miri/shims/os_str.rs
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367
use std::borrow::Cow;
use std::ffi::{OsStr, OsString};
#[cfg(unix)]
use std::os::unix::ffi::{OsStrExt, OsStringExt};
#[cfg(windows)]
use std::os::windows::ffi::{OsStrExt, OsStringExt};
use std::path::{Path, PathBuf};
use rustc_middle::ty::Ty;
use rustc_middle::ty::layout::LayoutOf;
use crate::*;
/// Represent how path separator conversion should be done.
pub enum PathConversion {
HostToTarget,
TargetToHost,
}
#[cfg(unix)]
pub fn bytes_to_os_str<'tcx>(bytes: &[u8]) -> InterpResult<'tcx, &OsStr> {
interp_ok(OsStr::from_bytes(bytes))
}
#[cfg(not(unix))]
pub fn bytes_to_os_str<'tcx>(bytes: &[u8]) -> InterpResult<'tcx, &OsStr> {
// We cannot use `from_encoded_bytes_unchecked` here since we can't trust `bytes`.
let s = std::str::from_utf8(bytes)
.map_err(|_| err_unsup_format!("{:?} is not a valid utf-8 string", bytes))?;
interp_ok(OsStr::new(s))
}
impl<'tcx> EvalContextExt<'tcx> for crate::MiriInterpCx<'tcx> {}
pub trait EvalContextExt<'tcx>: crate::MiriInterpCxExt<'tcx> {
/// Helper function to read an OsString from a null-terminated sequence of bytes, which is what
/// the Unix APIs usually handle.
fn read_os_str_from_c_str<'a>(&'a self, ptr: Pointer) -> InterpResult<'tcx, &'a OsStr>
where
'tcx: 'a,
{
let this = self.eval_context_ref();
let bytes = this.read_c_str(ptr)?;
bytes_to_os_str(bytes)
}
/// Helper function to read an OsString from a 0x0000-terminated sequence of u16,
/// which is what the Windows APIs usually handle.
fn read_os_str_from_wide_str<'a>(&'a self, ptr: Pointer) -> InterpResult<'tcx, OsString>
where
'tcx: 'a,
{
#[cfg(windows)]
pub fn u16vec_to_osstring<'tcx>(u16_vec: Vec<u16>) -> InterpResult<'tcx, OsString> {
interp_ok(OsString::from_wide(&u16_vec[..]))
}
#[cfg(not(windows))]
pub fn u16vec_to_osstring<'tcx>(u16_vec: Vec<u16>) -> InterpResult<'tcx, OsString> {
let s = String::from_utf16(&u16_vec[..])
.map_err(|_| err_unsup_format!("{:?} is not a valid utf-16 string", u16_vec))?;
interp_ok(s.into())
}
let u16_vec = self.eval_context_ref().read_wide_str(ptr)?;
u16vec_to_osstring(u16_vec)
}
/// Helper function to write an OsStr as a null-terminated sequence of bytes, which is what the
/// Unix APIs usually handle. Returns `(success, full_len)`, where length includes the null
/// terminator. On failure, nothing is written.
fn write_os_str_to_c_str(
&mut self,
os_str: &OsStr,
ptr: Pointer,
size: u64,
) -> InterpResult<'tcx, (bool, u64)> {
let bytes = os_str.as_encoded_bytes();
self.eval_context_mut().write_c_str(bytes, ptr, size)
}
/// Internal helper to share code between `write_os_str_to_wide_str` and
/// `write_os_str_to_wide_str_truncated`.
fn write_os_str_to_wide_str_helper(
&mut self,
os_str: &OsStr,
ptr: Pointer,
size: u64,
truncate: bool,
) -> InterpResult<'tcx, (bool, u64)> {
#[cfg(windows)]
fn os_str_to_u16vec<'tcx>(os_str: &OsStr) -> InterpResult<'tcx, Vec<u16>> {
interp_ok(os_str.encode_wide().collect())
}
#[cfg(not(windows))]
fn os_str_to_u16vec<'tcx>(os_str: &OsStr) -> InterpResult<'tcx, Vec<u16>> {
// On non-Windows platforms the best we can do to transform Vec<u16> from/to OS strings is to do the
// intermediate transformation into strings. Which invalidates non-utf8 paths that are actually
// valid.
os_str
.to_str()
.map(|s| s.encode_utf16().collect())
.ok_or_else(|| err_unsup_format!("{:?} is not a valid utf-8 string", os_str))
.into()
}
let u16_vec = os_str_to_u16vec(os_str)?;
let (written, size_needed) = self.eval_context_mut().write_wide_str(&u16_vec, ptr, size)?;
if truncate && !written && size > 0 {
// Write the truncated part that fits.
let truncated_data = &u16_vec[..size.saturating_sub(1).try_into().unwrap()];
let (written, written_len) =
self.eval_context_mut().write_wide_str(truncated_data, ptr, size)?;
assert!(written && written_len == size);
}
interp_ok((written, size_needed))
}
/// Helper function to write an OsStr as a 0x0000-terminated u16-sequence, which is what the
/// Windows APIs usually handle. Returns `(success, full_len)`, where length is measured
/// in units of `u16` and includes the null terminator. On failure, nothing is written.
fn write_os_str_to_wide_str(
&mut self,
os_str: &OsStr,
ptr: Pointer,
size: u64,
) -> InterpResult<'tcx, (bool, u64)> {
self.write_os_str_to_wide_str_helper(os_str, ptr, size, /*truncate*/ false)
}
/// Like `write_os_str_to_wide_str`, but on failure as much as possible is written into
/// the buffer (always with a null terminator).
fn write_os_str_to_wide_str_truncated(
&mut self,
os_str: &OsStr,
ptr: Pointer,
size: u64,
) -> InterpResult<'tcx, (bool, u64)> {
self.write_os_str_to_wide_str_helper(os_str, ptr, size, /*truncate*/ true)
}
/// Allocate enough memory to store the given `OsStr` as a null-terminated sequence of bytes.
fn alloc_os_str_as_c_str(
&mut self,
os_str: &OsStr,
memkind: MemoryKind,
) -> InterpResult<'tcx, Pointer> {
let size = u64::try_from(os_str.len()).unwrap().strict_add(1); // Make space for `0` terminator.
let this = self.eval_context_mut();
let arg_type = Ty::new_array(this.tcx.tcx, this.tcx.types.u8, size);
let arg_place = this.allocate(this.layout_of(arg_type).unwrap(), memkind)?;
let (written, _) = self.write_os_str_to_c_str(os_str, arg_place.ptr(), size).unwrap();
assert!(written);
interp_ok(arg_place.ptr())
}
/// Allocate enough memory to store the given `OsStr` as a null-terminated sequence of `u16`.
fn alloc_os_str_as_wide_str(
&mut self,
os_str: &OsStr,
memkind: MemoryKind,
) -> InterpResult<'tcx, Pointer> {
let size = u64::try_from(os_str.len()).unwrap().strict_add(1); // Make space for `0x0000` terminator.
let this = self.eval_context_mut();
let arg_type = Ty::new_array(this.tcx.tcx, this.tcx.types.u16, size);
let arg_place = this.allocate(this.layout_of(arg_type).unwrap(), memkind)?;
let (written, _) = self.write_os_str_to_wide_str(os_str, arg_place.ptr(), size).unwrap();
assert!(written);
interp_ok(arg_place.ptr())
}
/// Read a null-terminated sequence of bytes, and perform path separator conversion if needed.
fn read_path_from_c_str<'a>(&'a self, ptr: Pointer) -> InterpResult<'tcx, Cow<'a, Path>>
where
'tcx: 'a,
{
let this = self.eval_context_ref();
let os_str = this.read_os_str_from_c_str(ptr)?;
interp_ok(match this.convert_path(Cow::Borrowed(os_str), PathConversion::TargetToHost) {
Cow::Borrowed(x) => Cow::Borrowed(Path::new(x)),
Cow::Owned(y) => Cow::Owned(PathBuf::from(y)),
})
}
/// Read a null-terminated sequence of `u16`s, and perform path separator conversion if needed.
fn read_path_from_wide_str(&self, ptr: Pointer) -> InterpResult<'tcx, PathBuf> {
let this = self.eval_context_ref();
let os_str = this.read_os_str_from_wide_str(ptr)?;
interp_ok(
this.convert_path(Cow::Owned(os_str), PathConversion::TargetToHost).into_owned().into(),
)
}
/// Write a Path to the machine memory (as a null-terminated sequence of bytes),
/// adjusting path separators if needed.
fn write_path_to_c_str(
&mut self,
path: &Path,
ptr: Pointer,
size: u64,
) -> InterpResult<'tcx, (bool, u64)> {
let this = self.eval_context_mut();
let os_str =
this.convert_path(Cow::Borrowed(path.as_os_str()), PathConversion::HostToTarget);
this.write_os_str_to_c_str(&os_str, ptr, size)
}
/// Write a Path to the machine memory (as a null-terminated sequence of `u16`s),
/// adjusting path separators if needed.
fn write_path_to_wide_str(
&mut self,
path: &Path,
ptr: Pointer,
size: u64,
) -> InterpResult<'tcx, (bool, u64)> {
let this = self.eval_context_mut();
let os_str =
this.convert_path(Cow::Borrowed(path.as_os_str()), PathConversion::HostToTarget);
this.write_os_str_to_wide_str(&os_str, ptr, size)
}
/// Write a Path to the machine memory (as a null-terminated sequence of `u16`s),
/// adjusting path separators if needed.
fn write_path_to_wide_str_truncated(
&mut self,
path: &Path,
ptr: Pointer,
size: u64,
) -> InterpResult<'tcx, (bool, u64)> {
let this = self.eval_context_mut();
let os_str =
this.convert_path(Cow::Borrowed(path.as_os_str()), PathConversion::HostToTarget);
this.write_os_str_to_wide_str_truncated(&os_str, ptr, size)
}
/// Allocate enough memory to store a Path as a null-terminated sequence of bytes,
/// adjusting path separators if needed.
fn alloc_path_as_c_str(
&mut self,
path: &Path,
memkind: MemoryKind,
) -> InterpResult<'tcx, Pointer> {
let this = self.eval_context_mut();
let os_str =
this.convert_path(Cow::Borrowed(path.as_os_str()), PathConversion::HostToTarget);
this.alloc_os_str_as_c_str(&os_str, memkind)
}
/// Allocate enough memory to store a Path as a null-terminated sequence of `u16`s,
/// adjusting path separators if needed.
fn alloc_path_as_wide_str(
&mut self,
path: &Path,
memkind: MemoryKind,
) -> InterpResult<'tcx, Pointer> {
let this = self.eval_context_mut();
let os_str =
this.convert_path(Cow::Borrowed(path.as_os_str()), PathConversion::HostToTarget);
this.alloc_os_str_as_wide_str(&os_str, memkind)
}
fn convert_path<'a>(
&self,
os_str: Cow<'a, OsStr>,
direction: PathConversion,
) -> Cow<'a, OsStr> {
let this = self.eval_context_ref();
let target_os = &this.tcx.sess.target.os;
/// Adjust a Windows path to Unix conventions such that it un-does everything that
/// `unix_to_windows` did, and such that if the Windows input path was absolute, then the
/// Unix output path is absolute.
fn windows_to_unix<T>(path: &mut Vec<T>)
where
T: From<u8> + Copy + Eq,
{
let sep = T::from(b'/');
// Make sure all path separators are `/`.
for c in path.iter_mut() {
if *c == b'\\'.into() {
*c = sep;
}
}
// If this starts with `//?/`, it was probably produced by `unix_to_windows`` and we
// remove the `//?` that got added to get the Unix path back out.
if path.get(0..4) == Some(&[sep, sep, b'?'.into(), sep]) {
// Remove first 3 characters. It still starts with `/` so it is absolute on Unix.
path.splice(0..3, std::iter::empty());
}
// If it starts with a drive letter (`X:/`), convert it to an absolute Unix path.
else if path.get(1..3) == Some(&[b':'.into(), sep]) {
// We add a `/` at the beginning, to store the absolute Windows
// path in something that looks like an absolute Unix path.
path.insert(0, sep);
}
}
/// Adjust a Unix path to Windows conventions such that it un-does everything that
/// `windows_to_unix` did, and such that if the Unix input path was absolute, then the
/// Windows output path is absolute.
fn unix_to_windows<T>(path: &mut Vec<T>)
where
T: From<u8> + Copy + Eq,
{
let sep = T::from(b'\\');
// Make sure all path separators are `\`.
for c in path.iter_mut() {
if *c == b'/'.into() {
*c = sep;
}
}
// If the path is `\X:\`, the leading separator was probably added by `windows_to_unix`
// and we should get rid of it again.
if path.get(2..4) == Some(&[b':'.into(), sep]) && path[0] == sep {
// The new path is still absolute on Windows.
path.remove(0);
}
// If this starts withs a `\` but not a `\\`, then this was absolute on Unix but is
// relative on Windows (relative to "the root of the current directory", e.g. the
// drive letter).
else if path.first() == Some(&sep) && path.get(1) != Some(&sep) {
// We add `\\?` so it starts with `\\?\` which is some magic path on Windows
// that *is* considered absolute. This way we store the absolute Unix path
// in something that looks like an absolute Windows path.
path.splice(0..0, [sep, sep, b'?'.into()]);
}
}
// Below we assume that everything non-Windows works like Unix, at least
// when it comes to file system path conventions.
#[cfg(windows)]
return if target_os == "windows" {
// Windows-on-Windows, all fine.
os_str
} else {
// Unix target, Windows host.
let mut path: Vec<u16> = os_str.encode_wide().collect();
match direction {
PathConversion::HostToTarget => {
windows_to_unix(&mut path);
}
PathConversion::TargetToHost => {
unix_to_windows(&mut path);
}
}
Cow::Owned(OsString::from_wide(&path))
};
#[cfg(unix)]
return if target_os == "windows" {
// Windows target, Unix host.
let mut path: Vec<u8> = os_str.into_owned().into_encoded_bytes();
match direction {
PathConversion::HostToTarget => {
unix_to_windows(&mut path);
}
PathConversion::TargetToHost => {
windows_to_unix(&mut path);
}
}
Cow::Owned(OsString::from_vec(path))
} else {
// Unix-on-Unix, all is fine.
os_str
};
}
}