llvm/test/Transforms/LoopVectorize/uniform-shared-operands.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 6
; RUN: opt -passes=loop-vectorize -force-vector-width=2 -force-vector-interleave=2 -verify-each -S %s | FileCheck %s

; Shared operands must not cause exponential uniformity analysis during unrolling.
define i64 @uniform_shared_operands(i64 %x, i64 %n) {
; CHECK-LABEL: define i64 @uniform_shared_operands(
; CHECK-SAME: i64 [[X:%.*]], i64 [[N:%.*]]) {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N]], 4
; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    [[TMP0:%.*]] = and i64 [[N]], 3
; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 [[N]], [[TMP0]]
; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[X]], i64 0
; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
; CHECK-NEXT:    [[TMP1:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[BROADCAST_SPLAT]], <2 x i64> [[BROADCAST_SPLAT]], <2 x i64> [[BROADCAST_SPLAT]])
; CHECK-NEXT:    [[TMP2:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP1]], <2 x i64> [[TMP1]], <2 x i64> [[TMP1]])
; CHECK-NEXT:    [[TMP3:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP2]], <2 x i64> [[TMP2]], <2 x i64> [[TMP2]])
; CHECK-NEXT:    [[TMP4:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP3]], <2 x i64> [[TMP3]], <2 x i64> [[TMP3]])
; CHECK-NEXT:    [[TMP5:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP4]], <2 x i64> [[TMP4]], <2 x i64> [[TMP4]])
; CHECK-NEXT:    [[TMP6:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP5]], <2 x i64> [[TMP5]], <2 x i64> [[TMP5]])
; CHECK-NEXT:    [[TMP7:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP6]], <2 x i64> [[TMP6]], <2 x i64> [[TMP6]])
; CHECK-NEXT:    [[TMP8:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP7]], <2 x i64> [[TMP7]], <2 x i64> [[TMP7]])
; CHECK-NEXT:    [[TMP9:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP8]], <2 x i64> [[TMP8]], <2 x i64> [[TMP8]])
; CHECK-NEXT:    [[TMP10:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP9]], <2 x i64> [[TMP9]], <2 x i64> [[TMP9]])
; CHECK-NEXT:    [[TMP11:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP10]], <2 x i64> [[TMP10]], <2 x i64> [[TMP10]])
; CHECK-NEXT:    [[TMP12:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP11]], <2 x i64> [[TMP11]], <2 x i64> [[TMP11]])
; CHECK-NEXT:    [[TMP13:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP12]], <2 x i64> [[TMP12]], <2 x i64> [[TMP12]])
; CHECK-NEXT:    [[TMP14:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP13]], <2 x i64> [[TMP13]], <2 x i64> [[TMP13]])
; CHECK-NEXT:    [[TMP15:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP14]], <2 x i64> [[TMP14]], <2 x i64> [[TMP14]])
; CHECK-NEXT:    [[TMP16:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP15]], <2 x i64> [[TMP15]], <2 x i64> [[TMP15]])
; CHECK-NEXT:    [[TMP17:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP16]], <2 x i64> [[TMP16]], <2 x i64> [[TMP16]])
; CHECK-NEXT:    [[TMP18:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP17]], <2 x i64> [[TMP17]], <2 x i64> [[TMP17]])
; CHECK-NEXT:    [[TMP19:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP18]], <2 x i64> [[TMP18]], <2 x i64> [[TMP18]])
; CHECK-NEXT:    [[TMP20:%.*]] = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> [[TMP19]], <2 x i64> [[TMP19]], <2 x i64> [[TMP19]])
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x i64> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP23:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[VEC_PHI1:%.*]] = phi <2 x i64> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[TMP24:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[STEP_ADD:%.*]] = add nuw <2 x i64> [[VEC_IND]], splat (i64 2)
; CHECK-NEXT:    [[TMP21:%.*]] = and <2 x i64> [[TMP20]], [[VEC_IND]]
; CHECK-NEXT:    [[TMP22:%.*]] = and <2 x i64> [[TMP20]], [[STEP_ADD]]
; CHECK-NEXT:    [[TMP23]] = xor <2 x i64> [[VEC_PHI]], [[TMP21]]
; CHECK-NEXT:    [[TMP24]] = xor <2 x i64> [[VEC_PHI1]], [[TMP22]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[STEP_ADD]], splat (i64 2)
; CHECK-NEXT:    [[TMP25:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[TMP25]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[BIN_RDX:%.*]] = xor <2 x i64> [[TMP24]], [[TMP23]]
; CHECK-NEXT:    [[TMP26:%.*]] = call i64 @llvm.vector.reduce.xor.v2i64(<2 x i64> [[BIN_RDX]])
; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
; CHECK:       [[SCALAR_PH]]:
; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i64 [ [[TMP26]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    br label %[[LOOP:.*]]
; CHECK:       [[LOOP]]:
; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
; CHECK-NEXT:    [[ACC:%.*]] = phi i64 [ [[BC_MERGE_RDX]], %[[SCALAR_PH]] ], [ [[ACC_NEXT:%.*]], %[[LOOP]] ]
; CHECK-NEXT:    [[A1:%.*]] = call i64 @llvm.fshl.i64(i64 [[X]], i64 [[X]], i64 [[X]])
; CHECK-NEXT:    [[A2:%.*]] = call i64 @llvm.fshl.i64(i64 [[A1]], i64 [[A1]], i64 [[A1]])
; CHECK-NEXT:    [[A3:%.*]] = call i64 @llvm.fshl.i64(i64 [[A2]], i64 [[A2]], i64 [[A2]])
; CHECK-NEXT:    [[A4:%.*]] = call i64 @llvm.fshl.i64(i64 [[A3]], i64 [[A3]], i64 [[A3]])
; CHECK-NEXT:    [[A5:%.*]] = call i64 @llvm.fshl.i64(i64 [[A4]], i64 [[A4]], i64 [[A4]])
; CHECK-NEXT:    [[A6:%.*]] = call i64 @llvm.fshl.i64(i64 [[A5]], i64 [[A5]], i64 [[A5]])
; CHECK-NEXT:    [[A7:%.*]] = call i64 @llvm.fshl.i64(i64 [[A6]], i64 [[A6]], i64 [[A6]])
; CHECK-NEXT:    [[A8:%.*]] = call i64 @llvm.fshl.i64(i64 [[A7]], i64 [[A7]], i64 [[A7]])
; CHECK-NEXT:    [[A9:%.*]] = call i64 @llvm.fshl.i64(i64 [[A8]], i64 [[A8]], i64 [[A8]])
; CHECK-NEXT:    [[A10:%.*]] = call i64 @llvm.fshl.i64(i64 [[A9]], i64 [[A9]], i64 [[A9]])
; CHECK-NEXT:    [[A11:%.*]] = call i64 @llvm.fshl.i64(i64 [[A10]], i64 [[A10]], i64 [[A10]])
; CHECK-NEXT:    [[A12:%.*]] = call i64 @llvm.fshl.i64(i64 [[A11]], i64 [[A11]], i64 [[A11]])
; CHECK-NEXT:    [[A13:%.*]] = call i64 @llvm.fshl.i64(i64 [[A12]], i64 [[A12]], i64 [[A12]])
; CHECK-NEXT:    [[A14:%.*]] = call i64 @llvm.fshl.i64(i64 [[A13]], i64 [[A13]], i64 [[A13]])
; CHECK-NEXT:    [[A15:%.*]] = call i64 @llvm.fshl.i64(i64 [[A14]], i64 [[A14]], i64 [[A14]])
; CHECK-NEXT:    [[A16:%.*]] = call i64 @llvm.fshl.i64(i64 [[A15]], i64 [[A15]], i64 [[A15]])
; CHECK-NEXT:    [[A17:%.*]] = call i64 @llvm.fshl.i64(i64 [[A16]], i64 [[A16]], i64 [[A16]])
; CHECK-NEXT:    [[A18:%.*]] = call i64 @llvm.fshl.i64(i64 [[A17]], i64 [[A17]], i64 [[A17]])
; CHECK-NEXT:    [[A19:%.*]] = call i64 @llvm.fshl.i64(i64 [[A18]], i64 [[A18]], i64 [[A18]])
; CHECK-NEXT:    [[A20:%.*]] = call i64 @llvm.fshl.i64(i64 [[A19]], i64 [[A19]], i64 [[A19]])
; CHECK-NEXT:    [[M:%.*]] = and i64 [[A20]], [[IV]]
; CHECK-NEXT:    [[ACC_NEXT]] = xor i64 [[ACC]], [[M]]
; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
; CHECK-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
; CHECK-NEXT:    br i1 [[EC]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    [[ACC_NEXT_LCSSA:%.*]] = phi i64 [ [[ACC_NEXT]], %[[LOOP]] ], [ [[TMP26]], %[[MIDDLE_BLOCK]] ]
; CHECK-NEXT:    ret i64 [[ACC_NEXT_LCSSA]]
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %acc = phi i64 [ 0, %entry ], [ %acc.next, %loop ]
  %a1 = call i64 @llvm.fshl.i64(i64 %x, i64 %x, i64 %x)
  %a2 = call i64 @llvm.fshl.i64(i64 %a1, i64 %a1, i64 %a1)
  %a3 = call i64 @llvm.fshl.i64(i64 %a2, i64 %a2, i64 %a2)
  %a4 = call i64 @llvm.fshl.i64(i64 %a3, i64 %a3, i64 %a3)
  %a5 = call i64 @llvm.fshl.i64(i64 %a4, i64 %a4, i64 %a4)
  %a6 = call i64 @llvm.fshl.i64(i64 %a5, i64 %a5, i64 %a5)
  %a7 = call i64 @llvm.fshl.i64(i64 %a6, i64 %a6, i64 %a6)
  %a8 = call i64 @llvm.fshl.i64(i64 %a7, i64 %a7, i64 %a7)
  %a9 = call i64 @llvm.fshl.i64(i64 %a8, i64 %a8, i64 %a8)
  %a10 = call i64 @llvm.fshl.i64(i64 %a9, i64 %a9, i64 %a9)
  %a11 = call i64 @llvm.fshl.i64(i64 %a10, i64 %a10, i64 %a10)
  %a12 = call i64 @llvm.fshl.i64(i64 %a11, i64 %a11, i64 %a11)
  %a13 = call i64 @llvm.fshl.i64(i64 %a12, i64 %a12, i64 %a12)
  %a14 = call i64 @llvm.fshl.i64(i64 %a13, i64 %a13, i64 %a13)
  %a15 = call i64 @llvm.fshl.i64(i64 %a14, i64 %a14, i64 %a14)
  %a16 = call i64 @llvm.fshl.i64(i64 %a15, i64 %a15, i64 %a15)
  %a17 = call i64 @llvm.fshl.i64(i64 %a16, i64 %a16, i64 %a16)
  %a18 = call i64 @llvm.fshl.i64(i64 %a17, i64 %a17, i64 %a17)
  %a19 = call i64 @llvm.fshl.i64(i64 %a18, i64 %a18, i64 %a18)
  %a20 = call i64 @llvm.fshl.i64(i64 %a19, i64 %a19, i64 %a19)
  %m = and i64 %a20, %iv
  %acc.next = xor i64 %acc, %m
  %iv.next = add i64 %iv, 1
  %ec = icmp eq i64 %iv.next, %n
  br i1 %ec, label %exit, label %loop

exit:
  ret i64 %acc.next
}