Auto merge of #29316 - GBGamer:change-unchecked-div-generic, r=eddyb

Similarly to the simd intrinsics. I believe this is a better solution than #29288, and I could implement it as well for overflowing_add/sub/mul. Also rename from udiv/sdiv to div, and same for rem.

8 files changed, 605 insertions, 404 deletions

diff --git a/src/libcore/intrinsics.rs b/src/libcore/intrinsics.rs
index 36c523cabc9..ea4792bb495 100644
--- a/src/libcore/intrinsics.rs
+++ b/src/libcore/intrinsics.rs
@@ -511,114 +511,172 @@ extern "rust-intrinsic" {
     pub fn roundf64(x: f64) -> f64;
 
     /// Returns the number of bits set in a `u8`.
+    #[cfg(stage0)]
     pub fn ctpop8(x: u8) -> u8;
     /// Returns the number of bits set in a `u16`.
+    #[cfg(stage0)]
     pub fn ctpop16(x: u16) -> u16;
     /// Returns the number of bits set in a `u32`.
+    #[cfg(stage0)]
     pub fn ctpop32(x: u32) -> u32;
     /// Returns the number of bits set in a `u64`.
+    #[cfg(stage0)]
     pub fn ctpop64(x: u64) -> u64;
+    /// Returns the number of bits set in an integer type `T`
+    #[cfg(not(stage0))]
+    pub fn ctpop<T>(x: T) -> T;
 
     /// Returns the number of leading bits unset in a `u8`.
+    #[cfg(stage0)]
     pub fn ctlz8(x: u8) -> u8;
     /// Returns the number of leading bits unset in a `u16`.
+    #[cfg(stage0)]
     pub fn ctlz16(x: u16) -> u16;
     /// Returns the number of leading bits unset in a `u32`.
+    #[cfg(stage0)]
     pub fn ctlz32(x: u32) -> u32;
     /// Returns the number of leading bits unset in a `u64`.
+    #[cfg(stage0)]
     pub fn ctlz64(x: u64) -> u64;
+    /// Returns the number of leading bits unset in an integer type `T`
+    #[cfg(not(stage0))]
+    pub fn ctlz<T>(x: T) -> T;
 
     /// Returns the number of trailing bits unset in a `u8`.
+    #[cfg(stage0)]
     pub fn cttz8(x: u8) -> u8;
     /// Returns the number of trailing bits unset in a `u16`.
+    #[cfg(stage0)]
     pub fn cttz16(x: u16) -> u16;
     /// Returns the number of trailing bits unset in a `u32`.
+    #[cfg(stage0)]
     pub fn cttz32(x: u32) -> u32;
     /// Returns the number of trailing bits unset in a `u64`.
+    #[cfg(stage0)]
     pub fn cttz64(x: u64) -> u64;
+    /// Returns the number of trailing bits unset in an integer type `T`
+    #[cfg(not(stage0))]
+    pub fn cttz<T>(x: T) -> T;
 
     /// Reverses the bytes in a `u16`.
+    #[cfg(stage0)]
     pub fn bswap16(x: u16) -> u16;
     /// Reverses the bytes in a `u32`.
+    #[cfg(stage0)]
     pub fn bswap32(x: u32) -> u32;
     /// Reverses the bytes in a `u64`.
+    #[cfg(stage0)]
     pub fn bswap64(x: u64) -> u64;
+    /// Reverses the bytes in an integer type `T`.
+    #[cfg(not(stage0))]
+    pub fn bswap<T>(x: T) -> T;
 
     /// Performs checked `i8` addition.
+    #[cfg(stage0)]
     pub fn i8_add_with_overflow(x: i8, y: i8) -> (i8, bool);
     /// Performs checked `i16` addition.
+    #[cfg(stage0)]
     pub fn i16_add_with_overflow(x: i16, y: i16) -> (i16, bool);
     /// Performs checked `i32` addition.
+    #[cfg(stage0)]
     pub fn i32_add_with_overflow(x: i32, y: i32) -> (i32, bool);
     /// Performs checked `i64` addition.
+    #[cfg(stage0)]
     pub fn i64_add_with_overflow(x: i64, y: i64) -> (i64, bool);
 
     /// Performs checked `u8` addition.
+    #[cfg(stage0)]
     pub fn u8_add_with_overflow(x: u8, y: u8) -> (u8, bool);
     /// Performs checked `u16` addition.
+    #[cfg(stage0)]
     pub fn u16_add_with_overflow(x: u16, y: u16) -> (u16, bool);
     /// Performs checked `u32` addition.
+    #[cfg(stage0)]
     pub fn u32_add_with_overflow(x: u32, y: u32) -> (u32, bool);
     /// Performs checked `u64` addition.
+    #[cfg(stage0)]
     pub fn u64_add_with_overflow(x: u64, y: u64) -> (u64, bool);
 
+    /// Performs checked integer addition.
+    #[cfg(not(stage0))]
+    pub fn add_with_overflow<T>(x: T, y: T) -> (T, bool);
+
     /// Performs checked `i8` subtraction.
+    #[cfg(stage0)]
     pub fn i8_sub_with_overflow(x: i8, y: i8) -> (i8, bool);
     /// Performs checked `i16` subtraction.
+    #[cfg(stage0)]
     pub fn i16_sub_with_overflow(x: i16, y: i16) -> (i16, bool);
     /// Performs checked `i32` subtraction.
+    #[cfg(stage0)]
     pub fn i32_sub_with_overflow(x: i32, y: i32) -> (i32, bool);
     /// Performs checked `i64` subtraction.
+    #[cfg(stage0)]
     pub fn i64_sub_with_overflow(x: i64, y: i64) -> (i64, bool);
 
     /// Performs checked `u8` subtraction.
+    #[cfg(stage0)]
     pub fn u8_sub_with_overflow(x: u8, y: u8) -> (u8, bool);
     /// Performs checked `u16` subtraction.
+    #[cfg(stage0)]
     pub fn u16_sub_with_overflow(x: u16, y: u16) -> (u16, bool);
     /// Performs checked `u32` subtraction.
+    #[cfg(stage0)]
     pub fn u32_sub_with_overflow(x: u32, y: u32) -> (u32, bool);
     /// Performs checked `u64` subtraction.
+    #[cfg(stage0)]
     pub fn u64_sub_with_overflow(x: u64, y: u64) -> (u64, bool);
 
+    /// Performs checked integer subtraction
+    #[cfg(not(stage0))]
+    pub fn sub_with_overflow<T>(x: T, y: T) -> (T, bool);
+
     /// Performs checked `i8` multiplication.
+    #[cfg(stage0)]
     pub fn i8_mul_with_overflow(x: i8, y: i8) -> (i8, bool);
     /// Performs checked `i16` multiplication.
+    #[cfg(stage0)]
     pub fn i16_mul_with_overflow(x: i16, y: i16) -> (i16, bool);
     /// Performs checked `i32` multiplication.
+    #[cfg(stage0)]
     pub fn i32_mul_with_overflow(x: i32, y: i32) -> (i32, bool);
     /// Performs checked `i64` multiplication.
+    #[cfg(stage0)]
     pub fn i64_mul_with_overflow(x: i64, y: i64) -> (i64, bool);
 
     /// Performs checked `u8` multiplication.
+    #[cfg(stage0)]
     pub fn u8_mul_with_overflow(x: u8, y: u8) -> (u8, bool);
     /// Performs checked `u16` multiplication.
+    #[cfg(stage0)]
     pub fn u16_mul_with_overflow(x: u16, y: u16) -> (u16, bool);
     /// Performs checked `u32` multiplication.
+    #[cfg(stage0)]
     pub fn u32_mul_with_overflow(x: u32, y: u32) -> (u32, bool);
     /// Performs checked `u64` multiplication.
+    #[cfg(stage0)]
     pub fn u64_mul_with_overflow(x: u64, y: u64) -> (u64, bool);
 
-    /// Returns (a + b) mod 2^N, where N is the width of N in bits.
+    /// Performs checked integer multiplication
+    #[cfg(not(stage0))]
+    pub fn mul_with_overflow<T>(x: T, y: T) -> (T, bool);
+
+    /// Performs an unchecked division, resulting in undefined behavior
+    /// where y = 0 or x = `T::min_value()` and y = -1
+    #[cfg(not(stage0))]
+    pub fn unchecked_div<T>(x: T, y: T) -> T;
+    /// Returns the remainder of an unchecked division, resulting in
+    /// undefined behavior where y = 0 or x = `T::min_value()` and y = -1
+    #[cfg(not(stage0))]
+    pub fn unchecked_rem<T>(x: T, y: T) -> T;
+
+    /// Returns (a + b) mod 2^N, where N is the width of T in bits.
     pub fn overflowing_add<T>(a: T, b: T) -> T;
-    /// Returns (a - b) mod 2^N, where N is the width of N in bits.
+    /// Returns (a - b) mod 2^N, where N is the width of T in bits.
     pub fn overflowing_sub<T>(a: T, b: T) -> T;
-    /// Returns (a * b) mod 2^N, where N is the width of N in bits.
+    /// Returns (a * b) mod 2^N, where N is the width of T in bits.
     pub fn overflowing_mul<T>(a: T, b: T) -> T;
 
-    /// Performs an unchecked signed division, which results in undefined behavior,
-    /// in cases where y == 0, or x == isize::MIN and y == -1
-    pub fn unchecked_sdiv<T>(x: T, y: T) -> T;
-    /// Performs an unchecked unsigned division, which results in undefined behavior,
-    /// in cases where y == 0
-    pub fn unchecked_udiv<T>(x: T, y: T) -> T;
-
-    /// Returns the remainder of an unchecked signed division, which results in
-    /// undefined behavior, in cases where y == 0, or x == isize::MIN and y == -1
-    pub fn unchecked_srem<T>(x: T, y: T) -> T;
-    /// Returns the remainder of an unchecked unsigned division, which results in
-    /// undefined behavior, in cases where y == 0
-    pub fn unchecked_urem<T>(x: T, y: T) -> T;
-
     /// Returns the value of the discriminant for the variant in 'v',
     /// cast to a `u64`; if `T` has no discriminant, returns 0.
     pub fn discriminant_value<T>(v: &T) -> u64;
diff --git a/src/libcore/num/bignum.rs b/src/libcore/num/bignum.rs
index 18b34e24fcb..5d15ada4e75 100644
--- a/src/libcore/num/bignum.rs
+++ b/src/libcore/num/bignum.rs
@@ -55,6 +55,7 @@ macro_rules! impl_full_ops {
     ($($ty:ty: add($addfn:path), mul/div($bigty:ident);)*) => (
         $(
             impl FullOps for $ty {
+                #[cfg(stage0)]
                 fn full_add(self, other: $ty, carry: bool) -> (bool, $ty) {
                     // this cannot overflow, the output is between 0 and 2*2^nbits - 1
                     // FIXME will LLVM optimize this into ADC or similar???
@@ -62,6 +63,16 @@ macro_rules! impl_full_ops {
                     let (v, carry2) = unsafe { $addfn(v, if carry {1} else {0}) };
                     (carry1 || carry2, v)
                 }
+                #[cfg(not(stage0))]
+                fn full_add(self, other: $ty, carry: bool) -> (bool, $ty) {
+                    // this cannot overflow, the output is between 0 and 2*2^nbits - 1
+                    // FIXME will LLVM optimize this into ADC or similar???
+                    let (v, carry1) = unsafe { intrinsics::add_with_overflow(self, other) };
+                    let (v, carry2) = unsafe {
+                        intrinsics::add_with_overflow(v, if carry {1} else {0})
+                    };
+                    (carry1 || carry2, v)
+                }
 
                 fn full_mul(self, other: $ty, carry: $ty) -> ($ty, $ty) {
                     // this cannot overflow, the output is between 0 and 2^nbits * (2^nbits - 1)
diff --git a/src/libcore/num/mod.rs b/src/libcore/num/mod.rs
index 801315d86db..c397d02d245 100644
--- a/src/libcore/num/mod.rs
+++ b/src/libcore/num/mod.rs
@@ -103,6 +103,11 @@ macro_rules! zero_one_impl_float {
 }
 zero_one_impl_float! { f32 f64 }
 
+// Just for stage0; a byte swap on a byte is a no-op
+// Delete this once it becomes unused
+#[cfg(stage0)]
+unsafe fn bswap8(x: u8) -> u8 { x }
+
 macro_rules! checked_op {
     ($U:ty, $op:path, $x:expr, $y:expr) => {{
         let (result, overflowed) = unsafe { $op($x as $U, $y as $U) };
@@ -110,10 +115,6 @@ macro_rules! checked_op {
     }}
 }
 
-/// Swapping a single byte is a no-op. This is marked as `unsafe` for
-/// consistency with the other `bswap` intrinsics.
-unsafe fn bswap8(x: u8) -> u8 { x }
-
 // `Int` + `SignedInt` implemented for signed integers
 macro_rules! int_impl {
     ($ActualT:ty, $UnsignedT:ty, $BITS:expr,
@@ -611,54 +612,110 @@ macro_rules! int_impl {
 }
 
 #[lang = "i8"]
+#[cfg(stage0)]
 impl i8 {
     int_impl! { i8, u8, 8,
         intrinsics::i8_add_with_overflow,
         intrinsics::i8_sub_with_overflow,
         intrinsics::i8_mul_with_overflow }
 }
+#[lang = "i8"]
+#[cfg(not(stage0))]
+impl i8 {
+    int_impl! { i8, u8, 8,
+        intrinsics::add_with_overflow,
+        intrinsics::sub_with_overflow,
+        intrinsics::mul_with_overflow }
+}
 
 #[lang = "i16"]
+#[cfg(stage0)]
 impl i16 {
     int_impl! { i16, u16, 16,
         intrinsics::i16_add_with_overflow,
         intrinsics::i16_sub_with_overflow,
         intrinsics::i16_mul_with_overflow }
 }
+#[lang = "i16"]
+#[cfg(not(stage0))]
+impl i16 {
+    int_impl! { i16, u16, 16,
+        intrinsics::add_with_overflow,
+        intrinsics::sub_with_overflow,
+        intrinsics::mul_with_overflow }
+}
 
 #[lang = "i32"]
+#[cfg(stage0)]
 impl i32 {
     int_impl! { i32, u32, 32,
         intrinsics::i32_add_with_overflow,
         intrinsics::i32_sub_with_overflow,
         intrinsics::i32_mul_with_overflow }
 }
+#[lang = "i32"]
+#[cfg(not(stage0))]
+impl i32 {
+    int_impl! { i32, u32, 32,
+        intrinsics::add_with_overflow,
+        intrinsics::sub_with_overflow,
+        intrinsics::mul_with_overflow }
+}
 
 #[lang = "i64"]
+#[cfg(stage0)]
 impl i64 {
     int_impl! { i64, u64, 64,
         intrinsics::i64_add_with_overflow,
         intrinsics::i64_sub_with_overflow,
         intrinsics::i64_mul_with_overflow }
 }
+#[lang = "i64"]
+#[cfg(not(stage0))]
+impl i64 {
+    int_impl! { i64, u64, 64,
+        intrinsics::add_with_overflow,
+        intrinsics::sub_with_overflow,
+        intrinsics::mul_with_overflow }
+}
 
 #[cfg(target_pointer_width = "32")]
 #[lang = "isize"]
+#[cfg(stage0)]
 impl isize {
     int_impl! { i32, u32, 32,
         intrinsics::i32_add_with_overflow,
         intrinsics::i32_sub_with_overflow,
         intrinsics::i32_mul_with_overflow }
 }
+#[cfg(target_pointer_width = "32")]
+#[lang = "isize"]
+#[cfg(not(stage0))]
+impl isize {
+    int_impl! { i32, u32, 32,
+        intrinsics::add_with_overflow,
+        intrinsics::sub_with_overflow,
+        intrinsics::mul_with_overflow }
+}
 
 #[cfg(target_pointer_width = "64")]
 #[lang = "isize"]
+#[cfg(stage0)]
 impl isize {
     int_impl! { i64, u64, 64,
         intrinsics::i64_add_with_overflow,
         intrinsics::i64_sub_with_overflow,
         intrinsics::i64_mul_with_overflow }
 }
+#[cfg(target_pointer_width = "64")]
+#[lang = "isize"]
+#[cfg(not(stage0))]
+impl isize {
+    int_impl! { i64, u64, 64,
+        intrinsics::add_with_overflow,
+        intrinsics::sub_with_overflow,
+        intrinsics::mul_with_overflow }
+}
 
 // `Int` + `UnsignedInt` implemented for signed integers
 macro_rules! uint_impl {
@@ -744,6 +801,25 @@ macro_rules! uint_impl {
             unsafe { $ctlz(self as $ActualT) as u32 }
         }
 
+        #[stable(feature = "rust1", since = "1.0.0")]
+        #[cfg(stage0)]
+        #[inline]
+        pub fn trailing_zeros(self) -> u32 {
+            // As of LLVM 3.6 the codegen for the zero-safe cttz8 intrinsic
+            // emits two conditional moves on x86_64. By promoting the value to
+            // u16 and setting bit 8, we get better code without any conditional
+            // operations.
+            // FIXME: There's a LLVM patch (http://reviews.llvm.org/D9284)
+            // pending, remove this workaround once LLVM generates better code
+            // for cttz8.
+            unsafe {
+                if $BITS == 8 {
+                    intrinsics::cttz16(self as u16 | 0x100) as u32
+                } else {
+                    $cttz(self as $ActualT) as u32
+                }
+            }
+        }
         /// Returns the number of trailing zeros in the binary representation
         /// of `self`.
         ///
@@ -755,6 +831,7 @@ macro_rules! uint_impl {
         /// assert_eq!(n.trailing_zeros(), 3);
         /// ```
         #[stable(feature = "rust1", since = "1.0.0")]
+        #[cfg(not(stage0))]
         #[inline]
         pub fn trailing_zeros(self) -> u32 {
             // As of LLVM 3.6 the codegen for the zero-safe cttz8 intrinsic
@@ -766,9 +843,9 @@ macro_rules! uint_impl {
             // for cttz8.
             unsafe {
                 if $BITS == 8 {
-                    intrinsics::cttz16(self as u16 | 0x100) as u32
+                    intrinsics::cttz(self as u16 | 0x100) as u32
                 } else {
-                    $cttz(self as $ActualT) as u32
+                    intrinsics::cttz(self) as u32
                 }
             }
         }
@@ -1163,6 +1240,7 @@ macro_rules! uint_impl {
 }
 
 #[lang = "u8"]
+#[cfg(stage0)]
 impl u8 {
     uint_impl! { u8, 8,
         intrinsics::ctpop8,
@@ -1173,8 +1251,21 @@ impl u8 {
         intrinsics::u8_sub_with_overflow,
         intrinsics::u8_mul_with_overflow }
 }
+#[lang = "u8"]
+#[cfg(not(stage0))]
+impl u8 {
+    uint_impl! { u8, 8,
+        intrinsics::ctpop,
+        intrinsics::ctlz,
+        intrinsics::cttz,
+        intrinsics::bswap,
+        intrinsics::add_with_overflow,
+        intrinsics::sub_with_overflow,
+        intrinsics::mul_with_overflow }
+}
 
 #[lang = "u16"]
+#[cfg(stage0)]
 impl u16 {
     uint_impl! { u16, 16,
         intrinsics::ctpop16,
@@ -1185,8 +1276,21 @@ impl u16 {
         intrinsics::u16_sub_with_overflow,
         intrinsics::u16_mul_with_overflow }
 }
+#[lang = "u16"]
+#[cfg(not(stage0))]
+impl u16 {
+    uint_impl! { u16, 16,
+        intrinsics::ctpop,
+        intrinsics::ctlz,
+        intrinsics::cttz,
+        intrinsics::bswap,
+        intrinsics::add_with_overflow,
+        intrinsics::sub_with_overflow,
+        intrinsics::mul_with_overflow }
+}
 
 #[lang = "u32"]
+#[cfg(stage0)]
 impl u32 {
     uint_impl! { u32, 32,
         intrinsics::ctpop32,
@@ -1197,9 +1301,21 @@ impl u32 {
         intrinsics::u32_sub_with_overflow,
         intrinsics::u32_mul_with_overflow }
 }
-
+#[lang = "u32"]
+#[cfg(not(stage0))]
+impl u32 {
+    uint_impl! { u32, 32,
+        intrinsics::ctpop,
+        intrinsics::ctlz,
+        intrinsics::cttz,
+        intrinsics::bswap,
+        intrinsics::add_with_overflow,
+        intrinsics::sub_with_overflow,
+        intrinsics::mul_with_overflow }
+}
 
 #[lang = "u64"]
+#[cfg(stage0)]
 impl u64 {
     uint_impl! { u64, 64,
         intrinsics::ctpop64,
@@ -1210,9 +1326,22 @@ impl u64 {
         intrinsics::u64_sub_with_overflow,
         intrinsics::u64_mul_with_overflow }
 }
+#[lang = "u64"]
+#[cfg(not(stage0))]
+impl u64 {
+    uint_impl! { u64, 64,
+        intrinsics::ctpop,
+        intrinsics::ctlz,
+        intrinsics::cttz,
+        intrinsics::bswap,
+        intrinsics::add_with_overflow,
+        intrinsics::sub_with_overflow,
+        intrinsics::mul_with_overflow }
+}
 
 #[cfg(target_pointer_width = "32")]
 #[lang = "usize"]
+#[cfg(stage0)]
 impl usize {
     uint_impl! { u32, 32,
         intrinsics::ctpop32,
@@ -1223,9 +1352,23 @@ impl usize {
         intrinsics::u32_sub_with_overflow,
         intrinsics::u32_mul_with_overflow }
 }
+#[cfg(target_pointer_width = "32")]
+#[lang = "usize"]
+#[cfg(not(stage0))]
+impl usize {
+    uint_impl! { u32, 32,
+        intrinsics::ctpop,
+        intrinsics::ctlz,
+        intrinsics::cttz,
+        intrinsics::bswap,
+        intrinsics::add_with_overflow,
+        intrinsics::sub_with_overflow,
+        intrinsics::mul_with_overflow }
+}
 
 #[cfg(target_pointer_width = "64")]
 #[lang = "usize"]
+#[cfg(stage0)]
 impl usize {
     uint_impl! { u64, 64,
         intrinsics::ctpop64,
@@ -1236,6 +1379,19 @@ impl usize {
         intrinsics::u64_sub_with_overflow,
         intrinsics::u64_mul_with_overflow }
 }
+#[cfg(target_pointer_width = "64")]
+#[lang = "usize"]
+#[cfg(not(stage0))]
+impl usize {
+    uint_impl! { u64, 64,
+        intrinsics::ctpop,
+        intrinsics::ctlz,
+        intrinsics::cttz,
+        intrinsics::bswap,
+        intrinsics::add_with_overflow,
+        intrinsics::sub_with_overflow,
+        intrinsics::mul_with_overflow }
+}
 
 /// Used for representing the classification of floating point numbers
 #[derive(Copy, Clone, PartialEq, Debug)]
diff --git a/src/libcore/num/wrapping.rs b/src/libcore/num/wrapping.rs
index 0e8ced0aa19..88f71e63da6 100644
--- a/src/libcore/num/wrapping.rs
+++ b/src/libcore/num/wrapping.rs
@@ -12,23 +12,35 @@
 #![unstable(feature = "wrapping", reason = "may be removed or relocated",
             issue = "27755")]
 
+#[cfg(stage0)]
+pub use intrinsics::{
+    u8_add_with_overflow, i8_add_with_overflow,
+    u16_add_with_overflow, i16_add_with_overflow,
+    u32_add_with_overflow, i32_add_with_overflow,
+    u64_add_with_overflow, i64_add_with_overflow,
+
+    u8_sub_with_overflow, i8_sub_with_overflow,
+    u16_sub_with_overflow, i16_sub_with_overflow,
+    u32_sub_with_overflow, i32_sub_with_overflow,
+    u64_sub_with_overflow, i64_sub_with_overflow,
+
+    u8_mul_with_overflow, i8_mul_with_overflow,
+    u16_mul_with_overflow, i16_mul_with_overflow,
+    u32_mul_with_overflow, i32_mul_with_overflow,
+    u64_mul_with_overflow, i64_mul_with_overflow,
+};
+
+#[cfg(not(stage0))]
+pub use intrinsics::{
+    add_with_overflow,
+    sub_with_overflow,
+    mul_with_overflow,
+};
+
 use super::Wrapping;
 
 use ops::*;
 
-use intrinsics::{i8_add_with_overflow, u8_add_with_overflow};
-use intrinsics::{i16_add_with_overflow, u16_add_with_overflow};
-use intrinsics::{i32_add_with_overflow, u32_add_with_overflow};
-use intrinsics::{i64_add_with_overflow, u64_add_with_overflow};
-use intrinsics::{i8_sub_with_overflow, u8_sub_with_overflow};
-use intrinsics::{i16_sub_with_overflow, u16_sub_with_overflow};
-use intrinsics::{i32_sub_with_overflow, u32_sub_with_overflow};
-use intrinsics::{i64_sub_with_overflow, u64_sub_with_overflow};
-use intrinsics::{i8_mul_with_overflow, u8_mul_with_overflow};
-use intrinsics::{i16_mul_with_overflow, u16_mul_with_overflow};
-use intrinsics::{i32_mul_with_overflow, u32_mul_with_overflow};
-use intrinsics::{i64_mul_with_overflow, u64_mul_with_overflow};
-
 use ::{i8,i16,i32,i64};
 
 pub trait OverflowingOps {
@@ -191,23 +203,47 @@ macro_rules! signed_overflowing_impl {
     ($($t:ident)*) => ($(
         impl OverflowingOps for $t {
             #[inline(always)]
+            #[cfg(stage0)]
             fn overflowing_add(self, rhs: $t) -> ($t, bool) {
                 unsafe {
                     concat_idents!($t, _add_with_overflow)(self, rhs)
                 }
             }
             #[inline(always)]
+            #[cfg(not(stage0))]
+            fn overflowing_add(self, rhs: $t) -> ($t, bool) {
+                unsafe {
+                    add_with_overflow(self, rhs)
+                }
+            }
+            #[inline(always)]
+            #[cfg(stage0)]
             fn overflowing_sub(self, rhs: $t) -> ($t, bool) {
                 unsafe {
                     concat_idents!($t, _sub_with_overflow)(self, rhs)
                 }
             }
             #[inline(always)]
+            #[cfg(not(stage0))]
+            fn overflowing_sub(self, rhs: $t) -> ($t, bool) {
+                unsafe {
+                    sub_with_overflow(self, rhs)
+                }
+            }
+            #[inline(always)]
+            #[cfg(stage0)]
             fn overflowing_mul(self, rhs: $t) -> ($t, bool) {
                 unsafe {
                     concat_idents!($t, _mul_with_overflow)(self, rhs)
                 }
             }
+            #[inline(always)]
+            #[cfg(not(stage0))]
+            fn overflowing_mul(self, rhs: $t) -> ($t, bool) {
+                unsafe {
+                    mul_with_overflow(self, rhs)
+                }
+            }
 
             #[inline(always)]
             fn overflowing_div(self, rhs: $t) -> ($t, bool) {
@@ -253,23 +289,47 @@ macro_rules! unsigned_overflowing_impl {
     ($($t:ident)*) => ($(
         impl OverflowingOps for $t {
             #[inline(always)]
+            #[cfg(stage0)]
             fn overflowing_add(self, rhs: $t) -> ($t, bool) {
                 unsafe {
                     concat_idents!($t, _add_with_overflow)(self, rhs)
                 }
             }
             #[inline(always)]
+            #[cfg(not(stage0))]
+            fn overflowing_add(self, rhs: $t) -> ($t, bool) {
+                unsafe {
+                    add_with_overflow(self, rhs)
+                }
+            }
+            #[inline(always)]
+            #[cfg(stage0)]
             fn overflowing_sub(self, rhs: $t) -> ($t, bool) {
                 unsafe {
                     concat_idents!($t, _sub_with_overflow)(self, rhs)
                 }
             }
             #[inline(always)]
+            #[cfg(not(stage0))]
+            fn overflowing_sub(self, rhs: $t) -> ($t, bool) {
+                unsafe {
+                    sub_with_overflow(self, rhs)
+                }
+            }
+            #[inline(always)]
+            #[cfg(stage0)]
             fn overflowing_mul(self, rhs: $t) -> ($t, bool) {
                 unsafe {
                     concat_idents!($t, _mul_with_overflow)(self, rhs)
                 }
             }
+            #[inline(always)]
+            #[cfg(not(stage0))]
+            fn overflowing_mul(self, rhs: $t) -> ($t, bool) {
+                unsafe {
+                    mul_with_overflow(self, rhs)
+                }
+            }
 
             #[inline(always)]
             fn overflowing_div(self, rhs: $t) -> ($t, bool) {
@@ -305,6 +365,7 @@ unsigned_overflowing_impl! { u8 u16 u32 u64 }
 #[cfg(target_pointer_width = "64")]
 impl OverflowingOps for usize {
     #[inline(always)]
+    #[cfg(stage0)]
     fn overflowing_add(self, rhs: usize) -> (usize, bool) {
         unsafe {
             let res = u64_add_with_overflow(self as u64, rhs as u64);
@@ -312,6 +373,14 @@ impl OverflowingOps for usize {
         }
     }
     #[inline(always)]
+    #[cfg(not(stage0))]
+    fn overflowing_add(self, rhs: usize) -> (usize, bool) {
+        unsafe {
+            add_with_overflow(self, rhs)
+        }
+    }
+    #[inline(always)]
+    #[cfg(stage0)]
     fn overflowing_sub(self, rhs: usize) -> (usize, bool) {
         unsafe {
             let res = u64_sub_with_overflow(self as u64, rhs as u64);
@@ -319,6 +388,14 @@ impl OverflowingOps for usize {
         }
     }
     #[inline(always)]
+    #[cfg(not(stage0))]
+    fn overflowing_sub(self, rhs: usize) -> (usize, bool) {
+        unsafe {
+            sub_with_overflow(self, rhs)
+        }
+    }
+    #[inline(always)]
+    #[cfg(stage0)]
     fn overflowing_mul(self, rhs: usize) -> (usize, bool) {
         unsafe {
             let res = u64_mul_with_overflow(self as u64, rhs as u64);
@@ -326,6 +403,13 @@ impl OverflowingOps for usize {
         }
     }
     #[inline(always)]
+    #[cfg(not(stage0))]
+    fn overflowing_mul(self, rhs: usize) -> (usize, bool) {
+        unsafe {
+            mul_with_overflow(self, rhs)
+        }
+    }
+    #[inline(always)]
     fn overflowing_div(self, rhs: usize) -> (usize, bool) {
         let (r, f) = (self as u64).overflowing_div(rhs as u64);
         (r as usize, f)
@@ -355,6 +439,7 @@ impl OverflowingOps for usize {
 #[cfg(target_pointer_width = "32")]
 impl OverflowingOps for usize {
     #[inline(always)]
+    #[cfg(stage0)]
     fn overflowing_add(self, rhs: usize) -> (usize, bool) {
         unsafe {
             let res = u32_add_with_overflow(self as u32, rhs as u32);
@@ -362,6 +447,14 @@ impl OverflowingOps for usize {
         }
     }
     #[inline(always)]
+    #[cfg(not(stage0))]
+    fn overflowing_add(self, rhs: usize) -> (usize, bool) {
+        unsafe {
+            add_with_overflow(self, rhs)
+        }
+    }
+    #[inline(always)]
+    #[cfg(stage0)]
     fn overflowing_sub(self, rhs: usize) -> (usize, bool) {
         unsafe {
             let res = u32_sub_with_overflow(self as u32, rhs as u32);
@@ -369,6 +462,14 @@ impl OverflowingOps for usize {
         }
     }
     #[inline(always)]
+    #[cfg(not(stage0))]
+    fn overflowing_sub(self, rhs: usize) -> (usize, bool) {
+        unsafe {
+            sub_with_overflow(self, rhs)
+        }
+    }
+    #[inline(always)]
+    #[cfg(stage0)]
     fn overflowing_mul(self, rhs: usize) -> (usize, bool) {
         unsafe {
             let res = u32_mul_with_overflow(self as u32, rhs as u32);
@@ -376,6 +477,13 @@ impl OverflowingOps for usize {
         }
     }
     #[inline(always)]
+    #[cfg(not(stage0))]
+    fn overflowing_mul(self, rhs: usize) -> (usize, bool) {
+        unsafe {
+            mul_with_overflow(self, rhs)
+        }
+    }
+    #[inline(always)]
     fn overflowing_div(self, rhs: usize) -> (usize, bool) {
         let (r, f) = (self as u32).overflowing_div(rhs as u32);
         (r as usize, f)
@@ -405,6 +513,7 @@ impl OverflowingOps for usize {
 #[cfg(target_pointer_width = "64")]
 impl OverflowingOps for isize {
     #[inline(always)]
+    #[cfg(stage0)]
     fn overflowing_add(self, rhs: isize) -> (isize, bool) {
         unsafe {
             let res = i64_add_with_overflow(self as i64, rhs as i64);
@@ -412,6 +521,14 @@ impl OverflowingOps for isize {
         }
     }
     #[inline(always)]
+    #[cfg(not(stage0))]
+    fn overflowing_add(self, rhs: isize) -> (isize, bool) {
+        unsafe {
+            add_with_overflow(self, rhs)
+        }
+    }
+    #[inline(always)]
+    #[cfg(stage0)]
     fn overflowing_sub(self, rhs: isize) -> (isize, bool) {
         unsafe {
             let res = i64_sub_with_overflow(self as i64, rhs as i64);
@@ -419,6 +536,14 @@ impl OverflowingOps for isize {
         }
     }
     #[inline(always)]
+    #[cfg(not(stage0))]
+    fn overflowing_sub(self, rhs: isize) -> (isize, bool) {
+        unsafe {
+            sub_with_overflow(self, rhs)
+        }
+    }
+    #[inline(always)]
+    #[cfg(stage0)]
     fn overflowing_mul(self, rhs: isize) -> (isize, bool) {
         unsafe {
             let res = i64_mul_with_overflow(self as i64, rhs as i64);
@@ -426,6 +551,13 @@ impl OverflowingOps for isize {
         }
     }
     #[inline(always)]
+    #[cfg(not(stage0))]
+    fn overflowing_mul(self, rhs: isize) -> (isize, bool) {
+        unsafe {
+            mul_with_overflow(self, rhs)
+        }
+    }
+    #[inline(always)]
     fn overflowing_div(self, rhs: isize) -> (isize, bool) {
         let (r, f) = (self as i64).overflowing_div(rhs as i64);
         (r as isize, f)
@@ -455,6 +587,7 @@ impl OverflowingOps for isize {
 #[cfg(target_pointer_width = "32")]
 impl OverflowingOps for isize {
     #[inline(always)]
+    #[cfg(stage0)]
     fn overflowing_add(self, rhs: isize) -> (isize, bool) {
         unsafe {
             let res = i32_add_with_overflow(self as i32, rhs as i32);
@@ -462,6 +595,14 @@ impl OverflowingOps for isize {
         }
     }
     #[inline(always)]
+    #[cfg(not(stage0))]
+    fn overflowing_add(self, rhs: isize) -> (isize, bool) {
+        unsafe {
+            add_with_overflow(self, rhs)
+        }
+    }
+    #[inline(always)]
+    #[cfg(stage0)]
     fn overflowing_sub(self, rhs: isize) -> (isize, bool) {
         unsafe {
             let res = i32_sub_with_overflow(self as i32, rhs as i32);
@@ -469,6 +610,14 @@ impl OverflowingOps for isize {
         }
     }
     #[inline(always)]
+    #[cfg(not(stage0))]
+    fn overflowing_sub(self, rhs: isize) -> (isize, bool) {
+        unsafe {
+            sub_with_overflow(self, rhs)
+        }
+    }
+    #[inline(always)]
+    #[cfg(stage0)]
     fn overflowing_mul(self, rhs: isize) -> (isize, bool) {
         unsafe {
             let res = i32_mul_with_overflow(self as i32, rhs as i32);
@@ -476,6 +625,13 @@ impl OverflowingOps for isize {
         }
     }
     #[inline(always)]
+    #[cfg(not(stage0))]
+    fn overflowing_mul(self, rhs: isize) -> (isize, bool) {
+        unsafe {
+            mul_with_overflow(self, rhs)
+        }
+    }
+    #[inline(always)]
     fn overflowing_div(self, rhs: isize) -> (isize, bool) {
         let (r, f) = (self as i32).overflowing_div(rhs as i32);
         (r as isize, f)
diff --git a/src/librustc_trans/diagnostics.rs b/src/librustc_trans/diagnostics.rs
index 2ad2e7528e4..539b9a4171f 100644
--- a/src/librustc_trans/diagnostics.rs
+++ b/src/librustc_trans/diagnostics.rs
@@ -79,12 +79,10 @@ Transmute with two differently sized types was attempted. Erroneous code
 example:
 
 ```
-extern "rust-intrinsic" {
-    pub fn ctpop8(x: u8) -> u8;
-}
+fn takes_u8(_: u8) {}
 
 fn main() {
-    unsafe { ctpop8(::std::mem::transmute(0u16)); }
+    unsafe { takes_u8(::std::mem::transmute(0u16)); }
     // error: transmute called with differently sized types
 }
 ```
@@ -92,14 +90,12 @@ fn main() {
 Please use types with same size or use the expected type directly. Example:
 
 ```
-extern "rust-intrinsic" {
-    pub fn ctpop8(x: u8) -> u8;
-}
+fn takes_u8(_: u8) {}
 
 fn main() {
-    unsafe { ctpop8(::std::mem::transmute(0i8)); } // ok!
+    unsafe { takes_u8(::std::mem::transmute(0i8)); } // ok!
     // or:
-    unsafe { ctpop8(0u8); } // ok!
+    unsafe { takes_u8(0u8); } // ok!
 }
 ```
 "##,
@@ -118,5 +114,4 @@ Example:
 let x = &[0, 1, 2][2]; // ok
 ```
 "##,
-
 }
diff --git a/src/librustc_trans/trans/intrinsic.rs b/src/librustc_trans/trans/intrinsic.rs
index a8d85ae17e2..72ee53b0c80 100644
--- a/src/librustc_trans/trans/intrinsic.rs
+++ b/src/librustc_trans/trans/intrinsic.rs
@@ -89,13 +89,6 @@ pub fn get_simple_intrinsic(ccx: &CrateContext, item: &hir::ForeignItem) -> Opti
         "nearbyintf64" => "llvm.nearbyint.f64",
         "roundf32" => "llvm.round.f32",
         "roundf64" => "llvm.round.f64",
-        "ctpop8" => "llvm.ctpop.i8",
-        "ctpop16" => "llvm.ctpop.i16",
-        "ctpop32" => "llvm.ctpop.i32",
-        "ctpop64" => "llvm.ctpop.i64",
-        "bswap16" => "llvm.bswap.i16",
-        "bswap32" => "llvm.bswap.i32",
-        "bswap64" => "llvm.bswap.i64",
         "assume" => "llvm.assume",
         _ => return None
     };
@@ -589,217 +582,63 @@ pub fn trans_intrinsic_call<'a, 'blk, 'tcx>(mut bcx: Block<'blk, 'tcx>,
             C_nil(ccx)
         },
 
-        (_, "ctlz8") => count_zeros_intrinsic(bcx,
-                                              "llvm.ctlz.i8",
-                                              llargs[0],
-                                              call_debug_location),
-        (_, "ctlz16") => count_zeros_intrinsic(bcx,
-                                               "llvm.ctlz.i16",
-                                               llargs[0],
-                                               call_debug_location),
-        (_, "ctlz32") => count_zeros_intrinsic(bcx,
-                                               "llvm.ctlz.i32",
-                                               llargs[0],
-                                               call_debug_location),
-        (_, "ctlz64") => count_zeros_intrinsic(bcx,
-                                               "llvm.ctlz.i64",
-                                               llargs[0],
-                                               call_debug_location),
-        (_, "cttz8") => count_zeros_intrinsic(bcx,
-                                              "llvm.cttz.i8",
-                                              llargs[0],
-                                              call_debug_location),
-        (_, "cttz16") => count_zeros_intrinsic(bcx,
-                                               "llvm.cttz.i16",
-                                               llargs[0],
-                                               call_debug_location),
-        (_, "cttz32") => count_zeros_intrinsic(bcx,
-                                               "llvm.cttz.i32",
-                                               llargs[0],
-                                               call_debug_location),
-        (_, "cttz64") => count_zeros_intrinsic(bcx,
-                                               "llvm.cttz.i64",
-                                               llargs[0],
-                                               call_debug_location),
-
-        (_, "i8_add_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.sadd.with.overflow.i8",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "i16_add_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.sadd.with.overflow.i16",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "i32_add_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.sadd.with.overflow.i32",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "i64_add_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.sadd.with.overflow.i64",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-
-        (_, "u8_add_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.uadd.with.overflow.i8",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "u16_add_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.uadd.with.overflow.i16",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "u32_add_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.uadd.with.overflow.i32",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "u64_add_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.uadd.with.overflow.i64",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "i8_sub_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.ssub.with.overflow.i8",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "i16_sub_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.ssub.with.overflow.i16",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "i32_sub_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.ssub.with.overflow.i32",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "i64_sub_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.ssub.with.overflow.i64",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "u8_sub_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.usub.with.overflow.i8",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "u16_sub_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.usub.with.overflow.i16",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "u32_sub_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.usub.with.overflow.i32",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "u64_sub_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.usub.with.overflow.i64",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "i8_mul_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.smul.with.overflow.i8",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "i16_mul_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.smul.with.overflow.i16",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "i32_mul_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.smul.with.overflow.i32",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "i64_mul_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.smul.with.overflow.i64",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "u8_mul_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.umul.with.overflow.i8",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "u16_mul_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.umul.with.overflow.i16",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "u32_mul_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.umul.with.overflow.i32",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-        (_, "u64_mul_with_overflow") =>
-            with_overflow_intrinsic(bcx,
-                                    "llvm.umul.with.overflow.i64",
-                                    llargs[0],
-                                    llargs[1],
-                                    llresult,
-                                    call_debug_location),
-
-        (_, "unchecked_udiv") => UDiv(bcx, llargs[0], llargs[1], call_debug_location),
-        (_, "unchecked_sdiv") => SDiv(bcx, llargs[0], llargs[1], call_debug_location),
-        (_, "unchecked_urem") => URem(bcx, llargs[0], llargs[1], call_debug_location),
-        (_, "unchecked_srem") => SRem(bcx, llargs[0], llargs[1], call_debug_location),
-
-        (_, "overflowing_add") => Add(bcx, llargs[0], llargs[1], call_debug_location),
-        (_, "overflowing_sub") => Sub(bcx, llargs[0], llargs[1], call_debug_location),
-        (_, "overflowing_mul") => Mul(bcx, llargs[0], llargs[1], call_debug_location),
+        (_, "ctlz") | (_, "cttz") | (_, "ctpop") | (_, "bswap") |
+        (_, "add_with_overflow") | (_, "sub_with_overflow") | (_, "mul_with_overflow") |
+        (_, "overflowing_add") | (_, "overflowing_sub") | (_, "overflowing_mul") |
+        (_, "unchecked_div") | (_, "unchecked_rem") => {
+            let sty = &arg_tys[0].sty;
+            match int_type_width_signed(sty, ccx) {
+                Some((width, signed)) =>
+                    match &*name {
+                        "ctlz" => count_zeros_intrinsic(bcx, &format!("llvm.ctlz.i{}", width),
+                                                        llargs[0], call_debug_location),
+                        "cttz" => count_zeros_intrinsic(bcx, &format!("llvm.cttz.i{}", width),
+                                                        llargs[0], call_debug_location),
+                        "ctpop" => Call(bcx, ccx.get_intrinsic(&format!("llvm.ctpop.i{}", width)),
+                                        &llargs, None, call_debug_location),
+                        "bswap" => {
+                            if width == 8 {
+                                llargs[0] // byte swap a u8/i8 is just a no-op
+                            } else {
+                                Call(bcx, ccx.get_intrinsic(&format!("llvm.bswap.i{}", width)),
+                                        &llargs, None, call_debug_location)
+                            }
+                        }
+                        "add_with_overflow" | "sub_with_overflow" | "mul_with_overflow" => {
+                            let intrinsic = format!("llvm.{}{}.with.overflow.i{}",
+                                                    if signed { 's' } else { 'u' },
+                                                    &name[..3], width);
+                            with_overflow_intrinsic(bcx, &intrinsic, llargs[0], llargs[1], llresult,
+                                                    call_debug_location)
+                        },
+                        "overflowing_add" => Add(bcx, llargs[0], llargs[1], call_debug_location),
+                        "overflowing_sub" => Sub(bcx, llargs[0], llargs[1], call_debug_location),
+                        "overflowing_mul" => Mul(bcx, llargs[0], llargs[1], call_debug_location),
+                        "unchecked_div" =>
+                            if signed {
+                                SDiv(bcx, llargs[0], llargs[1], call_debug_location)
+                            } else {
+                                UDiv(bcx, llargs[0], llargs[1], call_debug_location)
+                            },
+                        "unchecked_rem" =>
+                            if signed {
+                                SRem(bcx, llargs[0], llargs[1], call_debug_location)
+                            } else {
+                                URem(bcx, llargs[0], llargs[1], call_debug_location)
+                            },
+                        _ => unreachable!(),
+                    },
+                None => {
+                    span_invalid_monomorphization_error(
+                        tcx.sess, call_info.span,
+                        &format!("invalid monomorphization of `{}` intrinsic: \
+                                  expected basic integer type, found `{}`", name, sty));
+                        C_null(llret_ty)
+                }
+            }
+
+        },
+
 
         (_, "return_address") => {
             if !fcx.caller_expects_out_pointer {
@@ -1174,7 +1013,7 @@ fn memset_intrinsic<'blk, 'tcx>(bcx: Block<'blk, 'tcx>,
 }
 
 fn count_zeros_intrinsic(bcx: Block,
-                         name: &'static str,
+                         name: &str,
                          val: ValueRef,
                          call_debug_location: DebugLoc)
                          -> ValueRef {
@@ -1184,7 +1023,7 @@ fn count_zeros_intrinsic(bcx: Block,
 }
 
 fn with_overflow_intrinsic<'blk, 'tcx>(bcx: Block<'blk, 'tcx>,
-                                       name: &'static str,
+                                       name: &str,
                                        a: ValueRef,
                                        b: ValueRef,
                                        out: ValueRef,
@@ -1716,3 +1555,39 @@ fn generic_simd_intrinsic<'blk, 'tcx, 'a>
     }
     bcx.sess().span_bug(call_info.span, "unknown SIMD intrinsic");
 }
+
+// Returns the width of an int TypeVariant, and if it's signed or not
+// Returns None if the type is not an integer
+fn int_type_width_signed<'tcx>(sty: &ty::TypeVariants<'tcx>, ccx: &CrateContext)
+        -> Option<(u64, bool)> {
+    use rustc::middle::ty::{TyInt, TyUint};
+    match *sty {
+        TyInt(t) => Some((match t {
+            ast::TyIs => {
+                match &ccx.tcx().sess.target.target.target_pointer_width[..] {
+                    "32" => 32,
+                    "64" => 64,
+                    tws => panic!("Unsupported target word size for isize: {}", tws),
+                }
+            },
+            ast::TyI8 => 8,
+            ast::TyI16 => 16,
+            ast::TyI32 => 32,
+            ast::TyI64 => 64,
+        }, true)),
+        TyUint(t) => Some((match t {
+            ast::TyUs => {
+                match &ccx.tcx().sess.target.target.target_pointer_width[..] {
+                    "32" => 32,
+                    "64" => 64,
+                    tws => panic!("Unsupported target word size for usize: {}", tws),
+                }
+            },
+            ast::TyU8 => 8,
+            ast::TyU16 => 16,
+            ast::TyU32 => 32,
+            ast::TyU64 => 64,
+        }, false)),
+        _ => None,
+    }
+}
diff --git a/src/librustc_typeck/check/intrinsic.rs b/src/librustc_typeck/check/intrinsic.rs
index 4afc610f63e..0ab0f1f9c1d 100644
--- a/src/librustc_typeck/check/intrinsic.rs
+++ b/src/librustc_typeck/check/intrinsic.rs
@@ -254,60 +254,19 @@ pub fn check_intrinsic_type(ccx: &CrateCtxt, it: &hir::ForeignItem) {
             "nearbyintf64" => (0, vec!( tcx.types.f64 ), tcx.types.f64),
             "roundf32"     => (0, vec!( tcx.types.f32 ), tcx.types.f32),
             "roundf64"     => (0, vec!( tcx.types.f64 ), tcx.types.f64),
-            "ctpop8"       => (0, vec!( tcx.types.u8  ), tcx.types.u8),
-            "ctpop16"      => (0, vec!( tcx.types.u16 ), tcx.types.u16),
-            "ctpop32"      => (0, vec!( tcx.types.u32 ), tcx.types.u32),
-            "ctpop64"      => (0, vec!( tcx.types.u64 ), tcx.types.u64),
-            "ctlz8"        => (0, vec!( tcx.types.u8  ), tcx.types.u8),
-            "ctlz16"       => (0, vec!( tcx.types.u16 ), tcx.types.u16),
-            "ctlz32"       => (0, vec!( tcx.types.u32 ), tcx.types.u32),
-            "ctlz64"       => (0, vec!( tcx.types.u64 ), tcx.types.u64),
-            "cttz8"        => (0, vec!( tcx.types.u8  ), tcx.types.u8),
-            "cttz16"       => (0, vec!( tcx.types.u16 ), tcx.types.u16),
-            "cttz32"       => (0, vec!( tcx.types.u32 ), tcx.types.u32),
-            "cttz64"       => (0, vec!( tcx.types.u64 ), tcx.types.u64),
-            "bswap16"      => (0, vec!( tcx.types.u16 ), tcx.types.u16),
-            "bswap32"      => (0, vec!( tcx.types.u32 ), tcx.types.u32),
-            "bswap64"      => (0, vec!( tcx.types.u64 ), tcx.types.u64),
 
             "volatile_load" =>
                 (1, vec!( tcx.mk_imm_ptr(param(ccx, 0)) ), param(ccx, 0)),
             "volatile_store" =>
                 (1, vec!( tcx.mk_mut_ptr(param(ccx, 0)), param(ccx, 0) ), tcx.mk_nil()),
 
-            "i8_add_with_overflow" | "i8_sub_with_overflow" | "i8_mul_with_overflow" =>
-                (0, vec!(tcx.types.i8, tcx.types.i8),
-                tcx.mk_tup(vec!(tcx.types.i8, tcx.types.bool))),
+            "ctpop" | "ctlz" | "cttz" | "bswap" => (1, vec!(param(ccx, 0)), param(ccx, 0)),
 
-            "i16_add_with_overflow" | "i16_sub_with_overflow" | "i16_mul_with_overflow" =>
-                (0, vec!(tcx.types.i16, tcx.types.i16),
-                tcx.mk_tup(vec!(tcx.types.i16, tcx.types.bool))),
+            "add_with_overflow" | "sub_with_overflow"  | "mul_with_overflow" =>
+                (1, vec!(param(ccx, 0), param(ccx, 0)),
+                tcx.mk_tup(vec!(param(ccx, 0), tcx.types.bool))),
 
-            "i32_add_with_overflow" | "i32_sub_with_overflow" | "i32_mul_with_overflow" =>
-                (0, vec!(tcx.types.i32, tcx.types.i32),
-                tcx.mk_tup(vec!(tcx.types.i32, tcx.types.bool))),
-
-            "i64_add_with_overflow" | "i64_sub_with_overflow" | "i64_mul_with_overflow" =>
-                (0, vec!(tcx.types.i64, tcx.types.i64),
-                tcx.mk_tup(vec!(tcx.types.i64, tcx.types.bool))),
-
-            "u8_add_with_overflow" | "u8_sub_with_overflow" | "u8_mul_with_overflow" =>
-                (0, vec!(tcx.types.u8, tcx.types.u8),
-                tcx.mk_tup(vec!(tcx.types.u8, tcx.types.bool))),
-
-            "u16_add_with_overflow" | "u16_sub_with_overflow" | "u16_mul_with_overflow" =>
-                (0, vec!(tcx.types.u16, tcx.types.u16),
-                tcx.mk_tup(vec!(tcx.types.u16, tcx.types.bool))),
-
-            "u32_add_with_overflow" | "u32_sub_with_overflow" | "u32_mul_with_overflow"=>
-                (0, vec!(tcx.types.u32, tcx.types.u32),
-                tcx.mk_tup(vec!(tcx.types.u32, tcx.types.bool))),
-
-            "u64_add_with_overflow" | "u64_sub_with_overflow"  | "u64_mul_with_overflow" =>
-                (0, vec!(tcx.types.u64, tcx.types.u64),
-                tcx.mk_tup(vec!(tcx.types.u64, tcx.types.bool))),
-
-            "unchecked_udiv" | "unchecked_sdiv" | "unchecked_urem" | "unchecked_srem" =>
+            "unchecked_div" | "unchecked_rem" =>
                 (1, vec![param(ccx, 0), param(ccx, 0)], param(ccx, 0)),
 
             "overflowing_add" | "overflowing_sub" | "overflowing_mul" =>
diff --git a/src/test/run-pass/intrinsics-integer.rs b/src/test/run-pass/intrinsics-integer.rs
index 8dbe927f06b..170a6c95aa8 100644
--- a/src/test/run-pass/intrinsics-integer.rs
+++ b/src/test/run-pass/intrinsics-integer.rs
@@ -14,24 +14,10 @@
 
 mod rusti {
     extern "rust-intrinsic" {
-        pub fn ctpop8(x: u8) -> u8;
-        pub fn ctpop16(x: u16) -> u16;
-        pub fn ctpop32(x: u32) -> u32;
-        pub fn ctpop64(x: u64) -> u64;
-
-        pub fn ctlz8(x: u8) -> u8;
-        pub fn ctlz16(x: u16) -> u16;
-        pub fn ctlz32(x: u32) -> u32;
-        pub fn ctlz64(x: u64) -> u64;
-
-        pub fn cttz8(x: u8) -> u8;
-        pub fn cttz16(x: u16) -> u16;
-        pub fn cttz32(x: u32) -> u32;
-        pub fn cttz64(x: u64) -> u64;
-
-        pub fn bswap16(x: u16) -> u16;
-        pub fn bswap32(x: u32) -> u32;
-        pub fn bswap64(x: u64) -> u64;
+        pub fn ctpop<T>(x: T) -> T;
+        pub fn ctlz<T>(x: T) -> T;
+        pub fn cttz<T>(x: T) -> T;
+        pub fn bswap<T>(x: T) -> T;
     }
 }
 
@@ -39,78 +25,83 @@ pub fn main() {
     unsafe {
         use rusti::*;
 
-        assert_eq!(ctpop8(0), 0);
-        assert_eq!(ctpop16(0), 0);
-        assert_eq!(ctpop32(0), 0);
-        assert_eq!(ctpop64(0), 0);
-
-        assert_eq!(ctpop8(1), 1);
-        assert_eq!(ctpop16(1), 1);
-        assert_eq!(ctpop32(1), 1);
-        assert_eq!(ctpop64(1), 1);
-
-        assert_eq!(ctpop8(10), 2);
-        assert_eq!(ctpop16(10), 2);
-        assert_eq!(ctpop32(10), 2);
-        assert_eq!(ctpop64(10), 2);
-
-        assert_eq!(ctpop8(100), 3);
-        assert_eq!(ctpop16(100), 3);
-        assert_eq!(ctpop32(100), 3);
-        assert_eq!(ctpop64(100), 3);
-
-        assert_eq!(ctpop8(-1), 8);
-        assert_eq!(ctpop16(-1), 16);
-        assert_eq!(ctpop32(-1), 32);
-        assert_eq!(ctpop64(-1), 64);
-
-        assert_eq!(ctlz8(0), 8);
-        assert_eq!(ctlz16(0), 16);
-        assert_eq!(ctlz32(0), 32);
-        assert_eq!(ctlz64(0), 64);
-
-        assert_eq!(ctlz8(1), 7);
-        assert_eq!(ctlz16(1), 15);
-        assert_eq!(ctlz32(1), 31);
-        assert_eq!(ctlz64(1), 63);
-
-        assert_eq!(ctlz8(10), 4);
-        assert_eq!(ctlz16(10), 12);
-        assert_eq!(ctlz32(10), 28);
-        assert_eq!(ctlz64(10), 60);
-
-        assert_eq!(ctlz8(100), 1);
-        assert_eq!(ctlz16(100), 9);
-        assert_eq!(ctlz32(100), 25);
-        assert_eq!(ctlz64(100), 57);
-
-        assert_eq!(cttz8(-1), 0);
-        assert_eq!(cttz16(-1), 0);
-        assert_eq!(cttz32(-1), 0);
-        assert_eq!(cttz64(-1), 0);
-
-        assert_eq!(cttz8(0), 8);
-        assert_eq!(cttz16(0), 16);
-        assert_eq!(cttz32(0), 32);
-        assert_eq!(cttz64(0), 64);
-
-        assert_eq!(cttz8(1), 0);
-        assert_eq!(cttz16(1), 0);
-        assert_eq!(cttz32(1), 0);
-        assert_eq!(cttz64(1), 0);
-
-        assert_eq!(cttz8(10), 1);
-        assert_eq!(cttz16(10), 1);
-        assert_eq!(cttz32(10), 1);
-        assert_eq!(cttz64(10), 1);
-
-        assert_eq!(cttz8(100), 2);
-        assert_eq!(cttz16(100), 2);
-        assert_eq!(cttz32(100), 2);
-        assert_eq!(cttz64(100), 2);
-
-        assert_eq!(bswap16(0x0A0B), 0x0B0A);
-        assert_eq!(bswap32(0x0ABBCC0D), 0x0DCCBB0A);
-        assert_eq!(bswap64(0x0122334455667708), 0x0877665544332201);
+        assert_eq!(ctpop(0u8), 0); assert_eq!(ctpop(0i8), 0);
+        assert_eq!(ctpop(0u16), 0); assert_eq!(ctpop(0i16), 0);
+        assert_eq!(ctpop(0u32), 0); assert_eq!(ctpop(0i32), 0);
+        assert_eq!(ctpop(0u64), 0); assert_eq!(ctpop(0i64), 0);
+
+        assert_eq!(ctpop(1u8), 1); assert_eq!(ctpop(1i8), 1);
+        assert_eq!(ctpop(1u16), 1); assert_eq!(ctpop(1i16), 1);
+        assert_eq!(ctpop(1u32), 1); assert_eq!(ctpop(1i32), 1);
+        assert_eq!(ctpop(1u64), 1); assert_eq!(ctpop(1i64), 1);
+
+        assert_eq!(ctpop(10u8), 2); assert_eq!(ctpop(10i8), 2);
+        assert_eq!(ctpop(10u16), 2); assert_eq!(ctpop(10i16), 2);
+        assert_eq!(ctpop(10u32), 2); assert_eq!(ctpop(10i32), 2);
+        assert_eq!(ctpop(10u64), 2); assert_eq!(ctpop(10i64), 2);
+
+        assert_eq!(ctpop(100u8), 3); assert_eq!(ctpop(100i8), 3);
+        assert_eq!(ctpop(100u16), 3); assert_eq!(ctpop(100i16), 3);
+        assert_eq!(ctpop(100u32), 3); assert_eq!(ctpop(100i32), 3);
+        assert_eq!(ctpop(100u64), 3); assert_eq!(ctpop(100i64), 3);
+
+        assert_eq!(ctpop(-1u8), 8); assert_eq!(ctpop(-1i8), 8);
+        assert_eq!(ctpop(-1u16), 16); assert_eq!(ctpop(-1i16), 16);
+        assert_eq!(ctpop(-1u32), 32); assert_eq!(ctpop(-1i32), 32);
+        assert_eq!(ctpop(-1u64), 64); assert_eq!(ctpop(-1i64), 64);
+
+        assert_eq!(ctlz(0u8), 8); assert_eq!(ctlz(0i8), 8);
+        assert_eq!(ctlz(0u16), 16); assert_eq!(ctlz(0i16), 16);
+        assert_eq!(ctlz(0u32), 32); assert_eq!(ctlz(0i32), 32);
+        assert_eq!(ctlz(0u64), 64); assert_eq!(ctlz(0i64), 64);
+
+        assert_eq!(ctlz(1u8), 7); assert_eq!(ctlz(1i8), 7);
+        assert_eq!(ctlz(1u16), 15); assert_eq!(ctlz(1i16), 15);
+        assert_eq!(ctlz(1u32), 31); assert_eq!(ctlz(1i32), 31);
+        assert_eq!(ctlz(1u64), 63); assert_eq!(ctlz(1i64), 63);
+
+        assert_eq!(ctlz(10u8), 4); assert_eq!(ctlz(10i8), 4);
+        assert_eq!(ctlz(10u16), 12); assert_eq!(ctlz(10i16), 12);
+        assert_eq!(ctlz(10u32), 28); assert_eq!(ctlz(10i32), 28);
+        assert_eq!(ctlz(10u64), 60); assert_eq!(ctlz(10i64), 60);
+
+        assert_eq!(ctlz(100u8), 1); assert_eq!(ctlz(100i8), 1);
+        assert_eq!(ctlz(100u16), 9); assert_eq!(ctlz(100i16), 9);
+        assert_eq!(ctlz(100u32), 25); assert_eq!(ctlz(100i32), 25);
+        assert_eq!(ctlz(100u64), 57); assert_eq!(ctlz(100i64), 57);
+
+        assert_eq!(cttz(-1u8), 0); assert_eq!(cttz(-1i8), 0);
+        assert_eq!(cttz(-1u16), 0); assert_eq!(cttz(-1i16), 0);
+        assert_eq!(cttz(-1u32), 0); assert_eq!(cttz(-1i32), 0);
+        assert_eq!(cttz(-1u64), 0); assert_eq!(cttz(-1i64), 0);
+
+        assert_eq!(cttz(0u8), 8); assert_eq!(cttz(0i8), 8);
+        assert_eq!(cttz(0u16), 16); assert_eq!(cttz(0i16), 16);
+        assert_eq!(cttz(0u32), 32); assert_eq!(cttz(0i32), 32);
+        assert_eq!(cttz(0u64), 64); assert_eq!(cttz(0i64), 64);
+
+        assert_eq!(cttz(1u8), 0); assert_eq!(cttz(1i8), 0);
+        assert_eq!(cttz(1u16), 0); assert_eq!(cttz(1i16), 0);
+        assert_eq!(cttz(1u32), 0); assert_eq!(cttz(1i32), 0);
+        assert_eq!(cttz(1u64), 0); assert_eq!(cttz(1i64), 0);
+
+        assert_eq!(cttz(10u8), 1); assert_eq!(cttz(10i8), 1);
+        assert_eq!(cttz(10u16), 1); assert_eq!(cttz(10i16), 1);
+        assert_eq!(cttz(10u32), 1); assert_eq!(cttz(10i32), 1);
+        assert_eq!(cttz(10u64), 1); assert_eq!(cttz(10i64), 1);
+
+        assert_eq!(cttz(100u8), 2); assert_eq!(cttz(100i8), 2);
+        assert_eq!(cttz(100u16), 2); assert_eq!(cttz(100i16), 2);
+        assert_eq!(cttz(100u32), 2); assert_eq!(cttz(100i32), 2);
+        assert_eq!(cttz(100u64), 2); assert_eq!(cttz(100i64), 2);
+
+        assert_eq!(bswap(0x0Au8), 0x0A); // no-op
+        assert_eq!(bswap(0x0Ai8), 0x0A); // no-op
+        assert_eq!(bswap(0x0A0Bu16), 0x0B0A);
+        assert_eq!(bswap(0x0A0Bi16), 0x0B0A);
+        assert_eq!(bswap(0x0ABBCC0Du32), 0x0DCCBB0A);
+        assert_eq!(bswap(0x0ABBCC0Di32), 0x0DCCBB0A);
+        assert_eq!(bswap(0x0122334455667708u64), 0x0877665544332201);
+        assert_eq!(bswap(0x0122334455667708i64), 0x0877665544332201);
     }
 }