diff options
Diffstat (limited to 'library/std/src')
32 files changed, 601 insertions, 293 deletions
diff --git a/library/std/src/collections/hash/map.rs b/library/std/src/collections/hash/map.rs index c722bad2e4f..f3316d97c5f 100644 --- a/library/std/src/collections/hash/map.rs +++ b/library/std/src/collections/hash/map.rs @@ -623,28 +623,27 @@ impl<K, V, S> HashMap<K, V, S> { /// If the closure returns false, or panics, the element remains in the map and will not be /// yielded. /// - /// Note that `drain_filter` lets you mutate every value in the filter closure, regardless of + /// Note that `extract_if` lets you mutate every value in the filter closure, regardless of /// whether you choose to keep or remove it. /// - /// If the iterator is only partially consumed or not consumed at all, each of the remaining - /// elements will still be subjected to the closure and removed and dropped if it returns true. + /// If the returned `ExtractIf` is not exhausted, e.g. because it is dropped without iterating + /// or the iteration short-circuits, then the remaining elements will be retained. + /// Use [`retain`] with a negated predicate if you do not need the returned iterator. /// - /// It is unspecified how many more elements will be subjected to the closure - /// if a panic occurs in the closure, or a panic occurs while dropping an element, - /// or if the `DrainFilter` value is leaked. + /// [`retain`]: HashMap::retain /// /// # Examples /// /// Splitting a map into even and odd keys, reusing the original map: /// /// ``` - /// #![feature(hash_drain_filter)] + /// #![feature(hash_extract_if)] /// use std::collections::HashMap; /// /// let mut map: HashMap<i32, i32> = (0..8).map(|x| (x, x)).collect(); - /// let drained: HashMap<i32, i32> = map.drain_filter(|k, _v| k % 2 == 0).collect(); + /// let extracted: HashMap<i32, i32> = map.extract_if(|k, _v| k % 2 == 0).collect(); /// - /// let mut evens = drained.keys().copied().collect::<Vec<_>>(); + /// let mut evens = extracted.keys().copied().collect::<Vec<_>>(); /// let mut odds = map.keys().copied().collect::<Vec<_>>(); /// evens.sort(); /// odds.sort(); @@ -654,12 +653,12 @@ impl<K, V, S> HashMap<K, V, S> { /// ``` #[inline] #[rustc_lint_query_instability] - #[unstable(feature = "hash_drain_filter", issue = "59618")] - pub fn drain_filter<F>(&mut self, pred: F) -> DrainFilter<'_, K, V, F> + #[unstable(feature = "hash_extract_if", issue = "59618")] + pub fn extract_if<F>(&mut self, pred: F) -> ExtractIf<'_, K, V, F> where F: FnMut(&K, &mut V) -> bool, { - DrainFilter { base: self.base.drain_filter(pred) } + ExtractIf { base: self.base.extract_if(pred) } } /// Retains only the elements specified by the predicate. @@ -1578,28 +1577,29 @@ impl<'a, K, V> Drain<'a, K, V> { /// A draining, filtering iterator over the entries of a `HashMap`. /// -/// This `struct` is created by the [`drain_filter`] method on [`HashMap`]. +/// This `struct` is created by the [`extract_if`] method on [`HashMap`]. /// -/// [`drain_filter`]: HashMap::drain_filter +/// [`extract_if`]: HashMap::extract_if /// /// # Example /// /// ``` -/// #![feature(hash_drain_filter)] +/// #![feature(hash_extract_if)] /// /// use std::collections::HashMap; /// /// let mut map = HashMap::from([ /// ("a", 1), /// ]); -/// let iter = map.drain_filter(|_k, v| *v % 2 == 0); +/// let iter = map.extract_if(|_k, v| *v % 2 == 0); /// ``` -#[unstable(feature = "hash_drain_filter", issue = "59618")] -pub struct DrainFilter<'a, K, V, F> +#[unstable(feature = "hash_extract_if", issue = "59618")] +#[must_use = "iterators are lazy and do nothing unless consumed"] +pub struct ExtractIf<'a, K, V, F> where F: FnMut(&K, &mut V) -> bool, { - base: base::DrainFilter<'a, K, V, F>, + base: base::ExtractIf<'a, K, V, F>, } /// A mutable iterator over the values of a `HashMap`. @@ -2479,8 +2479,8 @@ where } } -#[unstable(feature = "hash_drain_filter", issue = "59618")] -impl<K, V, F> Iterator for DrainFilter<'_, K, V, F> +#[unstable(feature = "hash_extract_if", issue = "59618")] +impl<K, V, F> Iterator for ExtractIf<'_, K, V, F> where F: FnMut(&K, &mut V) -> bool, { @@ -2496,16 +2496,16 @@ where } } -#[unstable(feature = "hash_drain_filter", issue = "59618")] -impl<K, V, F> FusedIterator for DrainFilter<'_, K, V, F> where F: FnMut(&K, &mut V) -> bool {} +#[unstable(feature = "hash_extract_if", issue = "59618")] +impl<K, V, F> FusedIterator for ExtractIf<'_, K, V, F> where F: FnMut(&K, &mut V) -> bool {} -#[unstable(feature = "hash_drain_filter", issue = "59618")] -impl<'a, K, V, F> fmt::Debug for DrainFilter<'a, K, V, F> +#[unstable(feature = "hash_extract_if", issue = "59618")] +impl<'a, K, V, F> fmt::Debug for ExtractIf<'a, K, V, F> where F: FnMut(&K, &mut V) -> bool, { fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - f.debug_struct("DrainFilter").finish_non_exhaustive() + f.debug_struct("ExtractIf").finish_non_exhaustive() } } diff --git a/library/std/src/collections/hash/map/tests.rs b/library/std/src/collections/hash/map/tests.rs index 6b89518e2e2..91a3776e7be 100644 --- a/library/std/src/collections/hash/map/tests.rs +++ b/library/std/src/collections/hash/map/tests.rs @@ -944,7 +944,7 @@ fn test_raw_entry() { } } -mod test_drain_filter { +mod test_extract_if { use super::*; use crate::panic::{catch_unwind, AssertUnwindSafe}; @@ -968,7 +968,7 @@ mod test_drain_filter { #[test] fn empty() { let mut map: HashMap<i32, i32> = HashMap::new(); - map.drain_filter(|_, _| unreachable!("there's nothing to decide on")); + map.extract_if(|_, _| unreachable!("there's nothing to decide on")).for_each(drop); assert!(map.is_empty()); } @@ -976,7 +976,7 @@ mod test_drain_filter { fn consuming_nothing() { let pairs = (0..3).map(|i| (i, i)); let mut map: HashMap<_, _> = pairs.collect(); - assert!(map.drain_filter(|_, _| false).eq_sorted(crate::iter::empty())); + assert!(map.extract_if(|_, _| false).eq_sorted(crate::iter::empty())); assert_eq!(map.len(), 3); } @@ -984,7 +984,7 @@ mod test_drain_filter { fn consuming_all() { let pairs = (0..3).map(|i| (i, i)); let mut map: HashMap<_, _> = pairs.clone().collect(); - assert!(map.drain_filter(|_, _| true).eq_sorted(pairs)); + assert!(map.extract_if(|_, _| true).eq_sorted(pairs)); assert!(map.is_empty()); } @@ -993,7 +993,7 @@ mod test_drain_filter { let pairs = (0..3).map(|i| (i, i)); let mut map: HashMap<_, _> = pairs.collect(); assert!( - map.drain_filter(|_, v| { + map.extract_if(|_, v| { *v += 6; false }) @@ -1008,7 +1008,7 @@ mod test_drain_filter { let pairs = (0..3).map(|i| (i, i)); let mut map: HashMap<_, _> = pairs.collect(); assert!( - map.drain_filter(|_, v| { + map.extract_if(|_, v| { *v += 6; true }) @@ -1034,14 +1034,15 @@ mod test_drain_filter { let mut map = (0..3).map(|i| (i, D)).collect::<HashMap<_, _>>(); catch_unwind(move || { - drop(map.drain_filter(|_, _| { + map.extract_if(|_, _| { PREDS.fetch_add(1, Ordering::SeqCst); true - })) + }) + .for_each(drop) }) .unwrap_err(); - assert_eq!(PREDS.load(Ordering::SeqCst), 3); + assert_eq!(PREDS.load(Ordering::SeqCst), 2); assert_eq!(DROPS.load(Ordering::SeqCst), 3); } @@ -1060,10 +1061,11 @@ mod test_drain_filter { let mut map = (0..3).map(|i| (i, D)).collect::<HashMap<_, _>>(); catch_unwind(AssertUnwindSafe(|| { - drop(map.drain_filter(|_, _| match PREDS.fetch_add(1, Ordering::SeqCst) { + map.extract_if(|_, _| match PREDS.fetch_add(1, Ordering::SeqCst) { 0 => true, _ => panic!(), - })) + }) + .for_each(drop) })) .unwrap_err(); @@ -1088,7 +1090,7 @@ mod test_drain_filter { let mut map = (0..3).map(|i| (i, D)).collect::<HashMap<_, _>>(); { - let mut it = map.drain_filter(|_, _| match PREDS.fetch_add(1, Ordering::SeqCst) { + let mut it = map.extract_if(|_, _| match PREDS.fetch_add(1, Ordering::SeqCst) { 0 => true, _ => panic!(), }); diff --git a/library/std/src/collections/hash/set.rs b/library/std/src/collections/hash/set.rs index ac906e682d5..ec59634df36 100644 --- a/library/std/src/collections/hash/set.rs +++ b/library/std/src/collections/hash/set.rs @@ -262,25 +262,24 @@ impl<T, S> HashSet<T, S> { /// If the closure returns false, the value will remain in the list and will not be yielded /// by the iterator. /// - /// If the iterator is only partially consumed or not consumed at all, each of the remaining - /// values will still be subjected to the closure and removed and dropped if it returns true. + /// If the returned `ExtractIf` is not exhausted, e.g. because it is dropped without iterating + /// or the iteration short-circuits, then the remaining elements will be retained. + /// Use [`retain`] with a negated predicate if you do not need the returned iterator. /// - /// It is unspecified how many more values will be subjected to the closure - /// if a panic occurs in the closure, or if a panic occurs while dropping a value, or if the - /// `DrainFilter` itself is leaked. + /// [`retain`]: HashSet::retain /// /// # Examples /// /// Splitting a set into even and odd values, reusing the original set: /// /// ``` - /// #![feature(hash_drain_filter)] + /// #![feature(hash_extract_if)] /// use std::collections::HashSet; /// /// let mut set: HashSet<i32> = (0..8).collect(); - /// let drained: HashSet<i32> = set.drain_filter(|v| v % 2 == 0).collect(); + /// let extracted: HashSet<i32> = set.extract_if(|v| v % 2 == 0).collect(); /// - /// let mut evens = drained.into_iter().collect::<Vec<_>>(); + /// let mut evens = extracted.into_iter().collect::<Vec<_>>(); /// let mut odds = set.into_iter().collect::<Vec<_>>(); /// evens.sort(); /// odds.sort(); @@ -290,12 +289,12 @@ impl<T, S> HashSet<T, S> { /// ``` #[inline] #[rustc_lint_query_instability] - #[unstable(feature = "hash_drain_filter", issue = "59618")] - pub fn drain_filter<F>(&mut self, pred: F) -> DrainFilter<'_, T, F> + #[unstable(feature = "hash_extract_if", issue = "59618")] + pub fn extract_if<F>(&mut self, pred: F) -> ExtractIf<'_, T, F> where F: FnMut(&T) -> bool, { - DrainFilter { base: self.base.drain_filter(pred) } + ExtractIf { base: self.base.extract_if(pred) } } /// Retains only the elements specified by the predicate. @@ -868,7 +867,9 @@ where /// Returns whether the value was newly inserted. That is: /// /// - If the set did not previously contain this value, `true` is returned. - /// - If the set already contained this value, `false` is returned. + /// - If the set already contained this value, `false` is returned, + /// and the set is not modified: original value is not replaced, + /// and the value passed as argument is dropped. /// /// # Examples /// @@ -1310,27 +1311,27 @@ pub struct Drain<'a, K: 'a> { /// A draining, filtering iterator over the items of a `HashSet`. /// -/// This `struct` is created by the [`drain_filter`] method on [`HashSet`]. +/// This `struct` is created by the [`extract_if`] method on [`HashSet`]. /// -/// [`drain_filter`]: HashSet::drain_filter +/// [`extract_if`]: HashSet::extract_if /// /// # Examples /// /// ``` -/// #![feature(hash_drain_filter)] +/// #![feature(hash_extract_if)] /// /// use std::collections::HashSet; /// /// let mut a = HashSet::from([1, 2, 3]); /// -/// let mut drain_filtered = a.drain_filter(|v| v % 2 == 0); +/// let mut extract_ifed = a.extract_if(|v| v % 2 == 0); /// ``` -#[unstable(feature = "hash_drain_filter", issue = "59618")] -pub struct DrainFilter<'a, K, F> +#[unstable(feature = "hash_extract_if", issue = "59618")] +pub struct ExtractIf<'a, K, F> where F: FnMut(&K) -> bool, { - base: base::DrainFilter<'a, K, F>, + base: base::ExtractIf<'a, K, F>, } /// A lazy iterator producing elements in the intersection of `HashSet`s. @@ -1576,8 +1577,8 @@ impl<K: fmt::Debug> fmt::Debug for Drain<'_, K> { } } -#[unstable(feature = "hash_drain_filter", issue = "59618")] -impl<K, F> Iterator for DrainFilter<'_, K, F> +#[unstable(feature = "hash_extract_if", issue = "59618")] +impl<K, F> Iterator for ExtractIf<'_, K, F> where F: FnMut(&K) -> bool, { @@ -1593,16 +1594,16 @@ where } } -#[unstable(feature = "hash_drain_filter", issue = "59618")] -impl<K, F> FusedIterator for DrainFilter<'_, K, F> where F: FnMut(&K) -> bool {} +#[unstable(feature = "hash_extract_if", issue = "59618")] +impl<K, F> FusedIterator for ExtractIf<'_, K, F> where F: FnMut(&K) -> bool {} -#[unstable(feature = "hash_drain_filter", issue = "59618")] -impl<'a, K, F> fmt::Debug for DrainFilter<'a, K, F> +#[unstable(feature = "hash_extract_if", issue = "59618")] +impl<'a, K, F> fmt::Debug for ExtractIf<'a, K, F> where F: FnMut(&K) -> bool, { fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - f.debug_struct("DrainFilter").finish_non_exhaustive() + f.debug_struct("ExtractIf").finish_non_exhaustive() } } diff --git a/library/std/src/collections/hash/set/tests.rs b/library/std/src/collections/hash/set/tests.rs index 941a0450cc7..e0cd80b44f8 100644 --- a/library/std/src/collections/hash/set/tests.rs +++ b/library/std/src/collections/hash/set/tests.rs @@ -3,6 +3,7 @@ use super::HashSet; use crate::panic::{catch_unwind, AssertUnwindSafe}; use crate::sync::atomic::{AtomicU32, Ordering}; +use crate::sync::Arc; #[test] fn test_zero_capacities() { @@ -418,18 +419,18 @@ fn test_retain() { } #[test] -fn test_drain_filter() { +fn test_extract_if() { let mut x: HashSet<_> = [1].iter().copied().collect(); let mut y: HashSet<_> = [1].iter().copied().collect(); - x.drain_filter(|_| true); - y.drain_filter(|_| false); + x.extract_if(|_| true).for_each(drop); + y.extract_if(|_| false).for_each(drop); assert_eq!(x.len(), 0); assert_eq!(y.len(), 1); } #[test] -fn test_drain_filter_drop_panic_leak() { +fn test_extract_if_drop_panic_leak() { static PREDS: AtomicU32 = AtomicU32::new(0); static DROPS: AtomicU32 = AtomicU32::new(0); @@ -446,19 +447,20 @@ fn test_drain_filter_drop_panic_leak() { let mut set = (0..3).map(|i| D(i)).collect::<HashSet<_>>(); catch_unwind(move || { - drop(set.drain_filter(|_| { + set.extract_if(|_| { PREDS.fetch_add(1, Ordering::SeqCst); true - })) + }) + .for_each(drop) }) .ok(); - assert_eq!(PREDS.load(Ordering::SeqCst), 3); + assert_eq!(PREDS.load(Ordering::SeqCst), 2); assert_eq!(DROPS.load(Ordering::SeqCst), 3); } #[test] -fn test_drain_filter_pred_panic_leak() { +fn test_extract_if_pred_panic_leak() { static PREDS: AtomicU32 = AtomicU32::new(0); static DROPS: AtomicU32 = AtomicU32::new(0); @@ -473,10 +475,11 @@ fn test_drain_filter_pred_panic_leak() { let mut set: HashSet<_> = (0..3).map(|_| D).collect(); catch_unwind(AssertUnwindSafe(|| { - drop(set.drain_filter(|_| match PREDS.fetch_add(1, Ordering::SeqCst) { + set.extract_if(|_| match PREDS.fetch_add(1, Ordering::SeqCst) { 0 => true, _ => panic!(), - })) + }) + .for_each(drop) })) .ok(); @@ -502,3 +505,22 @@ fn const_with_hasher() { const X: HashSet<(), ()> = HashSet::with_hasher(()); assert_eq!(X.len(), 0); } + +#[test] +fn test_insert_does_not_overwrite_the_value() { + let first_value = Arc::new(17); + let second_value = Arc::new(17); + + let mut set = HashSet::new(); + let inserted = set.insert(first_value.clone()); + assert!(inserted); + + let inserted = set.insert(second_value); + assert!(!inserted); + + assert!( + Arc::ptr_eq(set.iter().next().unwrap(), &first_value), + "Insert must not overwrite the value, so the contained value pointer \ + must be the same as first value pointer we inserted" + ); +} diff --git a/library/std/src/f32.rs b/library/std/src/f32.rs index 408244b2ce9..bed90418be1 100644 --- a/library/std/src/f32.rs +++ b/library/std/src/f32.rs @@ -528,7 +528,7 @@ impl f32 { /// The positive difference of two numbers. /// - /// * If `self <= other`: `0:0` + /// * If `self <= other`: `0.0` /// * Else: `self - other` /// /// # Examples diff --git a/library/std/src/f64.rs b/library/std/src/f64.rs index 6782b861f11..e72de05ca41 100644 --- a/library/std/src/f64.rs +++ b/library/std/src/f64.rs @@ -530,7 +530,7 @@ impl f64 { /// The positive difference of two numbers. /// - /// * If `self <= other`: `0:0` + /// * If `self <= other`: `0.0` /// * Else: `self - other` /// /// # Examples diff --git a/library/std/src/ffi/os_str.rs b/library/std/src/ffi/os_str.rs index bb577acf769..fbdf7f5ecac 100644 --- a/library/std/src/ffi/os_str.rs +++ b/library/std/src/ffi/os_str.rs @@ -745,7 +745,7 @@ impl OsStr { without modifying the original"] #[inline] pub fn to_str(&self) -> Option<&str> { - self.inner.to_str() + self.inner.to_str().ok() } /// Converts an `OsStr` to a <code>[Cow]<[str]></code>. @@ -1165,6 +1165,24 @@ impl<'a> From<Cow<'a, OsStr>> for OsString { } } +#[stable(feature = "str_tryfrom_osstr_impl", since = "CURRENT_RUSTC_VERSION")] +impl<'a> TryFrom<&'a OsStr> for &'a str { + type Error = crate::str::Utf8Error; + + /// Tries to convert an `&OsStr` to a `&str`. + /// + /// ``` + /// use std::ffi::OsStr; + /// + /// let os_str = OsStr::new("foo"); + /// let as_str = <&str>::try_from(os_str).unwrap(); + /// assert_eq!(as_str, "foo"); + /// ``` + fn try_from(value: &'a OsStr) -> Result<Self, Self::Error> { + value.inner.to_str() + } +} + #[stable(feature = "box_default_extra", since = "1.17.0")] impl Default for Box<OsStr> { #[inline] diff --git a/library/std/src/io/buffered/bufreader.rs b/library/std/src/io/buffered/bufreader.rs index 4f339a18a48..a66e6ccf673 100644 --- a/library/std/src/io/buffered/bufreader.rs +++ b/library/std/src/io/buffered/bufreader.rs @@ -47,9 +47,9 @@ use buffer::Buffer; /// } /// ``` #[stable(feature = "rust1", since = "1.0.0")] -pub struct BufReader<R> { - inner: R, +pub struct BufReader<R: ?Sized> { buf: Buffer, + inner: R, } impl<R: Read> BufReader<R> { @@ -95,7 +95,7 @@ impl<R: Read> BufReader<R> { } } -impl<R> BufReader<R> { +impl<R: ?Sized> BufReader<R> { /// Gets a reference to the underlying reader. /// /// It is inadvisable to directly read from the underlying reader. @@ -213,26 +213,29 @@ impl<R> BufReader<R> { /// } /// ``` #[stable(feature = "rust1", since = "1.0.0")] - pub fn into_inner(self) -> R { + pub fn into_inner(self) -> R + where + R: Sized, + { self.inner } /// Invalidates all data in the internal buffer. #[inline] - fn discard_buffer(&mut self) { + pub(in crate::io) fn discard_buffer(&mut self) { self.buf.discard_buffer() } } // This is only used by a test which asserts that the initialization-tracking is correct. #[cfg(test)] -impl<R> BufReader<R> { +impl<R: ?Sized> BufReader<R> { pub fn initialized(&self) -> usize { self.buf.initialized() } } -impl<R: Seek> BufReader<R> { +impl<R: ?Sized + Seek> BufReader<R> { /// Seeks relative to the current position. If the new position lies within the buffer, /// the buffer will not be flushed, allowing for more efficient seeks. /// This method does not return the location of the underlying reader, so the caller @@ -257,7 +260,7 @@ impl<R: Seek> BufReader<R> { } #[stable(feature = "rust1", since = "1.0.0")] -impl<R: Read> Read for BufReader<R> { +impl<R: ?Sized + Read> Read for BufReader<R> { fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> { // If we don't have any buffered data and we're doing a massive read // (larger than our internal buffer), bypass our internal buffer @@ -371,7 +374,7 @@ impl<R: Read> Read for BufReader<R> { } #[stable(feature = "rust1", since = "1.0.0")] -impl<R: Read> BufRead for BufReader<R> { +impl<R: ?Sized + Read> BufRead for BufReader<R> { fn fill_buf(&mut self) -> io::Result<&[u8]> { self.buf.fill_buf(&mut self.inner) } @@ -384,11 +387,11 @@ impl<R: Read> BufRead for BufReader<R> { #[stable(feature = "rust1", since = "1.0.0")] impl<R> fmt::Debug for BufReader<R> where - R: fmt::Debug, + R: ?Sized + fmt::Debug, { fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { fmt.debug_struct("BufReader") - .field("reader", &self.inner) + .field("reader", &&self.inner) .field( "buffer", &format_args!("{}/{}", self.buf.filled() - self.buf.pos(), self.capacity()), @@ -398,7 +401,7 @@ where } #[stable(feature = "rust1", since = "1.0.0")] -impl<R: Seek> Seek for BufReader<R> { +impl<R: ?Sized + Seek> Seek for BufReader<R> { /// Seek to an offset, in bytes, in the underlying reader. /// /// The position used for seeking with <code>[SeekFrom::Current]\(_)</code> is the @@ -491,7 +494,7 @@ impl<R: Seek> Seek for BufReader<R> { } } -impl<T> SizeHint for BufReader<T> { +impl<T: ?Sized> SizeHint for BufReader<T> { #[inline] fn lower_bound(&self) -> usize { SizeHint::lower_bound(self.get_ref()) + self.buffer().len() diff --git a/library/std/src/io/buffered/bufwriter.rs b/library/std/src/io/buffered/bufwriter.rs index 14c455d4fa3..0e2450655e5 100644 --- a/library/std/src/io/buffered/bufwriter.rs +++ b/library/std/src/io/buffered/bufwriter.rs @@ -67,8 +67,7 @@ use crate::ptr; /// [`TcpStream`]: crate::net::TcpStream /// [`flush`]: BufWriter::flush #[stable(feature = "rust1", since = "1.0.0")] -pub struct BufWriter<W: Write> { - inner: W, +pub struct BufWriter<W: ?Sized + Write> { // The buffer. Avoid using this like a normal `Vec` in common code paths. // That is, don't use `buf.push`, `buf.extend_from_slice`, or any other // methods that require bounds checking or the like. This makes an enormous @@ -78,6 +77,7 @@ pub struct BufWriter<W: Write> { // write the buffered data a second time in BufWriter's destructor. This // flag tells the Drop impl if it should skip the flush. panicked: bool, + inner: W, } impl<W: Write> BufWriter<W> { @@ -115,6 +115,69 @@ impl<W: Write> BufWriter<W> { BufWriter { inner, buf: Vec::with_capacity(capacity), panicked: false } } + /// Unwraps this `BufWriter<W>`, returning the underlying writer. + /// + /// The buffer is written out before returning the writer. + /// + /// # Errors + /// + /// An [`Err`] will be returned if an error occurs while flushing the buffer. + /// + /// # Examples + /// + /// ```no_run + /// use std::io::BufWriter; + /// use std::net::TcpStream; + /// + /// let mut buffer = BufWriter::new(TcpStream::connect("127.0.0.1:34254").unwrap()); + /// + /// // unwrap the TcpStream and flush the buffer + /// let stream = buffer.into_inner().unwrap(); + /// ``` + #[stable(feature = "rust1", since = "1.0.0")] + pub fn into_inner(mut self) -> Result<W, IntoInnerError<BufWriter<W>>> { + match self.flush_buf() { + Err(e) => Err(IntoInnerError::new(self, e)), + Ok(()) => Ok(self.into_parts().0), + } + } + + /// Disassembles this `BufWriter<W>`, returning the underlying writer, and any buffered but + /// unwritten data. + /// + /// If the underlying writer panicked, it is not known what portion of the data was written. + /// In this case, we return `WriterPanicked` for the buffered data (from which the buffer + /// contents can still be recovered). + /// + /// `into_parts` makes no attempt to flush data and cannot fail. + /// + /// # Examples + /// + /// ``` + /// use std::io::{BufWriter, Write}; + /// + /// let mut buffer = [0u8; 10]; + /// let mut stream = BufWriter::new(buffer.as_mut()); + /// write!(stream, "too much data").unwrap(); + /// stream.flush().expect_err("it doesn't fit"); + /// let (recovered_writer, buffered_data) = stream.into_parts(); + /// assert_eq!(recovered_writer.len(), 0); + /// assert_eq!(&buffered_data.unwrap(), b"ata"); + /// ``` + #[stable(feature = "bufwriter_into_parts", since = "1.56.0")] + pub fn into_parts(mut self) -> (W, Result<Vec<u8>, WriterPanicked>) { + let buf = mem::take(&mut self.buf); + let buf = if !self.panicked { Ok(buf) } else { Err(WriterPanicked { buf }) }; + + // SAFETY: forget(self) prevents double dropping inner + let inner = unsafe { ptr::read(&self.inner) }; + mem::forget(self); + + (inner, buf) + } +} + +impl<W: ?Sized + Write> BufWriter<W> { /// Send data in our local buffer into the inner writer, looping as /// necessary until either it's all been sent or an error occurs. /// @@ -284,67 +347,6 @@ impl<W: Write> BufWriter<W> { self.buf.capacity() } - /// Unwraps this `BufWriter<W>`, returning the underlying writer. - /// - /// The buffer is written out before returning the writer. - /// - /// # Errors - /// - /// An [`Err`] will be returned if an error occurs while flushing the buffer. - /// - /// # Examples - /// - /// ```no_run - /// use std::io::BufWriter; - /// use std::net::TcpStream; - /// - /// let mut buffer = BufWriter::new(TcpStream::connect("127.0.0.1:34254").unwrap()); - /// - /// // unwrap the TcpStream and flush the buffer - /// let stream = buffer.into_inner().unwrap(); - /// ``` - #[stable(feature = "rust1", since = "1.0.0")] - pub fn into_inner(mut self) -> Result<W, IntoInnerError<BufWriter<W>>> { - match self.flush_buf() { - Err(e) => Err(IntoInnerError::new(self, e)), - Ok(()) => Ok(self.into_parts().0), - } - } - - /// Disassembles this `BufWriter<W>`, returning the underlying writer, and any buffered but - /// unwritten data. - /// - /// If the underlying writer panicked, it is not known what portion of the data was written. - /// In this case, we return `WriterPanicked` for the buffered data (from which the buffer - /// contents can still be recovered). - /// - /// `into_parts` makes no attempt to flush data and cannot fail. - /// - /// # Examples - /// - /// ``` - /// use std::io::{BufWriter, Write}; - /// - /// let mut buffer = [0u8; 10]; - /// let mut stream = BufWriter::new(buffer.as_mut()); - /// write!(stream, "too much data").unwrap(); - /// stream.flush().expect_err("it doesn't fit"); - /// let (recovered_writer, buffered_data) = stream.into_parts(); - /// assert_eq!(recovered_writer.len(), 0); - /// assert_eq!(&buffered_data.unwrap(), b"ata"); - /// ``` - #[stable(feature = "bufwriter_into_parts", since = "1.56.0")] - pub fn into_parts(mut self) -> (W, Result<Vec<u8>, WriterPanicked>) { - let buf = mem::take(&mut self.buf); - let buf = if !self.panicked { Ok(buf) } else { Err(WriterPanicked { buf }) }; - - // SAFETY: forget(self) prevents double dropping inner - let inner = unsafe { ptr::read(&self.inner) }; - mem::forget(self); - - (inner, buf) - } - // Ensure this function does not get inlined into `write`, so that it // remains inlineable and its common path remains as short as possible. // If this function ends up being called frequently relative to `write`, @@ -511,7 +513,7 @@ impl fmt::Debug for WriterPanicked { } #[stable(feature = "rust1", since = "1.0.0")] -impl<W: Write> Write for BufWriter<W> { +impl<W: ?Sized + Write> Write for BufWriter<W> { #[inline] fn write(&mut self, buf: &[u8]) -> io::Result<usize> { // Use < instead of <= to avoid a needless trip through the buffer in some cases. @@ -640,20 +642,20 @@ impl<W: Write> Write for BufWriter<W> { } #[stable(feature = "rust1", since = "1.0.0")] -impl<W: Write> fmt::Debug for BufWriter<W> +impl<W: ?Sized + Write> fmt::Debug for BufWriter<W> where W: fmt::Debug, { fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { fmt.debug_struct("BufWriter") - .field("writer", &self.inner) + .field("writer", &&self.inner) .field("buffer", &format_args!("{}/{}", self.buf.len(), self.buf.capacity())) .finish() } } #[stable(feature = "rust1", since = "1.0.0")] -impl<W: Write + Seek> Seek for BufWriter<W> { +impl<W: ?Sized + Write + Seek> Seek for BufWriter<W> { /// Seek to the offset, in bytes, in the underlying writer. /// /// Seeking always writes out the internal buffer before seeking. @@ -664,7 +666,7 @@ impl<W: Write + Seek> Seek for BufWriter<W> { } #[stable(feature = "rust1", since = "1.0.0")] -impl<W: Write> Drop for BufWriter<W> { +impl<W: ?Sized + Write> Drop for BufWriter<W> { fn drop(&mut self) { if !self.panicked { // dtors should not panic, so we ignore a failed flush diff --git a/library/std/src/io/buffered/linewriter.rs b/library/std/src/io/buffered/linewriter.rs index a26a4ab330e..3d4ae704193 100644 --- a/library/std/src/io/buffered/linewriter.rs +++ b/library/std/src/io/buffered/linewriter.rs @@ -64,7 +64,7 @@ use crate::io::{self, buffered::LineWriterShim, BufWriter, IntoInnerError, IoSli /// } /// ``` #[stable(feature = "rust1", since = "1.0.0")] -pub struct LineWriter<W: Write> { +pub struct LineWriter<W: ?Sized + Write> { inner: BufWriter<W>, } @@ -109,27 +109,6 @@ impl<W: Write> LineWriter<W> { LineWriter { inner: BufWriter::with_capacity(capacity, inner) } } - /// Gets a reference to the underlying writer. - /// - /// # Examples - /// - /// ```no_run - /// use std::fs::File; - /// use std::io::LineWriter; - /// - /// fn main() -> std::io::Result<()> { - /// let file = File::create("poem.txt")?; - /// let file = LineWriter::new(file); - /// - /// let reference = file.get_ref(); - /// Ok(()) - /// } - /// ``` - #[stable(feature = "rust1", since = "1.0.0")] - pub fn get_ref(&self) -> &W { - self.inner.get_ref() - } - /// Gets a mutable reference to the underlying writer. /// /// Caution must be taken when calling methods on the mutable reference @@ -184,8 +163,31 @@ impl<W: Write> LineWriter<W> { } } +impl<W: ?Sized + Write> LineWriter<W> { + /// Gets a reference to the underlying writer. + /// + /// # Examples + /// + /// ```no_run + /// use std::fs::File; + /// use std::io::LineWriter; + /// + /// fn main() -> std::io::Result<()> { + /// let file = File::create("poem.txt")?; + /// let file = LineWriter::new(file); + /// + /// let reference = file.get_ref(); + /// Ok(()) + /// } + /// ``` + #[stable(feature = "rust1", since = "1.0.0")] + pub fn get_ref(&self) -> &W { + self.inner.get_ref() + } +} + #[stable(feature = "rust1", since = "1.0.0")] -impl<W: Write> Write for LineWriter<W> { +impl<W: ?Sized + Write> Write for LineWriter<W> { fn write(&mut self, buf: &[u8]) -> io::Result<usize> { LineWriterShim::new(&mut self.inner).write(buf) } @@ -216,7 +218,7 @@ impl<W: Write> Write for LineWriter<W> { } #[stable(feature = "rust1", since = "1.0.0")] -impl<W: Write> fmt::Debug for LineWriter<W> +impl<W: ?Sized + Write> fmt::Debug for LineWriter<W> where W: fmt::Debug, { diff --git a/library/std/src/io/buffered/linewritershim.rs b/library/std/src/io/buffered/linewritershim.rs index 0175d2693e8..f2a55da05b2 100644 --- a/library/std/src/io/buffered/linewritershim.rs +++ b/library/std/src/io/buffered/linewritershim.rs @@ -11,11 +11,11 @@ use crate::sys_common::memchr; /// `BufWriters` to be temporarily given line-buffering logic; this is what /// enables Stdout to be alternately in line-buffered or block-buffered mode. #[derive(Debug)] -pub struct LineWriterShim<'a, W: Write> { +pub struct LineWriterShim<'a, W: ?Sized + Write> { buffer: &'a mut BufWriter<W>, } -impl<'a, W: Write> LineWriterShim<'a, W> { +impl<'a, W: ?Sized + Write> LineWriterShim<'a, W> { pub fn new(buffer: &'a mut BufWriter<W>) -> Self { Self { buffer } } @@ -49,7 +49,7 @@ impl<'a, W: Write> LineWriterShim<'a, W> { } } -impl<'a, W: Write> Write for LineWriterShim<'a, W> { +impl<'a, W: ?Sized + Write> Write for LineWriterShim<'a, W> { /// Write some data into this BufReader with line buffering. This means /// that, if any newlines are present in the data, the data up to the last /// newline is sent directly to the underlying writer, and data after it diff --git a/library/std/src/io/copy.rs b/library/std/src/io/copy.rs index 1d9d93f5b64..ef1f4031ef2 100644 --- a/library/std/src/io/copy.rs +++ b/library/std/src/io/copy.rs @@ -1,6 +1,9 @@ -use super::{BorrowedBuf, BufWriter, ErrorKind, Read, Result, Write, DEFAULT_BUF_SIZE}; +use super::{BorrowedBuf, BufReader, BufWriter, ErrorKind, Read, Result, Write, DEFAULT_BUF_SIZE}; use crate::mem::MaybeUninit; +#[cfg(test)] +mod tests; + /// Copies the entire contents of a reader into a writer. /// /// This function will continuously read data from `reader` and then @@ -71,32 +74,113 @@ where R: Read, W: Write, { - BufferedCopySpec::copy_to(reader, writer) + let read_buf = BufferedReaderSpec::buffer_size(reader); + let write_buf = BufferedWriterSpec::buffer_size(writer); + + if read_buf >= DEFAULT_BUF_SIZE && read_buf >= write_buf { + return BufferedReaderSpec::copy_to(reader, writer); + } + + BufferedWriterSpec::copy_from(writer, reader) +} + +/// Specialization of the read-write loop that reuses the internal +/// buffer of a BufReader. If there's no buffer then the writer side +/// should be used intead. +trait BufferedReaderSpec { + fn buffer_size(&self) -> usize; + + fn copy_to(&mut self, to: &mut (impl Write + ?Sized)) -> Result<u64>; +} + +impl<T> BufferedReaderSpec for T +where + Self: Read, + T: ?Sized, +{ + #[inline] + default fn buffer_size(&self) -> usize { + 0 + } + + default fn copy_to(&mut self, _to: &mut (impl Write + ?Sized)) -> Result<u64> { + unimplemented!("only called from specializations"); + } +} + +impl<I> BufferedReaderSpec for BufReader<I> +where + Self: Read, + I: ?Sized, +{ + fn buffer_size(&self) -> usize { + self.capacity() + } + + fn copy_to(&mut self, to: &mut (impl Write + ?Sized)) -> Result<u64> { + let mut len = 0; + + loop { + // Hack: this relies on `impl Read for BufReader` always calling fill_buf + // if the buffer is empty, even for empty slices. + // It can't be called directly here since specialization prevents us + // from adding I: Read + match self.read(&mut []) { + Ok(_) => {} + Err(e) if e.kind() == ErrorKind::Interrupted => continue, + Err(e) => return Err(e), + } + let buf = self.buffer(); + if self.buffer().len() == 0 { + return Ok(len); + } + + // In case the writer side is a BufWriter then its write_all + // implements an optimization that passes through large + // buffers to the underlying writer. That code path is #[cold] + // but we're still avoiding redundant memcopies when doing + // a copy between buffered inputs and outputs. + to.write_all(buf)?; + len += buf.len() as u64; + self.discard_buffer(); + } + } } /// Specialization of the read-write loop that either uses a stack buffer /// or reuses the internal buffer of a BufWriter -trait BufferedCopySpec: Write { - fn copy_to<R: Read + ?Sized>(reader: &mut R, writer: &mut Self) -> Result<u64>; +trait BufferedWriterSpec: Write { + fn buffer_size(&self) -> usize; + + fn copy_from<R: Read + ?Sized>(&mut self, reader: &mut R) -> Result<u64>; } -impl<W: Write + ?Sized> BufferedCopySpec for W { - default fn copy_to<R: Read + ?Sized>(reader: &mut R, writer: &mut Self) -> Result<u64> { - stack_buffer_copy(reader, writer) +impl<W: Write + ?Sized> BufferedWriterSpec for W { + #[inline] + default fn buffer_size(&self) -> usize { + 0 + } + + default fn copy_from<R: Read + ?Sized>(&mut self, reader: &mut R) -> Result<u64> { + stack_buffer_copy(reader, self) } } -impl<I: Write> BufferedCopySpec for BufWriter<I> { - fn copy_to<R: Read + ?Sized>(reader: &mut R, writer: &mut Self) -> Result<u64> { - if writer.capacity() < DEFAULT_BUF_SIZE { - return stack_buffer_copy(reader, writer); +impl<I: Write + ?Sized> BufferedWriterSpec for BufWriter<I> { + fn buffer_size(&self) -> usize { + self.capacity() + } + + fn copy_from<R: Read + ?Sized>(&mut self, reader: &mut R) -> Result<u64> { + if self.capacity() < DEFAULT_BUF_SIZE { + return stack_buffer_copy(reader, self); } let mut len = 0; let mut init = 0; loop { - let buf = writer.buffer_mut(); + let buf = self.buffer_mut(); let mut read_buf: BorrowedBuf<'_> = buf.spare_capacity_mut().into(); unsafe { @@ -127,7 +211,7 @@ impl<I: Write> BufferedCopySpec for BufWriter<I> { Err(e) => return Err(e), } } else { - writer.flush_buf()?; + self.flush_buf()?; init = 0; } } diff --git a/library/std/src/io/copy/tests.rs b/library/std/src/io/copy/tests.rs new file mode 100644 index 00000000000..8c816af15b8 --- /dev/null +++ b/library/std/src/io/copy/tests.rs @@ -0,0 +1,108 @@ +use crate::cmp::{max, min}; +use crate::io::*; + +#[test] +fn copy_copies() { + let mut r = repeat(0).take(4); + let mut w = sink(); + assert_eq!(copy(&mut r, &mut w).unwrap(), 4); + + let mut r = repeat(0).take(1 << 17); + assert_eq!(copy(&mut r as &mut dyn Read, &mut w as &mut dyn Write).unwrap(), 1 << 17); +} + +struct ShortReader { + cap: usize, + read_size: usize, + observed_buffer: usize, +} + +impl Read for ShortReader { + fn read(&mut self, buf: &mut [u8]) -> Result<usize> { + let bytes = min(self.cap, self.read_size); + self.cap -= bytes; + self.observed_buffer = max(self.observed_buffer, buf.len()); + Ok(bytes) + } +} + +struct WriteObserver { + observed_buffer: usize, +} + +impl Write for WriteObserver { + fn write(&mut self, buf: &[u8]) -> Result<usize> { + self.observed_buffer = max(self.observed_buffer, buf.len()); + Ok(buf.len()) + } + + fn flush(&mut self) -> Result<()> { + Ok(()) + } +} + +#[test] +fn copy_specializes_bufwriter() { + let cap = 117 * 1024; + let buf_sz = 16 * 1024; + let mut r = ShortReader { cap, observed_buffer: 0, read_size: 1337 }; + let mut w = BufWriter::with_capacity(buf_sz, WriteObserver { observed_buffer: 0 }); + assert_eq!( + copy(&mut r, &mut w).unwrap(), + cap as u64, + "expected the whole capacity to be copied" + ); + assert_eq!(r.observed_buffer, buf_sz, "expected a large buffer to be provided to the reader"); + assert!(w.get_mut().observed_buffer > DEFAULT_BUF_SIZE, "expected coalesced writes"); +} + +#[test] +fn copy_specializes_bufreader() { + let mut source = vec![0; 768 * 1024]; + source[1] = 42; + let mut buffered = BufReader::with_capacity(256 * 1024, Cursor::new(&mut source)); + + let mut sink = Vec::new(); + assert_eq!(crate::io::copy(&mut buffered, &mut sink).unwrap(), source.len() as u64); + assert_eq!(source.as_slice(), sink.as_slice()); + + let buf_sz = 71 * 1024; + assert!(buf_sz > DEFAULT_BUF_SIZE, "test precondition"); + + let mut buffered = BufReader::with_capacity(buf_sz, Cursor::new(&mut source)); + let mut sink = WriteObserver { observed_buffer: 0 }; + assert_eq!(crate::io::copy(&mut buffered, &mut sink).unwrap(), source.len() as u64); + assert_eq!( + sink.observed_buffer, buf_sz, + "expected a large buffer to be provided to the writer" + ); +} + +#[cfg(unix)] +mod io_benches { + use crate::fs::File; + use crate::fs::OpenOptions; + use crate::io::prelude::*; + use crate::io::BufReader; + + use test::Bencher; + + #[bench] + fn bench_copy_buf_reader(b: &mut Bencher) { + let mut file_in = File::open("/dev/zero").expect("opening /dev/zero failed"); + // use dyn to avoid specializations unrelated to readbuf + let dyn_in = &mut file_in as &mut dyn Read; + let mut reader = BufReader::with_capacity(256 * 1024, dyn_in.take(0)); + let mut writer = + OpenOptions::new().write(true).open("/dev/null").expect("opening /dev/null failed"); + + const BYTES: u64 = 1024 * 1024; + + b.bytes = BYTES; + + b.iter(|| { + reader.get_mut().set_limit(BYTES); + crate::io::copy(&mut reader, &mut writer).unwrap() + }); + } +} diff --git a/library/std/src/io/mod.rs b/library/std/src/io/mod.rs index 8a007d095d5..5c1d2d8f46c 100644 --- a/library/std/src/io/mod.rs +++ b/library/std/src/io/mod.rs @@ -1416,17 +1416,18 @@ pub trait Write { /// /// This function will attempt to write the entire contents of `buf`, but /// the entire write might not succeed, or the write may also generate an - /// error. A call to `write` represents *at most one* attempt to write to + /// error. Typically, a call to `write` represents one attempt to write to /// any wrapped object. /// /// Calls to `write` are not guaranteed to block waiting for data to be /// written, and a write which would otherwise block can be indicated through /// an [`Err`] variant. /// - /// If the return value is [`Ok(n)`] then it must be guaranteed that - /// `n <= buf.len()`. A return value of `0` typically means that the - /// underlying object is no longer able to accept bytes and will likely not - /// be able to in the future as well, or that the buffer provided is empty. + /// If this method consumed `n > 0` bytes of `buf` it must return [`Ok(n)`]. + /// If the return value is `Ok(n)` then `n` must satisfy `n <= buf.len()`. + /// Unless `buf` is empty, this function shouldn’t return `Ok(0)` since the + /// caller may interpret that as an error. To indicate lack of space, + /// implementors should return [`ErrorKind::StorageFull`] error instead. /// /// # Errors /// @@ -2754,7 +2755,7 @@ trait SizeHint { } } -impl<T> SizeHint for T { +impl<T: ?Sized> SizeHint for T { #[inline] default fn lower_bound(&self) -> usize { 0 diff --git a/library/std/src/io/util/tests.rs b/library/std/src/io/util/tests.rs index ce5e2c9da1d..1baa94e64c9 100644 --- a/library/std/src/io/util/tests.rs +++ b/library/std/src/io/util/tests.rs @@ -1,68 +1,9 @@ -use crate::cmp::{max, min}; use crate::io::prelude::*; -use crate::io::{ - copy, empty, repeat, sink, BorrowedBuf, BufWriter, Empty, Repeat, Result, SeekFrom, Sink, - DEFAULT_BUF_SIZE, -}; +use crate::io::{empty, repeat, sink, BorrowedBuf, Empty, Repeat, SeekFrom, Sink}; use crate::mem::MaybeUninit; #[test] -fn copy_copies() { - let mut r = repeat(0).take(4); - let mut w = sink(); - assert_eq!(copy(&mut r, &mut w).unwrap(), 4); - - let mut r = repeat(0).take(1 << 17); - assert_eq!(copy(&mut r as &mut dyn Read, &mut w as &mut dyn Write).unwrap(), 1 << 17); -} - -struct ShortReader { - cap: usize, - read_size: usize, - observed_buffer: usize, -} - -impl Read for ShortReader { - fn read(&mut self, buf: &mut [u8]) -> Result<usize> { - let bytes = min(self.cap, self.read_size); - self.cap -= bytes; - self.observed_buffer = max(self.observed_buffer, buf.len()); - Ok(bytes) - } -} - -struct WriteObserver { - observed_buffer: usize, -} - -impl Write for WriteObserver { - fn write(&mut self, buf: &[u8]) -> Result<usize> { - self.observed_buffer = max(self.observed_buffer, buf.len()); - Ok(buf.len()) - } - - fn flush(&mut self) -> Result<()> { - Ok(()) - } -} - -#[test] -fn copy_specializes_bufwriter() { - let cap = 117 * 1024; - let buf_sz = 16 * 1024; - let mut r = ShortReader { cap, observed_buffer: 0, read_size: 1337 }; - let mut w = BufWriter::with_capacity(buf_sz, WriteObserver { observed_buffer: 0 }); - assert_eq!( - copy(&mut r, &mut w).unwrap(), - cap as u64, - "expected the whole capacity to be copied" - ); - assert_eq!(r.observed_buffer, buf_sz, "expected a large buffer to be provided to the reader"); - assert!(w.get_mut().observed_buffer > DEFAULT_BUF_SIZE, "expected coalesced writes"); -} - -#[test] fn sink_sinks() { let mut s = sink(); assert_eq!(s.write(&[]).unwrap(), 0); diff --git a/library/std/src/lib.rs b/library/std/src/lib.rs index d53f1a2b2ff..da08c018d0e 100644 --- a/library/std/src/lib.rs +++ b/library/std/src/lib.rs @@ -188,6 +188,13 @@ //! [array]: prim@array //! [slice]: prim@slice +// To run std tests without x.py without ending up with two copies of std, Miri needs to be +// able to "empty" this crate. See <https://github.com/rust-lang/miri-test-libstd/issues/4>. +// rustc itself never sets the feature, so this line has no affect there. +#![cfg(any(not(feature = "miri-test-libstd"), test, doctest))] +// miri-test-libstd also prefers to make std use the sysroot versions of the dependencies. +#![cfg_attr(feature = "miri-test-libstd", feature(rustc_private))] +// #![cfg_attr(not(feature = "restricted-std"), stable(feature = "rust1", since = "1.0.0"))] #![cfg_attr(feature = "restricted-std", unstable(feature = "restricted_std", issue = "none"))] #![doc( @@ -202,12 +209,6 @@ no_global_oom_handling, not(no_global_oom_handling) ))] -// To run std tests without x.py without ending up with two copies of std, Miri needs to be -// able to "empty" this crate. See <https://github.com/rust-lang/miri-test-libstd/issues/4>. -// rustc itself never sets the feature, so this line has no affect there. -#![cfg(any(not(feature = "miri-test-libstd"), test, doctest))] -// miri-test-libstd also prefers to make std use the sysroot versions of the dependencies. -#![cfg_attr(feature = "miri-test-libstd", feature(rustc_private))] // Don't link to std. We are std. #![no_std] // Tell the compiler to link to either panic_abort or panic_unwind diff --git a/library/std/src/os/unix/net/ancillary.rs b/library/std/src/os/unix/net/ancillary.rs index 7565fbc0d09..218536689fd 100644 --- a/library/std/src/os/unix/net/ancillary.rs +++ b/library/std/src/os/unix/net/ancillary.rs @@ -11,12 +11,19 @@ use crate::slice::from_raw_parts; use crate::sys::net::Socket; // FIXME(#43348): Make libc adapt #[doc(cfg(...))] so we don't need these fake definitions here? -#[cfg(all(doc, not(target_os = "linux"), not(target_os = "android"), not(target_os = "netbsd")))] +#[cfg(all( + doc, + not(target_os = "linux"), + not(target_os = "android"), + not(target_os = "netbsd"), + not(target_os = "freebsd") +))] #[allow(non_camel_case_types)] mod libc { pub use libc::c_int; pub struct ucred; pub struct cmsghdr; + pub struct sockcred2; pub type pid_t = i32; pub type gid_t = u32; pub type uid_t = u32; diff --git a/library/std/src/personality.rs b/library/std/src/personality.rs index 63f0ad4f16e..386a399f532 100644 --- a/library/std/src/personality.rs +++ b/library/std/src/personality.rs @@ -29,7 +29,7 @@ cfg_if::cfg_if! { all(target_family = "windows", target_env = "gnu"), target_os = "psp", target_os = "solid_asp3", - all(target_family = "unix", not(target_os = "espidf")), + all(target_family = "unix", not(target_os = "espidf"), not(target_os = "l4re")), all(target_vendor = "fortanix", target_env = "sgx"), ))] { mod gcc; diff --git a/library/std/src/primitive_docs.rs b/library/std/src/primitive_docs.rs index 8266e899011..80289ca08c3 100644 --- a/library/std/src/primitive_docs.rs +++ b/library/std/src/primitive_docs.rs @@ -308,7 +308,7 @@ mod prim_never {} /// /// ```no_run /// // Undefined behaviour -/// unsafe { char::from_u32_unchecked(0x110000) }; +/// let _ = unsafe { char::from_u32_unchecked(0x110000) }; /// ``` /// /// USVs are also the exact set of values that may be encoded in UTF-8. Because diff --git a/library/std/src/sys/unix/kernel_copy.rs b/library/std/src/sys/unix/kernel_copy.rs index 16c8e0c0ebf..7d49bbdcbe0 100644 --- a/library/std/src/sys/unix/kernel_copy.rs +++ b/library/std/src/sys/unix/kernel_copy.rs @@ -466,7 +466,7 @@ impl<T: CopyRead> CopyRead for Take<T> { } } -impl<T: CopyRead> CopyRead for BufReader<T> { +impl<T: ?Sized + CopyRead> CopyRead for BufReader<T> { fn drain_to<W: Write>(&mut self, writer: &mut W, outer_limit: u64) -> Result<u64> { let buf = self.buffer(); let buf = &buf[0..min(buf.len(), outer_limit.try_into().unwrap_or(usize::MAX))]; @@ -495,7 +495,7 @@ impl<T: CopyRead> CopyRead for BufReader<T> { } } -impl<T: CopyWrite> CopyWrite for BufWriter<T> { +impl<T: ?Sized + CopyWrite> CopyWrite for BufWriter<T> { fn properties(&self) -> CopyParams { self.get_ref().properties() } diff --git a/library/std/src/sys/unix/l4re.rs b/library/std/src/sys/unix/l4re.rs index ee016887e70..fe9559f2a56 100644 --- a/library/std/src/sys/unix/l4re.rs +++ b/library/std/src/sys/unix/l4re.rs @@ -10,7 +10,7 @@ macro_rules! unimpl { pub mod net { #![allow(warnings)] use crate::fmt; - use crate::io::{self, IoSlice, IoSliceMut}; + use crate::io::{self, BorrowedCursor, IoSlice, IoSliceMut}; use crate::net::{Ipv4Addr, Ipv6Addr, Shutdown, SocketAddr}; use crate::os::unix::io::{AsFd, AsRawFd, BorrowedFd, FromRawFd, IntoRawFd, RawFd}; use crate::sys::fd::FileDesc; @@ -218,6 +218,10 @@ pub mod net { unimpl!(); } + pub fn read_buf(&self, _: BorrowedCursor<'_>) -> io::Result<()> { + unimpl!(); + } + pub fn read_vectored(&self, _: &mut [IoSliceMut<'_>]) -> io::Result<usize> { unimpl!(); } diff --git a/library/std/src/sys/unix/os_str.rs b/library/std/src/sys/unix/os_str.rs index 142fcb9ed0b..f7333fd5a1f 100644 --- a/library/std/src/sys/unix/os_str.rs +++ b/library/std/src/sys/unix/os_str.rs @@ -207,8 +207,8 @@ impl Slice { unsafe { Slice::from_os_str_bytes_unchecked(s.as_bytes()) } } - pub fn to_str(&self) -> Option<&str> { - str::from_utf8(&self.inner).ok() + pub fn to_str(&self) -> Result<&str, crate::str::Utf8Error> { + str::from_utf8(&self.inner) } pub fn to_string_lossy(&self) -> Cow<'_, str> { diff --git a/library/std/src/sys/unix/process/process_unix.rs b/library/std/src/sys/unix/process/process_unix.rs index 3721988b405..22d9d6141f4 100644 --- a/library/std/src/sys/unix/process/process_unix.rs +++ b/library/std/src/sys/unix/process/process_unix.rs @@ -836,31 +836,47 @@ fn signal_string(signal: i32) -> &'static str { libc::SIGILL => " (SIGILL)", libc::SIGTRAP => " (SIGTRAP)", libc::SIGABRT => " (SIGABRT)", + #[cfg(not(target_os = "l4re"))] libc::SIGBUS => " (SIGBUS)", libc::SIGFPE => " (SIGFPE)", libc::SIGKILL => " (SIGKILL)", + #[cfg(not(target_os = "l4re"))] libc::SIGUSR1 => " (SIGUSR1)", libc::SIGSEGV => " (SIGSEGV)", + #[cfg(not(target_os = "l4re"))] libc::SIGUSR2 => " (SIGUSR2)", libc::SIGPIPE => " (SIGPIPE)", libc::SIGALRM => " (SIGALRM)", libc::SIGTERM => " (SIGTERM)", + #[cfg(not(target_os = "l4re"))] libc::SIGCHLD => " (SIGCHLD)", + #[cfg(not(target_os = "l4re"))] libc::SIGCONT => " (SIGCONT)", + #[cfg(not(target_os = "l4re"))] libc::SIGSTOP => " (SIGSTOP)", + #[cfg(not(target_os = "l4re"))] libc::SIGTSTP => " (SIGTSTP)", + #[cfg(not(target_os = "l4re"))] libc::SIGTTIN => " (SIGTTIN)", + #[cfg(not(target_os = "l4re"))] libc::SIGTTOU => " (SIGTTOU)", + #[cfg(not(target_os = "l4re"))] libc::SIGURG => " (SIGURG)", + #[cfg(not(target_os = "l4re"))] libc::SIGXCPU => " (SIGXCPU)", + #[cfg(not(target_os = "l4re"))] libc::SIGXFSZ => " (SIGXFSZ)", + #[cfg(not(target_os = "l4re"))] libc::SIGVTALRM => " (SIGVTALRM)", + #[cfg(not(target_os = "l4re"))] libc::SIGPROF => " (SIGPROF)", + #[cfg(not(target_os = "l4re"))] libc::SIGWINCH => " (SIGWINCH)", - #[cfg(not(target_os = "haiku"))] + #[cfg(not(any(target_os = "haiku", target_os = "l4re")))] libc::SIGIO => " (SIGIO)", #[cfg(target_os = "haiku")] libc::SIGPOLL => " (SIGPOLL)", + #[cfg(not(target_os = "l4re"))] libc::SIGSYS => " (SIGSYS)", // For information on Linux signals, run `man 7 signal` #[cfg(all( diff --git a/library/std/src/sys/unix/thread.rs b/library/std/src/sys/unix/thread.rs index 878af5088d9..010015667f7 100644 --- a/library/std/src/sys/unix/thread.rs +++ b/library/std/src/sys/unix/thread.rs @@ -344,6 +344,29 @@ pub fn available_parallelism() -> io::Result<NonZeroUsize> { } } + #[cfg(target_os = "netbsd")] + { + unsafe { + let set = libc::_cpuset_create(); + if !set.is_null() { + let mut count: usize = 0; + if libc::pthread_getaffinity_np(libc::pthread_self(), libc::_cpuset_size(set), set) == 0 { + for i in 0..u64::MAX { + match libc::_cpuset_isset(i, set) { + -1 => break, + 0 => continue, + _ => count = count + 1, + } + } + } + libc::_cpuset_destroy(set); + if let Some(count) = NonZeroUsize::new(count) { + return Ok(count); + } + } + } + } + let mut cpus: libc::c_uint = 0; let mut cpus_size = crate::mem::size_of_val(&cpus); diff --git a/library/std/src/sys/windows/c.rs b/library/std/src/sys/windows/c.rs index 07b0610d463..5fc6136ba1f 100644 --- a/library/std/src/sys/windows/c.rs +++ b/library/std/src/sys/windows/c.rs @@ -19,6 +19,7 @@ pub use windows_sys::*; pub type DWORD = c_ulong; pub type NonZeroDWORD = NonZero_c_ulong; pub type LARGE_INTEGER = c_longlong; +#[cfg_attr(target_vendor = "uwp", allow(unused))] pub type LONG = c_long; pub type UINT = c_uint; pub type WCHAR = u16; @@ -267,6 +268,8 @@ pub unsafe fn getaddrinfo( windows_sys::getaddrinfo(node.cast::<u8>(), service.cast::<u8>(), hints, res) } +cfg_if::cfg_if! { +if #[cfg(not(target_vendor = "uwp"))] { pub unsafe fn NtReadFile( filehandle: BorrowedHandle<'_>, event: HANDLE, @@ -313,6 +316,8 @@ pub unsafe fn NtWriteFile( key.map(|k| k as *const u32).unwrap_or(ptr::null()), ) } +} +} // Functions that aren't available on every version of Windows that we support, // but we still use them and just provide some form of a fallback implementation. @@ -376,4 +381,54 @@ compat_fn_with_fallback! { ) -> NTSTATUS { panic!("keyed events not available") } + + // These functions are available on UWP when lazily loaded. They will fail WACK if loaded statically. + #[cfg(target_vendor = "uwp")] + pub fn NtCreateFile( + filehandle: *mut HANDLE, + desiredaccess: FILE_ACCESS_RIGHTS, + objectattributes: *const OBJECT_ATTRIBUTES, + iostatusblock: *mut IO_STATUS_BLOCK, + allocationsize: *const i64, + fileattributes: FILE_FLAGS_AND_ATTRIBUTES, + shareaccess: FILE_SHARE_MODE, + createdisposition: NTCREATEFILE_CREATE_DISPOSITION, + createoptions: NTCREATEFILE_CREATE_OPTIONS, + eabuffer: *const ::core::ffi::c_void, + ealength: u32 + ) -> NTSTATUS { + STATUS_NOT_IMPLEMENTED + } + #[cfg(target_vendor = "uwp")] + pub fn NtReadFile( + filehandle: BorrowedHandle<'_>, + event: HANDLE, + apcroutine: PIO_APC_ROUTINE, + apccontext: *mut c_void, + iostatusblock: &mut IO_STATUS_BLOCK, + buffer: *mut crate::mem::MaybeUninit<u8>, + length: ULONG, + byteoffset: Option<&LARGE_INTEGER>, + key: Option<&ULONG> + ) -> NTSTATUS { + STATUS_NOT_IMPLEMENTED + } + #[cfg(target_vendor = "uwp")] + pub fn NtWriteFile( + filehandle: BorrowedHandle<'_>, + event: HANDLE, + apcroutine: PIO_APC_ROUTINE, + apccontext: *mut c_void, + iostatusblock: &mut IO_STATUS_BLOCK, + buffer: *const u8, + length: ULONG, + byteoffset: Option<&LARGE_INTEGER>, + key: Option<&ULONG> + ) -> NTSTATUS { + STATUS_NOT_IMPLEMENTED + } + #[cfg(target_vendor = "uwp")] + pub fn RtlNtStatusToDosError(Status: NTSTATUS) -> u32 { + Status as u32 + } } diff --git a/library/std/src/sys/windows/c/windows_sys.lst b/library/std/src/sys/windows/c/windows_sys.lst index 3e454199f13..2cf1ade99ce 100644 --- a/library/std/src/sys/windows/c/windows_sys.lst +++ b/library/std/src/sys/windows/c/windows_sys.lst @@ -1930,6 +1930,7 @@ Windows.Win32.Foundation.SetLastError Windows.Win32.Foundation.STATUS_DELETE_PENDING Windows.Win32.Foundation.STATUS_END_OF_FILE Windows.Win32.Foundation.STATUS_INVALID_PARAMETER +Windows.Win32.Foundation.STATUS_NOT_IMPLEMENTED Windows.Win32.Foundation.STATUS_PENDING Windows.Win32.Foundation.STATUS_SUCCESS Windows.Win32.Foundation.TRUE diff --git a/library/std/src/sys/windows/c/windows_sys.rs b/library/std/src/sys/windows/c/windows_sys.rs index 36a30f6ba56..a4294f336fe 100644 --- a/library/std/src/sys/windows/c/windows_sys.rs +++ b/library/std/src/sys/windows/c/windows_sys.rs @@ -3888,6 +3888,7 @@ pub type STARTUPINFOW_FLAGS = u32; pub const STATUS_DELETE_PENDING: NTSTATUS = -1073741738i32; pub const STATUS_END_OF_FILE: NTSTATUS = -1073741807i32; pub const STATUS_INVALID_PARAMETER: NTSTATUS = -1073741811i32; +pub const STATUS_NOT_IMPLEMENTED: NTSTATUS = -1073741822i32; pub const STATUS_PENDING: NTSTATUS = 259i32; pub const STATUS_SUCCESS: NTSTATUS = 0i32; pub const STD_ERROR_HANDLE: STD_HANDLE = 4294967284u32; @@ -4274,3 +4275,23 @@ impl ::core::clone::Clone for XSAVE_FORMAT { *self } } +// Begin of ARM32 shim +// The raw content of this file should be processed by `generate-windows-sys` +// to be merged with the generated binding. It is not supposed to be used as +// a normal Rust module. +cfg_if::cfg_if! { +if #[cfg(target_arch = "arm")] { +#[repr(C)] +pub struct WSADATA { + pub wVersion: u16, + pub wHighVersion: u16, + pub szDescription: [u8; 257], + pub szSystemStatus: [u8; 129], + pub iMaxSockets: u16, + pub iMaxUdpDg: u16, + pub lpVendorInfo: PSTR, +} +pub enum CONTEXT {} +} +} +// End of ARM32 shim diff --git a/library/std/src/sys/windows/os_str.rs b/library/std/src/sys/windows/os_str.rs index 611f0d040f0..16c4f55c687 100644 --- a/library/std/src/sys/windows/os_str.rs +++ b/library/std/src/sys/windows/os_str.rs @@ -166,7 +166,7 @@ impl Slice { unsafe { mem::transmute(Wtf8::from_str(s)) } } - pub fn to_str(&self) -> Option<&str> { + pub fn to_str(&self) -> Result<&str, crate::str::Utf8Error> { self.inner.as_str() } diff --git a/library/std/src/sys/windows/rand.rs b/library/std/src/sys/windows/rand.rs index bca4e38d9f6..5d8fd13785a 100644 --- a/library/std/src/sys/windows/rand.rs +++ b/library/std/src/sys/windows/rand.rs @@ -1,5 +1,3 @@ -use crate::ffi::c_void; -use crate::io; use crate::mem; use crate::ptr; use crate::sys::c; @@ -25,6 +23,9 @@ pub fn hashmap_random_keys() -> (u64, u64) { #[cfg(not(target_vendor = "uwp"))] #[inline(never)] fn fallback_rng() -> (u64, u64) { + use crate::ffi::c_void; + use crate::io; + let mut v = (0, 0); let ret = unsafe { c::RtlGenRandom(&mut v as *mut _ as *mut c_void, mem::size_of_val(&v) as c::ULONG) diff --git a/library/std/src/sys_common/wtf8.rs b/library/std/src/sys_common/wtf8.rs index 31bb0ad25a6..c9d3e13cf0c 100644 --- a/library/std/src/sys_common/wtf8.rs +++ b/library/std/src/sys_common/wtf8.rs @@ -626,13 +626,8 @@ impl Wtf8 { /// /// This does not copy the data. #[inline] - pub fn as_str(&self) -> Option<&str> { - // Well-formed WTF-8 is also well-formed UTF-8 - // if and only if it contains no surrogate. - match self.next_surrogate(0) { - None => Some(unsafe { str::from_utf8_unchecked(&self.bytes) }), - Some(_) => None, - } + pub fn as_str(&self) -> Result<&str, str::Utf8Error> { + str::from_utf8(&self.bytes) } /// Creates an owned `Wtf8Buf` from a borrowed `Wtf8`. diff --git a/library/std/src/sys_common/wtf8/tests.rs b/library/std/src/sys_common/wtf8/tests.rs index 1a302d64694..a07bbe6d7e4 100644 --- a/library/std/src/sys_common/wtf8/tests.rs +++ b/library/std/src/sys_common/wtf8/tests.rs @@ -521,11 +521,11 @@ fn wtf8_code_points() { #[test] fn wtf8_as_str() { - assert_eq!(Wtf8::from_str("").as_str(), Some("")); - assert_eq!(Wtf8::from_str("aé 💩").as_str(), Some("aé 💩")); + assert_eq!(Wtf8::from_str("").as_str(), Ok("")); + assert_eq!(Wtf8::from_str("aé 💩").as_str(), Ok("aé 💩")); let mut string = Wtf8Buf::new(); string.push(CodePoint::from_u32(0xD800).unwrap()); - assert_eq!(string.as_str(), None); + assert!(string.as_str().is_err()); } #[test] diff --git a/library/std/src/thread/mod.rs b/library/std/src/thread/mod.rs index d9973185bc4..1230bb5deed 100644 --- a/library/std/src/thread/mod.rs +++ b/library/std/src/thread/mod.rs @@ -195,7 +195,7 @@ mod local; cfg_if::cfg_if! { if #[cfg(test)] { - // Avoid duplicating the global state assoicated with thread-locals between this crate and + // Avoid duplicating the global state associated with thread-locals between this crate and // realstd. Miri relies on this. pub use realstd::thread::{local_impl, AccessError, LocalKey}; } else { |