diff --git a/src/libcollections/string.rs b/src/libcollections/string.rs
index 06952253ef3b0..51fb4f515b287 100644
--- a/src/libcollections/string.rs
+++ b/src/libcollections/string.rs
@@ -1895,9 +1895,24 @@ impl<'a> From<&'a [char]> for String {
 
 #[stable(feature = "stringfromchars", since = "1.12.0")]
 impl From<Vec<char>> for String {
-    #[inline]
-    fn from(v: Vec<char>) -> String {
-        String::from(v.as_slice())
+    fn from(mut v: Vec<char>) -> String {
+        unsafe {
+            let cap = v.capacity();
+            let ptr = v.as_mut_ptr() as *mut u8;
+            let mut bytes = 0usize;
+
+            for chr in v.iter() {
+                // Regular loop instead of copy_nonoverlapping, because LLVM insists on having a
+                // memcpy for slices shorter than 4 bytes.
+                for b in chr.encode_utf8().as_slice() {
+                    // Neither bytes nor ptr can overflow.
+                    *ptr.offset(bytes as isize) = *b;
+                    bytes = bytes.wrapping_add(1);
+                }
+            }
+            mem::forget(v);
+            String::from_raw_parts(ptr, bytes, cap * mem::size_of::<char>())
+        }
     }
 }
 
diff --git a/src/libcollectionstest/string.rs b/src/libcollectionstest/string.rs
index 1652fb5a88d80..aacdc2d1e0668 100644
--- a/src/libcollectionstest/string.rs
+++ b/src/libcollectionstest/string.rs
@@ -392,6 +392,22 @@ fn test_into_boxed_str() {
     assert_eq!(&*ys, "hello my name is bob");
 }
 
+#[test]
+fn test_string_from_vec_char() {
+    let str1 = String::from(vec!['a', 'b', '😃', 'a', 'b']);
+    let str2 = String::from(vec!['a', 'ą', 'あ', '🞎']);
+    let str3 = String::from(vec!['🞎', 'あ', 'ą', 'a']);
+    assert_eq!("ab😃ab", str1);
+    assert_eq!(str1.len(), 8);
+    assert!(str1.capacity() >= 20);
+    assert_eq!("aąあ🞎", str2);
+    assert_eq!(str2.len(), 10);
+    assert!(str2.capacity() >= 16);
+    assert_eq!("🞎あąa", str3);
+    assert_eq!(str3.len(), 10);
+    assert!(str3.capacity() >= 16);
+}
+
 #[bench]
 fn bench_with_capacity(b: &mut Bencher) {
     b.iter(|| String::with_capacity(100));
diff --git a/src/libcore/char.rs b/src/libcore/char.rs
index a3440fe8aa644..24aa7b1fc34ee 100644
--- a/src/libcore/char.rs
+++ b/src/libcore/char.rs
@@ -28,6 +28,7 @@ const TAG_FOUR_B: u8  = 0b1111_0000;
 const MAX_ONE_B: u32   =     0x80;
 const MAX_TWO_B: u32   =    0x800;
 const MAX_THREE_B: u32 =  0x10000;
+const UTF8_BUF_SIZE: usize = 4;
 
 /*
     Lu  Uppercase_Letter        an uppercase letter
@@ -644,15 +645,27 @@ impl ExactSizeIterator for EscapeDebug { }
 #[unstable(feature = "unicode", issue = "27784")]
 #[derive(Debug)]
 pub struct EncodeUtf8 {
-    buf: [u8; 4],
+    buf: [u8; UTF8_BUF_SIZE],
     pos: usize,
 }
 
 impl EncodeUtf8 {
     /// Returns the remaining bytes of this iterator as a slice.
     #[unstable(feature = "unicode", issue = "27784")]
+    #[inline(always)]
     pub fn as_slice(&self) -> &[u8] {
-        &self.buf[self.pos..]
+        // We know for sure this method cannot slice out-of-bounds because:
+        // * 0 ≤ self.pos ≤ 3
+        // * self.buf.len() = 4
+        //
+        // This way the slicing will always succeed, but LLVM is incapable of figuring out both
+        // these conditions hold, resulting in suboptimal code, especially after inlining.
+        // Ideally there would be a `slice_unchecked` method for slices, but there isn’t any,
+        // therefore we construct the slice manually.
+        unsafe {
+            ::slice::from_raw_parts(self.buf.as_ptr().offset(self.pos as isize),
+                                    UTF8_BUF_SIZE.wrapping_sub(self.pos))
+        }
     }
 }