auto merge of #13899 : bjz/rust/simd, r=pcwalton

cc. @pcwalton

Author: bors

src/librustc/middle/trans/base.rs

@@ -619,6 +619,44 @@ pub fn compare_scalar_values<'a>(
     }
 }
 
+pub fn compare_simd_types(
+                    cx: &Block,
+                    lhs: ValueRef,
+                    rhs: ValueRef,
+                    t: ty::t,
+                    size: uint,
+                    op: ast::BinOp)
+                    -> ValueRef {
+    match ty::get(t).sty {
+        ty::ty_float(_) => {
+            // The comparison operators for floating point vectors are challenging.
+            // LLVM outputs a `< size x i1 >`, but if we perform a sign extension
+            // then bitcast to a floating point vector, the result will be `-NaN`
+            // for each truth value. Because of this they are unsupported.
+            cx.sess().bug("compare_simd_types: comparison operators \
+                           not supported for floating point SIMD types")
+        },
+        ty::ty_uint(_) | ty::ty_int(_) => {
+            let cmp = match op {
+                ast::BiEq => lib::llvm::IntEQ,
+                ast::BiNe => lib::llvm::IntNE,
+                ast::BiLt => lib::llvm::IntSLT,
+                ast::BiLe => lib::llvm::IntSLE,
+                ast::BiGt => lib::llvm::IntSGT,
+                ast::BiGe => lib::llvm::IntSGE,
+                _ => cx.sess().bug("compare_simd_types: must be a comparison operator"),
+            };
+            let return_ty = Type::vector(&type_of(cx.ccx(), t), size as u64);
+            // LLVM outputs an `< size x i1 >`, so we need to perform a sign extension
+            // to get the correctly sized type. This will compile to a single instruction
+            // once the IR is converted to assembly if the SIMD instruction is supported
+            // by the target architecture.
+            SExt(cx, ICmp(cx, cmp, lhs, rhs), return_ty)
+        },
+        _ => cx.sess().bug("compare_simd_types: invalid SIMD type"),
+    }
+}
+
 pub type val_and_ty_fn<'r,'b> =
     |&'b Block<'b>, ValueRef, ty::t|: 'r -> &'b Block<'b>;
 

src/librustc/middle/trans/expr.rs

@@ -1259,16 +1259,15 @@ fn trans_eager_binop<'a>(
                      -> DatumBlock<'a, Expr> {
     let _icx = push_ctxt("trans_eager_binop");
 
-    let mut intype = {
+    let tcx = bcx.tcx();
+    let is_simd = ty::type_is_simd(tcx, lhs_t);
+    let intype = {
         if ty::type_is_bot(lhs_t) { rhs_t }
+        else if is_simd { ty::simd_type(tcx, lhs_t) }
         else { lhs_t }
     };
-    let tcx = bcx.tcx();
-    if ty::type_is_simd(tcx, intype) {
-        intype = ty::simd_type(tcx, intype);
-    }
     let is_float = ty::type_is_fp(intype);
-    let signed = ty::type_is_signed(intype);
+    let is_signed = ty::type_is_signed(intype);
 
     let rhs = base::cast_shift_expr_rhs(bcx, op, lhs, rhs);
 
@@ -1293,7 +1292,7 @@ fn trans_eager_binop<'a>(
             // Only zero-check integers; fp /0 is NaN
             bcx = base::fail_if_zero(bcx, binop_expr.span,
                                      op, rhs, rhs_t);
-            if signed {
+            if is_signed {
                 SDiv(bcx, lhs, rhs)
             } else {
                 UDiv(bcx, lhs, rhs)
@@ -1307,7 +1306,7 @@ fn trans_eager_binop<'a>(
             // Only zero-check integers; fp %0 is NaN
             bcx = base::fail_if_zero(bcx, binop_expr.span,
                                      op, rhs, rhs_t);
-            if signed {
+            if is_signed {
                 SRem(bcx, lhs, rhs)
             } else {
                 URem(bcx, lhs, rhs)
@@ -1319,21 +1318,21 @@ fn trans_eager_binop<'a>(
       ast::BiBitXor => Xor(bcx, lhs, rhs),
       ast::BiShl => Shl(bcx, lhs, rhs),
       ast::BiShr => {
-        if signed {
+        if is_signed {
             AShr(bcx, lhs, rhs)
         } else { LShr(bcx, lhs, rhs) }
       }
       ast::BiEq | ast::BiNe | ast::BiLt | ast::BiGe | ast::BiLe | ast::BiGt => {
         if ty::type_is_bot(rhs_t) {
             C_bool(bcx.ccx(), false)
-        } else {
-            if !ty::type_is_scalar(rhs_t) {
-                bcx.tcx().sess.span_bug(binop_expr.span,
-                                        "non-scalar comparison");
-            }
+        } else if ty::type_is_scalar(rhs_t) {
             let cmpr = base::compare_scalar_types(bcx, lhs, rhs, rhs_t, op);
             bcx = cmpr.bcx;
             ZExt(bcx, cmpr.val, Type::i8(bcx.ccx()))
+        } else if is_simd {
+            base::compare_simd_types(bcx, lhs, rhs, intype, ty::simd_size(tcx, lhs_t), op)
+        } else {
+            bcx.tcx().sess.span_bug(binop_expr.span, "comparison operator unsupported for type")
         }
       }
       _ => {

src/librustc/middle/typeck/check/mod.rs

@@ -2102,8 +2102,27 @@ fn check_expr_with_unifier(fcx: &FnCtxt,
 
             let result_t = match op {
                 ast::BiEq | ast::BiNe | ast::BiLt | ast::BiLe | ast::BiGe |
-                ast::BiGt => ty::mk_bool(),
-                _ => lhs_t
+                ast::BiGt => {
+                    if ty::type_is_simd(tcx, lhs_t) {
+                        if ty::type_is_fp(ty::simd_type(tcx, lhs_t)) {
+                            fcx.type_error_message(expr.span,
+                                |actual| {
+                                    format!("binary comparison operation `{}` not supported \
+                                            for floating point SIMD vector `{}`",
+                                            ast_util::binop_to_str(op), actual)
+                                },
+                                lhs_t,
+                                None
+                            );
+                            ty::mk_err()
+                        } else {
+                            lhs_t
+                        }
+                    } else {
+                        ty::mk_bool()
+                    }
+                },
+                _ => lhs_t,
             };
 
             fcx.write_ty(expr.id, result_t);

src/test/compile-fail/simd-binop.rs

@@ -0,0 +1,37 @@
+// Copyright 2014 The Rust Project Developers. See the COPYRIGHT
+// file at the top-level directory of this distribution and at
+// http://rust-lang.org/COPYRIGHT.
+//
+// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
+// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
+// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
+// option. This file may not be copied, modified, or distributed
+// except according to those terms.
+
+// ignore-tidy-linelength
+
+#![allow(experimental)]
+
+use std::unstable::simd::f32x4;
+
+fn main() {
+
+    let _ = f32x4(0.0, 0.0, 0.0, 0.0) == f32x4(0.0, 0.0, 0.0, 0.0);
+    //~^ ERROR binary comparison operation `==` not supported for floating point SIMD vector `std::unstable::simd::f32x4`
+
+    let _ = f32x4(0.0, 0.0, 0.0, 0.0) != f32x4(0.0, 0.0, 0.0, 0.0);
+    //~^ ERROR binary comparison operation `!=` not supported for floating point SIMD vector `std::unstable::simd::f32x4`
+
+    let _ = f32x4(0.0, 0.0, 0.0, 0.0) < f32x4(0.0, 0.0, 0.0, 0.0);
+    //~^ ERROR binary comparison operation `<` not supported for floating point SIMD vector `std::unstable::simd::f32x4`
+
+    let _ = f32x4(0.0, 0.0, 0.0, 0.0) <= f32x4(0.0, 0.0, 0.0, 0.0);
+    //~^ ERROR binary comparison operation `<=` not supported for floating point SIMD vector `std::unstable::simd::f32x4`
+
+    let _ = f32x4(0.0, 0.0, 0.0, 0.0) >= f32x4(0.0, 0.0, 0.0, 0.0);
+    //~^ ERROR binary comparison operation `>=` not supported for floating point SIMD vector `std::unstable::simd::f32x4`
+
+    let _ = f32x4(0.0, 0.0, 0.0, 0.0) > f32x4(0.0, 0.0, 0.0, 0.0);
+    //~^ ERROR binary comparison operation `>` not supported for floating point SIMD vector `std::unstable::simd::f32x4`
+
+}

src/test/run-pass/simd-binop.rs

@@ -12,27 +12,61 @@
 
 use std::unstable::simd::{i32x4, f32x4, u32x4};
 
-fn test_int(e: i32) -> i32 {
-    let v = i32x4(e, 0i32, 0i32, 0i32);
-    let i32x4(e2, _, _, _) = v * v + v - v;
-    e2
+fn eq_u32x4(u32x4(x0, x1, x2, x3): u32x4, u32x4(y0, y1, y2, y3): u32x4) -> bool {
+    (x0 == y0) && (x1 == y1) && (x2 == y2) && (x3 == y3)
 }
 
-fn test_float(e: f32) -> f32 {
-    let v = f32x4(e, 0f32, 0f32, 0f32);
-    let f32x4(e2, _, _, _) = v * v + v - v;
-    e2
+fn eq_f32x4(f32x4(x0, x1, x2, x3): f32x4, f32x4(y0, y1, y2, y3): f32x4) -> bool {
+    (x0 == y0) && (x1 == y1) && (x2 == y2) && (x3 == y3)
 }
 
-pub fn test_shift(e: u32) -> u32 {
-    let v = u32x4(e, 0u32, 0u32, 0u32);
-    let one = u32x4(1u32, 0u32, 0u32, 0u32);
-    let u32x4(e2, _, _, _) = v << one >> one;
-    e2
+fn eq_i32x4(i32x4(x0, x1, x2, x3): i32x4, i32x4(y0, y1, y2, y3): i32x4) -> bool {
+    (x0 == y0) && (x1 == y1) && (x2 == y2) && (x3 == y3)
 }
 
 pub fn main() {
-    assert_eq!(test_int(3i32), 9i32);
-    assert_eq!(test_float(3f32), 9f32);
-    assert_eq!(test_shift(3u32), 3u32);
+    // arithmetic operators
+
+    assert!(eq_u32x4(u32x4(1, 2, 3, 4) + u32x4(4, 3, 2, 1), u32x4(5, 5, 5, 5)));
+    assert!(eq_u32x4(u32x4(4, 5, 6, 7) - u32x4(4, 3, 2, 1), u32x4(0, 2, 4, 6)));
+    assert!(eq_u32x4(u32x4(1, 2, 3, 4) * u32x4(4, 3, 2, 1), u32x4(4, 6, 6, 4)));
+    assert!(eq_u32x4(u32x4(1, 2, 3, 4) & u32x4(4, 3, 2, 1), u32x4(0, 2, 2, 0)));
+    assert!(eq_u32x4(u32x4(1, 2, 3, 4) | u32x4(4, 3, 2, 1), u32x4(5, 3, 3, 5)));
+    assert!(eq_u32x4(u32x4(1, 2, 3, 4) ^ u32x4(4, 3, 2, 1), u32x4(5, 1, 1, 5)));
+    assert!(eq_u32x4(u32x4(1, 2, 3, 4) << u32x4(4, 3, 2, 1), u32x4(16, 16, 12, 8)));
+    assert!(eq_u32x4(u32x4(1, 2, 3, 4) >> u32x4(4, 3, 2, 1), u32x4(0, 0, 0, 2)));
+
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) + i32x4(4, 3, 2, 1), i32x4(5, 5, 5, 5)));
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) - i32x4(4, 3, 2, 1), i32x4(-3, -1, 1, 3)));
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) * i32x4(4, 3, 2, 1), i32x4(4, 6, 6, 4)));
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) & i32x4(4, 3, 2, 1), i32x4(0, 2, 2, 0)));
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) | i32x4(4, 3, 2, 1), i32x4(5, 3, 3, 5)));
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) ^ i32x4(4, 3, 2, 1), i32x4(5, 1, 1, 5)));
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) << i32x4(4, 3, 2, 1), i32x4(16, 16, 12, 8)));
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) >> i32x4(4, 3, 2, 1), i32x4(0, 0, 0, 2)));
+
+    assert!(eq_f32x4(f32x4(1.0, 2.0, 3.0, 4.0) + f32x4(4.0, 3.0, 2.0, 1.0),
+            f32x4(5.0, 5.0, 5.0, 5.0)));
+    assert!(eq_f32x4(f32x4(1.0, 2.0, 3.0, 4.0) - f32x4(4.0, 3.0, 2.0, 1.0),
+            f32x4(-3.0, -1.0, 1.0, 3.0)));
+    assert!(eq_f32x4(f32x4(1.0, 2.0, 3.0, 4.0) * f32x4(4.0, 3.0, 2.0, 1.0),
+            f32x4(4.0, 6.0, 6.0, 4.0)));
+    assert!(eq_f32x4(f32x4(1.0, 2.0, 3.0, 4.0) / f32x4(4.0, 4.0, 2.0, 1.0),
+            f32x4(0.25, 0.5, 1.5, 4.0)));
+
+    // comparison operators
+
+    assert!(eq_u32x4(u32x4(1, 2, 3, 4) == u32x4(3, 2, 1, 0), u32x4(0, !0, 0, 0)));
+    assert!(eq_u32x4(u32x4(1, 2, 3, 4) != u32x4(3, 2, 1, 0), u32x4(!0, 0, !0, !0)));
+    assert!(eq_u32x4(u32x4(1, 2, 3, 4) < u32x4(3, 2, 1, 0), u32x4(!0, 0, 0, 0)));
+    assert!(eq_u32x4(u32x4(1, 2, 3, 4) <= u32x4(3, 2, 1, 0), u32x4(!0, !0, 0, 0)));
+    assert!(eq_u32x4(u32x4(1, 2, 3, 4) >= u32x4(3, 2, 1, 0), u32x4(0, !0, !0, !0)));
+    assert!(eq_u32x4(u32x4(1, 2, 3, 4) > u32x4(3, 2, 1, 0), u32x4(0, 0, !0, !0)));
+
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) == i32x4(3, 2, 1, 0), i32x4(0, !0, 0, 0)));
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) != i32x4(3, 2, 1, 0), i32x4(!0, 0, !0, !0)));
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) < i32x4(3, 2, 1, 0), i32x4(!0, 0, 0, 0)));
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) <= i32x4(3, 2, 1, 0), i32x4(!0, !0, 0, 0)));
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) >= i32x4(3, 2, 1, 0), i32x4(0, !0, !0, !0)));
+    assert!(eq_i32x4(i32x4(1, 2, 3, 4) > i32x4(3, 2, 1, 0), i32x4(0, 0, !0, !0)));
 }