|
| 1 | +use std::{borrow::Cow, collections::HashMap}; |
| 2 | + |
| 3 | +use anyhow::{anyhow, bail, Result}; |
| 4 | +use object::{ |
| 5 | + elf, File, Object, ObjectSection, ObjectSymbol, Relocation, RelocationFlags, SectionIndex, |
| 6 | + SectionKind, Symbol, |
| 7 | +}; |
| 8 | +use unarm::{ |
| 9 | + args::{Argument, OffsetImm, OffsetReg, Register}, |
| 10 | + parse::{ArmVersion, ParseMode, Parser}, |
| 11 | + ParsedIns, |
| 12 | +}; |
| 13 | + |
| 14 | +use crate::{ |
| 15 | + arch::{ObjArch, ProcessCodeResult}, |
| 16 | + diff::DiffObjConfig, |
| 17 | + obj::{ObjInfo, ObjIns, ObjInsArg, ObjInsArgValue, ObjSection, SymbolRef}, |
| 18 | +}; |
| 19 | + |
| 20 | +pub struct ObjArchArm { |
| 21 | + /// Maps section index, to list of disasm modes (arm, thumb or data) sorted by address |
| 22 | + disasm_modes: HashMap<SectionIndex, Vec<DisasmMode>>, |
| 23 | +} |
| 24 | + |
| 25 | +impl ObjArchArm { |
| 26 | + pub fn new(file: &File) -> Result<Self> { |
| 27 | + match file { |
| 28 | + File::Elf32(_) => { |
| 29 | + let disasm_modes: HashMap<_, _> = file |
| 30 | + .sections() |
| 31 | + .filter(|s| s.kind() == SectionKind::Text) |
| 32 | + .map(|s| { |
| 33 | + let index = s.index(); |
| 34 | + let mut mapping_symbols: Vec<_> = file |
| 35 | + .symbols() |
| 36 | + .filter(|s| s.section_index().map(|i| i == index).unwrap_or(false)) |
| 37 | + .filter_map(|s| DisasmMode::from_symbol(&s)) |
| 38 | + .collect(); |
| 39 | + mapping_symbols.sort_unstable_by_key(|x| x.address); |
| 40 | + (s.index(), mapping_symbols) |
| 41 | + }) |
| 42 | + .collect(); |
| 43 | + Ok(Self { disasm_modes }) |
| 44 | + } |
| 45 | + _ => bail!("Unsupported file format {:?}", file.format()), |
| 46 | + } |
| 47 | + } |
| 48 | +} |
| 49 | + |
| 50 | +impl ObjArch for ObjArchArm { |
| 51 | + fn process_code( |
| 52 | + &self, |
| 53 | + obj: &ObjInfo, |
| 54 | + symbol_ref: SymbolRef, |
| 55 | + config: &DiffObjConfig, |
| 56 | + ) -> Result<ProcessCodeResult> { |
| 57 | + let (section, symbol) = obj.section_symbol(symbol_ref); |
| 58 | + let section = section.ok_or_else(|| anyhow!("Code symbol section not found"))?; |
| 59 | + let code = §ion.data |
| 60 | + [symbol.section_address as usize..(symbol.section_address + symbol.size) as usize]; |
| 61 | + |
| 62 | + let start_addr = symbol.address as u32; |
| 63 | + let end_addr = start_addr + symbol.size as u32; |
| 64 | + |
| 65 | + // Mapping symbols decide what kind of data comes after it. $a for ARM code, $t for Thumb code and $d for data. |
| 66 | + let fallback_mappings = |
| 67 | + [DisasmMode { address: symbol.address as u32, mapping: ParseMode::Arm }]; |
| 68 | + let mapping_symbols = self |
| 69 | + .disasm_modes |
| 70 | + .get(&SectionIndex(section.orig_index)) |
| 71 | + .map(|x| x.as_slice()) |
| 72 | + .unwrap_or(&fallback_mappings); |
| 73 | + let first_mapping_idx = |
| 74 | + match mapping_symbols.binary_search_by_key(&(symbol.address as u32), |x| x.address) { |
| 75 | + Ok(idx) => idx, |
| 76 | + Err(idx) => idx - 1, |
| 77 | + }; |
| 78 | + let first_mapping = mapping_symbols[first_mapping_idx].mapping; |
| 79 | + |
| 80 | + let mut mappings_iter = |
| 81 | + mapping_symbols.iter().skip(first_mapping_idx + 1).take_while(|x| x.address < end_addr); |
| 82 | + let mut next_mapping = mappings_iter.next(); |
| 83 | + |
| 84 | + let ins_count = code.len() / first_mapping.instruction_size(); |
| 85 | + let mut ops = Vec::<u16>::with_capacity(ins_count); |
| 86 | + let mut insts = Vec::<ObjIns>::with_capacity(ins_count); |
| 87 | + let mut parser = Parser::new(ArmVersion::V5Te, first_mapping, start_addr, code); |
| 88 | + |
| 89 | + while let Some((address, op, ins)) = parser.next() { |
| 90 | + if let Some(next) = next_mapping { |
| 91 | + let next_address = parser.address; |
| 92 | + if next_address >= next.address { |
| 93 | + // Change mapping |
| 94 | + parser.mode = next.mapping; |
| 95 | + next_mapping = mappings_iter.next(); |
| 96 | + } |
| 97 | + } |
| 98 | + |
| 99 | + let line = section.line_info.range(..=address as u64).last().map(|(_, &b)| b); |
| 100 | + |
| 101 | + let reloc = |
| 102 | + section.relocations.iter().find(|r| (r.address as u32 & !1) == address).cloned(); |
| 103 | + |
| 104 | + let mut reloc_arg = None; |
| 105 | + if let Some(reloc) = &reloc { |
| 106 | + match reloc.flags { |
| 107 | + RelocationFlags::Elf { r_type: elf::R_ARM_THM_XPC22 } |
| 108 | + | RelocationFlags::Elf { r_type: elf::R_ARM_PC24 } => { |
| 109 | + reloc_arg = |
| 110 | + ins.args.iter().rposition(|a| matches!(a, Argument::BranchDest(_))); |
| 111 | + } |
| 112 | + _ => (), |
| 113 | + } |
| 114 | + }; |
| 115 | + |
| 116 | + let (args, branch_dest) = if reloc.is_some() && parser.mode == ParseMode::Data { |
| 117 | + (vec![ObjInsArg::Reloc], None) |
| 118 | + } else { |
| 119 | + push_args(&ins, config, reloc_arg, address)? |
| 120 | + }; |
| 121 | + |
| 122 | + ops.push(op.id()); |
| 123 | + insts.push(ObjIns { |
| 124 | + address: address as u64, |
| 125 | + size: (parser.address - address) as u8, |
| 126 | + op: op.id(), |
| 127 | + mnemonic: ins.mnemonic.to_string(), |
| 128 | + args, |
| 129 | + reloc, |
| 130 | + branch_dest, |
| 131 | + line, |
| 132 | + formatted: ins.to_string(), |
| 133 | + orig: None, |
| 134 | + }); |
| 135 | + } |
| 136 | + |
| 137 | + Ok(ProcessCodeResult { ops, insts }) |
| 138 | + } |
| 139 | + |
| 140 | + fn implcit_addend( |
| 141 | + &self, |
| 142 | + _section: &ObjSection, |
| 143 | + address: u64, |
| 144 | + reloc: &Relocation, |
| 145 | + ) -> anyhow::Result<i64> { |
| 146 | + bail!("Unsupported ARM implicit relocation {:#x}{:?}", address, reloc.flags()) |
| 147 | + } |
| 148 | + |
| 149 | + fn demangle(&self, name: &str) -> Option<String> { |
| 150 | + cpp_demangle::Symbol::new(name) |
| 151 | + .ok() |
| 152 | + .and_then(|s| s.demangle(&cpp_demangle::DemangleOptions::default()).ok()) |
| 153 | + } |
| 154 | + |
| 155 | + fn display_reloc(&self, flags: RelocationFlags) -> Cow<'static, str> { |
| 156 | + Cow::Owned(format!("<{flags:?}>")) |
| 157 | + } |
| 158 | +} |
| 159 | + |
| 160 | +#[derive(Clone, Copy, Debug)] |
| 161 | +struct DisasmMode { |
| 162 | + address: u32, |
| 163 | + mapping: ParseMode, |
| 164 | +} |
| 165 | + |
| 166 | +impl DisasmMode { |
| 167 | + fn from_symbol<'a>(sym: &Symbol<'a, '_, &'a [u8]>) -> Option<Self> { |
| 168 | + if let Ok(name) = sym.name() { |
| 169 | + ParseMode::from_mapping_symbol(name) |
| 170 | + .map(|mapping| DisasmMode { address: sym.address() as u32, mapping }) |
| 171 | + } else { |
| 172 | + None |
| 173 | + } |
| 174 | + } |
| 175 | +} |
| 176 | + |
| 177 | +fn push_args( |
| 178 | + parsed_ins: &ParsedIns, |
| 179 | + config: &DiffObjConfig, |
| 180 | + reloc_arg: Option<usize>, |
| 181 | + cur_addr: u32, |
| 182 | +) -> Result<(Vec<ObjInsArg>, Option<u64>)> { |
| 183 | + let mut args = vec![]; |
| 184 | + let mut branch_dest = None; |
| 185 | + let mut writeback = false; |
| 186 | + let mut deref = false; |
| 187 | + for (i, arg) in parsed_ins.args_iter().enumerate() { |
| 188 | + // Emit punctuation before separator |
| 189 | + if deref { |
| 190 | + match arg { |
| 191 | + Argument::OffsetImm(OffsetImm { post_indexed: true, value: _ }) |
| 192 | + | Argument::OffsetReg(OffsetReg { add: _, post_indexed: true, reg: _ }) |
| 193 | + | Argument::CoOption(_) => { |
| 194 | + deref = false; |
| 195 | + args.push(ObjInsArg::PlainText("]".into())); |
| 196 | + if writeback { |
| 197 | + writeback = false; |
| 198 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque("!".into()))); |
| 199 | + } |
| 200 | + } |
| 201 | + _ => {} |
| 202 | + } |
| 203 | + } |
| 204 | + |
| 205 | + if i > 0 { |
| 206 | + args.push(ObjInsArg::PlainText(config.separator().into())); |
| 207 | + } |
| 208 | + |
| 209 | + if reloc_arg == Some(i) { |
| 210 | + args.push(ObjInsArg::Reloc); |
| 211 | + } else { |
| 212 | + match arg { |
| 213 | + Argument::Reg(reg) => { |
| 214 | + if reg.deref { |
| 215 | + deref = true; |
| 216 | + args.push(ObjInsArg::PlainText("[".into())); |
| 217 | + } |
| 218 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(reg.reg.to_string().into()))); |
| 219 | + if reg.writeback { |
| 220 | + if reg.deref { |
| 221 | + writeback = true; |
| 222 | + } else { |
| 223 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque("!".into()))); |
| 224 | + } |
| 225 | + } |
| 226 | + } |
| 227 | + Argument::RegList(reg_list) => { |
| 228 | + args.push(ObjInsArg::PlainText("{".into())); |
| 229 | + let mut first = true; |
| 230 | + for i in 0..16 { |
| 231 | + if (reg_list.regs & (1 << i)) != 0 { |
| 232 | + if !first { |
| 233 | + args.push(ObjInsArg::PlainText(config.separator().into())); |
| 234 | + } |
| 235 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque( |
| 236 | + Register::parse(i).to_string().into(), |
| 237 | + ))); |
| 238 | + first = false; |
| 239 | + } |
| 240 | + } |
| 241 | + args.push(ObjInsArg::PlainText("}".into())); |
| 242 | + if reg_list.user_mode { |
| 243 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque("^".to_string().into()))); |
| 244 | + } |
| 245 | + } |
| 246 | + Argument::UImm(value) | Argument::CoOpcode(value) => { |
| 247 | + args.push(ObjInsArg::PlainText("#".into())); |
| 248 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Unsigned(*value as u64))); |
| 249 | + } |
| 250 | + Argument::SImm(value) |
| 251 | + | Argument::OffsetImm(OffsetImm { post_indexed: _, value }) => { |
| 252 | + args.push(ObjInsArg::PlainText("#".into())); |
| 253 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Signed(*value as i64))); |
| 254 | + } |
| 255 | + Argument::BranchDest(value) => { |
| 256 | + let dest = cur_addr.wrapping_add_signed(*value) as u64; |
| 257 | + args.push(ObjInsArg::BranchDest(dest)); |
| 258 | + branch_dest = Some(dest); |
| 259 | + } |
| 260 | + Argument::CoOption(value) => { |
| 261 | + args.push(ObjInsArg::PlainText("{".into())); |
| 262 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Unsigned(*value as u64))); |
| 263 | + args.push(ObjInsArg::PlainText("}".into())); |
| 264 | + } |
| 265 | + Argument::CoprocNum(value) => { |
| 266 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(format!("p{}", value).into()))); |
| 267 | + } |
| 268 | + Argument::ShiftImm(shift) => { |
| 269 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(shift.op.to_string().into()))); |
| 270 | + args.push(ObjInsArg::PlainText(" #".into())); |
| 271 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Unsigned(shift.imm as u64))); |
| 272 | + } |
| 273 | + Argument::ShiftReg(shift) => { |
| 274 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(shift.op.to_string().into()))); |
| 275 | + args.push(ObjInsArg::PlainText(" ".into())); |
| 276 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(shift.reg.to_string().into()))); |
| 277 | + } |
| 278 | + Argument::OffsetReg(offset) => { |
| 279 | + if !offset.add { |
| 280 | + args.push(ObjInsArg::PlainText("-".into())); |
| 281 | + } |
| 282 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque( |
| 283 | + offset.reg.to_string().into(), |
| 284 | + ))); |
| 285 | + } |
| 286 | + _ => args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(arg.to_string().into()))), |
| 287 | + } |
| 288 | + } |
| 289 | + } |
| 290 | + if deref { |
| 291 | + args.push(ObjInsArg::PlainText("]".into())); |
| 292 | + if writeback { |
| 293 | + args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque("!".into()))); |
| 294 | + } |
| 295 | + } |
| 296 | + Ok((args, branch_dest)) |
| 297 | +} |
0 commit comments