kernel/str.rs
1// SPDX-License-Identifier: GPL-2.0
2
3//! String representations.
4
5use crate::alloc::{flags::*, AllocError, KVec};
6use core::fmt::{self, Write};
7use core::ops::{self, Deref, DerefMut, Index};
8
9use crate::error::{code::*, Error};
10
11/// Byte string without UTF-8 validity guarantee.
12#[repr(transparent)]
13pub struct BStr([u8]);
14
15impl BStr {
16 /// Returns the length of this string.
17 #[inline]
18 pub const fn len(&self) -> usize {
19 self.0.len()
20 }
21
22 /// Returns `true` if the string is empty.
23 #[inline]
24 pub const fn is_empty(&self) -> bool {
25 self.len() == 0
26 }
27
28 /// Creates a [`BStr`] from a `[u8]`.
29 #[inline]
30 pub const fn from_bytes(bytes: &[u8]) -> &Self {
31 // SAFETY: `BStr` is transparent to `[u8]`.
32 unsafe { &*(bytes as *const [u8] as *const BStr) }
33 }
34}
35
36impl fmt::Display for BStr {
37 /// Formats printable ASCII characters, escaping the rest.
38 ///
39 /// ```
40 /// # use kernel::{fmt, b_str, str::{BStr, CString}};
41 /// let ascii = b_str!("Hello, BStr!");
42 /// let s = CString::try_from_fmt(fmt!("{}", ascii))?;
43 /// assert_eq!(s.as_bytes(), "Hello, BStr!".as_bytes());
44 ///
45 /// let non_ascii = b_str!("🦀");
46 /// let s = CString::try_from_fmt(fmt!("{}", non_ascii))?;
47 /// assert_eq!(s.as_bytes(), "\\xf0\\x9f\\xa6\\x80".as_bytes());
48 /// # Ok::<(), kernel::error::Error>(())
49 /// ```
50 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
51 for &b in &self.0 {
52 match b {
53 // Common escape codes.
54 b'\t' => f.write_str("\\t")?,
55 b'\n' => f.write_str("\\n")?,
56 b'\r' => f.write_str("\\r")?,
57 // Printable characters.
58 0x20..=0x7e => f.write_char(b as char)?,
59 _ => write!(f, "\\x{:02x}", b)?,
60 }
61 }
62 Ok(())
63 }
64}
65
66impl fmt::Debug for BStr {
67 /// Formats printable ASCII characters with a double quote on either end,
68 /// escaping the rest.
69 ///
70 /// ```
71 /// # use kernel::{fmt, b_str, str::{BStr, CString}};
72 /// // Embedded double quotes are escaped.
73 /// let ascii = b_str!("Hello, \"BStr\"!");
74 /// let s = CString::try_from_fmt(fmt!("{:?}", ascii))?;
75 /// assert_eq!(s.as_bytes(), "\"Hello, \\\"BStr\\\"!\"".as_bytes());
76 ///
77 /// let non_ascii = b_str!("😺");
78 /// let s = CString::try_from_fmt(fmt!("{:?}", non_ascii))?;
79 /// assert_eq!(s.as_bytes(), "\"\\xf0\\x9f\\x98\\xba\"".as_bytes());
80 /// # Ok::<(), kernel::error::Error>(())
81 /// ```
82 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
83 f.write_char('"')?;
84 for &b in &self.0 {
85 match b {
86 // Common escape codes.
87 b'\t' => f.write_str("\\t")?,
88 b'\n' => f.write_str("\\n")?,
89 b'\r' => f.write_str("\\r")?,
90 // String escape characters.
91 b'\"' => f.write_str("\\\"")?,
92 b'\\' => f.write_str("\\\\")?,
93 // Printable characters.
94 0x20..=0x7e => f.write_char(b as char)?,
95 _ => write!(f, "\\x{:02x}", b)?,
96 }
97 }
98 f.write_char('"')
99 }
100}
101
102impl Deref for BStr {
103 type Target = [u8];
104
105 #[inline]
106 fn deref(&self) -> &Self::Target {
107 &self.0
108 }
109}
110
111/// Creates a new [`BStr`] from a string literal.
112///
113/// `b_str!` converts the supplied string literal to byte string, so non-ASCII
114/// characters can be included.
115///
116/// # Examples
117///
118/// ```
119/// # use kernel::b_str;
120/// # use kernel::str::BStr;
121/// const MY_BSTR: &BStr = b_str!("My awesome BStr!");
122/// ```
123#[macro_export]
124macro_rules! b_str {
125 ($str:literal) => {{
126 const S: &'static str = $str;
127 const C: &'static $crate::str::BStr = $crate::str::BStr::from_bytes(S.as_bytes());
128 C
129 }};
130}
131
132/// Possible errors when using conversion functions in [`CStr`].
133#[derive(Debug, Clone, Copy)]
134pub enum CStrConvertError {
135 /// Supplied bytes contain an interior `NUL`.
136 InteriorNul,
137
138 /// Supplied bytes are not terminated by `NUL`.
139 NotNulTerminated,
140}
141
142impl From<CStrConvertError> for Error {
143 #[inline]
144 fn from(_: CStrConvertError) -> Error {
145 EINVAL
146 }
147}
148
149/// A string that is guaranteed to have exactly one `NUL` byte, which is at the
150/// end.
151///
152/// Used for interoperability with kernel APIs that take C strings.
153#[repr(transparent)]
154pub struct CStr([u8]);
155
156impl CStr {
157 /// Returns the length of this string excluding `NUL`.
158 #[inline]
159 pub const fn len(&self) -> usize {
160 self.len_with_nul() - 1
161 }
162
163 /// Returns the length of this string with `NUL`.
164 #[inline]
165 pub const fn len_with_nul(&self) -> usize {
166 if self.0.is_empty() {
167 // SAFETY: This is one of the invariant of `CStr`.
168 // We add a `unreachable_unchecked` here to hint the optimizer that
169 // the value returned from this function is non-zero.
170 unsafe { core::hint::unreachable_unchecked() };
171 }
172 self.0.len()
173 }
174
175 /// Returns `true` if the string only includes `NUL`.
176 #[inline]
177 pub const fn is_empty(&self) -> bool {
178 self.len() == 0
179 }
180
181 /// Wraps a raw C string pointer.
182 ///
183 /// # Safety
184 ///
185 /// `ptr` must be a valid pointer to a `NUL`-terminated C string, and it must
186 /// last at least `'a`. When `CStr` is alive, the memory pointed by `ptr`
187 /// must not be mutated.
188 #[inline]
189 pub unsafe fn from_char_ptr<'a>(ptr: *const crate::ffi::c_char) -> &'a Self {
190 // SAFETY: The safety precondition guarantees `ptr` is a valid pointer
191 // to a `NUL`-terminated C string.
192 let len = unsafe { bindings::strlen(ptr) } + 1;
193 // SAFETY: Lifetime guaranteed by the safety precondition.
194 let bytes = unsafe { core::slice::from_raw_parts(ptr as _, len) };
195 // SAFETY: As `len` is returned by `strlen`, `bytes` does not contain interior `NUL`.
196 // As we have added 1 to `len`, the last byte is known to be `NUL`.
197 unsafe { Self::from_bytes_with_nul_unchecked(bytes) }
198 }
199
200 /// Creates a [`CStr`] from a `[u8]`.
201 ///
202 /// The provided slice must be `NUL`-terminated, does not contain any
203 /// interior `NUL` bytes.
204 pub const fn from_bytes_with_nul(bytes: &[u8]) -> Result<&Self, CStrConvertError> {
205 if bytes.is_empty() {
206 return Err(CStrConvertError::NotNulTerminated);
207 }
208 if bytes[bytes.len() - 1] != 0 {
209 return Err(CStrConvertError::NotNulTerminated);
210 }
211 let mut i = 0;
212 // `i + 1 < bytes.len()` allows LLVM to optimize away bounds checking,
213 // while it couldn't optimize away bounds checks for `i < bytes.len() - 1`.
214 while i + 1 < bytes.len() {
215 if bytes[i] == 0 {
216 return Err(CStrConvertError::InteriorNul);
217 }
218 i += 1;
219 }
220 // SAFETY: We just checked that all properties hold.
221 Ok(unsafe { Self::from_bytes_with_nul_unchecked(bytes) })
222 }
223
224 /// Creates a [`CStr`] from a `[u8]` without performing any additional
225 /// checks.
226 ///
227 /// # Safety
228 ///
229 /// `bytes` *must* end with a `NUL` byte, and should only have a single
230 /// `NUL` byte (or the string will be truncated).
231 #[inline]
232 pub const unsafe fn from_bytes_with_nul_unchecked(bytes: &[u8]) -> &CStr {
233 // SAFETY: Properties of `bytes` guaranteed by the safety precondition.
234 unsafe { core::mem::transmute(bytes) }
235 }
236
237 /// Creates a mutable [`CStr`] from a `[u8]` without performing any
238 /// additional checks.
239 ///
240 /// # Safety
241 ///
242 /// `bytes` *must* end with a `NUL` byte, and should only have a single
243 /// `NUL` byte (or the string will be truncated).
244 #[inline]
245 pub unsafe fn from_bytes_with_nul_unchecked_mut(bytes: &mut [u8]) -> &mut CStr {
246 // SAFETY: Properties of `bytes` guaranteed by the safety precondition.
247 unsafe { &mut *(bytes as *mut [u8] as *mut CStr) }
248 }
249
250 /// Returns a C pointer to the string.
251 #[inline]
252 pub const fn as_char_ptr(&self) -> *const crate::ffi::c_char {
253 self.0.as_ptr()
254 }
255
256 /// Convert the string to a byte slice without the trailing `NUL` byte.
257 #[inline]
258 pub fn as_bytes(&self) -> &[u8] {
259 &self.0[..self.len()]
260 }
261
262 /// Convert the string to a byte slice containing the trailing `NUL` byte.
263 #[inline]
264 pub const fn as_bytes_with_nul(&self) -> &[u8] {
265 &self.0
266 }
267
268 /// Yields a [`&str`] slice if the [`CStr`] contains valid UTF-8.
269 ///
270 /// If the contents of the [`CStr`] are valid UTF-8 data, this
271 /// function will return the corresponding [`&str`] slice. Otherwise,
272 /// it will return an error with details of where UTF-8 validation failed.
273 ///
274 /// # Examples
275 ///
276 /// ```
277 /// # use kernel::str::CStr;
278 /// let cstr = CStr::from_bytes_with_nul(b"foo\0")?;
279 /// assert_eq!(cstr.to_str(), Ok("foo"));
280 /// # Ok::<(), kernel::error::Error>(())
281 /// ```
282 #[inline]
283 pub fn to_str(&self) -> Result<&str, core::str::Utf8Error> {
284 core::str::from_utf8(self.as_bytes())
285 }
286
287 /// Unsafely convert this [`CStr`] into a [`&str`], without checking for
288 /// valid UTF-8.
289 ///
290 /// # Safety
291 ///
292 /// The contents must be valid UTF-8.
293 ///
294 /// # Examples
295 ///
296 /// ```
297 /// # use kernel::c_str;
298 /// # use kernel::str::CStr;
299 /// let bar = c_str!("ツ");
300 /// // SAFETY: String literals are guaranteed to be valid UTF-8
301 /// // by the Rust compiler.
302 /// assert_eq!(unsafe { bar.as_str_unchecked() }, "ツ");
303 /// ```
304 #[inline]
305 pub unsafe fn as_str_unchecked(&self) -> &str {
306 // SAFETY: TODO.
307 unsafe { core::str::from_utf8_unchecked(self.as_bytes()) }
308 }
309
310 /// Convert this [`CStr`] into a [`CString`] by allocating memory and
311 /// copying over the string data.
312 pub fn to_cstring(&self) -> Result<CString, AllocError> {
313 CString::try_from(self)
314 }
315
316 /// Converts this [`CStr`] to its ASCII lower case equivalent in-place.
317 ///
318 /// ASCII letters 'A' to 'Z' are mapped to 'a' to 'z',
319 /// but non-ASCII letters are unchanged.
320 ///
321 /// To return a new lowercased value without modifying the existing one, use
322 /// [`to_ascii_lowercase()`].
323 ///
324 /// [`to_ascii_lowercase()`]: #method.to_ascii_lowercase
325 pub fn make_ascii_lowercase(&mut self) {
326 // INVARIANT: This doesn't introduce or remove NUL bytes in the C
327 // string.
328 self.0.make_ascii_lowercase();
329 }
330
331 /// Converts this [`CStr`] to its ASCII upper case equivalent in-place.
332 ///
333 /// ASCII letters 'a' to 'z' are mapped to 'A' to 'Z',
334 /// but non-ASCII letters are unchanged.
335 ///
336 /// To return a new uppercased value without modifying the existing one, use
337 /// [`to_ascii_uppercase()`].
338 ///
339 /// [`to_ascii_uppercase()`]: #method.to_ascii_uppercase
340 pub fn make_ascii_uppercase(&mut self) {
341 // INVARIANT: This doesn't introduce or remove NUL bytes in the C
342 // string.
343 self.0.make_ascii_uppercase();
344 }
345
346 /// Returns a copy of this [`CString`] where each character is mapped to its
347 /// ASCII lower case equivalent.
348 ///
349 /// ASCII letters 'A' to 'Z' are mapped to 'a' to 'z',
350 /// but non-ASCII letters are unchanged.
351 ///
352 /// To lowercase the value in-place, use [`make_ascii_lowercase`].
353 ///
354 /// [`make_ascii_lowercase`]: str::make_ascii_lowercase
355 pub fn to_ascii_lowercase(&self) -> Result<CString, AllocError> {
356 let mut s = self.to_cstring()?;
357
358 s.make_ascii_lowercase();
359
360 Ok(s)
361 }
362
363 /// Returns a copy of this [`CString`] where each character is mapped to its
364 /// ASCII upper case equivalent.
365 ///
366 /// ASCII letters 'a' to 'z' are mapped to 'A' to 'Z',
367 /// but non-ASCII letters are unchanged.
368 ///
369 /// To uppercase the value in-place, use [`make_ascii_uppercase`].
370 ///
371 /// [`make_ascii_uppercase`]: str::make_ascii_uppercase
372 pub fn to_ascii_uppercase(&self) -> Result<CString, AllocError> {
373 let mut s = self.to_cstring()?;
374
375 s.make_ascii_uppercase();
376
377 Ok(s)
378 }
379}
380
381impl fmt::Display for CStr {
382 /// Formats printable ASCII characters, escaping the rest.
383 ///
384 /// ```
385 /// # use kernel::c_str;
386 /// # use kernel::fmt;
387 /// # use kernel::str::CStr;
388 /// # use kernel::str::CString;
389 /// let penguin = c_str!("🐧");
390 /// let s = CString::try_from_fmt(fmt!("{}", penguin))?;
391 /// assert_eq!(s.as_bytes_with_nul(), "\\xf0\\x9f\\x90\\xa7\0".as_bytes());
392 ///
393 /// let ascii = c_str!("so \"cool\"");
394 /// let s = CString::try_from_fmt(fmt!("{}", ascii))?;
395 /// assert_eq!(s.as_bytes_with_nul(), "so \"cool\"\0".as_bytes());
396 /// # Ok::<(), kernel::error::Error>(())
397 /// ```
398 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
399 for &c in self.as_bytes() {
400 if (0x20..0x7f).contains(&c) {
401 // Printable character.
402 f.write_char(c as char)?;
403 } else {
404 write!(f, "\\x{:02x}", c)?;
405 }
406 }
407 Ok(())
408 }
409}
410
411impl fmt::Debug for CStr {
412 /// Formats printable ASCII characters with a double quote on either end, escaping the rest.
413 ///
414 /// ```
415 /// # use kernel::c_str;
416 /// # use kernel::fmt;
417 /// # use kernel::str::CStr;
418 /// # use kernel::str::CString;
419 /// let penguin = c_str!("🐧");
420 /// let s = CString::try_from_fmt(fmt!("{:?}", penguin))?;
421 /// assert_eq!(s.as_bytes_with_nul(), "\"\\xf0\\x9f\\x90\\xa7\"\0".as_bytes());
422 ///
423 /// // Embedded double quotes are escaped.
424 /// let ascii = c_str!("so \"cool\"");
425 /// let s = CString::try_from_fmt(fmt!("{:?}", ascii))?;
426 /// assert_eq!(s.as_bytes_with_nul(), "\"so \\\"cool\\\"\"\0".as_bytes());
427 /// # Ok::<(), kernel::error::Error>(())
428 /// ```
429 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
430 f.write_str("\"")?;
431 for &c in self.as_bytes() {
432 match c {
433 // Printable characters.
434 b'\"' => f.write_str("\\\"")?,
435 0x20..=0x7e => f.write_char(c as char)?,
436 _ => write!(f, "\\x{:02x}", c)?,
437 }
438 }
439 f.write_str("\"")
440 }
441}
442
443impl AsRef<BStr> for CStr {
444 #[inline]
445 fn as_ref(&self) -> &BStr {
446 BStr::from_bytes(self.as_bytes())
447 }
448}
449
450impl Deref for CStr {
451 type Target = BStr;
452
453 #[inline]
454 fn deref(&self) -> &Self::Target {
455 self.as_ref()
456 }
457}
458
459impl Index<ops::RangeFrom<usize>> for CStr {
460 type Output = CStr;
461
462 #[inline]
463 fn index(&self, index: ops::RangeFrom<usize>) -> &Self::Output {
464 // Delegate bounds checking to slice.
465 // Assign to _ to mute clippy's unnecessary operation warning.
466 let _ = &self.as_bytes()[index.start..];
467 // SAFETY: We just checked the bounds.
468 unsafe { Self::from_bytes_with_nul_unchecked(&self.0[index.start..]) }
469 }
470}
471
472impl Index<ops::RangeFull> for CStr {
473 type Output = CStr;
474
475 #[inline]
476 fn index(&self, _index: ops::RangeFull) -> &Self::Output {
477 self
478 }
479}
480
481mod private {
482 use core::ops;
483
484 // Marker trait for index types that can be forward to `BStr`.
485 pub trait CStrIndex {}
486
487 impl CStrIndex for usize {}
488 impl CStrIndex for ops::Range<usize> {}
489 impl CStrIndex for ops::RangeInclusive<usize> {}
490 impl CStrIndex for ops::RangeToInclusive<usize> {}
491}
492
493impl<Idx> Index<Idx> for CStr
494where
495 Idx: private::CStrIndex,
496 BStr: Index<Idx>,
497{
498 type Output = <BStr as Index<Idx>>::Output;
499
500 #[inline]
501 fn index(&self, index: Idx) -> &Self::Output {
502 &self.as_ref()[index]
503 }
504}
505
506/// Creates a new [`CStr`] from a string literal.
507///
508/// The string literal should not contain any `NUL` bytes.
509///
510/// # Examples
511///
512/// ```
513/// # use kernel::c_str;
514/// # use kernel::str::CStr;
515/// const MY_CSTR: &CStr = c_str!("My awesome CStr!");
516/// ```
517#[macro_export]
518macro_rules! c_str {
519 ($str:expr) => {{
520 const S: &str = concat!($str, "\0");
521 const C: &$crate::str::CStr = match $crate::str::CStr::from_bytes_with_nul(S.as_bytes()) {
522 Ok(v) => v,
523 Err(_) => panic!("string contains interior NUL"),
524 };
525 C
526 }};
527}
528
529#[cfg(test)]
530#[expect(clippy::items_after_test_module)]
531mod tests {
532 use super::*;
533
534 struct String(CString);
535
536 impl String {
537 fn from_fmt(args: fmt::Arguments<'_>) -> Self {
538 String(CString::try_from_fmt(args).unwrap())
539 }
540 }
541
542 impl Deref for String {
543 type Target = str;
544
545 fn deref(&self) -> &str {
546 self.0.to_str().unwrap()
547 }
548 }
549
550 macro_rules! format {
551 ($($f:tt)*) => ({
552 &*String::from_fmt(kernel::fmt!($($f)*))
553 })
554 }
555
556 const ALL_ASCII_CHARS: &str =
557 "\\x01\\x02\\x03\\x04\\x05\\x06\\x07\\x08\\x09\\x0a\\x0b\\x0c\\x0d\\x0e\\x0f\
558 \\x10\\x11\\x12\\x13\\x14\\x15\\x16\\x17\\x18\\x19\\x1a\\x1b\\x1c\\x1d\\x1e\\x1f \
559 !\"#$%&'()*+,-./0123456789:;<=>?@\
560 ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~\\x7f\
561 \\x80\\x81\\x82\\x83\\x84\\x85\\x86\\x87\\x88\\x89\\x8a\\x8b\\x8c\\x8d\\x8e\\x8f\
562 \\x90\\x91\\x92\\x93\\x94\\x95\\x96\\x97\\x98\\x99\\x9a\\x9b\\x9c\\x9d\\x9e\\x9f\
563 \\xa0\\xa1\\xa2\\xa3\\xa4\\xa5\\xa6\\xa7\\xa8\\xa9\\xaa\\xab\\xac\\xad\\xae\\xaf\
564 \\xb0\\xb1\\xb2\\xb3\\xb4\\xb5\\xb6\\xb7\\xb8\\xb9\\xba\\xbb\\xbc\\xbd\\xbe\\xbf\
565 \\xc0\\xc1\\xc2\\xc3\\xc4\\xc5\\xc6\\xc7\\xc8\\xc9\\xca\\xcb\\xcc\\xcd\\xce\\xcf\
566 \\xd0\\xd1\\xd2\\xd3\\xd4\\xd5\\xd6\\xd7\\xd8\\xd9\\xda\\xdb\\xdc\\xdd\\xde\\xdf\
567 \\xe0\\xe1\\xe2\\xe3\\xe4\\xe5\\xe6\\xe7\\xe8\\xe9\\xea\\xeb\\xec\\xed\\xee\\xef\
568 \\xf0\\xf1\\xf2\\xf3\\xf4\\xf5\\xf6\\xf7\\xf8\\xf9\\xfa\\xfb\\xfc\\xfd\\xfe\\xff";
569
570 #[test]
571 fn test_cstr_to_str() {
572 let good_bytes = b"\xf0\x9f\xa6\x80\0";
573 let checked_cstr = CStr::from_bytes_with_nul(good_bytes).unwrap();
574 let checked_str = checked_cstr.to_str().unwrap();
575 assert_eq!(checked_str, "🦀");
576 }
577
578 #[test]
579 #[should_panic]
580 fn test_cstr_to_str_panic() {
581 let bad_bytes = b"\xc3\x28\0";
582 let checked_cstr = CStr::from_bytes_with_nul(bad_bytes).unwrap();
583 checked_cstr.to_str().unwrap();
584 }
585
586 #[test]
587 fn test_cstr_as_str_unchecked() {
588 let good_bytes = b"\xf0\x9f\x90\xA7\0";
589 let checked_cstr = CStr::from_bytes_with_nul(good_bytes).unwrap();
590 // SAFETY: The contents come from a string literal which contains valid UTF-8.
591 let unchecked_str = unsafe { checked_cstr.as_str_unchecked() };
592 assert_eq!(unchecked_str, "🐧");
593 }
594
595 #[test]
596 fn test_cstr_display() {
597 let hello_world = CStr::from_bytes_with_nul(b"hello, world!\0").unwrap();
598 assert_eq!(format!("{}", hello_world), "hello, world!");
599 let non_printables = CStr::from_bytes_with_nul(b"\x01\x09\x0a\0").unwrap();
600 assert_eq!(format!("{}", non_printables), "\\x01\\x09\\x0a");
601 let non_ascii = CStr::from_bytes_with_nul(b"d\xe9j\xe0 vu\0").unwrap();
602 assert_eq!(format!("{}", non_ascii), "d\\xe9j\\xe0 vu");
603 let good_bytes = CStr::from_bytes_with_nul(b"\xf0\x9f\xa6\x80\0").unwrap();
604 assert_eq!(format!("{}", good_bytes), "\\xf0\\x9f\\xa6\\x80");
605 }
606
607 #[test]
608 fn test_cstr_display_all_bytes() {
609 let mut bytes: [u8; 256] = [0; 256];
610 // fill `bytes` with [1..=255] + [0]
611 for i in u8::MIN..=u8::MAX {
612 bytes[i as usize] = i.wrapping_add(1);
613 }
614 let cstr = CStr::from_bytes_with_nul(&bytes).unwrap();
615 assert_eq!(format!("{}", cstr), ALL_ASCII_CHARS);
616 }
617
618 #[test]
619 fn test_cstr_debug() {
620 let hello_world = CStr::from_bytes_with_nul(b"hello, world!\0").unwrap();
621 assert_eq!(format!("{:?}", hello_world), "\"hello, world!\"");
622 let non_printables = CStr::from_bytes_with_nul(b"\x01\x09\x0a\0").unwrap();
623 assert_eq!(format!("{:?}", non_printables), "\"\\x01\\x09\\x0a\"");
624 let non_ascii = CStr::from_bytes_with_nul(b"d\xe9j\xe0 vu\0").unwrap();
625 assert_eq!(format!("{:?}", non_ascii), "\"d\\xe9j\\xe0 vu\"");
626 let good_bytes = CStr::from_bytes_with_nul(b"\xf0\x9f\xa6\x80\0").unwrap();
627 assert_eq!(format!("{:?}", good_bytes), "\"\\xf0\\x9f\\xa6\\x80\"");
628 }
629
630 #[test]
631 fn test_bstr_display() {
632 let hello_world = BStr::from_bytes(b"hello, world!");
633 assert_eq!(format!("{}", hello_world), "hello, world!");
634 let escapes = BStr::from_bytes(b"_\t_\n_\r_\\_\'_\"_");
635 assert_eq!(format!("{}", escapes), "_\\t_\\n_\\r_\\_'_\"_");
636 let others = BStr::from_bytes(b"\x01");
637 assert_eq!(format!("{}", others), "\\x01");
638 let non_ascii = BStr::from_bytes(b"d\xe9j\xe0 vu");
639 assert_eq!(format!("{}", non_ascii), "d\\xe9j\\xe0 vu");
640 let good_bytes = BStr::from_bytes(b"\xf0\x9f\xa6\x80");
641 assert_eq!(format!("{}", good_bytes), "\\xf0\\x9f\\xa6\\x80");
642 }
643
644 #[test]
645 fn test_bstr_debug() {
646 let hello_world = BStr::from_bytes(b"hello, world!");
647 assert_eq!(format!("{:?}", hello_world), "\"hello, world!\"");
648 let escapes = BStr::from_bytes(b"_\t_\n_\r_\\_\'_\"_");
649 assert_eq!(format!("{:?}", escapes), "\"_\\t_\\n_\\r_\\\\_'_\\\"_\"");
650 let others = BStr::from_bytes(b"\x01");
651 assert_eq!(format!("{:?}", others), "\"\\x01\"");
652 let non_ascii = BStr::from_bytes(b"d\xe9j\xe0 vu");
653 assert_eq!(format!("{:?}", non_ascii), "\"d\\xe9j\\xe0 vu\"");
654 let good_bytes = BStr::from_bytes(b"\xf0\x9f\xa6\x80");
655 assert_eq!(format!("{:?}", good_bytes), "\"\\xf0\\x9f\\xa6\\x80\"");
656 }
657}
658
659/// Allows formatting of [`fmt::Arguments`] into a raw buffer.
660///
661/// It does not fail if callers write past the end of the buffer so that they can calculate the
662/// size required to fit everything.
663///
664/// # Invariants
665///
666/// The memory region between `pos` (inclusive) and `end` (exclusive) is valid for writes if `pos`
667/// is less than `end`.
668pub(crate) struct RawFormatter {
669 // Use `usize` to use `saturating_*` functions.
670 beg: usize,
671 pos: usize,
672 end: usize,
673}
674
675impl RawFormatter {
676 /// Creates a new instance of [`RawFormatter`] with an empty buffer.
677 fn new() -> Self {
678 // INVARIANT: The buffer is empty, so the region that needs to be writable is empty.
679 Self {
680 beg: 0,
681 pos: 0,
682 end: 0,
683 }
684 }
685
686 /// Creates a new instance of [`RawFormatter`] with the given buffer pointers.
687 ///
688 /// # Safety
689 ///
690 /// If `pos` is less than `end`, then the region between `pos` (inclusive) and `end`
691 /// (exclusive) must be valid for writes for the lifetime of the returned [`RawFormatter`].
692 pub(crate) unsafe fn from_ptrs(pos: *mut u8, end: *mut u8) -> Self {
693 // INVARIANT: The safety requirements guarantee the type invariants.
694 Self {
695 beg: pos as _,
696 pos: pos as _,
697 end: end as _,
698 }
699 }
700
701 /// Creates a new instance of [`RawFormatter`] with the given buffer.
702 ///
703 /// # Safety
704 ///
705 /// The memory region starting at `buf` and extending for `len` bytes must be valid for writes
706 /// for the lifetime of the returned [`RawFormatter`].
707 pub(crate) unsafe fn from_buffer(buf: *mut u8, len: usize) -> Self {
708 let pos = buf as usize;
709 // INVARIANT: We ensure that `end` is never less then `buf`, and the safety requirements
710 // guarantees that the memory region is valid for writes.
711 Self {
712 pos,
713 beg: pos,
714 end: pos.saturating_add(len),
715 }
716 }
717
718 /// Returns the current insert position.
719 ///
720 /// N.B. It may point to invalid memory.
721 pub(crate) fn pos(&self) -> *mut u8 {
722 self.pos as _
723 }
724
725 /// Returns the number of bytes written to the formatter.
726 pub(crate) fn bytes_written(&self) -> usize {
727 self.pos - self.beg
728 }
729}
730
731impl fmt::Write for RawFormatter {
732 fn write_str(&mut self, s: &str) -> fmt::Result {
733 // `pos` value after writing `len` bytes. This does not have to be bounded by `end`, but we
734 // don't want it to wrap around to 0.
735 let pos_new = self.pos.saturating_add(s.len());
736
737 // Amount that we can copy. `saturating_sub` ensures we get 0 if `pos` goes past `end`.
738 let len_to_copy = core::cmp::min(pos_new, self.end).saturating_sub(self.pos);
739
740 if len_to_copy > 0 {
741 // SAFETY: If `len_to_copy` is non-zero, then we know `pos` has not gone past `end`
742 // yet, so it is valid for write per the type invariants.
743 unsafe {
744 core::ptr::copy_nonoverlapping(
745 s.as_bytes().as_ptr(),
746 self.pos as *mut u8,
747 len_to_copy,
748 )
749 };
750 }
751
752 self.pos = pos_new;
753 Ok(())
754 }
755}
756
757/// Allows formatting of [`fmt::Arguments`] into a raw buffer.
758///
759/// Fails if callers attempt to write more than will fit in the buffer.
760pub(crate) struct Formatter(RawFormatter);
761
762impl Formatter {
763 /// Creates a new instance of [`Formatter`] with the given buffer.
764 ///
765 /// # Safety
766 ///
767 /// The memory region starting at `buf` and extending for `len` bytes must be valid for writes
768 /// for the lifetime of the returned [`Formatter`].
769 pub(crate) unsafe fn from_buffer(buf: *mut u8, len: usize) -> Self {
770 // SAFETY: The safety requirements of this function satisfy those of the callee.
771 Self(unsafe { RawFormatter::from_buffer(buf, len) })
772 }
773}
774
775impl Deref for Formatter {
776 type Target = RawFormatter;
777
778 fn deref(&self) -> &Self::Target {
779 &self.0
780 }
781}
782
783impl fmt::Write for Formatter {
784 fn write_str(&mut self, s: &str) -> fmt::Result {
785 self.0.write_str(s)?;
786
787 // Fail the request if we go past the end of the buffer.
788 if self.0.pos > self.0.end {
789 Err(fmt::Error)
790 } else {
791 Ok(())
792 }
793 }
794}
795
796/// An owned string that is guaranteed to have exactly one `NUL` byte, which is at the end.
797///
798/// Used for interoperability with kernel APIs that take C strings.
799///
800/// # Invariants
801///
802/// The string is always `NUL`-terminated and contains no other `NUL` bytes.
803///
804/// # Examples
805///
806/// ```
807/// use kernel::{str::CString, fmt};
808///
809/// let s = CString::try_from_fmt(fmt!("{}{}{}", "abc", 10, 20))?;
810/// assert_eq!(s.as_bytes_with_nul(), "abc1020\0".as_bytes());
811///
812/// let tmp = "testing";
813/// let s = CString::try_from_fmt(fmt!("{tmp}{}", 123))?;
814/// assert_eq!(s.as_bytes_with_nul(), "testing123\0".as_bytes());
815///
816/// // This fails because it has an embedded `NUL` byte.
817/// let s = CString::try_from_fmt(fmt!("a\0b{}", 123));
818/// assert_eq!(s.is_ok(), false);
819/// # Ok::<(), kernel::error::Error>(())
820/// ```
821pub struct CString {
822 buf: KVec<u8>,
823}
824
825impl CString {
826 /// Creates an instance of [`CString`] from the given formatted arguments.
827 pub fn try_from_fmt(args: fmt::Arguments<'_>) -> Result<Self, Error> {
828 // Calculate the size needed (formatted string plus `NUL` terminator).
829 let mut f = RawFormatter::new();
830 f.write_fmt(args)?;
831 f.write_str("\0")?;
832 let size = f.bytes_written();
833
834 // Allocate a vector with the required number of bytes, and write to it.
835 let mut buf = KVec::with_capacity(size, GFP_KERNEL)?;
836 // SAFETY: The buffer stored in `buf` is at least of size `size` and is valid for writes.
837 let mut f = unsafe { Formatter::from_buffer(buf.as_mut_ptr(), size) };
838 f.write_fmt(args)?;
839 f.write_str("\0")?;
840
841 // SAFETY: The number of bytes that can be written to `f` is bounded by `size`, which is
842 // `buf`'s capacity. The contents of the buffer have been initialised by writes to `f`.
843 unsafe { buf.set_len(f.bytes_written()) };
844
845 // Check that there are no `NUL` bytes before the end.
846 // SAFETY: The buffer is valid for read because `f.bytes_written()` is bounded by `size`
847 // (which the minimum buffer size) and is non-zero (we wrote at least the `NUL` terminator)
848 // so `f.bytes_written() - 1` doesn't underflow.
849 let ptr = unsafe { bindings::memchr(buf.as_ptr().cast(), 0, f.bytes_written() - 1) };
850 if !ptr.is_null() {
851 return Err(EINVAL);
852 }
853
854 // INVARIANT: We wrote the `NUL` terminator and checked above that no other `NUL` bytes
855 // exist in the buffer.
856 Ok(Self { buf })
857 }
858}
859
860impl Deref for CString {
861 type Target = CStr;
862
863 fn deref(&self) -> &Self::Target {
864 // SAFETY: The type invariants guarantee that the string is `NUL`-terminated and that no
865 // other `NUL` bytes exist.
866 unsafe { CStr::from_bytes_with_nul_unchecked(self.buf.as_slice()) }
867 }
868}
869
870impl DerefMut for CString {
871 fn deref_mut(&mut self) -> &mut Self::Target {
872 // SAFETY: A `CString` is always NUL-terminated and contains no other
873 // NUL bytes.
874 unsafe { CStr::from_bytes_with_nul_unchecked_mut(self.buf.as_mut_slice()) }
875 }
876}
877
878impl<'a> TryFrom<&'a CStr> for CString {
879 type Error = AllocError;
880
881 fn try_from(cstr: &'a CStr) -> Result<CString, AllocError> {
882 let mut buf = KVec::new();
883
884 buf.extend_from_slice(cstr.as_bytes_with_nul(), GFP_KERNEL)?;
885
886 // INVARIANT: The `CStr` and `CString` types have the same invariants for
887 // the string data, and we copied it over without changes.
888 Ok(CString { buf })
889 }
890}
891
892impl fmt::Debug for CString {
893 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
894 fmt::Debug::fmt(&**self, f)
895 }
896}
897
898/// A convenience alias for [`core::format_args`].
899#[macro_export]
900macro_rules! fmt {
901 ($($f:tt)*) => ( core::format_args!($($f)*) )
902}