fix several incorrect tests and docs in 64- and 32-bit modes

6 files changed, 90 insertions, 51 deletions

diff --git a/src/lib.rs b/src/lib.rs
index cb879fd..057c125 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -84,8 +84,8 @@ mod real_mode;
 pub use real_mode::Arch as x86_16;
 
 const MEM_SIZE_STRINGS: [&'static str; 64] = [
-    "byte", "word", "BUG", "dword", "far", "ptr", "BUG", "qword",
-    "far", "mword", "BUG", "BUG", "BUG", "BUG", "BUG", "xmmword",
+    "byte", "word", "BUG", "dword", "ptr", "far", "BUG", "qword",
+    "BUG", "mword", "BUG", "BUG", "BUG", "BUG", "BUG", "xmmword",
     "BUG", "BUG", "BUG", "BUG", "BUG", "BUG", "BUG", "BUG",
     "BUG", "BUG", "BUG", "BUG", "BUG", "BUG", "BUG", "ymmword",
     "BUG", "BUG", "BUG", "BUG", "BUG", "BUG", "BUG", "BUG",
@@ -98,6 +98,11 @@ pub struct MemoryAccessSize {
     size: u8,
 }
 impl MemoryAccessSize {
+    /// return the number of bytes referenced by this memory access.
+    ///
+    /// if the number of bytes cannot be confidently known by the instruction in isolation (as is
+    /// the case for `xsave`/`xrstor`-style "operate on all processor state" instructions), this
+    /// function will return `None`.
     pub fn bytes_size(&self) -> Option<u8> {
         if self.size == 63 {
             None
@@ -106,6 +111,31 @@ impl MemoryAccessSize {
         }
     }
 
+    /// a human-friendly label for the number of bytes this memory access references.
+    ///
+    /// there are some differences from size names that may be expected elsewhere; `yaxpeax-x86`
+    /// prefers to use consistent names for a width even if the way those bytes are used varies.
+    ///
+    /// the sizes `yaxpeax-x86` knows are as follows:
+    /// | size (bytes) | name       |
+    /// |--------------|------------|
+    /// | 1            | `byte`     |
+    /// | 2            | `word`     |
+    /// | 4            | `dword`    |
+    /// | 6            | `far`      |
+    /// | 8            | `qword`    |
+    /// | 10           | `mword`    |
+    /// | 16           | `xmmword`  |
+    /// | 32           | `ymmword`  |
+    /// | 64           | `zmmword`  |
+    /// | variable     | `ptr`      |
+    ///
+    /// "mword" refers to an mmx-sized access - 80 bits, or 10 bytes. `mword` is also used for
+    /// 64-bit far calls, because they reference a contiguous ten bytes; two bytes of segment
+    /// selector and eight bytes of address.
+    ///
+    /// "variable" accesses access a number of bytes dependent on the physical processor and its
+    /// operating mode. this is particularly relevant for `xsave`/`xrstor`-style instructions.
     pub fn size_name(&self) -> &'static str {
         MEM_SIZE_STRINGS[self.size as usize - 1]
     }
diff --git a/src/long_mode/mod.rs b/src/long_mode/mod.rs
index 600a81a..79f3fee 100644
--- a/src/long_mode/mod.rs
+++ b/src/long_mode/mod.rs
@@ -7673,7 +7673,7 @@ fn read_operands<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch as yaxpe
                 if opcode == Opcode::CALL || opcode == Opcode::JMP || opcode == Opcode::PUSH || opcode == Opcode::POP {
                     instruction.mem_size = 8;
                 } else if opcode == Opcode::CALLF || opcode == Opcode::JMPF {
-                    instruction.mem_size = 9;
+                    instruction.mem_size = 10;
                 }
             }
             instruction.opcode = opcode;
diff --git a/src/long_mode/vex.rs b/src/long_mode/vex.rs
index 67c4965..5695b17 100644
--- a/src/long_mode/vex.rs
+++ b/src/long_mode/vex.rs
@@ -411,7 +411,7 @@ fn read_vex_operands<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch as y
             instruction.opcode = if modrm & 0xc0 == 0xc0 {
                 Opcode::VMOVHLPS
             } else {
-                instruction.mem_size = 4;
+                instruction.mem_size = 8;
                 Opcode::VMOVLPS
             };
             instruction.regs[0] =
@@ -1925,7 +1925,7 @@ fn read_vex_instruction<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch a
                         } else {
                             VEXOperandCode::G_V_E_xmm
                         }),
-                        0xDA => (Opcode::VPMINSW, if L {
+                        0xDA => (Opcode::VPMINUB, if L {
                             VEXOperandCode::G_V_E_ymm
                         } else {
                             VEXOperandCode::G_V_E_xmm
@@ -3201,12 +3201,12 @@ fn read_vex_instruction<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch a
                         VEXOperandCode::G_E_xmm_imm8
                     }),
                     0x0A => (Opcode::VROUNDSS, if L {
-                        VEXOperandCode::G_V_E_ymm_imm8
+                        VEXOperandCode::G_V_E_xmm_imm8
                     } else {
                         VEXOperandCode::G_V_E_xmm_imm8
                     }),
                     0x0B => (Opcode::VROUNDSD, if L {
-                        VEXOperandCode::G_V_E_ymm_imm8
+                        VEXOperandCode::G_V_E_xmm_imm8
                     } else {
                         VEXOperandCode::G_V_E_xmm_imm8
                     }),
@@ -3269,7 +3269,7 @@ fn read_vex_instruction<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch a
                         return Err(DecodeError::InvalidOpcode);
                     } else {
                         (Opcode::VINSERTF128, if L {
-                            VEXOperandCode::G_V_E_ymm_imm8
+                            VEXOperandCode::G_ymm_V_ymm_E_xmm_imm8
                         } else {
                             instruction.opcode = Opcode::Invalid;
                             return Err(DecodeError::InvalidOpcode);
diff --git a/src/protected_mode/display.rs b/src/protected_mode/display.rs
index baca1a2..c5892c2 100644
--- a/src/protected_mode/display.rs
+++ b/src/protected_mode/display.rs
@@ -1328,7 +1328,7 @@ const MNEMONICS: &[&'static str] = &[
     "loopnz",
     "loopz",
     "loop",
-    "jrcxz",
+    "jecxz",
     "pusha",
     "popa",
     "bound",
@@ -1337,8 +1337,8 @@ const MNEMONICS: &[&'static str] = &[
     "aaa",
     "das",
     "daa",
-    "amx",
-    "adx",
+    "aam",
+    "aad",
     "movdir64b",
     "movdiri",
     "aesdec128kl",
@@ -2332,8 +2332,8 @@ impl <T: fmt::Write, Y: YaxColors> Colorize<T, Y> for Opcode {
             Opcode::AAS |
             Opcode::DAS |
             Opcode::DAA |
-            Opcode::ADX |
-            Opcode::AMX |
+            Opcode::AAD |
+            Opcode::AAM |
             Opcode::KADDB |
             Opcode::KANDB |
             Opcode::KANDNB |
@@ -2403,7 +2403,7 @@ impl <T: fmt::Write, Y: YaxColors> Colorize<T, Y> for Opcode {
             Opcode::LOOPNZ |
             Opcode::LOOPZ |
             Opcode::LOOP |
-            Opcode::JRCXZ |
+            Opcode::JECXZ |
             Opcode::CALL |
             Opcode::CALLF |
             Opcode::JMP |
diff --git a/src/protected_mode/mod.rs b/src/protected_mode/mod.rs
index 3678bf8..a6ad2ee 100644
--- a/src/protected_mode/mod.rs
+++ b/src/protected_mode/mod.rs
@@ -888,7 +888,7 @@ const REGISTER_CLASS_NAMES: &[&'static str] = &[
     "eflags",
 ];
 
-/// high-level register classes in an x86 machine, such as "8-byte general purpose", "xmm", "x87",
+/// high-level register classes in an x86 machine, such as "4-byte general purpose", "xmm", "x87",
 /// and so on. constants in this module are useful for inspecting the register class of a decoded
 /// instruction. as an example:
 /// ```
@@ -2078,7 +2078,7 @@ pub enum Opcode {
     LOOPNZ,
     LOOPZ,
     LOOP,
-    JRCXZ,
+    JECXZ,
 
     PUSHA,
     POPA,
@@ -2088,8 +2088,8 @@ pub enum Opcode {
     AAA,
     DAS,
     DAA,
-    AMX,
-    ADX,
+    AAM,
+    AAD,
 
     // started shipping in Tremont, 2020 sept 23
     MOVDIR64B,
@@ -5471,8 +5471,8 @@ const OPCODES: [OpcodeRecord; 256] = [
     OpcodeRecord(Interpretation::Instruction(Opcode::Invalid), OperandCode::ModRM_0xd1_Ev_1),
     OpcodeRecord(Interpretation::Instruction(Opcode::Invalid), OperandCode::ModRM_0xd2_Eb_CL),
     OpcodeRecord(Interpretation::Instruction(Opcode::Invalid), OperandCode::ModRM_0xd3_Ev_CL),
-    OpcodeRecord(Interpretation::Instruction(Opcode::AMX), OperandCode::Ib),
-    OpcodeRecord(Interpretation::Instruction(Opcode::ADX), OperandCode::Ib),
+    OpcodeRecord(Interpretation::Instruction(Opcode::AAM), OperandCode::Ib),
+    OpcodeRecord(Interpretation::Instruction(Opcode::AAD), OperandCode::Ib),
     OpcodeRecord(Interpretation::Instruction(Opcode::SALC), OperandCode::Nothing),
     // XLAT
     OpcodeRecord(Interpretation::Instruction(Opcode::XLAT), OperandCode::Nothing),
@@ -5496,7 +5496,7 @@ const OPCODES: [OpcodeRecord; 256] = [
     OpcodeRecord(Interpretation::Instruction(Opcode::LOOPNZ), OperandCode::Ibs),
     OpcodeRecord(Interpretation::Instruction(Opcode::LOOPZ), OperandCode::Ibs),
     OpcodeRecord(Interpretation::Instruction(Opcode::LOOP), OperandCode::Ibs),
-    OpcodeRecord(Interpretation::Instruction(Opcode::JRCXZ), OperandCode::Ibs),
+    OpcodeRecord(Interpretation::Instruction(Opcode::JECXZ), OperandCode::Ibs),
     OpcodeRecord(Interpretation::Instruction(Opcode::IN), OperandCode::AL_Ib),
     OpcodeRecord(Interpretation::Instruction(Opcode::IN), OperandCode::AX_Ib),
     OpcodeRecord(Interpretation::Instruction(Opcode::OUT), OperandCode::Ib_AL),
@@ -5721,19 +5721,35 @@ fn read_M_16bit<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch as yaxpea
             }
         },
         0b01 => {
-            instr.disp = read_num(words, 1)?;
+            instr.disp = read_num(words, 1)? as i8 as i32 as u32;
             if mmm > 3 {
-                Ok(OperandSpec::RegDisp)
+                if instr.disp != 0 {
+                    Ok(OperandSpec::RegDisp)
+                } else {
+                    Ok(OperandSpec::Deref)
+                }
             } else {
-                Ok(OperandSpec::RegIndexBaseDisp)
+                if instr.disp != 0 {
+                    Ok(OperandSpec::RegIndexBaseDisp)
+                } else {
+                    Ok(OperandSpec::RegIndexBase)
+                }
             }
         },
         0b10 => {
-            instr.disp = read_num(words, 2)?;
+            instr.disp = read_num(words, 2)? as i16 as i32 as u32;
             if mmm > 3 {
-                Ok(OperandSpec::RegDisp)
+                if instr.disp != 0 {
+                    Ok(OperandSpec::RegDisp)
+                } else {
+                    Ok(OperandSpec::Deref)
+                }
             } else {
-                Ok(OperandSpec::RegIndexBaseDisp)
+                if instr.disp != 0 {
+                    Ok(OperandSpec::RegIndexBaseDisp)
+                } else {
+                    Ok(OperandSpec::RegIndexBase)
+                }
             }
         },
         _ => {
@@ -7528,7 +7544,7 @@ fn read_operands<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch as yaxpe
                 if opcode == Opcode::CALL || opcode == Opcode::JMP || opcode == Opcode::PUSH || opcode == Opcode::POP {
                     instruction.mem_size = 4;
                 } else if opcode == Opcode::CALLF || opcode == Opcode::JMPF {
-                    instruction.mem_size = 5;
+                    instruction.mem_size = 6;
                 }
             }
             instruction.opcode = opcode;
@@ -7782,6 +7798,9 @@ fn unlikely_operands<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch as y
             instruction.regs[0] = RegSpec { bank: RegisterBank::X, num: (modrm >> 3) & 7 };
             instruction.operands[0] = OperandSpec::RegRRR;
             instruction.operands[1] = read_E_xmm(words, instruction, modrm)?;
+            if instruction.opcode == Opcode::CVTTSD2SI || instruction.opcode == Opcode::CVTSD2SI {
+                instruction.regs[0].bank = RegisterBank::D;
+            }
             if instruction.operands[1] != OperandSpec::RegMMM {
                 if [Opcode::PMOVSXBQ, Opcode::PMOVZXBQ].contains(&instruction.opcode) {
                     instruction.mem_size = 2;
@@ -8462,7 +8481,7 @@ fn unlikely_operands<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch as y
                     }
                     7 => {
                         instruction.opcode = Opcode::RDPID;
-                        instruction.operands[0] = read_E(words, instruction, modrm, opwidth)?;
+                        instruction.operands[0] = read_E(words, instruction, modrm, 4)?;
                         if instruction.operands[0] != OperandSpec::RegMMM {
                             return Err(DecodeError::InvalidOperand);
                         }
@@ -9668,7 +9687,11 @@ fn unlikely_operands<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch as y
                         }
                         6 => {
                             instruction.opcode = Opcode::UMONITOR;
-                            instruction.regs[1] = RegSpec::from_parts(m, RegisterBank::D);
+                            if instruction.prefixes.address_size() {
+                                instruction.regs[1] = RegSpec::from_parts(m, RegisterBank::W);
+                            } else {
+                                instruction.regs[1] = RegSpec::from_parts(m, RegisterBank::D);
+                            };
                             instruction.operands[0] = OperandSpec::RegMMM;
                             instruction.operand_count = 1;
                         }
diff --git a/src/protected_mode/vex.rs b/src/protected_mode/vex.rs
index 053d1aa..3550f77 100644
--- a/src/protected_mode/vex.rs
+++ b/src/protected_mode/vex.rs
@@ -412,7 +412,7 @@ fn read_vex_operands<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch as y
             instruction.opcode = if modrm & 0xc0 == 0xc0 {
                 Opcode::VMOVHLPS
             } else {
-                instruction.mem_size = 4;
+                instruction.mem_size = 8;
                 Opcode::VMOVLPS
             };
             instruction.regs[0] =
@@ -1854,7 +1854,7 @@ fn read_vex_instruction<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch a
                         } else {
                             VEXOperandCode::G_V_E_xmm
                         }),
-                        0xDA => (Opcode::VPMINSW, if L {
+                        0xDA => (Opcode::VPMINUB, if L {
                             VEXOperandCode::G_V_E_ymm
                         } else {
                             VEXOperandCode::G_V_E_xmm
@@ -3130,12 +3130,12 @@ fn read_vex_instruction<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch a
                         VEXOperandCode::G_E_xmm_imm8
                     }),
                     0x0A => (Opcode::VROUNDSS, if L {
-                        VEXOperandCode::G_V_E_ymm_imm8
+                        VEXOperandCode::G_V_E_xmm_imm8
                     } else {
                         VEXOperandCode::G_V_E_xmm_imm8
                     }),
                     0x0B => (Opcode::VROUNDSD, if L {
-                        VEXOperandCode::G_V_E_ymm_imm8
+                        VEXOperandCode::G_V_E_xmm_imm8
                     } else {
                         VEXOperandCode::G_V_E_xmm_imm8
                     }),
@@ -3171,14 +3171,7 @@ fn read_vex_instruction<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch a
                     } else {
                         VEXOperandCode::Ev_G_xmm_imm8
                     }),
-                    0x16 => if instruction.prefixes.vex_unchecked().w() {
-                        (Opcode::VPEXTRQ, if L {
-                            instruction.opcode = Opcode::Invalid;
-                            return Err(DecodeError::InvalidOpcode);
-                        } else {
-                            VEXOperandCode::Ev_G_xmm_imm8
-                        })
-                    } else {
+                    0x16 => {
                         (Opcode::VPEXTRD, if L {
                             instruction.opcode = Opcode::Invalid;
                             return Err(DecodeError::InvalidOpcode);
@@ -3198,7 +3191,7 @@ fn read_vex_instruction<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch a
                         return Err(DecodeError::InvalidOpcode);
                     } else {
                         (Opcode::VINSERTF128, if L {
-                            VEXOperandCode::G_V_E_ymm_imm8
+                            VEXOperandCode::G_ymm_V_ymm_E_xmm_imm8
                         } else {
                             instruction.opcode = Opcode::Invalid;
                             return Err(DecodeError::InvalidOpcode);
@@ -3232,14 +3225,7 @@ fn read_vex_instruction<T: Reader<<Arch as yaxpeax_arch::Arch>::Address, <Arch a
                     } else {
                         VEXOperandCode::G_V_E_xmm_imm8
                     }),
-                    0x22 => if instruction.prefixes.vex_unchecked().w() {
-                        (Opcode::VPINSRQ, if L {
-                            instruction.opcode = Opcode::Invalid;
-                            return Err(DecodeError::InvalidOpcode);
-                        } else {
-                            VEXOperandCode::G_V_xmm_Ev_imm8
-                        })
-                    } else {
+                    0x22 => {
                         (Opcode::VPINSRD, if L {
                             instruction.opcode = Opcode::Invalid;
                             return Err(DecodeError::InvalidOpcode);