about summary refs log tree commit diff stats
path: root/linux/bootstrap/033check_operands.cc
diff options
context:
space:
mode:
authorKartik K. Agaram <vc@akkartik.com>2021-03-03 22:09:50 -0800
committerKartik K. Agaram <vc@akkartik.com>2021-03-03 22:21:03 -0800
commit71e4f3812982dba2efb471283d310224e8db363e (patch)
treeea111a1acb8b8845dbda39c0e1b4bac1d198143b /linux/bootstrap/033check_operands.cc
parentc6b928be29ac8cdb4e4d6e1eaa20420ff03e5a4c (diff)
downloadmu-71e4f3812982dba2efb471283d310224e8db363e.tar.gz
7842 - new directory organization
Baremetal is now the default build target and therefore has its sources
at the top-level. Baremetal programs build using the phase-2 Mu toolchain
that requires a Linux kernel. This phase-2 codebase which used to be at
the top-level is now under the linux/ directory. Finally, the phase-2 toolchain,
while self-hosting, has a way to bootstrap from a C implementation, which
is now stored in linux/bootstrap. The bootstrap C implementation uses some
literate programming tools that are now in linux/bootstrap/tools.

So the whole thing has gotten inverted. Each directory should build one
artifact and include the main sources (along with standard library). Tools
used for building it are relegated to sub-directories, even though those
tools are often useful in their own right, and have had lots of interesting
programs written using them.

A couple of things have gotten dropped in this process:
  - I had old ways to run on just a Linux kernel, or with a Soso kernel.
    No more.
  - I had some old tooling for running a single test at the cursor. I haven't
    used that lately. Maybe I'll bring it back one day.

The reorg isn't done yet. Still to do:
  - redo documentation everywhere. All the README files, all other markdown,
    particularly vocabulary.md.
  - clean up how-to-run comments at the start of programs everywhere
  - rethink what to do with the html/ directory. Do we even want to keep
    supporting it?

In spite of these shortcomings, all the scripts at the top-level, linux/
and linux/bootstrap are working. The names of the scripts also feel reasonable.
This is a good milestone to take stock at.
Diffstat (limited to 'linux/bootstrap/033check_operands.cc')
-rw-r--r--linux/bootstrap/033check_operands.cc786
1 files changed, 786 insertions, 0 deletions
diff --git a/linux/bootstrap/033check_operands.cc b/linux/bootstrap/033check_operands.cc
new file mode 100644
index 00000000..c764fc61
--- /dev/null
+++ b/linux/bootstrap/033check_operands.cc
@@ -0,0 +1,786 @@
+//: Since we're tagging arguments with their types, let's start checking these
+//: argument types for each instruction.
+
+void test_check_missing_imm8_argument() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "cd\n"  // interrupt ??
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: 'cd' (software interrupt): missing imm8 argument\n"
+  );
+}
+
+:(before "Pack Operands(segment code)")
+check_arguments(code);
+if (trace_contains_errors()) return;
+
+:(code)
+void check_arguments(const segment& code) {
+  trace(3, "transform") << "-- check arguments" << end();
+  for (int i = 0;  i < SIZE(code.lines);  ++i) {
+    check_arguments(code.lines.at(i));
+    if (trace_contains_errors()) return;  // stop at the first mal-formed instruction
+  }
+}
+
+void check_arguments(const line& inst) {
+  word op = preprocess_op(inst.words.at(0));
+  if (op.data == "0f") {
+    check_arguments_0f(inst);
+    return;
+  }
+  if (op.data == "f3") {
+    check_arguments_f3(inst);
+    return;
+  }
+  check_arguments(inst, op);
+}
+
+word preprocess_op(word/*copy*/ op) {
+  op.data = tolower(op.data.c_str());
+  // opcodes can't be negative
+  if (starts_with(op.data, "0x"))
+    op.data = op.data.substr(2);
+  if (SIZE(op.data) == 1)
+    op.data = string("0")+op.data;
+  return op;
+}
+
+void test_preprocess_op() {
+  word w1;  w1.data = "0xf";
+  word w2;  w2.data = "0f";
+  CHECK_EQ(preprocess_op(w1).data, preprocess_op(w2).data);
+}
+
+//: To check the arguments for an opcode, we'll track the permitted arguments
+//: for each supported opcode in a bitvector. That way we can often compute the
+//: 'received' argument bitvector for each instruction's arguments and compare
+//: it with the 'expected' bitvector.
+//:
+//: The 'expected' and 'received' bitvectors can be different; the MODRM bit
+//: in the 'expected' bitvector maps to multiple 'received' argument types in
+//: an instruction. We deal in expected bitvectors throughout.
+
+:(before "End Types")
+enum expected_argument_type {
+  // start from the least significant bit
+  MODRM,  // more complex, may also involve disp8 or disp32
+  SUBOP,
+  DISP8,
+  DISP16,
+  DISP32,
+  IMM8,
+  IMM32,
+  NUM_OPERAND_TYPES
+};
+:(before "End Globals")
+vector<string> Operand_type_name;
+map<string, expected_argument_type> Operand_type;
+:(before "End One-time Setup")
+init_op_types();
+:(code)
+void init_op_types() {
+  assert(NUM_OPERAND_TYPES <= /*bits in a uint8_t*/8);
+  Operand_type_name.resize(NUM_OPERAND_TYPES);
+  #define DEF(type) Operand_type_name.at(type) = tolower(#type), put(Operand_type, tolower(#type), type);
+  DEF(MODRM);
+  DEF(SUBOP);
+  DEF(DISP8);
+  DEF(DISP16);
+  DEF(DISP32);
+  DEF(IMM8);
+  DEF(IMM32);
+  #undef DEF
+}
+
+:(before "End Globals")
+map</*op*/string, /*bitvector*/uint8_t> Permitted_arguments;
+const uint8_t INVALID_OPERANDS = 0xff;  // no instruction uses all the argument types
+:(before "End One-time Setup")
+init_permitted_arguments();
+:(code)
+void init_permitted_arguments() {
+  //// Class A: just op, no arguments
+  // halt
+  put(Permitted_arguments, "f4", 0x00);
+  // inc
+  put(Permitted_arguments, "40", 0x00);
+  put(Permitted_arguments, "41", 0x00);
+  put(Permitted_arguments, "42", 0x00);
+  put(Permitted_arguments, "43", 0x00);
+  put(Permitted_arguments, "44", 0x00);
+  put(Permitted_arguments, "45", 0x00);
+  put(Permitted_arguments, "46", 0x00);
+  put(Permitted_arguments, "47", 0x00);
+  // dec
+  put(Permitted_arguments, "48", 0x00);
+  put(Permitted_arguments, "49", 0x00);
+  put(Permitted_arguments, "4a", 0x00);
+  put(Permitted_arguments, "4b", 0x00);
+  put(Permitted_arguments, "4c", 0x00);
+  put(Permitted_arguments, "4d", 0x00);
+  put(Permitted_arguments, "4e", 0x00);
+  put(Permitted_arguments, "4f", 0x00);
+  // push
+  put(Permitted_arguments, "50", 0x00);
+  put(Permitted_arguments, "51", 0x00);
+  put(Permitted_arguments, "52", 0x00);
+  put(Permitted_arguments, "53", 0x00);
+  put(Permitted_arguments, "54", 0x00);
+  put(Permitted_arguments, "55", 0x00);
+  put(Permitted_arguments, "56", 0x00);
+  put(Permitted_arguments, "57", 0x00);
+  // pop
+  put(Permitted_arguments, "58", 0x00);
+  put(Permitted_arguments, "59", 0x00);
+  put(Permitted_arguments, "5a", 0x00);
+  put(Permitted_arguments, "5b", 0x00);
+  put(Permitted_arguments, "5c", 0x00);
+  put(Permitted_arguments, "5d", 0x00);
+  put(Permitted_arguments, "5e", 0x00);
+  put(Permitted_arguments, "5f", 0x00);
+  // sign-extend EAX into EDX
+  put(Permitted_arguments, "99", 0x00);
+  // return
+  put(Permitted_arguments, "c3", 0x00);
+
+  //// Class B: just op and disp8
+  //  imm32 imm8  disp32 |disp16  disp8 subop modrm
+  //  0     0     0      |0       1     0     0
+
+  // jump
+  put(Permitted_arguments, "eb", 0x04);
+  put(Permitted_arguments, "72", 0x04);
+  put(Permitted_arguments, "73", 0x04);
+  put(Permitted_arguments, "74", 0x04);
+  put(Permitted_arguments, "75", 0x04);
+  put(Permitted_arguments, "76", 0x04);
+  put(Permitted_arguments, "77", 0x04);
+  put(Permitted_arguments, "7c", 0x04);
+  put(Permitted_arguments, "7d", 0x04);
+  put(Permitted_arguments, "7e", 0x04);
+  put(Permitted_arguments, "7f", 0x04);
+
+  //// Class D: just op and disp32
+  //  imm32 imm8  disp32 |disp16  disp8 subop modrm
+  //  0     0     1      |0       0     0     0
+  put(Permitted_arguments, "e8", 0x10);  // call
+  put(Permitted_arguments, "e9", 0x10);  // jump
+
+  //// Class E: just op and imm8
+  //  imm32 imm8  disp32 |disp16  disp8 subop modrm
+  //  0     1     0      |0       0     0     0
+  put(Permitted_arguments, "cd", 0x20);  // software interrupt
+
+  //// Class F: just op and imm32
+  //  imm32 imm8  disp32 |disp16  disp8 subop modrm
+  //  1     0     0      |0       0     0     0
+  put(Permitted_arguments, "05", 0x40);  // add
+  put(Permitted_arguments, "2d", 0x40);  // subtract
+  put(Permitted_arguments, "25", 0x40);  // and
+  put(Permitted_arguments, "0d", 0x40);  // or
+  put(Permitted_arguments, "35", 0x40);  // xor
+  put(Permitted_arguments, "3d", 0x40);  // compare
+  put(Permitted_arguments, "68", 0x40);  // push
+  // copy
+  put(Permitted_arguments, "b8", 0x40);
+  put(Permitted_arguments, "b9", 0x40);
+  put(Permitted_arguments, "ba", 0x40);
+  put(Permitted_arguments, "bb", 0x40);
+  put(Permitted_arguments, "bc", 0x40);
+  put(Permitted_arguments, "bd", 0x40);
+  put(Permitted_arguments, "be", 0x40);
+  put(Permitted_arguments, "bf", 0x40);
+
+  //// Class M: using ModR/M byte
+  //  imm32 imm8  disp32 |disp16  disp8 subop modrm
+  //  0     0     0      |0       0     0     1
+
+  // add
+  put(Permitted_arguments, "01", 0x01);
+  put(Permitted_arguments, "03", 0x01);
+  // subtract
+  put(Permitted_arguments, "29", 0x01);
+  put(Permitted_arguments, "2b", 0x01);
+  // and
+  put(Permitted_arguments, "21", 0x01);
+  put(Permitted_arguments, "23", 0x01);
+  // or
+  put(Permitted_arguments, "09", 0x01);
+  put(Permitted_arguments, "0b", 0x01);
+  // xor
+  put(Permitted_arguments, "31", 0x01);
+  put(Permitted_arguments, "33", 0x01);
+  // compare
+  put(Permitted_arguments, "39", 0x01);
+  put(Permitted_arguments, "3b", 0x01);
+  // copy
+  put(Permitted_arguments, "88", 0x01);
+  put(Permitted_arguments, "89", 0x01);
+  put(Permitted_arguments, "8a", 0x01);
+  put(Permitted_arguments, "8b", 0x01);
+  // swap
+  put(Permitted_arguments, "87", 0x01);
+  // copy address (lea)
+  put(Permitted_arguments, "8d", 0x01);
+
+  //// Class N: op, ModR/M and subop (not r32)
+  //  imm32 imm8  disp32 |disp16  disp8 subop modrm
+  //  0     0     0      |0       0     1     1
+  put(Permitted_arguments, "8f", 0x03);  // pop
+  put(Permitted_arguments, "d3", 0x03);  // shift
+  put(Permitted_arguments, "f7", 0x03);  // test/not/mul/div
+  put(Permitted_arguments, "ff", 0x03);  // jump/push/call
+
+  //// Class O: op, ModR/M, subop (not r32) and imm8
+  //  imm32 imm8  disp32 |disp16  disp8 subop modrm
+  //  0     1     0      |0       0     1     1
+  put(Permitted_arguments, "c1", 0x23);  // combine
+  put(Permitted_arguments, "c6", 0x23);  // copy
+
+  //// Class P: op, ModR/M, subop (not r32) and imm32
+  //  imm32 imm8  disp32 |disp16  disp8 subop modrm
+  //  1     0     0      |0       0     1     1
+  put(Permitted_arguments, "81", 0x43);  // combine
+  put(Permitted_arguments, "c7", 0x43);  // copy
+
+  //// Class Q: op, ModR/M and imm32
+  //  imm32 imm8  disp32 |disp16  disp8 subop modrm
+  //  1     0     0      |0       0     0     1
+  put(Permitted_arguments, "69", 0x41);  // multiply
+
+  // End Init Permitted Operands
+}
+
+#define HAS(bitvector, bit)  ((bitvector) & (1 << (bit)))
+#define SET(bitvector, bit)  ((bitvector) | (1 << (bit)))
+#define CLEAR(bitvector, bit)  ((bitvector) & (~(1 << (bit))))
+
+void check_arguments(const line& inst, const word& op) {
+  if (!is_hex_byte(op)) return;
+  uint8_t expected_bitvector = get(Permitted_arguments, op.data);
+  if (HAS(expected_bitvector, MODRM)) {
+    check_arguments_modrm(inst, op);
+    compare_bitvector_modrm(inst, expected_bitvector, maybe_name(op));
+  }
+  else {
+    compare_bitvector(inst, expected_bitvector, maybe_name(op));
+  }
+}
+
+//: Many instructions can be checked just by comparing bitvectors.
+
+void compare_bitvector(const line& inst, uint8_t expected, const string& maybe_op_name) {
+  if (all_hex_bytes(inst) && has_arguments(inst)) return;  // deliberately programming in raw hex; we'll raise a warning elsewhere
+  uint8_t bitvector = compute_expected_argument_bitvector(inst);
+  if (trace_contains_errors()) return;  // duplicate argument type
+  if (bitvector == expected) return;  // all good with this instruction
+  for (int i = 0;  i < NUM_OPERAND_TYPES;  ++i, bitvector >>= 1, expected >>= 1) {
+//?     cerr << "comparing " << HEXBYTE << NUM(bitvector) << " with " << NUM(expected) << '\n';
+    if ((bitvector & 0x1) == (expected & 0x1)) continue;  // all good with this argument
+    const string& optype = Operand_type_name.at(i);
+    if ((bitvector & 0x1) > (expected & 0x1))
+      raise << "'" << to_string(inst) << "'" << maybe_op_name << ": unexpected " << optype << " argument\n" << end();
+    else
+      raise << "'" << to_string(inst) << "'" << maybe_op_name << ": missing " << optype << " argument\n" << end();
+    // continue giving all errors for a single instruction
+  }
+  // ignore settings in any unused bits
+}
+
+string maybe_name(const word& op) {
+  if (!is_hex_byte(op)) return "";
+  if (!contains_key(Name, op.data)) return "";
+  // strip stuff in parens from the name
+  const string& s = get(Name, op.data);
+  return " ("+s.substr(0, s.find(" ("))+')';
+}
+
+uint32_t compute_expected_argument_bitvector(const line& inst) {
+  set<string> arguments_found;
+  uint32_t bitvector = 0;
+  for (int i = /*skip op*/1;  i < SIZE(inst.words);  ++i) {
+    bitvector = bitvector | expected_bit_for_received_argument(inst.words.at(i), arguments_found, inst);
+    if (trace_contains_errors()) return INVALID_OPERANDS;  // duplicate argument type
+  }
+  return bitvector;
+}
+
+bool has_arguments(const line& inst) {
+  return SIZE(inst.words) > first_argument(inst);
+}
+
+int first_argument(const line& inst) {
+  if (inst.words.at(0).data == "0f") return 2;
+  if (inst.words.at(0).data == "f2" || inst.words.at(0).data == "f3") {
+    if (inst.words.at(1).data == "0f")
+      return 3;
+    else
+      return 2;
+  }
+  return 1;
+}
+
+// Scan the metadata of 'w' and return the expected bit corresponding to any argument type.
+// Also raise an error if metadata contains multiple argument types.
+uint32_t expected_bit_for_received_argument(const word& w, set<string>& instruction_arguments, const line& inst) {
+  uint32_t bv = 0;
+  bool found = false;
+  for (int i = 0;  i < SIZE(w.metadata);  ++i) {
+    string/*copy*/ curr = w.metadata.at(i);
+    string expected_metadata = curr;
+    if (curr == "mod" || curr == "rm32" || curr == "r32" || curr == "xm32" || curr == "x32" || curr == "scale" || curr == "index" || curr == "base")
+      expected_metadata = "modrm";
+    else if (!contains_key(Operand_type, curr)) continue;  // ignore unrecognized metadata
+    if (found) {
+      raise << "'" << w.original << "' has conflicting argument types; it should have only one\n" << end();
+      return INVALID_OPERANDS;
+    }
+    if (instruction_arguments.find(curr) != instruction_arguments.end()) {
+      raise << "'" << to_string(inst) << "': duplicate " << curr << " argument\n" << end();
+      return INVALID_OPERANDS;
+    }
+    instruction_arguments.insert(curr);
+    bv = (1 << get(Operand_type, expected_metadata));
+    found = true;
+  }
+  return bv;
+}
+
+void test_conflicting_argument_type() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "cd/software-interrupt 80/imm8/imm32\n"
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '80/imm8/imm32' has conflicting argument types; it should have only one\n"
+  );
+}
+
+//: Instructions computing effective addresses have more complex rules, so
+//: we'll hard-code a common set of instruction-decoding rules.
+
+void test_check_missing_mod_argument() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "81 0/add/subop       3/rm32/ebx 1/imm32\n"
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '81 0/add/subop 3/rm32/ebx 1/imm32' (combine rm32 with imm32 based on subop): missing mod argument\n"
+  );
+}
+
+void check_arguments_modrm(const line& inst, const word& op) {
+  if (all_hex_bytes(inst)) return;  // deliberately programming in raw hex; we'll raise a warning elsewhere
+  check_argument_metadata_present(inst, "mod", op);
+  if (!has_argument_metadata(inst, "rm32") && !has_argument_metadata(inst, "xm32"))
+    raise << "'" << to_string(inst) << "'" << maybe_name(op) << ": missing rm32 (or xm32) argument\n" << end();
+  // no check for r32; some instructions don't use it; just assume it's 0 if missing
+  if (op.data == "81" || op.data == "8f" || op.data == "f7" || op.data == "ff") {  // keep sync'd with 'help subop'
+    check_argument_metadata_present(inst, "subop", op);
+    check_argument_metadata_absent(inst, "r32", op, "should be replaced by subop");
+    check_argument_metadata_absent(inst, "x32", op, "should be replaced by subop");
+  }
+  if (trace_contains_errors()) return;
+  if (metadata_m32(inst).data != "4") return;
+  // SIB byte checks
+  uint8_t mod = hex_byte(metadata(inst, "mod").data);
+  if (mod != /*direct*/3) {
+    check_argument_metadata_present(inst, "base", op);
+    check_argument_metadata_present(inst, "index", op);  // otherwise why go to SIB?
+  }
+  else {
+    check_argument_metadata_absent(inst, "base", op, "direct mode");
+    check_argument_metadata_absent(inst, "index", op, "direct mode");
+  }
+  // no check for scale; 0 (2**0 = 1) by default
+}
+
+word metadata_m32(const line& inst) {
+  for (int i = 0;  i < SIZE(inst.words);  ++i)
+    if (has_argument_metadata(inst.words.at(i), "rm32") || has_argument_metadata(inst.words.at(i), "xm32"))
+      return inst.words.at(i);
+  assert(false);
+}
+
+// same as compare_bitvector, with one additional exception for modrm-based
+// instructions: they may use an extra displacement on occasion
+void compare_bitvector_modrm(const line& inst, uint8_t expected, const string& maybe_op_name) {
+  if (all_hex_bytes(inst) && has_arguments(inst)) return;  // deliberately programming in raw hex; we'll raise a warning elsewhere
+  uint8_t bitvector = compute_expected_argument_bitvector(inst);
+  if (trace_contains_errors()) return;  // duplicate argument type
+  // update 'expected' bitvector for the additional exception
+  if (has_argument_metadata(inst, "mod")) {
+    int32_t mod = parse_int(metadata(inst, "mod").data);
+    switch (mod) {
+    case 0:
+      if (has_argument_metadata(inst, "rm32") && parse_int(metadata(inst, "rm32").data) == 5)
+        expected |= (1<<DISP32);
+      break;
+    case 1:
+      expected |= (1<<DISP8);
+      break;
+    case 2:
+      expected |= (1<<DISP32);
+      break;
+    }
+  }
+  if (bitvector == expected) return;  // all good with this instruction
+  for (int i = 0;  i < NUM_OPERAND_TYPES;  ++i, bitvector >>= 1, expected >>= 1) {
+//?     cerr << "comparing for modrm " << HEXBYTE << NUM(bitvector) << " with " << NUM(expected) << '\n';
+    if ((bitvector & 0x1) == (expected & 0x1)) continue;  // all good with this argument
+    const string& optype = Operand_type_name.at(i);
+    if ((bitvector & 0x1) > (expected & 0x1))
+      raise << "'" << to_string(inst) << "'" << maybe_op_name << ": unexpected " << optype << " argument\n" << end();
+    else
+      raise << "'" << to_string(inst) << "'" << maybe_op_name << ": missing " << optype << " argument\n" << end();
+    // continue giving all errors for a single instruction
+  }
+  // ignore settings in any unused bits
+}
+
+void check_argument_metadata_present(const line& inst, const string& type, const word& op) {
+  if (!has_argument_metadata(inst, type))
+    raise << "'" << to_string(inst) << "'" << maybe_name(op) << ": missing " << type << " argument\n" << end();
+}
+
+void check_argument_metadata_absent(const line& inst, const string& type, const word& op, const string& msg) {
+  if (has_argument_metadata(inst, type))
+    raise << "'" << to_string(inst) << "'" << maybe_name(op) << ": unexpected " << type << " argument (" << msg << ")\n" << end();
+}
+
+void test_modrm_with_displacement() {
+  Reg[EAX].u = 0x1;
+  transform(
+      "== code 0x1\n"
+      // just avoid null pointer
+      "8b/copy 1/mod/lookup+disp8 0/rm32/EAX 2/r32/EDX 4/disp8\n"  // copy *(EAX+4) to EDX
+  );
+  CHECK_TRACE_COUNT("error", 0);
+}
+
+void test_check_missing_disp8() {
+  Hide_errors = true;
+  transform(
+      "== code 0x1\n"
+      "89/copy 1/mod/lookup+disp8 0/rm32/EAX 1/r32/ECX\n"  // missing disp8
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '89/copy 1/mod/lookup+disp8 0/rm32/EAX 1/r32/ECX' (copy r32 to rm32): missing disp8 argument\n"
+  );
+}
+
+void test_check_missing_disp32() {
+  Hide_errors = true;
+  transform(
+      "== code 0x1\n"
+      "8b/copy 0/mod/indirect 5/rm32/.disp32 2/r32/EDX\n"  // missing disp32
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '8b/copy 0/mod/indirect 5/rm32/.disp32 2/r32/EDX' (copy rm32 to r32): missing disp32 argument\n"
+  );
+}
+
+void test_conflicting_arguments_in_modrm_instruction() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "01/add 0/mod 3/mod\n"
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '01/add 0/mod 3/mod' has conflicting mod arguments\n"
+  );
+}
+
+void test_conflicting_argument_type_modrm() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "01/add 0/mod 3/rm32/r32\n"
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '3/rm32/r32' has conflicting argument types; it should have only one\n"
+  );
+}
+
+void test_check_missing_rm32_argument() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "81 0/add/subop 0/mod            1/imm32\n"
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '81 0/add/subop 0/mod 1/imm32' (combine rm32 with imm32 based on subop): missing rm32 (or xm32) argument\n"
+  );
+}
+
+void test_check_missing_subop_argument() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "81             0/mod 3/rm32/ebx 1/imm32\n"
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '81 0/mod 3/rm32/ebx 1/imm32' (combine rm32 with imm32 based on subop): missing subop argument\n"
+  );
+}
+
+void test_check_missing_base_argument() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "81 0/add/subop 0/mod/indirect 4/rm32/use-sib 1/imm32\n"
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '81 0/add/subop 0/mod/indirect 4/rm32/use-sib 1/imm32' (combine rm32 with imm32 based on subop): missing base argument\n"
+  );
+}
+
+void test_check_missing_index_argument() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "81 0/add/subop 0/mod/indirect 4/rm32/use-sib 0/base 1/imm32\n"
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '81 0/add/subop 0/mod/indirect 4/rm32/use-sib 0/base 1/imm32' (combine rm32 with imm32 based on subop): missing index argument\n"
+  );
+}
+
+void test_check_missing_base_argument_2() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "81 0/add/subop 0/mod/indirect 4/rm32/use-sib 2/index 3/scale 1/imm32\n"
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '81 0/add/subop 0/mod/indirect 4/rm32/use-sib 2/index 3/scale 1/imm32' (combine rm32 with imm32 based on subop): missing base argument\n"
+  );
+}
+
+void test_check_extra_displacement() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "89/copy 0/mod/indirect 0/rm32/EAX 1/r32/ECX 4/disp8\n"
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '89/copy 0/mod/indirect 0/rm32/EAX 1/r32/ECX 4/disp8' (copy r32 to rm32): unexpected disp8 argument\n"
+  );
+}
+
+void test_check_duplicate_argument() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "89/copy 0/mod/indirect 0/rm32/EAX 1/r32/ECX 1/r32\n"
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '89/copy 0/mod/indirect 0/rm32/EAX 1/r32/ECX 1/r32': duplicate r32 argument\n"
+  );
+}
+
+void test_check_base_argument_not_needed_in_direct_mode() {
+  run(
+      "== code 0x1\n"
+      "81 0/add/subop 3/mod/indirect 4/rm32/use-sib 1/imm32\n"
+  );
+  CHECK_TRACE_COUNT("error", 0);
+}
+
+void test_extra_modrm() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "59/pop-to-ECX  3/mod/direct 1/rm32/ECX 4/r32/ESP\n"
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '59/pop-to-ECX 3/mod/direct 1/rm32/ECX 4/r32/ESP' (pop top of stack to ECX): unexpected modrm argument\n"
+  );
+}
+
+//:: similarly handle multi-byte opcodes
+
+void check_arguments_0f(const line& inst) {
+  assert(inst.words.at(0).data == "0f");
+  if (SIZE(inst.words) == 1) {
+    raise << "opcode '0f' requires a second opcode\n" << end();
+    return;
+  }
+  word op = preprocess_op(inst.words.at(1));
+  if (!contains_key(Name_0f, op.data)) {
+    raise << "unknown 2-byte opcode '0f " << op.data << "'\n" << end();
+    return;
+  }
+  check_arguments_0f(inst, op);
+}
+
+void check_arguments_f3(const line& inst) {
+  assert(inst.words.at(0).data == "f3");
+  if (SIZE(inst.words) == 1) {
+    raise << "opcode 'f3' requires a second opcode\n" << end();
+    return;
+  }
+  word op = preprocess_op(inst.words.at(1));
+  if (op.data == "0f") {
+    word op2 = preprocess_op(inst.words.at(2));
+    check_arguments_f3_0f(inst, op2);
+    return;
+  }
+  if (!contains_key(Name_f3, op.data)) {
+    raise << "unknown 2-byte opcode 'f3 " << op.data << "'\n" << end();
+    return;
+  }
+  check_arguments_f3(inst, op);
+}
+
+void test_check_missing_disp32_argument() {
+  Hide_errors = true;
+  run(
+      "== code 0x1\n"
+      "  0f 84  # jmp if ZF to ??\n"
+  );
+  CHECK_TRACE_CONTENTS(
+      "error: '0f 84' (jump disp32 bytes away if equal, if ZF is set): missing disp32 argument\n"
+  );
+}
+
+void test_0f_opcode_with_modrm() {
+  transform(
+      "== code 0x1\n"
+      "0f af/multiply 2/mod/*+disp32 5/rm32/ebp 8/disp32 0/r32\n"
+  );
+  CHECK_TRACE_DOESNT_CONTAIN_ERRORS();
+}
+
+:(before "End Globals")
+map</*op*/string, /*bitvector*/uint8_t> Permitted_arguments_0f;
+:(before "End Init Permitted Operands")
+//// Class D: just op and disp32
+//  imm32 imm8  disp32 |disp16  disp8 subop modrm
+//  0     0     1      |0       0     0     0
+put_new(Permitted_arguments_0f, "82", 0x10);
+put_new(Permitted_arguments_0f, "83", 0x10);
+put_new(Permitted_arguments_0f, "84", 0x10);
+put_new(Permitted_arguments_0f, "85", 0x10);
+put_new(Permitted_arguments_0f, "86", 0x10);
+put_new(Permitted_arguments_0f, "87", 0x10);
+put_new(Permitted_arguments_0f, "8c", 0x10);
+put_new(Permitted_arguments_0f, "8d", 0x10);
+put_new(Permitted_arguments_0f, "8e", 0x10);
+put_new(Permitted_arguments_0f, "8f", 0x10);
+
+//// Class M: using ModR/M byte
+//  imm32 imm8  disp32 |disp16  disp8 subop modrm
+//  0     0     0      |0       0     0     1
+put_new(Permitted_arguments_0f, "2f", 0x01);  // compare floats
+put_new(Permitted_arguments_0f, "af", 0x01);  // multiply ints
+// setcc
+put_new(Permitted_arguments_0f, "92", 0x01);
+put_new(Permitted_arguments_0f, "93", 0x01);
+put_new(Permitted_arguments_0f, "94", 0x01);
+put_new(Permitted_arguments_0f, "95", 0x01);
+put_new(Permitted_arguments_0f, "96", 0x01);
+put_new(Permitted_arguments_0f, "97", 0x01);
+put_new(Permitted_arguments_0f, "9c", 0x01);
+put_new(Permitted_arguments_0f, "9d", 0x01);
+put_new(Permitted_arguments_0f, "9e", 0x01);
+put_new(Permitted_arguments_0f, "9f", 0x01);
+
+:(before "End Globals")
+map</*op*/string, /*bitvector*/uint8_t> Permitted_arguments_f3;
+map</*op*/string, /*bitvector*/uint8_t> Permitted_arguments_f3_0f;
+:(before "End Init Permitted Operands")
+//// Class M: using ModR/M byte
+//  imm32 imm8  disp32 |disp16  disp8 subop modrm
+//  0     0     0      |0       0     0     1
+put_new(Permitted_arguments_f3_0f, "10", 0x01);  // copy xm32 to x32
+put_new(Permitted_arguments_f3_0f, "11", 0x01);  // copy x32 to xm32
+put_new(Permitted_arguments_f3_0f, "2a", 0x01);  // convert-to-float
+put_new(Permitted_arguments_f3_0f, "2c", 0x01);  // truncate-to-int
+put_new(Permitted_arguments_f3_0f, "2d", 0x01);  // convert-to-int
+put_new(Permitted_arguments_f3_0f, "51", 0x01);  // square root
+put_new(Permitted_arguments_f3_0f, "52", 0x01);  // inverse square root
+put_new(Permitted_arguments_f3_0f, "53", 0x01);  // reciprocal
+put_new(Permitted_arguments_f3_0f, "58", 0x01);  // add floats
+put_new(Permitted_arguments_f3_0f, "59", 0x01);  // multiply floats
+put_new(Permitted_arguments_f3_0f, "5c", 0x01);  // subtract floats
+put_new(Permitted_arguments_f3_0f, "5d", 0x01);  // minimum of floats
+put_new(Permitted_arguments_f3_0f, "5e", 0x01);  // divide floats
+put_new(Permitted_arguments_f3_0f, "5f", 0x01);  // maximum of floats
+
+:(code)
+void check_arguments_0f(const line& inst, const word& op) {
+  uint8_t expected_bitvector = get(Permitted_arguments_0f, op.data);
+  if (HAS(expected_bitvector, MODRM)) {
+    check_arguments_modrm(inst, op);
+    compare_bitvector_modrm(inst, expected_bitvector, maybe_name_0f(op));
+  }
+  else {
+    compare_bitvector(inst, CLEAR(expected_bitvector, MODRM), maybe_name_0f(op));
+  }
+}
+
+void check_arguments_f3(const line& inst, const word& op) {
+  uint8_t expected_bitvector = get(Permitted_arguments_f3, op.data);
+  if (HAS(expected_bitvector, MODRM)) {
+    check_arguments_modrm(inst, op);
+    compare_bitvector_modrm(inst, expected_bitvector, maybe_name_f3(op));
+  }
+  else {
+    compare_bitvector(inst, CLEAR(expected_bitvector, MODRM), maybe_name_f3(op));
+  }
+}
+
+void check_arguments_f3_0f(const line& inst, const word& op) {
+  uint8_t expected_bitvector = get(Permitted_arguments_f3_0f, op.data);
+  if (HAS(expected_bitvector, MODRM)) {
+    check_arguments_modrm(inst, op);
+    compare_bitvector_modrm(inst, expected_bitvector, maybe_name_f3_0f(op));
+  }
+  else {
+    compare_bitvector(inst, CLEAR(expected_bitvector, MODRM), maybe_name_f3_0f(op));
+  }
+}
+
+string maybe_name_0f(const word& op) {
+  if (!is_hex_byte(op)) return "";
+  if (!contains_key(Name_0f, op.data)) return "";
+  // strip stuff in parens from the name
+  const string& s = get(Name_0f, op.data);
+  return " ("+s.substr(0, s.find(" ("))+')';
+}
+
+string maybe_name_f3(const word& op) {
+  if (!is_hex_byte(op)) return "";
+  if (!contains_key(Name_f3, op.data)) return "";
+  // strip stuff in parens from the name
+  const string& s = get(Name_f3, op.data);
+  return " ("+s.substr(0, s.find(" ("))+')';
+}
+
+string maybe_name_f3_0f(const word& op) {
+  if (!is_hex_byte(op)) return "";
+  if (!contains_key(Name_f3_0f, op.data)) return "";
+  // strip stuff in parens from the name
+  const string& s = get(Name_f3_0f, op.data);
+  return " ("+s.substr(0, s.find(" ("))+')';
+}
+
+string tolower(const char* s) {
+  ostringstream out;
+  for (/*nada*/;  *s;  ++s)
+    out << static_cast<char>(tolower(*s));
+  return out.str();
+}
+
+#undef HAS
+#undef SET
+#undef CLEAR
+
+:(before "End Includes")
+#include<cctype>