Auto merge of #128330 - matthiaskrgr:rollup-auairjd, r=matthiaskrgr

Rollup of 5 pull requests

Successful merges:

 - #109174 (Replace `io::Cursor::{remaining_slice, is_empty}`)
 - #127290 (Fully document `rustdoc-json-types`)
 - #128055 (std: unsafe-wrap personality::dwarf::eh)
 - #128269 (improve cargo invocations on bootstrap)
 - #128310 (Add missing periods on `BTreeMap` cursor `peek_next` docs)

Failed merges:

 - #127543 (More unsafe attr verification)
 - #128182 (handle no_std targets on std builds)

r? `@ghost`
`@rustbot` modify labels: rollup

library/alloc/src/collections/btree/map.rs

@@ -2921,7 +2921,7 @@ impl<'a, K, V> Cursor<'a, K, V> {
     /// Returns a reference to the key and value of the next element without
     /// moving the cursor.
     ///
-    /// If the cursor is at the end of the map then `None` is returned
+    /// If the cursor is at the end of the map then `None` is returned.
     #[unstable(feature = "btree_cursors", issue = "107540")]
     pub fn peek_next(&self) -> Option<(&'a K, &'a V)> {
         self.clone().next()
@@ -2963,7 +2963,7 @@ impl<'a, K, V, A> CursorMut<'a, K, V, A> {
     /// Returns a reference to the key and value of the next element without
     /// moving the cursor.
     ///
-    /// If the cursor is at the end of the map then `None` is returned
+    /// If the cursor is at the end of the map then `None` is returned.
     #[unstable(feature = "btree_cursors", issue = "107540")]
     pub fn peek_next(&mut self) -> Option<(&K, &mut V)> {
         let (k, v) = self.inner.peek_next()?;
@@ -3061,7 +3061,7 @@ impl<'a, K, V, A> CursorMutKey<'a, K, V, A> {
     /// Returns a reference to the key and value of the next element without
     /// moving the cursor.
     ///
-    /// If the cursor is at the end of the map then `None` is returned
+    /// If the cursor is at the end of the map then `None` is returned.
     #[unstable(feature = "btree_cursors", issue = "107540")]
     pub fn peek_next(&mut self) -> Option<(&mut K, &mut V)> {
         let current = self.current.as_mut()?;

library/std/src/io/cursor.rs

@@ -209,55 +209,60 @@ impl<T> Cursor<T>
 where
     T: AsRef<[u8]>,
 {
-    /// Returns the remaining slice.
+    /// Splits the underlying slice at the cursor position and returns them.
     ///
     /// # Examples
     ///
     /// ```
-    /// #![feature(cursor_remaining)]
+    /// #![feature(cursor_split)]
     /// use std::io::Cursor;
     ///
     /// let mut buff = Cursor::new(vec![1, 2, 3, 4, 5]);
     ///
-    /// assert_eq!(buff.remaining_slice(), &[1, 2, 3, 4, 5]);
+    /// assert_eq!(buff.split(), ([].as_slice(), [1, 2, 3, 4, 5].as_slice()));
     ///
     /// buff.set_position(2);
-    /// assert_eq!(buff.remaining_slice(), &[3, 4, 5]);
-    ///
-    /// buff.set_position(4);
-    /// assert_eq!(buff.remaining_slice(), &[5]);
+    /// assert_eq!(buff.split(), ([1, 2].as_slice(), [3, 4, 5].as_slice()));
     ///
     /// buff.set_position(6);
-    /// assert_eq!(buff.remaining_slice(), &[]);
+    /// assert_eq!(buff.split(), ([1, 2, 3, 4, 5].as_slice(), [].as_slice()));
     /// ```
-    #[unstable(feature = "cursor_remaining", issue = "86369")]
-    pub fn remaining_slice(&self) -> &[u8] {
-        let len = self.pos.min(self.inner.as_ref().len() as u64);
-        &self.inner.as_ref()[(len as usize)..]
+    #[unstable(feature = "cursor_split", issue = "86369")]
+    pub fn split(&self) -> (&[u8], &[u8]) {
+        let slice = self.inner.as_ref();
+        let pos = self.pos.min(slice.len() as u64);
+        slice.split_at(pos as usize)
     }
+}
 
-    /// Returns `true` if the remaining slice is empty.
+impl<T> Cursor<T>
+where
+    T: AsMut<[u8]>,
+{
+    /// Splits the underlying slice at the cursor position and returns them
+    /// mutably.
     ///
     /// # Examples
     ///
     /// ```
-    /// #![feature(cursor_remaining)]
+    /// #![feature(cursor_split)]
     /// use std::io::Cursor;
     ///
     /// let mut buff = Cursor::new(vec![1, 2, 3, 4, 5]);
     ///
-    /// buff.set_position(2);
-    /// assert!(!buff.is_empty());
+    /// assert_eq!(buff.split_mut(), ([].as_mut_slice(), [1, 2, 3, 4, 5].as_mut_slice()));
     ///
-    /// buff.set_position(5);
-    /// assert!(buff.is_empty());
+    /// buff.set_position(2);
+    /// assert_eq!(buff.split_mut(), ([1, 2].as_mut_slice(), [3, 4, 5].as_mut_slice()));
     ///
-    /// buff.set_position(10);
-    /// assert!(buff.is_empty());
+    /// buff.set_position(6);
+    /// assert_eq!(buff.split_mut(), ([1, 2, 3, 4, 5].as_mut_slice(), [].as_mut_slice()));
     /// ```
-    #[unstable(feature = "cursor_remaining", issue = "86369")]
-    pub fn is_empty(&self) -> bool {
-        self.pos >= self.inner.as_ref().len() as u64
+    #[unstable(feature = "cursor_split", issue = "86369")]
+    pub fn split_mut(&mut self) -> (&mut [u8], &mut [u8]) {
+        let slice = self.inner.as_mut();
+        let pos = self.pos.min(slice.len() as u64);
+        slice.split_at_mut(pos as usize)
     }
 }
 
@@ -319,15 +324,15 @@ where
     T: AsRef<[u8]>,
 {
     fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> {
-        let n = Read::read(&mut self.remaining_slice(), buf)?;
+        let n = Read::read(&mut Cursor::split(self).1, buf)?;
         self.pos += n as u64;
         Ok(n)
     }
 
     fn read_buf(&mut self, mut cursor: BorrowedCursor<'_>) -> io::Result<()> {
         let prev_written = cursor.written();
 
-        Read::read_buf(&mut self.remaining_slice(), cursor.reborrow())?;
+        Read::read_buf(&mut Cursor::split(self).1, cursor.reborrow())?;
 
         self.pos += (cursor.written() - prev_written) as u64;
 
@@ -351,7 +356,7 @@ where
     }
 
     fn read_exact(&mut self, buf: &mut [u8]) -> io::Result<()> {
-        let result = Read::read_exact(&mut self.remaining_slice(), buf);
+        let result = Read::read_exact(&mut Cursor::split(self).1, buf);
 
         match result {
             Ok(_) => self.pos += buf.len() as u64,
@@ -365,14 +370,14 @@ where
     fn read_buf_exact(&mut self, mut cursor: BorrowedCursor<'_>) -> io::Result<()> {
         let prev_written = cursor.written();
 
-        let result = Read::read_buf_exact(&mut self.remaining_slice(), cursor.reborrow());
+        let result = Read::read_buf_exact(&mut Cursor::split(self).1, cursor.reborrow());
         self.pos += (cursor.written() - prev_written) as u64;
 
         result
     }
 
     fn read_to_end(&mut self, buf: &mut Vec<u8>) -> io::Result<usize> {
-        let content = self.remaining_slice();
+        let content = Cursor::split(self).1;
         let len = content.len();
         buf.try_reserve(len)?;
         buf.extend_from_slice(content);
@@ -383,7 +388,7 @@ where
 
     fn read_to_string(&mut self, buf: &mut String) -> io::Result<usize> {
         let content =
-            crate::str::from_utf8(self.remaining_slice()).map_err(|_| io::Error::INVALID_UTF8)?;
+            crate::str::from_utf8(Cursor::split(self).1).map_err(|_| io::Error::INVALID_UTF8)?;
         let len = content.len();
         buf.try_reserve(len)?;
         buf.push_str(content);
@@ -399,7 +404,7 @@ where
     T: AsRef<[u8]>,
 {
     fn fill_buf(&mut self) -> io::Result<&[u8]> {
-        Ok(self.remaining_slice())
+        Ok(Cursor::split(self).1)
     }
     fn consume(&mut self, amt: usize) {
         self.pos += amt as u64;

library/std/src/io/tests.rs

@@ -676,13 +676,13 @@ fn cursor_read_exact_eof() {
 
     let mut r = slice.clone();
     assert!(r.read_exact(&mut [0; 10]).is_err());
-    assert!(r.is_empty());
+    assert!(Cursor::split(&r).1.is_empty());
 
     let mut r = slice;
     let buf = &mut [0; 10];
     let mut buf = BorrowedBuf::from(buf.as_mut_slice());
     assert!(r.read_buf_exact(buf.unfilled()).is_err());
-    assert!(r.is_empty());
+    assert!(Cursor::split(&r).1.is_empty());
     assert_eq!(buf.filled(), b"123456");
 }
 

library/std/src/sys/personality/dwarf/eh.rs

@@ -70,45 +70,51 @@ pub unsafe fn find_eh_action(lsda: *const u8, context: &EHContext<'_>) -> Result
 
     let func_start = context.func_start;
     let mut reader = DwarfReader::new(lsda);
-
-    let start_encoding = reader.read::<u8>();
-    // base address for landing pad offsets
-    let lpad_base = if start_encoding != DW_EH_PE_omit {
-        read_encoded_pointer(&mut reader, context, start_encoding)?
-    } else {
-        func_start
+    let lpad_base = unsafe {
+        let start_encoding = reader.read::<u8>();
+        // base address for landing pad offsets
+        if start_encoding != DW_EH_PE_omit {
+            read_encoded_pointer(&mut reader, context, start_encoding)?
+        } else {
+            func_start
+        }
     };
+    let call_site_encoding = unsafe {
+        let ttype_encoding = reader.read::<u8>();
+        if ttype_encoding != DW_EH_PE_omit {
+            // Rust doesn't analyze exception types, so we don't care about the type table
+            reader.read_uleb128();
+        }
 
-    let ttype_encoding = reader.read::<u8>();
-    if ttype_encoding != DW_EH_PE_omit {
-        // Rust doesn't analyze exception types, so we don't care about the type table
-        reader.read_uleb128();
-    }
-
-    let call_site_encoding = reader.read::<u8>();
-    let call_site_table_length = reader.read_uleb128();
-    let action_table = reader.ptr.add(call_site_table_length as usize);
+        reader.read::<u8>()
+    };
+    let action_table = unsafe {
+        let call_site_table_length = reader.read_uleb128();
+        reader.ptr.add(call_site_table_length as usize)
+    };
     let ip = context.ip;
 
     if !USING_SJLJ_EXCEPTIONS {
         // read the callsite table
         while reader.ptr < action_table {
-            // these are offsets rather than pointers;
-            let cs_start = read_encoded_offset(&mut reader, call_site_encoding)?;
-            let cs_len = read_encoded_offset(&mut reader, call_site_encoding)?;
-            let cs_lpad = read_encoded_offset(&mut reader, call_site_encoding)?;
-            let cs_action_entry = reader.read_uleb128();
-            // Callsite table is sorted by cs_start, so if we've passed the ip, we
-            // may stop searching.
-            if ip < func_start.wrapping_add(cs_start) {
-                break;
-            }
-            if ip < func_start.wrapping_add(cs_start + cs_len) {
-                if cs_lpad == 0 {
-                    return Ok(EHAction::None);
-                } else {
-                    let lpad = lpad_base.wrapping_add(cs_lpad);
-                    return Ok(interpret_cs_action(action_table, cs_action_entry, lpad));
+            unsafe {
+                // these are offsets rather than pointers;
+                let cs_start = read_encoded_offset(&mut reader, call_site_encoding)?;
+                let cs_len = read_encoded_offset(&mut reader, call_site_encoding)?;
+                let cs_lpad = read_encoded_offset(&mut reader, call_site_encoding)?;
+                let cs_action_entry = reader.read_uleb128();
+                // Callsite table is sorted by cs_start, so if we've passed the ip, we
+                // may stop searching.
+                if ip < func_start.wrapping_add(cs_start) {
+                    break;
+                }
+                if ip < func_start.wrapping_add(cs_start + cs_len) {
+                    if cs_lpad == 0 {
+                        return Ok(EHAction::None);
+                    } else {
+                        let lpad = lpad_base.wrapping_add(cs_lpad);
+                        return Ok(interpret_cs_action(action_table, cs_action_entry, lpad));
+                    }
                 }
             }
         }
@@ -125,15 +131,15 @@ pub unsafe fn find_eh_action(lsda: *const u8, context: &EHContext<'_>) -> Result
         }
         let mut idx = ip.addr();
         loop {
-            let cs_lpad = reader.read_uleb128();
-            let cs_action_entry = reader.read_uleb128();
+            let cs_lpad = unsafe { reader.read_uleb128() };
+            let cs_action_entry = unsafe { reader.read_uleb128() };
             idx -= 1;
             if idx == 0 {
                 // Can never have null landing pad for sjlj -- that would have
                 // been indicated by a -1 call site index.
                 // FIXME(strict provenance)
                 let lpad = ptr::with_exposed_provenance((cs_lpad + 1) as usize);
-                return Ok(interpret_cs_action(action_table, cs_action_entry, lpad));
+                return Ok(unsafe { interpret_cs_action(action_table, cs_action_entry, lpad) });
             }
         }
     }
@@ -151,9 +157,9 @@ unsafe fn interpret_cs_action(
     } else {
         // If lpad != 0 and cs_action_entry != 0, we have to check ttype_index.
         // If ttype_index == 0 under the condition, we take cleanup action.
-        let action_record = action_table.offset(cs_action_entry as isize - 1);
+        let action_record = unsafe { action_table.offset(cs_action_entry as isize - 1) };
         let mut action_reader = DwarfReader::new(action_record);
-        let ttype_index = action_reader.read_sleb128();
+        let ttype_index = unsafe { action_reader.read_sleb128() };
         if ttype_index == 0 {
             EHAction::Cleanup(lpad)
         } else if ttype_index > 0 {
@@ -186,18 +192,20 @@ unsafe fn read_encoded_offset(reader: &mut DwarfReader, encoding: u8) -> Result<
     if encoding == DW_EH_PE_omit || encoding & 0xF0 != 0 {
         return Err(());
     }
-    let result = match encoding & 0x0F {
-        // despite the name, LLVM also uses absptr for offsets instead of pointers
-        DW_EH_PE_absptr => reader.read::<usize>(),
-        DW_EH_PE_uleb128 => reader.read_uleb128() as usize,
-        DW_EH_PE_udata2 => reader.read::<u16>() as usize,
-        DW_EH_PE_udata4 => reader.read::<u32>() as usize,
-        DW_EH_PE_udata8 => reader.read::<u64>() as usize,
-        DW_EH_PE_sleb128 => reader.read_sleb128() as usize,
-        DW_EH_PE_sdata2 => reader.read::<i16>() as usize,
-        DW_EH_PE_sdata4 => reader.read::<i32>() as usize,
-        DW_EH_PE_sdata8 => reader.read::<i64>() as usize,
-        _ => return Err(()),
+    let result = unsafe {
+        match encoding & 0x0F {
+            // despite the name, LLVM also uses absptr for offsets instead of pointers
+            DW_EH_PE_absptr => reader.read::<usize>(),
+            DW_EH_PE_uleb128 => reader.read_uleb128() as usize,
+            DW_EH_PE_udata2 => reader.read::<u16>() as usize,
+            DW_EH_PE_udata4 => reader.read::<u32>() as usize,
+            DW_EH_PE_udata8 => reader.read::<u64>() as usize,
+            DW_EH_PE_sleb128 => reader.read_sleb128() as usize,
+            DW_EH_PE_sdata2 => reader.read::<i16>() as usize,
+            DW_EH_PE_sdata4 => reader.read::<i32>() as usize,
+            DW_EH_PE_sdata8 => reader.read::<i64>() as usize,
+            _ => return Err(()),
+        }
     };
     Ok(result)
 }
@@ -250,14 +258,14 @@ unsafe fn read_encoded_pointer(
         if encoding & 0x0F != DW_EH_PE_absptr {
             return Err(());
         }
-        reader.read::<*const u8>()
+        unsafe { reader.read::<*const u8>() }
     } else {
-        let offset = read_encoded_offset(reader, encoding & 0x0F)?;
+        let offset = unsafe { read_encoded_offset(reader, encoding & 0x0F)? };
         base_ptr.wrapping_add(offset)
     };
 
     if encoding & DW_EH_PE_indirect != 0 {
-        ptr = *(ptr.cast::<*const u8>());
+        ptr = unsafe { *(ptr.cast::<*const u8>()) };
     }
 
     Ok(ptr)

library/std/src/sys/personality/dwarf/mod.rs

@@ -5,6 +5,7 @@
 // This module is used only by x86_64-pc-windows-gnu for now, but we
 // are compiling it everywhere to avoid regressions.
 #![allow(unused)]
+#![forbid(unsafe_op_in_unsafe_fn)]
 
 #[cfg(test)]
 mod tests;
@@ -17,7 +18,6 @@ pub struct DwarfReader {
     pub ptr: *const u8,
 }
 
-#[forbid(unsafe_op_in_unsafe_fn)]
 impl DwarfReader {
     pub fn new(ptr: *const u8) -> DwarfReader {
         DwarfReader { ptr }