forked from OSchip/llvm-project
				
			[AArch64] Fix isLegalAddImmediate() to return true for valid negative values.
Reviewers: t.p.northover, jmolloy Subscribers: mcrosier, aemerson, llvm-commits, rengolin Differential Revision: http://reviews.llvm.org/D17463 llvm-svn: 262248
This commit is contained in:
		
							parent
							
								
									fb2add2be1
								
							
						
					
					
						commit
						f5ba61d18c
					
				| 
						 | 
				
			
			@ -7280,6 +7280,8 @@ EVT AArch64TargetLowering::getOptimalMemOpType(uint64_t Size, unsigned DstAlign,
 | 
			
		|||
 | 
			
		||||
// 12-bit optionally shifted immediates are legal for adds.
 | 
			
		||||
bool AArch64TargetLowering::isLegalAddImmediate(int64_t Immed) const {
 | 
			
		||||
  // Same encoding for add/sub, just flip the sign.
 | 
			
		||||
  Immed = std::abs(Immed);
 | 
			
		||||
  if ((Immed >> 12) == 0 || ((Immed & 0xfff) == 0 && Immed >> 24 == 0))
 | 
			
		||||
    return true;
 | 
			
		||||
  return false;
 | 
			
		||||
| 
						 | 
				
			
			@ -7288,8 +7290,6 @@ bool AArch64TargetLowering::isLegalAddImmediate(int64_t Immed) const {
 | 
			
		|||
// Integer comparisons are implemented with ADDS/SUBS, so the range of valid
 | 
			
		||||
// immediates is the same as for an add or a sub.
 | 
			
		||||
bool AArch64TargetLowering::isLegalICmpImmediate(int64_t Immed) const {
 | 
			
		||||
  if (Immed < 0)
 | 
			
		||||
    Immed *= -1;
 | 
			
		||||
  return isLegalAddImmediate(Immed);
 | 
			
		||||
}
 | 
			
		||||
 | 
			
		||||
| 
						 | 
				
			
			
 | 
			
		|||
| 
						 | 
				
			
			@ -0,0 +1,46 @@
 | 
			
		|||
; RUN: llc -mtriple=aarch64-linux-gnu -verify-machineinstrs -o - %s | FileCheck %s
 | 
			
		||||
; LSR used to pick a sub-optimal solution due to the target responding
 | 
			
		||||
; conservatively to isLegalAddImmediate for negative values.
 | 
			
		||||
 | 
			
		||||
declare void @foo(i32)
 | 
			
		||||
 | 
			
		||||
define void @test(i32 %px) {
 | 
			
		||||
; CHECK-LABEL: test:
 | 
			
		||||
; CHECK: // %entry
 | 
			
		||||
; CHECK: subs
 | 
			
		||||
; CHECK-NEXT: csel
 | 
			
		||||
entry:
 | 
			
		||||
  %sub = add nsw i32 %px, -1
 | 
			
		||||
  %cmp = icmp slt i32 %px, 1
 | 
			
		||||
  %.sub = select i1 %cmp, i32 0, i32 %sub
 | 
			
		||||
  br label %for.body
 | 
			
		||||
 | 
			
		||||
for.body:
 | 
			
		||||
; CHECK: // %for.body
 | 
			
		||||
; CHECK:  cmp
 | 
			
		||||
; CHECK-NEXT:  b.eq
 | 
			
		||||
; CHECK: // %if.then3
 | 
			
		||||
  %x.015 = phi i32 [ %inc, %for.inc ], [ %.sub, %entry ]
 | 
			
		||||
  %cmp2 = icmp eq i32 %x.015, %px
 | 
			
		||||
  br i1 %cmp2, label %for.inc, label %if.then3
 | 
			
		||||
 | 
			
		||||
if.then3:
 | 
			
		||||
  tail call void @foo(i32 %x.015)
 | 
			
		||||
  br label %for.inc
 | 
			
		||||
 | 
			
		||||
for.inc:
 | 
			
		||||
; CHECK: // %for.inc
 | 
			
		||||
; CHECK:  add
 | 
			
		||||
; CHECK-NEXT:  cmp
 | 
			
		||||
; CHECK:  b.le
 | 
			
		||||
; CHECK: // %for.cond.cleanup
 | 
			
		||||
  %inc = add nsw i32 %x.015, 1
 | 
			
		||||
  %cmp1 = icmp sgt i32 %x.015, %px
 | 
			
		||||
  br i1 %cmp1, label %for.cond.cleanup.loopexit, label %for.body
 | 
			
		||||
 | 
			
		||||
for.cond.cleanup.loopexit:
 | 
			
		||||
  br label %for.cond.cleanup
 | 
			
		||||
 | 
			
		||||
for.cond.cleanup:
 | 
			
		||||
  ret void
 | 
			
		||||
}
 | 
			
		||||
		Loading…
	
		Reference in New Issue