diff --git a/src/librustc_metadata/index.rs b/src/librustc_metadata/index.rs
index 18f30383090cd..f9543a18c68a7 100644
--- a/src/librustc_metadata/index.rs
+++ b/src/librustc_metadata/index.rs
@@ -2,10 +2,71 @@ use crate::schema::*;
 
 use rustc::hir::def_id::{DefId, DefIndex, DefIndexAddressSpace};
 use rustc_serialize::opaque::Encoder;
-use std::slice;
 use std::u32;
 use log::debug;
 
+/// Helper trait, for encoding to, and decoding from, a fixed number of bytes.
+pub trait FixedSizeEncoding {
+    const BYTE_LEN: usize;
+
+    // FIXME(eddyb) convert to and from `[u8; Self::BYTE_LEN]` instead,
+    // once that starts being allowed by the compiler (i.e. lazy normalization).
+    fn from_bytes(b: &[u8]) -> Self;
+    fn write_to_bytes(self, b: &mut [u8]);
+
+    // FIXME(eddyb) make these generic functions, or at least defaults here.
+    // (same problem as above, needs `[u8; Self::BYTE_LEN]`)
+    // For now, a macro (`fixed_size_encoding_byte_len_and_defaults`) is used.
+    fn read_from_bytes_at(b: &[u8], i: usize) -> Self;
+    fn write_to_bytes_at(self, b: &mut [u8], i: usize);
+}
+
+// HACK(eddyb) this shouldn't be needed (see comments on the methods above).
+macro_rules! fixed_size_encoding_byte_len_and_defaults {
+    ($byte_len:expr) => {
+        const BYTE_LEN: usize = $byte_len;
+        fn read_from_bytes_at(b: &[u8], i: usize) -> Self {
+            const BYTE_LEN: usize = $byte_len;
+            // HACK(eddyb) ideally this would be done with fully safe code,
+            // but slicing `[u8]` with `i * N..` is optimized worse, due to the
+            // possibility of `i * N` overflowing, than indexing `[[u8; N]]`.
+            let b = unsafe {
+                std::slice::from_raw_parts(
+                    b.as_ptr() as *const [u8; BYTE_LEN],
+                    b.len() / BYTE_LEN,
+                )
+            };
+            Self::from_bytes(&b[i])
+        }
+        fn write_to_bytes_at(self, b: &mut [u8], i: usize) {
+            const BYTE_LEN: usize = $byte_len;
+            // HACK(eddyb) ideally this would be done with fully safe code,
+            // see similar comment in `read_from_bytes_at` for why it can't yet.
+            let b = unsafe {
+                std::slice::from_raw_parts_mut(
+                    b.as_mut_ptr() as *mut [u8; BYTE_LEN],
+                    b.len() / BYTE_LEN,
+                )
+            };
+            self.write_to_bytes(&mut b[i]);
+        }
+    }
+}
+
+impl FixedSizeEncoding for u32 {
+    fixed_size_encoding_byte_len_and_defaults!(4);
+
+    fn from_bytes(b: &[u8]) -> Self {
+        let mut bytes = [0; Self::BYTE_LEN];
+        bytes.copy_from_slice(&b[..Self::BYTE_LEN]);
+        Self::from_le_bytes(bytes)
+    }
+
+    fn write_to_bytes(self, b: &mut [u8]) {
+        b[..Self::BYTE_LEN].copy_from_slice(&self.to_le_bytes());
+    }
+}
+
 /// While we are generating the metadata, we also track the position
 /// of each DefIndex. It is not required that all definitions appear
 /// in the metadata, nor that they are serialized in order, and
@@ -14,14 +75,14 @@ use log::debug;
 /// appropriate spot by calling `record_position`. We should never
 /// visit the same index twice.
 pub struct Index {
-    positions: [Vec<u32>; 2]
+    positions: [Vec<u8>; 2]
 }
 
 impl Index {
     pub fn new((max_index_lo, max_index_hi): (usize, usize)) -> Index {
         Index {
-            positions: [vec![u32::MAX; max_index_lo],
-                        vec![u32::MAX; max_index_hi]],
+            positions: [vec![0xff; max_index_lo * 4],
+                        vec![0xff; max_index_hi * 4]],
         }
     }
 
@@ -36,26 +97,27 @@ impl Index {
         let space_index = item.address_space().index();
         let array_index = item.as_array_index();
 
-        assert!(self.positions[space_index][array_index] == u32::MAX,
+        let positions = &mut self.positions[space_index];
+        assert!(u32::read_from_bytes_at(positions, array_index) == u32::MAX,
                 "recorded position for item {:?} twice, first at {:?} and now at {:?}",
                 item,
-                self.positions[space_index][array_index],
+                u32::read_from_bytes_at(positions, array_index),
                 position);
 
-        self.positions[space_index][array_index] = position.to_le();
+        position.write_to_bytes_at(positions, array_index)
     }
 
     pub fn write_index(&self, buf: &mut Encoder) -> LazySeq<Index> {
         let pos = buf.position();
 
         // First we write the length of the lower range ...
-        buf.emit_raw_bytes(words_to_bytes(&[(self.positions[0].len() as u32).to_le()]));
+        buf.emit_raw_bytes(&(self.positions[0].len() as u32 / 4).to_le_bytes());
         // ... then the values in the lower range ...
-        buf.emit_raw_bytes(words_to_bytes(&self.positions[0][..]));
+        buf.emit_raw_bytes(&self.positions[0]);
         // ... then the values in the higher range.
-        buf.emit_raw_bytes(words_to_bytes(&self.positions[1][..]));
+        buf.emit_raw_bytes(&self.positions[1]);
         LazySeq::with_position_and_length(pos as usize,
-            self.positions[0].len() + self.positions[1].len() + 1)
+            (self.positions[0].len() + self.positions[1].len()) / 4 + 1)
     }
 }
 
@@ -64,24 +126,21 @@ impl<'tcx> LazySeq<Index> {
     /// DefIndex (if any).
     #[inline(never)]
     pub fn lookup(&self, bytes: &[u8], def_index: DefIndex) -> Option<Lazy<Entry<'tcx>>> {
-        let words = &bytes_to_words(&bytes[self.position..])[..self.len];
-
-        debug!("Index::lookup: index={:?} words.len={:?}",
+        let bytes = &bytes[self.position..];
+        debug!("Index::lookup: index={:?} len={:?}",
                def_index,
-               words.len());
+               self.len);
 
-        let positions = match def_index.address_space() {
-            DefIndexAddressSpace::Low => &words[1..],
+        let i = def_index.as_array_index() + match def_index.address_space() {
+            DefIndexAddressSpace::Low => 0,
             DefIndexAddressSpace::High => {
                 // This is a DefIndex in the higher range, so find out where
                 // that starts:
-                let lo_count = u32::from_le(words[0].get()) as usize;
-                &words[lo_count + 1 .. ]
+                u32::read_from_bytes_at(bytes, 0) as usize
             }
         };
 
-        let array_index = def_index.as_array_index();
-        let position = u32::from_le(positions[array_index].get());
+        let position = u32::read_from_bytes_at(bytes, 1 + i);
         if position == u32::MAX {
             debug!("Index::lookup: position=u32::MAX");
             None
@@ -91,27 +150,3 @@ impl<'tcx> LazySeq<Index> {
         }
     }
 }
-
-#[repr(packed)]
-#[derive(Copy)]
-struct Unaligned<T>(T);
-
-// The derived Clone impl is unsafe for this packed struct since it needs to pass a reference to
-// the field to `T::clone`, but this reference may not be properly aligned.
-impl<T: Copy> Clone for Unaligned<T> {
-    fn clone(&self) -> Self {
-        *self
-    }
-}
-
-impl<T> Unaligned<T> {
-    fn get(self) -> T { self.0 }
-}
-
-fn bytes_to_words(b: &[u8]) -> &[Unaligned<u32>] {
-    unsafe { slice::from_raw_parts(b.as_ptr() as *const Unaligned<u32>, b.len() / 4) }
-}
-
-fn words_to_bytes(w: &[u32]) -> &[u8] {
-    unsafe { slice::from_raw_parts(w.as_ptr() as *const u8, w.len() * 4) }
-}