Add ARM64 CLS patterns

This patch adds patterns to generate the cls instruction ARM64. Includes tests
for 64 bit and 32 bit operands.

rdar://15611957

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206079 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/ARM64/ARM64InstrInfo.td b/lib/Target/ARM64/ARM64InstrInfo.td
index 7d319dd61eb294d3fbe5cff387b2064f791a7b55..2af217fe91043da891cd33611eaf10da4d55f2e5 100644 (file)
--- a/lib/Target/ARM64/ARM64InstrInfo.td
+++ b/lib/Target/ARM64/ARM64InstrInfo.td
@@ -652,6 +652,12 @@ def : Pat<(cttz GPR32:$Rn),
           (CLZWr (RBITWr GPR32:$Rn))>;
 def : Pat<(cttz GPR64:$Rn),
           (CLZXr (RBITXr GPR64:$Rn))>;
+def : Pat<(ctlz (or (shl (xor (sra GPR32:$Rn, (i64 31)), GPR32:$Rn), (i64 1)),
+                (i32 1))),
+          (CLSWr GPR32:$Rn)>;
+def : Pat<(ctlz (or (shl (xor (sra GPR64:$Rn, (i64 63)), GPR64:$Rn), (i64 1)),
+                (i64 1))),
+          (CLSXr GPR64:$Rn)>;
 
 // Unlike the other one operand instructions, the instructions with the "rev"
 // mnemonic do *not* just different in the size bit, but actually use different

diff --git a/test/CodeGen/ARM64/clrsb.ll b/test/CodeGen/ARM64/clrsb.ll
new file mode 100644 (file)
index 0000000..042e52e
--- /dev/null
+++ b/test/CodeGen/ARM64/clrsb.ll
@@ -0,0 +1,36 @@
+; RUN: llc < %s -march=arm64 |  FileCheck %s
+
+target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
+target triple = "arm64-apple-ios7.0.0"
+
+; Function Attrs: nounwind readnone
+declare i32 @llvm.ctlz.i32(i32, i1) #0
+declare i64 @llvm.ctlz.i64(i64, i1) #1
+
+; Function Attrs: nounwind ssp
+define i32 @clrsb32(i32 %x) #2 {
+entry:
+  %shr = ashr i32 %x, 31
+  %xor = xor i32 %shr, %x
+  %mul = shl i32 %xor, 1
+  %add = or i32 %mul, 1
+  %0 = tail call i32 @llvm.ctlz.i32(i32 %add, i1 false)
+
+  ret i32 %0
+; CHECK-LABEL: clrsb32
+; CHECK:   cls [[TEMP:w[0-9]+]], [[TEMP]]
+}
+
+; Function Attrs: nounwind ssp
+define i64 @clrsb64(i64 %x) #3 {
+entry:
+  %shr = ashr i64 %x, 63
+  %xor = xor i64 %shr, %x
+  %mul = shl nsw i64 %xor, 1
+  %add = or i64 %mul, 1
+  %0 = tail call i64 @llvm.ctlz.i64(i64 %add, i1 false)
+
+  ret i64 %0
+; CHECK-LABEL: clrsb64
+; CHECK:   cls [[TEMP:x[0-9]+]], [[TEMP]]
+}