Rollup merge of #67569 - Mark-Simulacrum:opt-char-encode, r=oli-obk

Clean up unsafety in char::encode_utf8

This originally started as an attempt to allow LLVM to optimize through
encode_utf8 to detect the try_encode_utf8 case (#52579, #52580), but due to a
typo my conclusion that my optimizations were successful was incorrect.

Furthermore, as far as I can tell, this optimization is probably just not
possible with LLVM today.  This [code](https://rust.godbolt.org/z/JggRj4)
compiles down to a long series of compares, notably, two identical series of
compares. That essentially means that LLVM is today unable to see that these two
ifs are identical and as such can be merged and then realize that no value of
the if condition can result in a call to `please_delete`. As such, for now, we
do not attempt to specifically optimize for that case.

src/libcore/char/methods.rs

@@ -434,36 +434,35 @@ impl char {
     #[inline]
     pub fn encode_utf8(self, dst: &mut [u8]) -> &mut str {
         let code = self as u32;
-        // SAFETY: each arm checks the size of the slice and only uses `get_unchecked` unsafe ops
-        unsafe {
-            let len = if code < MAX_ONE_B && !dst.is_empty() {
-                *dst.get_unchecked_mut(0) = code as u8;
-                1
-            } else if code < MAX_TWO_B && dst.len() >= 2 {
-                *dst.get_unchecked_mut(0) = (code >> 6 & 0x1F) as u8 | TAG_TWO_B;
-                *dst.get_unchecked_mut(1) = (code & 0x3F) as u8 | TAG_CONT;
-                2
-            } else if code < MAX_THREE_B && dst.len() >= 3 {
-                *dst.get_unchecked_mut(0) = (code >> 12 & 0x0F) as u8 | TAG_THREE_B;
-                *dst.get_unchecked_mut(1) = (code >> 6 & 0x3F) as u8 | TAG_CONT;
-                *dst.get_unchecked_mut(2) = (code & 0x3F) as u8 | TAG_CONT;
-                3
-            } else if dst.len() >= 4 {
-                *dst.get_unchecked_mut(0) = (code >> 18 & 0x07) as u8 | TAG_FOUR_B;
-                *dst.get_unchecked_mut(1) = (code >> 12 & 0x3F) as u8 | TAG_CONT;
-                *dst.get_unchecked_mut(2) = (code >> 6 & 0x3F) as u8 | TAG_CONT;
-                *dst.get_unchecked_mut(3) = (code & 0x3F) as u8 | TAG_CONT;
-                4
-            } else {
-                panic!(
-                    "encode_utf8: need {} bytes to encode U+{:X}, but the buffer has {}",
-                    from_u32_unchecked(code).len_utf8(),
-                    code,
-                    dst.len(),
-                )
-            };
-            from_utf8_unchecked_mut(dst.get_unchecked_mut(..len))
-        }
+        let len = self.len_utf8();
+        match (len, &mut dst[..]) {
+            (1, [a, ..]) => {
+                *a = code as u8;
+            }
+            (2, [a, b, ..]) => {
+                *a = (code >> 6 & 0x1F) as u8 | TAG_TWO_B;
+                *b = (code & 0x3F) as u8 | TAG_CONT;
+            }
+            (3, [a, b, c, ..]) => {
+                *a = (code >> 12 & 0x0F) as u8 | TAG_THREE_B;
+                *b = (code >> 6 & 0x3F) as u8 | TAG_CONT;
+                *c = (code & 0x3F) as u8 | TAG_CONT;
+            }
+            (4, [a, b, c, d, ..]) => {
+                *a = (code >> 18 & 0x07) as u8 | TAG_FOUR_B;
+                *b = (code >> 12 & 0x3F) as u8 | TAG_CONT;
+                *c = (code >> 6 & 0x3F) as u8 | TAG_CONT;
+                *d = (code & 0x3F) as u8 | TAG_CONT;
+            }
+            _ => panic!(
+                "encode_utf8: need {} bytes to encode U+{:X}, but the buffer has {}",
+                len,
+                code,
+                dst.len(),
+            ),
+        };
+        // SAFETY: We just wrote UTF-8 content in, so converting to str is fine.
+        unsafe { from_utf8_unchecked_mut(&mut dst[..len]) }
     }
 
     /// Encodes this character as UTF-16 into the provided `u16` buffer,

src/libcore/lib.rs

@@ -129,6 +129,7 @@
 #![feature(associated_type_bounds)]
 #![feature(const_type_id)]
 #![feature(const_caller_location)]
+#![feature(slice_patterns)]
 
 #[prelude_import]
 #[allow(unused)]