Before:
movabsq $
4294967296, %rax ## encoding: [0x48,0xb8,0x00,0x00,0x00,0x00,0x01,0x00,0x00,0x00]
testq %rax, %rdi ## encoding: [0x48,0x85,0xf8]
jne LBB0_2 ## encoding: [0x75,A]
After:
btq $32, %rdi ## encoding: [0x48,0x0f,0xba,0xe7,0x20]
jb LBB0_2 ## encoding: [0x72,A]
btq is usually slower than testq because it doesn't fuse with the jump, but here we're better off
saving one register and a giant movabsq.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145103
91177308-0d34-0410-b5e6-
96231b3b80d8
}
} else if (Op1.getOpcode() == ISD::Constant) {
ConstantSDNode *AndRHS = cast<ConstantSDNode>(Op1);
+ uint64_t AndRHSVal = AndRHS->getZExtValue();
SDValue AndLHS = Op0;
- if (AndRHS->getZExtValue() == 1 && AndLHS.getOpcode() == ISD::SRL) {
+
+ if (AndRHSVal == 1 && AndLHS.getOpcode() == ISD::SRL) {
LHS = AndLHS.getOperand(0);
RHS = AndLHS.getOperand(1);
}
+
+ // Use BT if the immediate can't be encoded in a TEST instruction.
+ if (!isUInt<32>(AndRHSVal) && isPowerOf2_64(AndRHSVal)) {
+ LHS = AndLHS;
+ RHS = DAG.getConstant(Log2_64_Ceil(AndRHSVal), LHS.getValueType());
+ }
}
if (LHS.getNode()) {
--- /dev/null
+; RUN: llc < %s -march=x86-64 | FileCheck %s
+
+declare void @bar()
+
+define void @test1(i64 %foo) nounwind {
+ %and = and i64 %foo, 4294967296
+ %tobool = icmp eq i64 %and, 0
+ br i1 %tobool, label %if.end, label %if.then
+
+; CHECK: test1:
+; CHECK: btq $32
+
+if.then:
+ tail call void @bar() nounwind
+ br label %if.end
+
+if.end:
+ ret void
+}
+
+define void @test2(i64 %foo) nounwind {
+ %and = and i64 %foo, 2147483648
+ %tobool = icmp eq i64 %and, 0
+ br i1 %tobool, label %if.end, label %if.then
+
+; CHECK: test2:
+; CHECK: testl $-2147483648
+
+if.then:
+ tail call void @bar() nounwind
+ br label %if.end
+
+if.end:
+ ret void
+}