[llvm] [InstCombine] Don't shrink AND mask if it grows the immediate (PR #199360)
via llvm-commits
llvm-commits at lists.llvm.org
Sat May 23 08:41:33 PDT 2026
https://github.com/mygitljf updated https://github.com/llvm/llvm-project/pull/199360
>From 36ca6efce4c69f4a62757a096cd5180c4c0ec0b7 Mon Sep 17 00:00:00 2001
From: mygitljf <2410316423 at qq.com>
Date: Sat, 23 May 2026 15:22:06 +0000
Subject: [PATCH] [InstCombine] Don't shrink AND mask if it grows the immediate
Skip rewrite if new constant exceeds 32-bit sign-extended immediate range. Fixes #199095
---
.../InstCombineSimplifyDemanded.cpp | 8 ++-
.../InstCombine/assume-and-negative-mask.ll | 59 +++++++++++++++
llvm/test/Transforms/InstCombine/cast.ll | 4 +-
.../InstCombine/gep-combine-loop-invariant.ll | 6 +-
.../LoopUnroll/runtime-unroll-remainder.ll | 2 +-
...86-interleaved-store-accesses-with-gaps.ll | 4 +-
.../LoopVectorize/float-induction.ll | 52 +++++++-------
.../Transforms/LoopVectorize/if-conversion.ll | 2 +-
.../Transforms/LoopVectorize/induction.ll | 72 +++++++++----------
.../LoopVectorize/interleaved-accesses.ll | 4 +-
.../invariant-store-vectorization-2.ll | 12 ++--
.../invariant-store-vectorization.ll | 14 ++--
.../Transforms/LoopVectorize/loop-scalars.ll | 14 ++--
.../Transforms/LoopVectorize/vector-geps.ll | 10 +--
.../X86/pr48844-br-to-switch-vectorization.ll | 4 +-
15 files changed, 166 insertions(+), 101 deletions(-)
create mode 100644 llvm/test/Transforms/InstCombine/assume-and-negative-mask.ll
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp b/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
index 83ea2d2932ac9..5aa3e04cf1f17 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
@@ -55,8 +55,14 @@ static bool ShrinkDemandedConstant(Instruction *I, unsigned OpNo,
if (C->isSubsetOf(Demanded))
return false;
+ APInt NewC = *C & Demanded;
+
+ // Don't grow a sign-extendable 32-bit immediate into a wider one.
+ if (C->isSignedIntN(32) && !NewC.isSignedIntN(32))
+ return false;
+
// This instruction is producing bits that are not demanded. Shrink the RHS.
- I->setOperand(OpNo, ConstantInt::get(Op->getType(), *C & Demanded));
+ I->setOperand(OpNo, ConstantInt::get(Op->getType(), NewC));
return true;
}
diff --git a/llvm/test/Transforms/InstCombine/assume-and-negative-mask.ll b/llvm/test/Transforms/InstCombine/assume-and-negative-mask.ll
new file mode 100644
index 0000000000000..833c77a12993d
--- /dev/null
+++ b/llvm/test/Transforms/InstCombine/assume-and-negative-mask.ll
@@ -0,0 +1,59 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -passes=instcombine -S | FileCheck %s
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+declare void @llvm.assume(i1) #0
+
+; PR199095: with `assume %x sge 0` the sign bit of %x is known-zero, so
+; SimplifyDemandedBits would shrink `and %x, -4` to
+; `and %x, 0x7FFFFFFFFFFFFFFC`, replacing a sign-extendable small immediate
+; with one that requires an i64 movabs on x86. The fix in
+; ShrinkDemandedConstant must keep the original `-4` mask.
+
+define i64 @pr199095_align4(i64 %x) {
+; CHECK-LABEL: @pr199095_align4(
+; CHECK-NEXT: [[NN:%.*]] = icmp sgt i64 [[X:%.*]], -1
+; CHECK-NEXT: tail call void @llvm.assume(i1 [[NN]])
+; CHECK-NEXT: [[R:%.*]] = and i64 [[X]], -4
+; CHECK-NEXT: ret i64 [[R]]
+;
+ %nn = icmp sge i64 %x, 0
+ tail call void @llvm.assume(i1 %nn)
+ %r = and i64 %x, -4
+ ret i64 %r
+}
+
+define i64 @pr199095_align16(i64 %x) {
+; CHECK-LABEL: @pr199095_align16(
+; CHECK-NEXT: [[NN:%.*]] = icmp sgt i64 [[X:%.*]], -1
+; CHECK-NEXT: tail call void @llvm.assume(i1 [[NN]])
+; CHECK-NEXT: [[R:%.*]] = and i64 [[X]], -16
+; CHECK-NEXT: ret i64 [[R]]
+;
+ %nn = icmp sge i64 %x, 0
+ tail call void @llvm.assume(i1 %nn)
+ %r = and i64 %x, -16
+ ret i64 %r
+}
+
+define <2 x i64> @pr199095_vector(<2 x i64> %x) {
+; CHECK-LABEL: @pr199095_vector(
+; CHECK-NEXT: [[NN:%.*]] = icmp sgt <2 x i64> [[X:%.*]], splat (i64 -1)
+; CHECK-NEXT: [[NN0:%.*]] = extractelement <2 x i1> [[NN]], i64 0
+; CHECK-NEXT: tail call void @llvm.assume(i1 [[NN0]])
+; CHECK-NEXT: [[NN1:%.*]] = extractelement <2 x i1> [[NN]], i64 1
+; CHECK-NEXT: tail call void @llvm.assume(i1 [[NN1]])
+; CHECK-NEXT: [[R:%.*]] = and <2 x i64> [[X]], splat (i64 -4)
+; CHECK-NEXT: ret <2 x i64> [[R]]
+;
+ %nn = icmp sge <2 x i64> %x, zeroinitializer
+ %nn0 = extractelement <2 x i1> %nn, i64 0
+ tail call void @llvm.assume(i1 %nn0)
+ %nn1 = extractelement <2 x i1> %nn, i64 1
+ tail call void @llvm.assume(i1 %nn1)
+ %r = and <2 x i64> %x, splat (i64 -4)
+ ret <2 x i64> %r
+}
+
+attributes #0 = { nounwind }
diff --git a/llvm/test/Transforms/InstCombine/cast.ll b/llvm/test/Transforms/InstCombine/cast.ll
index 8fdf9033b4e4a..d37de82997a96 100644
--- a/llvm/test/Transforms/InstCombine/cast.ll
+++ b/llvm/test/Transforms/InstCombine/cast.ll
@@ -688,11 +688,11 @@ define i64 @test50(i64 %x) {
define i64 @test51(i64 %A, i1 %cond) {
; ALL-LABEL: @test51(
-; ALL-NEXT: [[C:%.*]] = and i64 [[A:%.*]], 4294967294
+; ALL-NEXT: [[C:%.*]] = and i64 [[A:%.*]], -2
; ALL-NEXT: [[NOT_COND:%.*]] = xor i1 [[COND:%.*]], true
; ALL-NEXT: [[MASKSEL:%.*]] = zext i1 [[NOT_COND]] to i64
; ALL-NEXT: [[E:%.*]] = or disjoint i64 [[C]], [[MASKSEL]]
-; ALL-NEXT: [[SEXT:%.*]] = shl nuw i64 [[E]], 32
+; ALL-NEXT: [[SEXT:%.*]] = shl i64 [[E]], 32
; ALL-NEXT: [[F:%.*]] = ashr exact i64 [[SEXT]], 32
; ALL-NEXT: ret i64 [[F]]
;
diff --git a/llvm/test/Transforms/InstCombine/gep-combine-loop-invariant.ll b/llvm/test/Transforms/InstCombine/gep-combine-loop-invariant.ll
index 081ba679bf53e..8a4c45a882423 100644
--- a/llvm/test/Transforms/InstCombine/gep-combine-loop-invariant.ll
+++ b/llvm/test/Transforms/InstCombine/gep-combine-loop-invariant.ll
@@ -98,7 +98,7 @@ define void @PR37005(ptr %base, ptr %in) {
; CHECK-NEXT: [[E4:%.*]] = getelementptr inbounds [8 x i8], ptr [[E2]], <2 x i64> <i64 0, i64 1>
; CHECK-NEXT: [[PI1:%.*]] = ptrtoint <2 x ptr> [[E4]] to <2 x i64>
; CHECK-NEXT: [[TMP0:%.*]] = lshr <2 x i64> [[PI1]], splat (i64 14)
-; CHECK-NEXT: [[SL1:%.*]] = and <2 x i64> [[TMP0]], splat (i64 1125899906842496)
+; CHECK-NEXT: [[SL1:%.*]] = and <2 x i64> [[TMP0]], splat (i64 -128)
; CHECK-NEXT: [[E5:%.*]] = getelementptr inbounds i8, ptr [[BASE:%.*]], <2 x i64> [[SL1]]
; CHECK-NEXT: [[E6:%.*]] = getelementptr inbounds i8, <2 x ptr> [[E5]], i64 80
; CHECK-NEXT: call void @blackhole(<2 x ptr> [[E6]])
@@ -129,7 +129,7 @@ define void @PR37005_2(ptr %base, ptr %in) {
; CHECK-NEXT: [[E2:%.*]] = getelementptr inbounds nuw i8, ptr [[E1]], i64 48
; CHECK-NEXT: [[PI1:%.*]] = ptrtoint ptr [[E2]] to i64
; CHECK-NEXT: [[TMP0:%.*]] = lshr i64 [[PI1]], 14
-; CHECK-NEXT: [[SL1:%.*]] = and i64 [[TMP0]], 1125899906842496
+; CHECK-NEXT: [[SL1:%.*]] = and i64 [[TMP0]], -128
; CHECK-NEXT: [[E5:%.*]] = getelementptr inbounds nuw i8, ptr [[BASE:%.*]], i64 [[SL1]]
; CHECK-NEXT: [[E6:%.*]] = getelementptr inbounds i8, ptr [[E5]], <2 x i64> <i64 80, i64 60>
; CHECK-NEXT: call void @blackhole(<2 x ptr> [[E6]])
@@ -160,7 +160,7 @@ define void @PR37005_3(<2 x ptr> %base, ptr %in) {
; CHECK-NEXT: [[E4:%.*]] = getelementptr inbounds [8 x i8], ptr [[E2]], <2 x i64> <i64 0, i64 1>
; CHECK-NEXT: [[PI1:%.*]] = ptrtoint <2 x ptr> [[E4]] to <2 x i64>
; CHECK-NEXT: [[TMP0:%.*]] = lshr <2 x i64> [[PI1]], splat (i64 14)
-; CHECK-NEXT: [[SL1:%.*]] = and <2 x i64> [[TMP0]], splat (i64 1125899906842496)
+; CHECK-NEXT: [[SL1:%.*]] = and <2 x i64> [[TMP0]], splat (i64 -128)
; CHECK-NEXT: [[E5:%.*]] = getelementptr inbounds i8, <2 x ptr> [[BASE:%.*]], <2 x i64> [[SL1]]
; CHECK-NEXT: [[E6:%.*]] = getelementptr inbounds i8, <2 x ptr> [[E5]], i64 80
; CHECK-NEXT: call void @blackhole(<2 x ptr> [[E6]])
diff --git a/llvm/test/Transforms/LoopUnroll/runtime-unroll-remainder.ll b/llvm/test/Transforms/LoopUnroll/runtime-unroll-remainder.ll
index 3c9c3bd10a93c..5a2c6037461a3 100644
--- a/llvm/test/Transforms/LoopUnroll/runtime-unroll-remainder.ll
+++ b/llvm/test/Transforms/LoopUnroll/runtime-unroll-remainder.ll
@@ -13,7 +13,7 @@ define i32 @unroll(ptr nocapture readonly %a, ptr nocapture readonly %b, i32 %N)
; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 [[N]], 4
; CHECK-NEXT: br i1 [[TMP0]], label [[FOR_BODY_EPIL_PREHEADER:%.*]], label [[FOR_BODY_LR_PH_NEW:%.*]]
; CHECK: for.body.lr.ph.new:
-; CHECK-NEXT: [[UNROLL_ITER:%.*]] = and i64 [[WIDE_TRIP_COUNT]], 4294967292
+; CHECK-NEXT: [[UNROLL_ITER:%.*]] = and i64 [[WIDE_TRIP_COUNT]], -4
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.cond.cleanup.loopexit.unr-lcssa:
; CHECK-NEXT: [[LCMP_MOD_NOT:%.*]] = icmp eq i64 [[XTRAITER]], 0
diff --git a/llvm/test/Transforms/LoopVectorize/X86/x86-interleaved-store-accesses-with-gaps.ll b/llvm/test/Transforms/LoopVectorize/X86/x86-interleaved-store-accesses-with-gaps.ll
index 545b845bead7d..969e8fe8a3d87 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/x86-interleaved-store-accesses-with-gaps.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/x86-interleaved-store-accesses-with-gaps.ll
@@ -141,7 +141,7 @@ define void @test2(ptr noalias nocapture %points, i32 %numPoints, ptr noalias no
; DISABLED_MASKED_STRIDED: [[VECTOR_PH]]:
; DISABLED_MASKED_STRIDED-NEXT: [[WIDE_TRIP_COUNT:%.*]] = zext nneg i32 [[NUMPOINTS]] to i64
; DISABLED_MASKED_STRIDED-NEXT: [[N_RND_UP:%.*]] = add nuw nsw i64 [[WIDE_TRIP_COUNT]], 3
-; DISABLED_MASKED_STRIDED-NEXT: [[N_VEC:%.*]] = and i64 [[N_RND_UP]], 4294967292
+; DISABLED_MASKED_STRIDED-NEXT: [[N_VEC:%.*]] = and i64 [[N_RND_UP]], -4
; DISABLED_MASKED_STRIDED-NEXT: [[TRIP_COUNT_MINUS_1:%.*]] = add nsw i64 [[WIDE_TRIP_COUNT]], -1
; DISABLED_MASKED_STRIDED-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[TRIP_COUNT_MINUS_1]], i64 0
; DISABLED_MASKED_STRIDED-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i64> [[BROADCAST_SPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer
@@ -245,7 +245,7 @@ define void @test2(ptr noalias nocapture %points, i32 %numPoints, ptr noalias no
; ENABLED_MASKED_STRIDED: [[VECTOR_PH]]:
; ENABLED_MASKED_STRIDED-NEXT: [[WIDE_TRIP_COUNT:%.*]] = zext nneg i32 [[NUMPOINTS]] to i64
; ENABLED_MASKED_STRIDED-NEXT: [[N_RND_UP:%.*]] = add nuw nsw i64 [[WIDE_TRIP_COUNT]], 3
-; ENABLED_MASKED_STRIDED-NEXT: [[N_VEC:%.*]] = and i64 [[N_RND_UP]], 4294967292
+; ENABLED_MASKED_STRIDED-NEXT: [[N_VEC:%.*]] = and i64 [[N_RND_UP]], -4
; ENABLED_MASKED_STRIDED-NEXT: [[TRIP_COUNT_MINUS_1:%.*]] = add nsw i64 [[WIDE_TRIP_COUNT]], -1
; ENABLED_MASKED_STRIDED-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[TRIP_COUNT_MINUS_1]], i64 0
; ENABLED_MASKED_STRIDED-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i64> [[BROADCAST_SPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer
diff --git a/llvm/test/Transforms/LoopVectorize/float-induction.ll b/llvm/test/Transforms/LoopVectorize/float-induction.ll
index f20b1cf8ed5ce..1a54a1f48b026 100644
--- a/llvm/test/Transforms/LoopVectorize/float-induction.ll
+++ b/llvm/test/Transforms/LoopVectorize/float-induction.ll
@@ -1062,13 +1062,13 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca
; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[CMP_N]], label %[[FOR_END_LOOPEXIT:.*]], label %[[SCALAR_PH]]
; VEC2_INTERL1_PRED_STORE: [[SCALAR_PH]]:
; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[FOR_BODY_LR_PH]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL10:%.*]] = phi float [ [[TMP3]], %[[MIDDLE_BLOCK]] ], [ 1.000000e-01, %[[FOR_BODY_LR_PH]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL11:%.*]] = phi float [ [[TMP5]], %[[MIDDLE_BLOCK]] ], [ [[INIT]], %[[FOR_BODY_LR_PH]] ]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL9:%.*]] = phi float [ [[TMP3]], %[[MIDDLE_BLOCK]] ], [ 1.000000e-01, %[[FOR_BODY_LR_PH]] ]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL10:%.*]] = phi float [ [[TMP5]], %[[MIDDLE_BLOCK]] ], [ [[INIT]], %[[FOR_BODY_LR_PH]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: br label %[[FOR_BODY:.*]]
; VEC2_INTERL1_PRED_STORE: [[FOR_BODY]]:
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[Y_012:%.*]] = phi float [ [[BC_RESUME_VAL10]], %[[SCALAR_PH]] ], [ [[CONV1:%.*]], %[[FOR_BODY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[X_011:%.*]] = phi float [ [[BC_RESUME_VAL11]], %[[SCALAR_PH]] ], [ [[ADD:%.*]], %[[FOR_BODY]] ]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[Y_012:%.*]] = phi float [ [[BC_RESUME_VAL9]], %[[SCALAR_PH]] ], [ [[CONV1:%.*]], %[[FOR_BODY]] ]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[X_011:%.*]] = phi float [ [[BC_RESUME_VAL10]], %[[SCALAR_PH]] ], [ [[ADD:%.*]], %[[FOR_BODY]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[A]], i64 [[INDVARS_IV]]
; VEC2_INTERL1_PRED_STORE-NEXT: store float [[X_011]], ptr [[ARRAYIDX]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[ADD]] = fadd fast float [[X_011]], [[TMP0]]
@@ -1357,7 +1357,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL1-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 4
; VEC4_INTERL1-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VEC4_INTERL1: [[VECTOR_PH]]:
-; VEC4_INTERL1-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775804
+; VEC4_INTERL1-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -4
; VEC4_INTERL1-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
; VEC4_INTERL1-NEXT: br label %[[VECTOR_BODY:.*]]
; VEC4_INTERL1: [[VECTOR_BODY]]:
@@ -1412,7 +1412,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL1: [[FOR_BODY]]:
; VEC4_INTERL1-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], %[[FOR_INC:.*]] ], [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ]
; VEC4_INTERL1-NEXT: [[J:%.*]] = phi float [ [[J_NEXT:%.*]], %[[FOR_INC]] ], [ [[BC_RESUME_VAL7]], %[[SCALAR_PH]] ]
-; VEC4_INTERL1-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[A]], i64 [[I]]
+; VEC4_INTERL1-NEXT: [[VAR0:%.*]] = getelementptr inbounds [4 x i8], ptr [[A]], i64 [[I]]
; VEC4_INTERL1-NEXT: [[VAR1:%.*]] = load float, ptr [[VAR0]], align 4
; VEC4_INTERL1-NEXT: [[VAR2:%.*]] = fcmp fast oeq float [[VAR1]], 0.000000e+00
; VEC4_INTERL1-NEXT: br i1 [[VAR2]], label %[[IF_PRED:.*]], label %[[FOR_INC]]
@@ -1434,7 +1434,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL2-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 8
; VEC4_INTERL2-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VEC4_INTERL2: [[VECTOR_PH]]:
-; VEC4_INTERL2-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775800
+; VEC4_INTERL2-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -8
; VEC4_INTERL2-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
; VEC4_INTERL2-NEXT: br label %[[VECTOR_BODY:.*]]
; VEC4_INTERL2: [[VECTOR_BODY]]:
@@ -1528,7 +1528,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL2: [[FOR_BODY]]:
; VEC4_INTERL2-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], %[[FOR_INC:.*]] ], [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ]
; VEC4_INTERL2-NEXT: [[J:%.*]] = phi float [ [[J_NEXT:%.*]], %[[FOR_INC]] ], [ [[BC_RESUME_VAL16]], %[[SCALAR_PH]] ]
-; VEC4_INTERL2-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[A]], i64 [[I]]
+; VEC4_INTERL2-NEXT: [[VAR0:%.*]] = getelementptr inbounds [4 x i8], ptr [[A]], i64 [[I]]
; VEC4_INTERL2-NEXT: [[VAR1:%.*]] = load float, ptr [[VAR0]], align 4
; VEC4_INTERL2-NEXT: [[VAR2:%.*]] = fcmp fast oeq float [[VAR1]], 0.000000e+00
; VEC4_INTERL2-NEXT: br i1 [[VAR2]], label %[[IF_PRED:.*]], label %[[FOR_INC]]
@@ -1550,7 +1550,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC1_INTERL2-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 2
; VEC1_INTERL2-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VEC1_INTERL2: [[VECTOR_PH]]:
-; VEC1_INTERL2-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775806
+; VEC1_INTERL2-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -2
; VEC1_INTERL2-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
; VEC1_INTERL2-NEXT: br label %[[VECTOR_BODY:.*]]
; VEC1_INTERL2: [[VECTOR_BODY]]:
@@ -1587,7 +1587,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC1_INTERL2: [[FOR_BODY]]:
; VEC1_INTERL2-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], %[[FOR_INC:.*]] ], [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ]
; VEC1_INTERL2-NEXT: [[J:%.*]] = phi float [ [[J_NEXT:%.*]], %[[FOR_INC]] ], [ [[BC_RESUME_VAL3]], %[[SCALAR_PH]] ]
-; VEC1_INTERL2-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[A]], i64 [[I]]
+; VEC1_INTERL2-NEXT: [[VAR0:%.*]] = getelementptr inbounds [4 x i8], ptr [[A]], i64 [[I]]
; VEC1_INTERL2-NEXT: [[VAR1:%.*]] = load float, ptr [[VAR0]], align 4
; VEC1_INTERL2-NEXT: [[VAR2:%.*]] = fcmp fast oeq float [[VAR1]], 0.000000e+00
; VEC1_INTERL2-NEXT: br i1 [[VAR2]], label %[[IF_PRED:.*]], label %[[FOR_INC]]
@@ -1609,7 +1609,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC2_INTERL1_PRED_STORE-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 2
; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VEC2_INTERL1_PRED_STORE: [[VECTOR_PH]]:
-; VEC2_INTERL1_PRED_STORE-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775806
+; VEC2_INTERL1_PRED_STORE-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -2
; VEC2_INTERL1_PRED_STORE-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
; VEC2_INTERL1_PRED_STORE-NEXT: br label %[[VECTOR_BODY:.*]]
; VEC2_INTERL1_PRED_STORE: [[VECTOR_BODY]]:
@@ -1641,12 +1641,12 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[CMP_N]], label %[[FOR_END:.*]], label %[[SCALAR_PH]]
; VEC2_INTERL1_PRED_STORE: [[SCALAR_PH]]:
; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL4:%.*]] = phi float [ [[DOTCAST]], %[[MIDDLE_BLOCK]] ], [ 0.000000e+00, %[[ENTRY]] ]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL3:%.*]] = phi float [ [[DOTCAST]], %[[MIDDLE_BLOCK]] ], [ 0.000000e+00, %[[ENTRY]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: br label %[[FOR_BODY:.*]]
; VEC2_INTERL1_PRED_STORE: [[FOR_BODY]]:
; VEC2_INTERL1_PRED_STORE-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], %[[FOR_INC:.*]] ], [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[J:%.*]] = phi float [ [[J_NEXT:%.*]], %[[FOR_INC]] ], [ [[BC_RESUME_VAL4]], %[[SCALAR_PH]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[A]], i64 [[I]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[J:%.*]] = phi float [ [[J_NEXT:%.*]], %[[FOR_INC]] ], [ [[BC_RESUME_VAL3]], %[[SCALAR_PH]] ]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[VAR0:%.*]] = getelementptr inbounds [4 x i8], ptr [[A]], i64 [[I]]
; VEC2_INTERL1_PRED_STORE-NEXT: [[VAR1:%.*]] = load float, ptr [[VAR0]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[VAR2:%.*]] = fcmp fast oeq float [[VAR1]], 0.000000e+00
; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[VAR2]], label %[[IF_PRED:.*]], label %[[FOR_INC]]
@@ -1794,7 +1794,7 @@ define void @fp_iv_used_in_gep_fadd(float %init, ptr noalias nocapture %A, float
; VEC4_INTERL1-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 3
; VEC4_INTERL1-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VEC4_INTERL1: [[VECTOR_PH]]:
-; VEC4_INTERL1-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934588
+; VEC4_INTERL1-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -4
; VEC4_INTERL1-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
; VEC4_INTERL1-NEXT: [[TMP3:%.*]] = fmul fast float [[FPINC]], [[DOTCAST]]
; VEC4_INTERL1-NEXT: [[TMP4:%.*]] = fadd fast float [[INIT]], [[TMP3]]
@@ -1871,7 +1871,7 @@ define void @fp_iv_used_in_gep_fadd(float %init, ptr noalias nocapture %A, float
; VEC4_INTERL2-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 7
; VEC4_INTERL2-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VEC4_INTERL2: [[VECTOR_PH]]:
-; VEC4_INTERL2-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934584
+; VEC4_INTERL2-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -8
; VEC4_INTERL2-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x float> poison, float [[FPINC]], i64 0
; VEC4_INTERL2-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x float> [[BROADCAST_SPLATINSERT]], <4 x float> poison, <4 x i32> zeroinitializer
; VEC4_INTERL2-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
@@ -1972,7 +1972,7 @@ define void @fp_iv_used_in_gep_fadd(float %init, ptr noalias nocapture %A, float
; VEC1_INTERL2-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp eq i32 [[TMP0]], 0
; VEC1_INTERL2-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VEC1_INTERL2: [[VECTOR_PH]]:
-; VEC1_INTERL2-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934590
+; VEC1_INTERL2-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -2
; VEC1_INTERL2-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
; VEC1_INTERL2-NEXT: [[TMP3:%.*]] = fmul fast float [[FPINC]], [[DOTCAST]]
; VEC1_INTERL2-NEXT: [[TMP4:%.*]] = fadd fast float [[INIT]], [[TMP3]]
@@ -2025,7 +2025,7 @@ define void @fp_iv_used_in_gep_fadd(float %init, ptr noalias nocapture %A, float
; VEC2_INTERL1_PRED_STORE-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp eq i32 [[TMP0]], 0
; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VEC2_INTERL1_PRED_STORE: [[VECTOR_PH]]:
-; VEC2_INTERL1_PRED_STORE-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934590
+; VEC2_INTERL1_PRED_STORE-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -2
; VEC2_INTERL1_PRED_STORE-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP3:%.*]] = fmul fast float [[FPINC]], [[DOTCAST]]
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP4:%.*]] = fadd fast float [[INIT]], [[TMP3]]
@@ -2064,11 +2064,11 @@ define void @fp_iv_used_in_gep_fadd(float %init, ptr noalias nocapture %A, float
; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
; VEC2_INTERL1_PRED_STORE: [[SCALAR_PH]]:
; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL7:%.*]] = phi float [ [[TMP4]], %[[MIDDLE_BLOCK]] ], [ [[INIT]], %[[ENTRY]] ]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL5:%.*]] = phi float [ [[TMP4]], %[[MIDDLE_BLOCK]] ], [ [[INIT]], %[[ENTRY]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: br label %[[FOR_BODY:.*]]
; VEC2_INTERL1_PRED_STORE: [[FOR_BODY]]:
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL7]], %[[SCALAR_PH]] ], [ [[ADD:%.*]], %[[FOR_BODY]] ]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL5]], %[[SCALAR_PH]] ], [ [[ADD:%.*]], %[[FOR_BODY]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[C:%.*]] = fptoui float [[X_05]] to i32
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP17:%.*]] = sext i32 [[C]] to i64
; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [4 x i8], ptr [[A]], i64 [[TMP17]]
@@ -2110,7 +2110,7 @@ define void @fp_iv_used_in_gep_fsub(float %init, ptr noalias nocapture %A, float
; VEC4_INTERL1-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 3
; VEC4_INTERL1-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VEC4_INTERL1: [[VECTOR_PH]]:
-; VEC4_INTERL1-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934588
+; VEC4_INTERL1-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -4
; VEC4_INTERL1-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
; VEC4_INTERL1-NEXT: [[TMP3:%.*]] = fmul fast float [[FPINC]], [[DOTCAST]]
; VEC4_INTERL1-NEXT: [[TMP4:%.*]] = fsub fast float [[INIT]], [[TMP3]]
@@ -2187,7 +2187,7 @@ define void @fp_iv_used_in_gep_fsub(float %init, ptr noalias nocapture %A, float
; VEC4_INTERL2-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 7
; VEC4_INTERL2-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VEC4_INTERL2: [[VECTOR_PH]]:
-; VEC4_INTERL2-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934584
+; VEC4_INTERL2-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -8
; VEC4_INTERL2-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x float> poison, float [[FPINC]], i64 0
; VEC4_INTERL2-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x float> [[BROADCAST_SPLATINSERT]], <4 x float> poison, <4 x i32> zeroinitializer
; VEC4_INTERL2-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
@@ -2287,7 +2287,7 @@ define void @fp_iv_used_in_gep_fsub(float %init, ptr noalias nocapture %A, float
; VEC1_INTERL2-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp eq i32 [[TMP0]], 0
; VEC1_INTERL2-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VEC1_INTERL2: [[VECTOR_PH]]:
-; VEC1_INTERL2-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934590
+; VEC1_INTERL2-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -2
; VEC1_INTERL2-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
; VEC1_INTERL2-NEXT: [[TMP3:%.*]] = fmul fast float [[FPINC]], [[DOTCAST]]
; VEC1_INTERL2-NEXT: [[TMP4:%.*]] = fsub fast float [[INIT]], [[TMP3]]
@@ -2340,7 +2340,7 @@ define void @fp_iv_used_in_gep_fsub(float %init, ptr noalias nocapture %A, float
; VEC2_INTERL1_PRED_STORE-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp eq i32 [[TMP0]], 0
; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; VEC2_INTERL1_PRED_STORE: [[VECTOR_PH]]:
-; VEC2_INTERL1_PRED_STORE-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934590
+; VEC2_INTERL1_PRED_STORE-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -2
; VEC2_INTERL1_PRED_STORE-NEXT: [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP3:%.*]] = fmul fast float [[FPINC]], [[DOTCAST]]
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP4:%.*]] = fsub fast float [[INIT]], [[TMP3]]
@@ -2379,11 +2379,11 @@ define void @fp_iv_used_in_gep_fsub(float %init, ptr noalias nocapture %A, float
; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
; VEC2_INTERL1_PRED_STORE: [[SCALAR_PH]]:
; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL7:%.*]] = phi float [ [[TMP4]], %[[MIDDLE_BLOCK]] ], [ [[INIT]], %[[ENTRY]] ]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[BC_RESUME_VAL5:%.*]] = phi float [ [[TMP4]], %[[MIDDLE_BLOCK]] ], [ [[INIT]], %[[ENTRY]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: br label %[[FOR_BODY:.*]]
; VEC2_INTERL1_PRED_STORE: [[FOR_BODY]]:
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL7]], %[[SCALAR_PH]] ], [ [[ADD:%.*]], %[[FOR_BODY]] ]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL5]], %[[SCALAR_PH]] ], [ [[ADD:%.*]], %[[FOR_BODY]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[C:%.*]] = fptoui float [[X_05]] to i32
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP17:%.*]] = sext i32 [[C]] to i64
; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [4 x i8], ptr [[A]], i64 [[TMP17]]
diff --git a/llvm/test/Transforms/LoopVectorize/if-conversion.ll b/llvm/test/Transforms/LoopVectorize/if-conversion.ll
index f3b8305e71630..499bd3874c804 100644
--- a/llvm/test/Transforms/LoopVectorize/if-conversion.ll
+++ b/llvm/test/Transforms/LoopVectorize/if-conversion.ll
@@ -48,7 +48,7 @@ define void @function0(ptr nocapture %a, ptr nocapture %b, i32 %start, i32 %end)
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP4]], 8589934588
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP4]], -4
; CHECK-NEXT: [[TMP12:%.*]] = add nsw i64 [[N_VEC]], [[TMP0]]
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
diff --git a/llvm/test/Transforms/LoopVectorize/induction.ll b/llvm/test/Transforms/LoopVectorize/induction.ll
index 94d6f3a27141b..ca907e5be1b90 100644
--- a/llvm/test/Transforms/LoopVectorize/induction.ll
+++ b/llvm/test/Transforms/LoopVectorize/induction.ll
@@ -60,7 +60,7 @@ define void @multi_int_induction(ptr %A, i32 %N) {
; IND-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp eq i32 [[TMP0]], 0
; IND-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; IND: vector.ph:
-; IND-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934590
+; IND-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -2
; IND-NEXT: [[DOTCAST:%.*]] = trunc i64 [[N_VEC]] to i32
; IND-NEXT: [[IND_END:%.*]] = add i32 [[DOTCAST]], 190
; IND-NEXT: br label [[VECTOR_BODY:%.*]]
@@ -101,7 +101,7 @@ define void @multi_int_induction(ptr %A, i32 %N) {
; UNROLL-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 3
; UNROLL-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; UNROLL: vector.ph:
-; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934588
+; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -4
; UNROLL-NEXT: [[DOTCAST:%.*]] = trunc i64 [[N_VEC]] to i32
; UNROLL-NEXT: [[IND_END:%.*]] = add i32 [[DOTCAST]], 190
; UNROLL-NEXT: br label [[VECTOR_BODY:%.*]]
@@ -190,7 +190,7 @@ define void @multi_int_induction(ptr %A, i32 %N) {
; INTERLEAVE-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 7
; INTERLEAVE-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; INTERLEAVE: vector.ph:
-; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934584
+; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -8
; INTERLEAVE-NEXT: [[DOTCAST:%.*]] = trunc i64 [[N_VEC]] to i32
; INTERLEAVE-NEXT: [[IND_END:%.*]] = add i32 [[DOTCAST]], 190
; INTERLEAVE-NEXT: br label [[VECTOR_BODY:%.*]]
@@ -654,7 +654,7 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; IND-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 2
; IND-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; IND: vector.ph:
-; IND-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775806
+; IND-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -2
; IND-NEXT: br label [[VECTOR_BODY:%.*]]
; IND: vector.body:
; IND-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -676,7 +676,7 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; IND: for.body:
; IND-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; IND-NEXT: [[SUM:%.*]] = phi i64 [ [[TMP6:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; IND-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [8 x i8], ptr [[A]], i64 [[I]]
+; IND-NEXT: [[TMP4:%.*]] = getelementptr inbounds [8 x i8], ptr [[A]], i64 [[I]]
; IND-NEXT: [[TMP5:%.*]] = load i64, ptr [[TMP4]], align 8
; IND-NEXT: [[TMP6]] = add i64 [[TMP5]], [[SUM]]
; IND-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -692,7 +692,7 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; UNROLL-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 4
; UNROLL-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; UNROLL: vector.ph:
-; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775804
+; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -4
; UNROLL-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL: vector.body:
; UNROLL-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -719,7 +719,7 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; UNROLL: for.body:
; UNROLL-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[SUM:%.*]] = phi i64 [ [[TMP8:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; UNROLL-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [8 x i8], ptr [[A]], i64 [[I]]
+; UNROLL-NEXT: [[TMP6:%.*]] = getelementptr inbounds [8 x i8], ptr [[A]], i64 [[I]]
; UNROLL-NEXT: [[TMP7:%.*]] = load i64, ptr [[TMP6]], align 8
; UNROLL-NEXT: [[TMP8]] = add i64 [[TMP7]], [[SUM]]
; UNROLL-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -779,7 +779,7 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; INTERLEAVE-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 8
; INTERLEAVE-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; INTERLEAVE: vector.ph:
-; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775800
+; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -8
; INTERLEAVE-NEXT: br label [[VECTOR_BODY:%.*]]
; INTERLEAVE: vector.body:
; INTERLEAVE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -806,7 +806,7 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; INTERLEAVE: for.body:
; INTERLEAVE-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[SUM:%.*]] = phi i64 [ [[TMP8:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; INTERLEAVE-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [8 x i8], ptr [[A]], i64 [[I]]
+; INTERLEAVE-NEXT: [[TMP6:%.*]] = getelementptr inbounds [8 x i8], ptr [[A]], i64 [[I]]
; INTERLEAVE-NEXT: [[TMP7:%.*]] = load i64, ptr [[TMP6]], align 8
; INTERLEAVE-NEXT: [[TMP8]] = add i64 [[TMP7]], [[SUM]]
; INTERLEAVE-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -916,7 +916,7 @@ define float @scalarize_induction_variable_02(ptr %a, ptr %b, i64 %n) {
; IND-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp eq i64 [[TMP1]], 0
; IND-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; IND: vector.ph:
-; IND-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 4611686018427387902
+; IND-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -2
; IND-NEXT: [[IND_END:%.*]] = shl i64 [[N_VEC]], 3
; IND-NEXT: br label [[VECTOR_BODY:%.*]]
; IND: vector.body:
@@ -976,7 +976,7 @@ define float @scalarize_induction_variable_02(ptr %a, ptr %b, i64 %n) {
; UNROLL-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 25
; UNROLL-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; UNROLL: vector.ph:
-; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 4611686018427387900
+; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -4
; UNROLL-NEXT: [[IND_END:%.*]] = shl i64 [[N_VEC]], 3
; UNROLL-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL: vector.body:
@@ -1274,7 +1274,7 @@ define void @scalarize_induction_variable_03(ptr %p, i32 %y, i64 %n) {
; IND-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 2
; IND-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; IND: vector.ph:
-; IND-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775806
+; IND-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -2
; IND-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[Y:%.*]], i64 0
; IND-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i32> [[BROADCAST_SPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer
; IND-NEXT: br label [[VECTOR_BODY:%.*]]
@@ -1304,7 +1304,7 @@ define void @scalarize_induction_variable_03(ptr %p, i32 %y, i64 %n) {
; IND-NEXT: br label [[FOR_BODY:%.*]]
; IND: for.body:
; IND-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; IND-NEXT: [[F_SPLIT:%.*]] = getelementptr inbounds nuw [8 x i8], ptr [[P]], i64 [[I]]
+; IND-NEXT: [[F_SPLIT:%.*]] = getelementptr inbounds [8 x i8], ptr [[P]], i64 [[I]]
; IND-NEXT: [[F:%.*]] = getelementptr inbounds nuw i8, ptr [[F_SPLIT]], i64 4
; IND-NEXT: [[TMP11:%.*]] = load i32, ptr [[F]], align 8
; IND-NEXT: [[TMP12:%.*]] = xor i32 [[TMP11]], [[Y]]
@@ -1321,7 +1321,7 @@ define void @scalarize_induction_variable_03(ptr %p, i32 %y, i64 %n) {
; UNROLL-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 4
; UNROLL-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; UNROLL: vector.ph:
-; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775804
+; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -4
; UNROLL-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[Y:%.*]], i64 0
; UNROLL-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i32> [[BROADCAST_SPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer
; UNROLL-NEXT: br label [[VECTOR_BODY:%.*]]
@@ -1364,7 +1364,7 @@ define void @scalarize_induction_variable_03(ptr %p, i32 %y, i64 %n) {
; UNROLL-NEXT: br label [[FOR_BODY:%.*]]
; UNROLL: for.body:
; UNROLL-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; UNROLL-NEXT: [[F_SPLIT:%.*]] = getelementptr inbounds nuw [8 x i8], ptr [[P]], i64 [[I]]
+; UNROLL-NEXT: [[F_SPLIT:%.*]] = getelementptr inbounds [8 x i8], ptr [[P]], i64 [[I]]
; UNROLL-NEXT: [[F:%.*]] = getelementptr inbounds nuw i8, ptr [[F_SPLIT]], i64 4
; UNROLL-NEXT: [[TMP22:%.*]] = load i32, ptr [[F]], align 8
; UNROLL-NEXT: [[TMP23:%.*]] = xor i32 [[TMP22]], [[Y]]
@@ -1618,7 +1618,7 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) {
; IND-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; IND-NEXT: br i1 [[FOUND_CONFLICT]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]]
; IND: vector.ph:
-; IND-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934590
+; IND-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -2
; IND-NEXT: br label [[VECTOR_BODY:%.*]]
; IND: vector.body:
; IND-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -1649,9 +1649,9 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) {
; IND: for.body:
; IND-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; IND-NEXT: [[DOTIDX:%.*]] = shl nsw i64 [[I]], 4
-; IND-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[DOTIDX]]
+; IND-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[DOTIDX]]
; IND-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 1
-; IND-NEXT: [[DOTSPLIT4:%.*]] = getelementptr inbounds nuw [8 x i8], ptr [[P]], i64 [[I]]
+; IND-NEXT: [[DOTSPLIT4:%.*]] = getelementptr inbounds [8 x i8], ptr [[P]], i64 [[I]]
; IND-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT4]], i64 4
; IND-NEXT: store i32 [[TMP21]], ptr [[TMP22]], align 1
; IND-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -1683,7 +1683,7 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) {
; UNROLL-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; UNROLL-NEXT: br i1 [[FOUND_CONFLICT]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]]
; UNROLL: vector.ph:
-; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934588
+; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -4
; UNROLL-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL: vector.body:
; UNROLL-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -1728,9 +1728,9 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) {
; UNROLL: for.body:
; UNROLL-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[DOTIDX:%.*]] = shl nsw i64 [[I]], 4
-; UNROLL-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[DOTIDX]]
+; UNROLL-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[DOTIDX]]
; UNROLL-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 1
-; UNROLL-NEXT: [[DOTSPLIT6:%.*]] = getelementptr inbounds nuw [8 x i8], ptr [[P]], i64 [[I]]
+; UNROLL-NEXT: [[DOTSPLIT6:%.*]] = getelementptr inbounds [8 x i8], ptr [[P]], i64 [[I]]
; UNROLL-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT6]], i64 4
; UNROLL-NEXT: store i32 [[TMP32]], ptr [[TMP33]], align 1
; UNROLL-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -2461,7 +2461,7 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) {
; IND-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp eq i32 [[TMP0]], 0
; IND-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; IND: vector.ph:
-; IND-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934590
+; IND-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -2
; IND-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[A:%.*]], i64 0
; IND-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i32> [[BROADCAST_SPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer
; IND-NEXT: br label [[VECTOR_BODY:%.*]]
@@ -2493,7 +2493,7 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) {
; IND-NEXT: [[TMP11:%.*]] = trunc i64 [[I]] to i32
; IND-NEXT: [[TMP12:%.*]] = add i32 [[A]], [[TMP11]]
; IND-NEXT: [[TMP13:%.*]] = trunc i32 [[TMP12]] to i16
-; IND-NEXT: [[DOTSPLIT2:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[P]], i64 [[I]]
+; IND-NEXT: [[DOTSPLIT2:%.*]] = getelementptr inbounds [4 x i8], ptr [[P]], i64 [[I]]
; IND-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT2]], i64 2
; IND-NEXT: store i16 [[TMP13]], ptr [[TMP14]], align 2
; IND-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -2511,7 +2511,7 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) {
; UNROLL-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 3
; UNROLL-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; UNROLL: vector.ph:
-; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934588
+; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -4
; UNROLL-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i32> poison, i32 [[A:%.*]], i64 0
; UNROLL-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i32> [[BROADCAST_SPLATINSERT]], <2 x i32> poison, <2 x i32> zeroinitializer
; UNROLL-NEXT: br label [[VECTOR_BODY:%.*]]
@@ -2554,7 +2554,7 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) {
; UNROLL-NEXT: [[TMP19:%.*]] = trunc i64 [[I]] to i32
; UNROLL-NEXT: [[TMP20:%.*]] = add i32 [[A]], [[TMP19]]
; UNROLL-NEXT: [[TMP21:%.*]] = trunc i32 [[TMP20]] to i16
-; UNROLL-NEXT: [[DOTSPLIT4:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[P]], i64 [[I]]
+; UNROLL-NEXT: [[DOTSPLIT4:%.*]] = getelementptr inbounds [4 x i8], ptr [[P]], i64 [[I]]
; UNROLL-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT4]], i64 2
; UNROLL-NEXT: store i16 [[TMP21]], ptr [[TMP22]], align 2
; UNROLL-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -2632,7 +2632,7 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) {
; INTERLEAVE-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 7
; INTERLEAVE-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; INTERLEAVE: vector.ph:
-; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934584
+; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -8
; INTERLEAVE-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[A:%.*]], i64 0
; INTERLEAVE-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
; INTERLEAVE-NEXT: br label [[VECTOR_BODY:%.*]]
@@ -2691,7 +2691,7 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) {
; INTERLEAVE-NEXT: [[TMP31:%.*]] = trunc i64 [[I]] to i32
; INTERLEAVE-NEXT: [[TMP32:%.*]] = add i32 [[A]], [[TMP31]]
; INTERLEAVE-NEXT: [[TMP33:%.*]] = trunc i32 [[TMP32]] to i16
-; INTERLEAVE-NEXT: [[DOTSPLIT8:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[P]], i64 [[I]]
+; INTERLEAVE-NEXT: [[DOTSPLIT8:%.*]] = getelementptr inbounds [4 x i8], ptr [[P]], i64 [[I]]
; INTERLEAVE-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT8]], i64 2
; INTERLEAVE-NEXT: store i16 [[TMP33]], ptr [[TMP34]], align 2
; INTERLEAVE-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -4267,7 +4267,7 @@ define void @trunciv(ptr nocapture %a, i32 %start, i64 %k) {
; IND-NEXT: [[DOTNOT:%.*]] = icmp ult i64 [[K]], 2147483649
; IND-NEXT: br i1 [[DOTNOT]], label [[VECTOR_PH:%.*]], label [[SCALAR_PH]]
; IND: vector.ph:
-; IND-NEXT: [[N_VEC:%.*]] = and i64 [[K]], 4294967294
+; IND-NEXT: [[N_VEC:%.*]] = and i64 [[K]], -2
; IND-NEXT: br label [[VECTOR_BODY:%.*]]
; IND: vector.body:
; IND-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -4307,7 +4307,7 @@ define void @trunciv(ptr nocapture %a, i32 %start, i64 %k) {
; UNROLL-NEXT: [[DOTNOT:%.*]] = icmp ult i64 [[K]], 2147483649
; UNROLL-NEXT: br i1 [[DOTNOT]], label [[VECTOR_PH:%.*]], label [[SCALAR_PH]]
; UNROLL: vector.ph:
-; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[K]], 4294967292
+; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[K]], -4
; UNROLL-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL: vector.body:
; UNROLL-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -4395,7 +4395,7 @@ define void @trunciv(ptr nocapture %a, i32 %start, i64 %k) {
; INTERLEAVE-NEXT: [[DOTNOT:%.*]] = icmp ult i64 [[K]], 2147483649
; INTERLEAVE-NEXT: br i1 [[DOTNOT]], label [[VECTOR_PH:%.*]], label [[SCALAR_PH]]
; INTERLEAVE: vector.ph:
-; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i64 [[K]], 4294967288
+; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i64 [[K]], -8
; INTERLEAVE-NEXT: br label [[VECTOR_BODY:%.*]]
; INTERLEAVE: vector.body:
; INTERLEAVE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -4717,7 +4717,7 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; IND-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 2
; IND-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; IND: vector.ph:
-; IND-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775806
+; IND-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -2
; IND-NEXT: [[IND_END:%.*]] = shl nuw i64 [[N_VEC]], 1
; IND-NEXT: br label [[VECTOR_BODY:%.*]]
; IND: vector.body:
@@ -4739,7 +4739,7 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; IND: for.body:
; IND-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; IND-NEXT: [[J:%.*]] = phi i64 [ [[J_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; IND-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[A]], i64 [[I]]
+; IND-NEXT: [[VAR0:%.*]] = getelementptr inbounds [4 x i8], ptr [[A]], i64 [[I]]
; IND-NEXT: [[VAR1:%.*]] = trunc i64 [[J]] to i32
; IND-NEXT: store i32 [[VAR1]], ptr [[VAR0]], align 4
; IND-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -4755,7 +4755,7 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; UNROLL-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 4
; UNROLL-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; UNROLL: vector.ph:
-; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775804
+; UNROLL-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -4
; UNROLL-NEXT: [[IND_END:%.*]] = shl nuw i64 [[N_VEC]], 1
; UNROLL-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL: vector.body:
@@ -4780,7 +4780,7 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; UNROLL: for.body:
; UNROLL-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[J:%.*]] = phi i64 [ [[J_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; UNROLL-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[A]], i64 [[I]]
+; UNROLL-NEXT: [[VAR0:%.*]] = getelementptr inbounds [4 x i8], ptr [[A]], i64 [[I]]
; UNROLL-NEXT: [[VAR1:%.*]] = trunc i64 [[J]] to i32
; UNROLL-NEXT: store i32 [[VAR1]], ptr [[VAR0]], align 4
; UNROLL-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -4838,7 +4838,7 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; INTERLEAVE-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 8
; INTERLEAVE-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; INTERLEAVE: vector.ph:
-; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775800
+; INTERLEAVE-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -8
; INTERLEAVE-NEXT: [[IND_END:%.*]] = shl nuw i64 [[N_VEC]], 1
; INTERLEAVE-NEXT: br label [[VECTOR_BODY:%.*]]
; INTERLEAVE: vector.body:
@@ -4863,7 +4863,7 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; INTERLEAVE: for.body:
; INTERLEAVE-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[J:%.*]] = phi i64 [ [[J_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; INTERLEAVE-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[A]], i64 [[I]]
+; INTERLEAVE-NEXT: [[VAR0:%.*]] = getelementptr inbounds [4 x i8], ptr [[A]], i64 [[I]]
; INTERLEAVE-NEXT: [[VAR1:%.*]] = trunc i64 [[J]] to i32
; INTERLEAVE-NEXT: store i32 [[VAR1]], ptr [[VAR0]], align 4
; INTERLEAVE-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
diff --git a/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll b/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
index 45b693e455ea9..b1c71be2cf36b 100644
--- a/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
+++ b/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
@@ -1252,7 +1252,7 @@ define void @PR27626_4(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 7
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 9223372036854775804
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -4
; CHECK-NEXT: [[IND_END:%.*]] = shl nuw i64 [[N_VEC]], 1
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[Y:%.*]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
@@ -1338,7 +1338,7 @@ define void @PR27626_5(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 10
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 9223372036854775804
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -4
; CHECK-NEXT: [[TMP3:%.*]] = shl nuw i64 [[N_VEC]], 1
; CHECK-NEXT: [[IND_END:%.*]] = or disjoint i64 [[TMP3]], 3
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
diff --git a/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization-2.ll b/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization-2.ll
index 392328137f089..078b9e2895f77 100644
--- a/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization-2.ll
+++ b/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization-2.ll
@@ -33,7 +33,7 @@ define void @inv_val_store_to_inv_address_conditional_diff_values_ic(ptr %a, i64
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX2]], 9223372036854775804
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX2]], -4
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[NTRUNC]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
; CHECK-NEXT: [[BROADCAST_SPLATINSERT3:%.*]] = insertelement <4 x i32> poison, i32 [[K:%.*]], i64 0
@@ -59,7 +59,7 @@ define void @inv_val_store_to_inv_address_conditional_diff_values_ic(ptr %a, i64
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[LATCH:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds [4 x i8], ptr [[B]], i64 [[I]]
; CHECK-NEXT: [[I2:%.*]] = load i32, ptr [[I1]], align 8
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[I2]], [[K]]
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[I1]], align 4
@@ -133,7 +133,7 @@ define void @inv_val_store_to_inv_address_conditional_inv(ptr %a, i64 %n, ptr %b
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX2]], 9223372036854775804
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX2]], -4
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[NTRUNC]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
@@ -153,7 +153,7 @@ define void @inv_val_store_to_inv_address_conditional_inv(ptr %a, i64 %n, ptr %b
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[LATCH:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds [4 x i8], ptr [[B]], i64 [[I]]
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[I1]], align 4
; CHECK-NEXT: br i1 [[CMP]], label [[COND_STORE:%.*]], label [[COND_STORE_K:%.*]]
; CHECK: cond_store:
@@ -218,7 +218,7 @@ define i32 @variant_val_store_to_inv_address(ptr %a, i64 %n, ptr %b, i32 %k) {
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX2]], 9223372036854775804
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX2]], -4
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -243,7 +243,7 @@ define i32 @variant_val_store_to_inv_address(ptr %a, i64 %n, ptr %b, i32 %k) {
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[I0:%.*]] = phi i32 [ [[I3:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds [4 x i8], ptr [[B]], i64 [[I]]
; CHECK-NEXT: [[I2:%.*]] = load i32, ptr [[I1]], align 8
; CHECK-NEXT: store i32 [[I2]], ptr [[A]], align 4
; CHECK-NEXT: [[I3]] = add i32 [[I0]], [[I2]]
diff --git a/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization.ll b/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization.ll
index 601323ae35345..418030a2d1798 100644
--- a/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization.ll
+++ b/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization.ll
@@ -32,7 +32,7 @@ define i32 @inv_val_store_to_inv_address_with_reduction(ptr %a, i64 %n, ptr %b)
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX2]], 9223372036854775804
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX2]], -4
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -56,7 +56,7 @@ define i32 @inv_val_store_to_inv_address_with_reduction(ptr %a, i64 %n, ptr %b)
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[I0:%.*]] = phi i32 [ [[I3:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds [4 x i8], ptr [[B]], i64 [[I]]
; CHECK-NEXT: [[I2:%.*]] = load i32, ptr [[I1]], align 8
; CHECK-NEXT: [[I3]] = add i32 [[I0]], [[I2]]
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[A]], align 4
@@ -106,7 +106,7 @@ define void @inv_val_store_to_inv_address(ptr %a, i64 %n, ptr %b) {
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX2]], 9223372036854775804
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX2]], -4
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[NTRUNC]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
@@ -126,7 +126,7 @@ define void @inv_val_store_to_inv_address(ptr %a, i64 %n, ptr %b) {
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds [4 x i8], ptr [[B]], i64 [[I]]
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[A]], align 4
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[I1]], align 4
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -181,7 +181,7 @@ define void @inv_val_store_to_inv_address_conditional(ptr %a, i64 %n, ptr %b, i3
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX2]], 9223372036854775804
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX2]], -4
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[NTRUNC]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
; CHECK-NEXT: [[BROADCAST_SPLATINSERT3:%.*]] = insertelement <4 x i32> poison, i32 [[K:%.*]], i64 0
@@ -228,7 +228,7 @@ define void @inv_val_store_to_inv_address_conditional(ptr %a, i64 %n, ptr %b, i3
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[LATCH:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds [4 x i8], ptr [[B]], i64 [[I]]
; CHECK-NEXT: [[I2:%.*]] = load i32, ptr [[I1]], align 8
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[I2]], [[K]]
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[I1]], align 4
@@ -379,7 +379,7 @@ define void @multiple_uniform_stores(ptr nocapture %var1, ptr nocapture readonly
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP8]], 8589934588
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP8]], -4
; CHECK-NEXT: [[IND_END:%.*]] = add nuw nsw i64 [[N_VEC]], [[TMP4]]
; CHECK-NEXT: [[TMP15:%.*]] = insertelement <4 x i32> <i32 poison, i32 0, i32 0, i32 0>, i32 [[ARRAYIDX5_PROMOTED]], i64 0
; CHECK-NEXT: [[INVARIANT_GEP:%.*]] = getelementptr [4 x i8], ptr [[VAR2]], i64 [[TMP4]]
diff --git a/llvm/test/Transforms/LoopVectorize/loop-scalars.ll b/llvm/test/Transforms/LoopVectorize/loop-scalars.ll
index 1cfe27f3f8db1..79dfb784d3806 100644
--- a/llvm/test/Transforms/LoopVectorize/loop-scalars.ll
+++ b/llvm/test/Transforms/LoopVectorize/loop-scalars.ll
@@ -11,7 +11,7 @@ define void @vector_gep(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 2
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775806
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -2
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -31,8 +31,8 @@ define void @vector_gep(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[B]], i64 [[I]]
-; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds nuw [8 x i8], ptr [[A]], i64 [[I]]
+; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds [4 x i8], ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds [8 x i8], ptr [[A]], i64 [[I]]
; CHECK-NEXT: store ptr [[VAR0]], ptr [[VAR1]], align 8
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
@@ -67,7 +67,7 @@ define void @scalar_store(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp eq i64 [[TMP1]], 0
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 9223372036854775806
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -2
; CHECK-NEXT: [[IND_END:%.*]] = shl nuw i64 [[N_VEC]], 1
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
@@ -127,7 +127,7 @@ define void @expansion(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp eq i64 [[TMP1]], 0
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], 9223372036854775806
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], -2
; CHECK-NEXT: [[IND_END:%.*]] = shl nuw i64 [[N_VEC]], 1
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
@@ -184,7 +184,7 @@ define void @no_gep_or_bitcast(ptr noalias %a, i64 %n) {
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 2
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775806
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -2
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -205,7 +205,7 @@ define void @no_gep_or_bitcast(ptr noalias %a, i64 %n) {
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw [8 x i8], ptr [[A]], i64 [[I]]
+; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds [8 x i8], ptr [[A]], i64 [[I]]
; CHECK-NEXT: [[VAR1:%.*]] = load ptr, ptr [[VAR0]], align 8
; CHECK-NEXT: store i32 0, ptr [[VAR1]], align 8
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
diff --git a/llvm/test/Transforms/LoopVectorize/vector-geps.ll b/llvm/test/Transforms/LoopVectorize/vector-geps.ll
index 8c2f23cd71d98..fe879c646354b 100644
--- a/llvm/test/Transforms/LoopVectorize/vector-geps.ll
+++ b/llvm/test/Transforms/LoopVectorize/vector-geps.ll
@@ -11,7 +11,7 @@ define void @vector_gep_stored(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 4
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775804
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -4
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -31,8 +31,8 @@ define void @vector_gep_stored(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw [4 x i8], ptr [[B]], i64 [[I]]
-; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds nuw [8 x i8], ptr [[A]], i64 [[I]]
+; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds [4 x i8], ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds [8 x i8], ptr [[A]], i64 [[I]]
; CHECK-NEXT: store ptr [[VAR0]], ptr [[VAR1]], align 8
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
@@ -64,7 +64,7 @@ define void @uniform_vector_gep_stored(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp slt i64 [[N]], 4
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775804
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[SMAX]], -4
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 4
; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <4 x ptr> poison, ptr [[TMP0]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <4 x ptr> [[DOTSPLATINSERT]], <4 x ptr> poison, <4 x i32> zeroinitializer
@@ -85,7 +85,7 @@ define void @uniform_vector_gep_stored(ptr %a, ptr %b, i64 %n) {
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 4
-; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds nuw [8 x i8], ptr [[A]], i64 [[I]]
+; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds [8 x i8], ptr [[A]], i64 [[I]]
; CHECK-NEXT: store ptr [[VAR0]], ptr [[VAR1]], align 8
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/pr48844-br-to-switch-vectorization.ll b/llvm/test/Transforms/PhaseOrdering/X86/pr48844-br-to-switch-vectorization.ll
index b2e07e77b05c5..5968b29ef6845 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/pr48844-br-to-switch-vectorization.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/pr48844-br-to-switch-vectorization.ll
@@ -47,7 +47,7 @@ define dso_local void @test(ptr %start, ptr %end) #0 {
; AVX2-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[BB12_PREHEADER11:%.*]], label [[VECTOR_PH:%.*]]
; AVX2: vector.ph:
; AVX2-NEXT: [[N_VEC_REMAINING:%.*]] = and i64 [[TMP3]], 24
-; AVX2-NEXT: [[N_VEC:%.*]] = and i64 [[TMP3]], 9223372036854775776
+; AVX2-NEXT: [[N_VEC:%.*]] = and i64 [[TMP3]], -32
; AVX2-NEXT: [[TMP4:%.*]] = shl i64 [[N_VEC]], 2
; AVX2-NEXT: [[IND_END11:%.*]] = getelementptr i8, ptr [[START]], i64 [[TMP4]]
; AVX2-NEXT: br label [[VECTOR_BODY:%.*]]
@@ -89,7 +89,7 @@ define dso_local void @test(ptr %start, ptr %end) #0 {
; AVX2-NEXT: br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[BB12_PREHEADER1]], label [[BB12_PREHEADER11]], !prof [[PROF3:![0-9]+]]
; AVX2: vec.epilog.ph:
; AVX2-NEXT: [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ]
-; AVX2-NEXT: [[N_VEC10:%.*]] = and i64 [[TMP3]], 9223372036854775800
+; AVX2-NEXT: [[N_VEC10:%.*]] = and i64 [[TMP3]], -8
; AVX2-NEXT: [[TMP21:%.*]] = shl i64 [[N_VEC10]], 2
; AVX2-NEXT: [[IND_END:%.*]] = getelementptr i8, ptr [[START]], i64 [[TMP21]]
; AVX2-NEXT: br label [[BB12:%.*]]
More information about the llvm-commits
mailing list