Add Intel encodings for the fcmp instruction.

Not all floating point condition codes are directly supported by the
ucimiss/ucomisd instructions. Some inequalities need to be reversed and
eq+ne require two separate tests.

cranelift/filetests/wasm/f64-compares.cton

@@ -0,0 +1,50 @@
+; Test code generation for WebAssembly f64 comparison operators.
+test compile
+
+set is_64bit=0
+isa intel haswell
+
+set is_64bit=1
+isa intel haswell
+
+function %f64_eq(f64, f64) -> i32 {
+ebb0(v0: f64, v1: f64):
+    v2 = fcmp eq v0, v1
+    v3 = bint.i32 v2
+    return v3
+}
+
+function %f64_ne(f64, f64) -> i32 {
+ebb0(v0: f64, v1: f64):
+    v2 = fcmp ne v0, v1
+    v3 = bint.i32 v2
+    return v3
+}
+
+function %f64_lt(f64, f64) -> i32 {
+ebb0(v0: f64, v1: f64):
+    v2 = fcmp lt v0, v1
+    v3 = bint.i32 v2
+    return v3
+}
+
+function %f64_gt(f64, f64) -> i32 {
+ebb0(v0: f64, v1: f64):
+    v2 = fcmp gt v0, v1
+    v3 = bint.i32 v2
+    return v3
+}
+
+function %f64_le(f64, f64) -> i32 {
+ebb0(v0: f64, v1: f64):
+    v2 = fcmp le v0, v1
+    v3 = bint.i32 v2
+    return v3
+}
+
+function %f64_ge(f64, f64) -> i32 {
+ebb0(v0: f64, v1: f64):
+    v2 = fcmp ge v0, v1
+    v3 = bint.i32 v2
+    return v3
+}