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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582
//! Non-operator trait implementations. use crate::{ domain::Domain, mem::BitMemory, order::{ BitOrder, Lsb0, Msb0, }, slice::BitSlice, store::BitStore, view::BitView, }; use core::{ any::TypeId, cmp, convert::TryFrom, fmt::{ self, Binary, Debug, Display, Formatter, LowerHex, Octal, UpperHex, }, hash::{ Hash, Hasher, }, str, }; use tap::pipe::Pipe; #[cfg(feature = "alloc")] use crate::vec::BitVec; #[cfg(feature = "alloc")] use alloc::borrow::ToOwned; impl<O, T> Eq for BitSlice<O, T> where O: BitOrder, T: BitStore, { } impl<O, T> Ord for BitSlice<O, T> where O: BitOrder, T: BitStore, { fn cmp(&self, rhs: &Self) -> cmp::Ordering { self.partial_cmp(rhs) .expect("BitSlice has a total ordering") } } /** Tests if two `BitSlice`s are semantically — not bitwise — equal. It is valid to compare slices of different ordering or memory types. The equality condition requires that they have the same length and that at each index, the two slices have the same bit value. **/ impl<O1, O2, T1, T2> PartialEq<BitSlice<O2, T2>> for BitSlice<O1, T1> where O1: BitOrder, O2: BitOrder, T1: BitStore, T2: BitStore, { fn eq(&self, rhs: &BitSlice<O2, T2>) -> bool { let fallback = || { if self.len() != rhs.len() { return false; } self.iter() .by_val() .zip(rhs.iter().by_val()) .all(|(l, r)| l == r) }; if TypeId::of::<O1>() == TypeId::of::<O2>() && TypeId::of::<T1>() == TypeId::of::<T2>() { if TypeId::of::<O1>() == TypeId::of::<Lsb0>() { let this: &BitSlice<Lsb0, T1> = unsafe { &*(self as *const _ as *const _) }; let that: &BitSlice<Lsb0, T1> = unsafe { &*(rhs as *const _ as *const _) }; this.sp_eq(that) } else if TypeId::of::<O1>() == TypeId::of::<Msb0>() { let this: &BitSlice<Msb0, T1> = unsafe { &*(self as *const _ as *const _) }; let that: &BitSlice<Msb0, T1> = unsafe { &*(rhs as *const _ as *const _) }; this.sp_eq(that) } else { fallback() } } else { fallback() } } } // ref-to-val equality impl<O1, O2, T1, T2> PartialEq<BitSlice<O2, T2>> for &BitSlice<O1, T1> where O1: BitOrder, O2: BitOrder, T1: BitStore, T2: BitStore, { fn eq(&self, rhs: &BitSlice<O2, T2>) -> bool { **self == rhs } } impl<O1, O2, T1, T2> PartialEq<BitSlice<O2, T2>> for &mut BitSlice<O1, T1> where O1: BitOrder, O2: BitOrder, T1: BitStore, T2: BitStore, { fn eq(&self, rhs: &BitSlice<O2, T2>) -> bool { **self == rhs } } // val-to-ref equality impl<O1, O2, T1, T2> PartialEq<&BitSlice<O2, T2>> for BitSlice<O1, T1> where O1: BitOrder, O2: BitOrder, T1: BitStore, T2: BitStore, { fn eq(&self, rhs: &&BitSlice<O2, T2>) -> bool { *self == **rhs } } impl<O1, O2, T1, T2> PartialEq<&mut BitSlice<O2, T2>> for BitSlice<O1, T1> where O1: BitOrder, O2: BitOrder, T1: BitStore, T2: BitStore, { fn eq(&self, rhs: &&mut BitSlice<O2, T2>) -> bool { *self == **rhs } } /** Compares two `BitSlice`s by semantic — not bitwise — ordering. The comparison sorts by testing at each index if one slice has a high bit where the other has a low. At the first index where the slices differ, the slice with the high bit is greater. If the slices are equal until at least one terminates, then they are compared by length. **/ impl<O1, O2, T1, T2> PartialOrd<BitSlice<O2, T2>> for BitSlice<O1, T1> where O1: BitOrder, O2: BitOrder, T1: BitStore, T2: BitStore, { fn partial_cmp(&self, rhs: &BitSlice<O2, T2>) -> Option<cmp::Ordering> { for (l, r) in self.iter().zip(rhs.iter()) { match (*l, *r) { (true, false) => return Some(cmp::Ordering::Greater), (false, true) => return Some(cmp::Ordering::Less), _ => continue, } } self.len().partial_cmp(&rhs.len()) } } // ref-to-val ordering impl<O1, O2, T1, T2> PartialOrd<BitSlice<O2, T2>> for &BitSlice<O1, T1> where O1: BitOrder, O2: BitOrder, T1: BitStore, T2: BitStore, { fn partial_cmp(&self, rhs: &BitSlice<O2, T2>) -> Option<cmp::Ordering> { (*self).partial_cmp(rhs) } } impl<O1, O2, T1, T2> PartialOrd<BitSlice<O2, T2>> for &mut BitSlice<O1, T1> where O1: BitOrder, O2: BitOrder, T1: BitStore, T2: BitStore, { fn partial_cmp(&self, rhs: &BitSlice<O2, T2>) -> Option<cmp::Ordering> { (**self).partial_cmp(rhs) } } // val-to-ref ordering impl<O1, O2, T1, T2> PartialOrd<&BitSlice<O2, T2>> for BitSlice<O1, T1> where O1: BitOrder, O2: BitOrder, T1: BitStore, T2: BitStore, { fn partial_cmp(&self, rhs: &&BitSlice<O2, T2>) -> Option<cmp::Ordering> { (*self).partial_cmp(&**rhs) } } impl<O1, O2, T1, T2> PartialOrd<&mut BitSlice<O2, T2>> for BitSlice<O1, T1> where O1: BitOrder, O2: BitOrder, T1: BitStore, T2: BitStore, { fn partial_cmp(&self, rhs: &&mut BitSlice<O2, T2>) -> Option<cmp::Ordering> { (*self).partial_cmp(&**rhs) } } // &mut-to-& ordering impl<O1, O2, T1, T2> PartialOrd<&mut BitSlice<O2, T2>> for &BitSlice<O1, T1> where O1: BitOrder, O2: BitOrder, T1: BitStore, T2: BitStore, { fn partial_cmp(&self, rhs: &&mut BitSlice<O2, T2>) -> Option<cmp::Ordering> { (**self).partial_cmp(&**rhs) } } impl<O1, O2, T1, T2> PartialOrd<&BitSlice<O2, T2>> for &mut BitSlice<O1, T1> where O1: BitOrder, O2: BitOrder, T1: BitStore, T2: BitStore, { fn partial_cmp(&self, rhs: &&BitSlice<O2, T2>) -> Option<cmp::Ordering> { (**self).partial_cmp(&**rhs) } } impl<'a, O, T> TryFrom<&'a [T]> for &'a BitSlice<O, T> where O: BitOrder, T: BitStore, { type Error = &'a [T]; fn try_from(slice: &'a [T]) -> Result<Self, Self::Error> { BitSlice::from_slice(slice).map_err(|_| slice) } } impl<'a, O, T> TryFrom<&'a mut [T]> for &'a mut BitSlice<O, T> where O: BitOrder, T: BitStore, { type Error = &'a mut [T]; fn try_from(slice: &'a mut [T]) -> Result<Self, Self::Error> { let slice_ptr = slice as *mut [T]; BitSlice::from_slice_mut(slice).map_err(|_| unsafe { &mut *slice_ptr }) } } impl<O, T> Default for &BitSlice<O, T> where O: BitOrder, T: BitStore, { fn default() -> Self { BitSlice::empty() } } impl<O, T> Default for &mut BitSlice<O, T> where O: BitOrder, T: BitStore, { fn default() -> Self { BitSlice::empty_mut() } } impl<O, T> Debug for BitSlice<O, T> where O: BitOrder, T: BitStore, { fn fmt(&self, fmt: &mut Formatter) -> fmt::Result { self.as_bitspan().render(fmt, "Slice", None)?; fmt.write_str(" ")?; Display::fmt(self, fmt) } } impl<O, T> Display for BitSlice<O, T> where O: BitOrder, T: BitStore, { fn fmt(&self, fmt: &mut Formatter) -> fmt::Result { Binary::fmt(self, fmt) } } /// Constructs numeric formatting implementations. macro_rules! fmt { ($trait:ident, $base:expr, $pfx:expr, $blksz:expr) => { /// Render the contents of a `BitSlice` in a numeric format. /// /// These implementations render the bits of memory contained in a /// `BitSlice` as one of the three numeric bases that the Rust format /// system supports: /// /// - `Binary` renders each bit individually as `0` or `1`, /// - `Octal` renders clusters of three bits as the numbers `0` through /// `7`, /// - and `UpperHex` and `LowerHex` render clusters of four bits as the /// numbers `0` through `9` and `A` through `F`. /// /// The formatters produce a “word” for each element `T` of memory. The /// chunked formats (octal and hexadecimal) operate somewhat peculiarly: /// they show the semantic value of the memory, as interpreted by the /// ordering parameter’s implementation rather than the raw value of /// memory you might observe with a debugger. In order to ease the /// process of expanding numbers back into bits, each digit is grouped to /// the right edge of the memory element. So, for example, the byte /// `0xFF` would be rendered in as `0o377` rather than `0o773`. /// /// Rendered words are chunked by memory elements, rather than by as /// clean as possible a number of digits, in order to aid visualization /// of the slice’s place in memory. impl<O, T> $trait for BitSlice<O, T> where O: BitOrder, T: BitStore, { fn fmt(&self, fmt: &mut Formatter) -> fmt::Result { /// Renders an accumulated text buffer as UTF-8. struct Seq<'a>(&'a [u8]); impl Debug for Seq<'_> { fn fmt(&self, fmt: &mut Formatter) -> fmt::Result { fmt.write_str(unsafe { str::from_utf8_unchecked(self.0) }) } } // If the alternate flag is set, include the radix prefix. let start = if fmt.alternate() { 0 } else { 2 }; // Create a list format accumulator. let mut dbg = fmt.debug_list(); /* Create a static buffer sized for the maximum number of UTF-8 bytes needed to render a `usize` in the selected radix. Rust does not yet grant access to trait constants for use in constant expressions within generics. */ const D: usize = <usize as BitMemory>::BITS as usize / $blksz; #[allow(clippy::modulo_one)] const M: usize = <usize as BitMemory>::BITS as usize % $blksz; const W: usize = D + (M != 0) as usize; let mut w: [u8; W + 2] = [b'0'; W + 2]; // Write the prefix symbol into the buffer. w[1] = $pfx; /* This closure does the main work of rendering a bit-slice as text. It will be called on each memory element of the slice being formatted. */ let mut writer = |bits: &BitSlice<O, T::Mem>| { // Set the end index of the text accumulator. let mut end = 2; /* Taking `rchunks` clusters the bits to the right edge, so that any remainder is in the left-most (first-rendered) digit, in the same manner as English digit clusters in ordinary writing. Since `rchunks` takes from back to front, it must be reversed in order to traverse the slice from front to back. The enumeration provides the offset from the buffer start for writing the computed digit into the text accumulator. */ for chunk in bits.rchunks($blksz).rev() { /* Copy the bits of the slice into the temporary, in Msb0 order, at the LSedge of the temporary. This will translate the bit sequence into the binary digit that represents it. */ let mut val = 0u8; for bit in chunk { val <<= 1; val |= *bit as u8; } /* Translate the accumulator digit into the matching ASCII hexadecimal glyph, and write the glyph into the text accumulator. */ w[end] = match val { v @ 0 ..= 9 => b'0' + v, v @ 10 ..= 16 => $base + (v - 10), _ => unsafe { core::hint::unreachable_unchecked() }, }; end += 1; } // View the text accumulator as UTF-8 and write it into the // main formatter. dbg.entry(&Seq(&w[start .. end])); }; /* Break the source `BitSlice` into its aliased sub-regions. This is necessary in order to load each element into local memory for formatting. */ match self.domain() { Domain::Enclave { head, elem, tail } => { // Load a copy of `*elem` into the stack, let tmp = elem.load_value(); // View the whole element as bits, narrow it to the // live span, and render. let bits = tmp.view_bits::<O>(); unsafe { bits.get_unchecked( head.value() as usize .. tail.value() as usize, ) } .pipe(writer); }, // Same process as above, but at different truncations. Domain::Region { head, body, tail } => { if let Some((head, elem)) = head { let tmp = elem.load_value(); let bits = tmp.view_bits::<O>(); unsafe { bits.get_unchecked(head.value() as usize ..) } .pipe(&mut writer); } for elem in body.iter().map(BitStore::load_value) { elem.view_bits::<O>().pipe(&mut writer); } if let Some((elem, tail)) = tail { let tmp = elem.load_value(); let bits = tmp.view_bits::<O>(); unsafe { bits.get_unchecked(.. tail.value() as usize) } .pipe(&mut writer); } }, } dbg.finish() } } }; } fmt!(Binary, b'0', b'b', 1); fmt!(Octal, b'0', b'o', 3); fmt!(LowerHex, b'a', b'x', 4); fmt!(UpperHex, b'A', b'x', 4); /// Writes the contents of the `BitSlice`, in semantic bit order, into a hasher. #[cfg(not(tarpaulin_include))] impl<O, T> Hash for BitSlice<O, T> where O: BitOrder, T: BitStore, { fn hash<H>(&self, hasher: &mut H) where H: Hasher { for bit in self { hasher.write_u8(*bit as u8); } } } /** Conditionally mark `BitSlice` as `Send` based on its `T` type argument. In order for `BitSlice` to be `Send` (that is, `&mut BitSlice` can be moved across thread boundaries), it must be capable of writing to memory without invalidating any other `&BitSlice` handles that alias the same memory address. This is true when `T` is one of the fundamental integers, because no other `&BitSlice` handle is able to observe mutations, or when `T` is a `BitSafe` type that implements atomic read-modify-write instructions, because other `&BitSlice` types will be protected from data races by the hardware. When `T` is a non-atomic `BitSafe` type, `BitSlice` cannot be `Send`, because one `&mut BitSlice` moved across a thread boundary may cause mutation that another `&BitSlice` may observe, but the instructions used to access memory do not guard against data races. A `&mut BitSlice` over aliased memory addresses is equivalent to either a `&Cell` or `&AtomicT`, depending on what the [`radium`] crate makes available for the register width. [`radium`]: radium::types **/ unsafe impl<O, T> Send for BitSlice<O, T> where O: BitOrder, T: BitStore + Sync, { } /** Conditionally mark `BitSlice` as `Sync` based on its `T` type argument. In order for `BitSlice` to be `Sync` (that is, `&BitSlice` can be copied across thread boundaries), it must be capable of reading from memory without being invalidated by any other `&mut BitSlice` handles that alias the same memory address. This is true when `T` is one of the fundamental integers, because no other `&mut BitSlice` handle can exist to effect mutations, or when `T` is a `BitSafe` type that implements atomic read-modify-write instructions, because it will guard against other `&mut BitSlice` modifications in hardware. When `T` is a non-atomic `BitSafe` type, `BitSlice` cannot be `Sync`, because one `&BitSlice` moved across a thread boundary may read from memory that is modified by the originally-owning thread, but the instructions used to access memory do not guard against such data races. A `&BitSlice` over aliased memory addresses is equivalent to either a `&Cell` or `&AtomicT`, depending on what the [`radium`] crate makes available for the register width. [`radium`]: radium::types **/ unsafe impl<O, T> Sync for BitSlice<O, T> where O: BitOrder, T: BitStore + Sync, { } #[cfg(feature = "alloc")] impl<O, T> ToOwned for BitSlice<O, T> where O: BitOrder, T: BitStore, { type Owned = BitVec<O, T>; fn to_owned(&self) -> Self::Owned { BitVec::from_bitslice(self) } }