Index: llvm/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
===================================================================
--- llvm/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
+++ llvm/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
@@ -223,9 +223,17 @@
   // Simplify mul instructions with a constant RHS.
   if (isa<Constant>(Op1)) {
     // Canonicalize (X+C1)*CI -> X*CI+C1*CI.
+    // Canonicalize (X|C1)*CI -> X*CI+C1*CI.
     Value *X;
     Constant *C1;
-    if (match(Op0, m_OneUse(m_Add(m_Value(X), m_Constant(C1))))) {
+    const APInt *C1Val, *C2;
+
+    if ((match(Op0, m_OneUse(m_Add(m_Value(X), m_Constant(C1))))) ||
+        (match(Op0, m_OneUse(m_Or(m_Value(X), m_Constant(C1)))) &&
+         match(C1, m_APInt(C1Val)) &&
+         ((match(X, m_Shl(m_Value(), m_APInt(C2))) &&
+           (C1Val->getZExtValue() < (1 << C2->getZExtValue()))) ||
+          match(X, m_Mul(m_Value(), m_APInt(C2)))))) {
       Value *Mul = Builder.CreateMul(C1, Op1);
       // Only go forward with the transform if C1*CI simplifies to a tidier
       // constant.
Index: llvm/test/Transforms/InstCombine/distributive_const.ll
===================================================================
--- /dev/null
+++ llvm/test/Transforms/InstCombine/distributive_const.ll
@@ -0,0 +1,208 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -passes=instcombine -S | FileCheck %s
+
+define void @foo_1(float* noalias noundef %fi, float* noalias noundef %f, i32 noundef %ci) {
+; CHECK-LABEL: @foo_1(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load float, float* [[FI:%.*]], align 4
+; CHECK-NEXT:    [[MUL1:%.*]] = mul nsw i32 [[CI:%.*]], 12
+; CHECK-NEXT:    [[IDXPROM:%.*]] = sext i32 [[MUL1]] to i64
+; CHECK-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds float, float* [[F:%.*]], i64 [[IDXPROM]]
+; CHECK-NEXT:    [[TMP1:%.*]] = load float, float* [[ARRAYIDX3]], align 4
+; CHECK-NEXT:    [[ADD4:%.*]] = fadd fast float [[TMP1]], [[TMP0]]
+; CHECK-NEXT:    store float [[ADD4]], float* [[ARRAYIDX3]], align 4
+; CHECK-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds float, float* [[FI]], i64 1
+; CHECK-NEXT:    [[TMP2:%.*]] = load float, float* [[ARRAYIDX5]], align 4
+; CHECK-NEXT:    [[ADD9:%.*]] = or i32 [[MUL1]], 1
+; CHECK-NEXT:    [[IDXPROM10:%.*]] = sext i32 [[ADD9]] to i64
+; CHECK-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds float, float* [[F]], i64 [[IDXPROM10]]
+; CHECK-NEXT:    [[TMP3:%.*]] = load float, float* [[ARRAYIDX11]], align 4
+; CHECK-NEXT:    [[ADD12:%.*]] = fadd fast float [[TMP3]], [[TMP2]]
+; CHECK-NEXT:    store float [[ADD12]], float* [[ARRAYIDX11]], align 4
+; CHECK-NEXT:    [[ARRAYIDX13:%.*]] = getelementptr inbounds float, float* [[FI]], i64 2
+; CHECK-NEXT:    [[TMP4:%.*]] = load float, float* [[ARRAYIDX13]], align 4
+; CHECK-NEXT:    [[ADD17:%.*]] = or i32 [[MUL1]], 2
+; CHECK-NEXT:    [[IDXPROM18:%.*]] = sext i32 [[ADD17]] to i64
+; CHECK-NEXT:    [[ARRAYIDX19:%.*]] = getelementptr inbounds float, float* [[F]], i64 [[IDXPROM18]]
+; CHECK-NEXT:    [[TMP5:%.*]] = load float, float* [[ARRAYIDX19]], align 4
+; CHECK-NEXT:    [[ADD20:%.*]] = fadd fast float [[TMP5]], [[TMP4]]
+; CHECK-NEXT:    store float [[ADD20]], float* [[ARRAYIDX19]], align 4
+; CHECK-NEXT:    [[ARRAYIDX21:%.*]] = getelementptr inbounds float, float* [[FI]], i64 3
+; CHECK-NEXT:    [[TMP6:%.*]] = load float, float* [[ARRAYIDX21]], align 4
+; CHECK-NEXT:    [[TMP7:%.*]] = mul i32 [[CI]], 12
+; CHECK-NEXT:    [[MUL24:%.*]] = or i32 [[TMP7]], 3
+; CHECK-NEXT:    [[IDXPROM26:%.*]] = sext i32 [[MUL24]] to i64
+; CHECK-NEXT:    [[ARRAYIDX27:%.*]] = getelementptr inbounds float, float* [[F]], i64 [[IDXPROM26]]
+; CHECK-NEXT:    [[TMP8:%.*]] = load float, float* [[ARRAYIDX27]], align 4
+; CHECK-NEXT:    [[ADD28:%.*]] = fadd fast float [[TMP8]], [[TMP6]]
+; CHECK-NEXT:    store float [[ADD28]], float* [[ARRAYIDX27]], align 4
+; CHECK-NEXT:    [[ARRAYIDX29:%.*]] = getelementptr inbounds float, float* [[FI]], i64 4
+; CHECK-NEXT:    [[TMP9:%.*]] = load float, float* [[ARRAYIDX29]], align 4
+; CHECK-NEXT:    [[ADD33:%.*]] = add i32 [[TMP7]], 4
+; CHECK-NEXT:    [[IDXPROM34:%.*]] = sext i32 [[ADD33]] to i64
+; CHECK-NEXT:    [[ARRAYIDX35:%.*]] = getelementptr inbounds float, float* [[F]], i64 [[IDXPROM34]]
+; CHECK-NEXT:    [[TMP10:%.*]] = load float, float* [[ARRAYIDX35]], align 4
+; CHECK-NEXT:    [[ADD36:%.*]] = fadd fast float [[TMP10]], [[TMP9]]
+; CHECK-NEXT:    store float [[ADD36]], float* [[ARRAYIDX35]], align 4
+; CHECK-NEXT:    [[ARRAYIDX37:%.*]] = getelementptr inbounds float, float* [[FI]], i64 5
+; CHECK-NEXT:    [[TMP11:%.*]] = load float, float* [[ARRAYIDX37]], align 4
+; CHECK-NEXT:    [[ADD41:%.*]] = add i32 [[TMP7]], 5
+; CHECK-NEXT:    [[IDXPROM42:%.*]] = sext i32 [[ADD41]] to i64
+; CHECK-NEXT:    [[ARRAYIDX43:%.*]] = getelementptr inbounds float, float* [[F]], i64 [[IDXPROM42]]
+; CHECK-NEXT:    [[TMP12:%.*]] = load float, float* [[ARRAYIDX43]], align 4
+; CHECK-NEXT:    [[ADD44:%.*]] = fadd fast float [[TMP12]], [[TMP11]]
+; CHECK-NEXT:    store float [[ADD44]], float* [[ARRAYIDX43]], align 4
+; CHECK-NEXT:    [[ARRAYIDX45:%.*]] = getelementptr inbounds float, float* [[FI]], i64 6
+; CHECK-NEXT:    [[TMP13:%.*]] = load float, float* [[ARRAYIDX45]], align 4
+; CHECK-NEXT:    [[TMP14:%.*]] = mul i32 [[CI]], 12
+; CHECK-NEXT:    [[MUL48:%.*]] = add i32 [[TMP14]], 6
+; CHECK-NEXT:    [[IDXPROM50:%.*]] = sext i32 [[MUL48]] to i64
+; CHECK-NEXT:    [[ARRAYIDX51:%.*]] = getelementptr inbounds float, float* [[F]], i64 [[IDXPROM50]]
+; CHECK-NEXT:    [[TMP15:%.*]] = load float, float* [[ARRAYIDX51]], align 4
+; CHECK-NEXT:    [[ADD52:%.*]] = fadd fast float [[TMP15]], [[TMP13]]
+; CHECK-NEXT:    store float [[ADD52]], float* [[ARRAYIDX51]], align 4
+; CHECK-NEXT:    [[ARRAYIDX53:%.*]] = getelementptr inbounds float, float* [[FI]], i64 7
+; CHECK-NEXT:    [[TMP16:%.*]] = load float, float* [[ARRAYIDX53]], align 4
+; CHECK-NEXT:    [[ADD57:%.*]] = add i32 [[TMP14]], 7
+; CHECK-NEXT:    [[IDXPROM58:%.*]] = sext i32 [[ADD57]] to i64
+; CHECK-NEXT:    [[ARRAYIDX59:%.*]] = getelementptr inbounds float, float* [[F]], i64 [[IDXPROM58]]
+; CHECK-NEXT:    [[TMP17:%.*]] = load float, float* [[ARRAYIDX59]], align 4
+; CHECK-NEXT:    [[ADD60:%.*]] = fadd fast float [[TMP17]], [[TMP16]]
+; CHECK-NEXT:    store float [[ADD60]], float* [[ARRAYIDX59]], align 4
+; CHECK-NEXT:    [[ARRAYIDX61:%.*]] = getelementptr inbounds float, float* [[FI]], i64 8
+; CHECK-NEXT:    [[TMP18:%.*]] = load float, float* [[ARRAYIDX61]], align 4
+; CHECK-NEXT:    [[ADD65:%.*]] = add i32 [[TMP14]], 8
+; CHECK-NEXT:    [[IDXPROM66:%.*]] = sext i32 [[ADD65]] to i64
+; CHECK-NEXT:    [[ARRAYIDX67:%.*]] = getelementptr inbounds float, float* [[F]], i64 [[IDXPROM66]]
+; CHECK-NEXT:    [[TMP19:%.*]] = load float, float* [[ARRAYIDX67]], align 4
+; CHECK-NEXT:    [[ADD68:%.*]] = fadd fast float [[TMP19]], [[TMP18]]
+; CHECK-NEXT:    store float [[ADD68]], float* [[ARRAYIDX67]], align 4
+; CHECK-NEXT:    [[ARRAYIDX69:%.*]] = getelementptr inbounds float, float* [[FI]], i64 9
+; CHECK-NEXT:    [[TMP20:%.*]] = load float, float* [[ARRAYIDX69]], align 4
+; CHECK-NEXT:    [[TMP21:%.*]] = mul i32 [[CI]], 12
+; CHECK-NEXT:    [[MUL72:%.*]] = add i32 [[TMP21]], 9
+; CHECK-NEXT:    [[IDXPROM74:%.*]] = sext i32 [[MUL72]] to i64
+; CHECK-NEXT:    [[ARRAYIDX75:%.*]] = getelementptr inbounds float, float* [[F]], i64 [[IDXPROM74]]
+; CHECK-NEXT:    [[TMP22:%.*]] = load float, float* [[ARRAYIDX75]], align 4
+; CHECK-NEXT:    [[ADD76:%.*]] = fadd fast float [[TMP22]], [[TMP20]]
+; CHECK-NEXT:    store float [[ADD76]], float* [[ARRAYIDX75]], align 4
+; CHECK-NEXT:    [[ARRAYIDX77:%.*]] = getelementptr inbounds float, float* [[FI]], i64 10
+; CHECK-NEXT:    [[TMP23:%.*]] = load float, float* [[ARRAYIDX77]], align 4
+; CHECK-NEXT:    [[ADD81:%.*]] = add i32 [[TMP21]], 10
+; CHECK-NEXT:    [[IDXPROM82:%.*]] = sext i32 [[ADD81]] to i64
+; CHECK-NEXT:    [[ARRAYIDX83:%.*]] = getelementptr inbounds float, float* [[F]], i64 [[IDXPROM82]]
+; CHECK-NEXT:    [[TMP24:%.*]] = load float, float* [[ARRAYIDX83]], align 4
+; CHECK-NEXT:    [[ADD84:%.*]] = fadd fast float [[TMP24]], [[TMP23]]
+; CHECK-NEXT:    store float [[ADD84]], float* [[ARRAYIDX83]], align 4
+; CHECK-NEXT:    [[ARRAYIDX85:%.*]] = getelementptr inbounds float, float* [[FI]], i64 11
+; CHECK-NEXT:    [[TMP25:%.*]] = load float, float* [[ARRAYIDX85]], align 4
+; CHECK-NEXT:    [[ADD89:%.*]] = add i32 [[TMP21]], 11
+; CHECK-NEXT:    [[IDXPROM90:%.*]] = sext i32 [[ADD89]] to i64
+; CHECK-NEXT:    [[ARRAYIDX91:%.*]] = getelementptr inbounds float, float* [[F]], i64 [[IDXPROM90]]
+; CHECK-NEXT:    [[TMP26:%.*]] = load float, float* [[ARRAYIDX91]], align 4
+; CHECK-NEXT:    [[ADD92:%.*]] = fadd fast float [[TMP26]], [[TMP25]]
+; CHECK-NEXT:    store float [[ADD92]], float* [[ARRAYIDX91]], align 4
+; CHECK-NEXT:    ret void
+;
+entry:
+  %0 = load float, float* %fi, align 4
+  %mul = mul nsw i32 %ci, 4
+  %mul1 = mul nsw i32 %mul, 3
+  %idxprom = sext i32 %mul1 to i64
+  %arrayidx3 = getelementptr inbounds float, float* %f, i64 %idxprom
+  %1 = load float, float* %arrayidx3, align 4
+  %add4 = fadd fast float %1, %0
+  store float %add4, float* %arrayidx3, align 4
+  %arrayidx5 = getelementptr inbounds float, float* %fi, i64 1
+  %2 = load float, float* %arrayidx5, align 4
+  %add9 = add nsw i32 %mul1, 1
+  %idxprom10 = sext i32 %add9 to i64
+  %arrayidx11 = getelementptr inbounds float, float* %f, i64 %idxprom10
+  %3 = load float, float* %arrayidx11, align 4
+  %add12 = fadd fast float %3, %2
+  store float %add12, float* %arrayidx11, align 4
+  %arrayidx13 = getelementptr inbounds float, float* %fi, i64 2
+  %4 = load float, float* %arrayidx13, align 4
+  %add17 = add nsw i32 %mul1, 2
+  %idxprom18 = sext i32 %add17 to i64
+  %arrayidx19 = getelementptr inbounds float, float* %f, i64 %idxprom18
+  %5 = load float, float* %arrayidx19, align 4
+  %add20 = fadd fast float %5, %4
+  store float %add20, float* %arrayidx19, align 4
+  %arrayidx21 = getelementptr inbounds float, float* %fi, i64 3
+  %6 = load float, float* %arrayidx21, align 4
+  %add23 = add nsw i32 %mul, 1
+  %mul24 = mul nsw i32 %add23, 3
+  %idxprom26 = sext i32 %mul24 to i64
+  %arrayidx27 = getelementptr inbounds float, float* %f, i64 %idxprom26
+  %7 = load float, float* %arrayidx27, align 4
+  %add28 = fadd fast float %7, %6
+  store float %add28, float* %arrayidx27, align 4
+  %arrayidx29 = getelementptr inbounds float, float* %fi, i64 4
+  %8 = load float, float* %arrayidx29, align 4
+  %add33 = add nsw i32 %mul24, 1
+  %idxprom34 = sext i32 %add33 to i64
+  %arrayidx35 = getelementptr inbounds float, float* %f, i64 %idxprom34
+  %9 = load float, float* %arrayidx35, align 4
+  %add36 = fadd fast float %9, %8
+  store float %add36, float* %arrayidx35, align 4
+  %arrayidx37 = getelementptr inbounds float, float* %fi, i64 5
+  %10 = load float, float* %arrayidx37, align 4
+  %add41 = add nsw i32 %mul24, 2
+  %idxprom42 = sext i32 %add41 to i64
+  %arrayidx43 = getelementptr inbounds float, float* %f, i64 %idxprom42
+  %11 = load float, float* %arrayidx43, align 4
+  %add44 = fadd fast float %11, %10
+  store float %add44, float* %arrayidx43, align 4
+  %arrayidx45 = getelementptr inbounds float, float* %fi, i64 6
+  %12 = load float, float* %arrayidx45, align 4
+  %add47 = add nsw i32 %mul, 2
+  %mul48 = mul nsw i32 %add47, 3
+  %idxprom50 = sext i32 %mul48 to i64
+  %arrayidx51 = getelementptr inbounds float, float* %f, i64 %idxprom50
+  %13 = load float, float* %arrayidx51, align 4
+  %add52 = fadd fast float %13, %12
+  store float %add52, float* %arrayidx51, align 4
+  %arrayidx53 = getelementptr inbounds float, float* %fi, i64 7
+  %14 = load float, float* %arrayidx53, align 4
+  %add57 = add nsw i32 %mul48, 1
+  %idxprom58 = sext i32 %add57 to i64
+  %arrayidx59 = getelementptr inbounds float, float* %f, i64 %idxprom58
+  %15 = load float, float* %arrayidx59, align 4
+  %add60 = fadd fast float %15, %14
+  store float %add60, float* %arrayidx59, align 4
+  %arrayidx61 = getelementptr inbounds float, float* %fi, i64 8
+  %16 = load float, float* %arrayidx61, align 4
+  %add65 = add nsw i32 %mul48, 2
+  %idxprom66 = sext i32 %add65 to i64
+  %arrayidx67 = getelementptr inbounds float, float* %f, i64 %idxprom66
+  %17 = load float, float* %arrayidx67, align 4
+  %add68 = fadd fast float %17, %16
+  store float %add68, float* %arrayidx67, align 4
+  %arrayidx69 = getelementptr inbounds float, float* %fi, i64 9
+  %18 = load float, float* %arrayidx69, align 4
+  %add71 = add nsw i32 %mul, 3
+  %mul72 = mul nsw i32 %add71, 3
+  %idxprom74 = sext i32 %mul72 to i64
+  %arrayidx75 = getelementptr inbounds float, float* %f, i64 %idxprom74
+  %19 = load float, float* %arrayidx75, align 4
+  %add76 = fadd fast float %19, %18
+  store float %add76, float* %arrayidx75, align 4
+  %arrayidx77 = getelementptr inbounds float, float* %fi, i64 10
+  %20 = load float, float* %arrayidx77, align 4
+  %add81 = add nsw i32 %mul72, 1
+  %idxprom82 = sext i32 %add81 to i64
+  %arrayidx83 = getelementptr inbounds float, float* %f, i64 %idxprom82
+  %21 = load float, float* %arrayidx83, align 4
+  %add84 = fadd fast float %21, %20
+  store float %add84, float* %arrayidx83, align 4
+  %arrayidx85 = getelementptr inbounds float, float* %fi, i64 11
+  %22 = load float, float* %arrayidx85, align 4
+  %add89 = add nsw i32 %mul72, 2
+  %idxprom90 = sext i32 %add89 to i64
+  %arrayidx91 = getelementptr inbounds float, float* %f, i64 %idxprom90
+  %23 = load float, float* %arrayidx91, align 4
+  %add92 = fadd fast float %23, %22
+  store float %add92, float* %arrayidx91, align 4
+  ret void
+}