);
+int adrp_imm_set(uint32_t *inst, uint32_t offset);
+uint32_t adrp_imm_get(uint32_t inst);