[llvm] [msan] Improve packed multiply-add instrumentation (PR #152941)
Thurston Dang via llvm-commits
llvm-commits at lists.llvm.org
Sun Aug 10 14:03:58 PDT 2025
https://github.com/thurstond created https://github.com/llvm/llvm-project/pull/152941
The current instrumentation has false positives: if there is a single uninitialized bit in any of the operands, the entire output is poisoned. This does not take into account that multiplying an uninitialized value with zero results in an initialized zero value.
This patch improves the instrumentation by underapproximating the multiplication as bitwise AND, with correct handling of ANDing with zero. This has no false positives but does have significant false negatives.
The horizontal add step is modeled precisely.
It also applies the handler to the AVX512 equivalents, which are currently strictly handled: llvm.avx512.pmaddw.d.512, llvm.avx512.pmaddubs.w.512
>From 538cd21f3dd082b9635f501c5ccb8df98f4bdd32 Mon Sep 17 00:00:00 2001
From: Thurston Dang <thurston at google.com>
Date: Sun, 10 Aug 2025 19:55:42 +0000
Subject: [PATCH 1/2] [msan] Improve packed multiply-add instrumentation
The current instrumentation has false positives: if there is a single
uninitialized bit in any of the operands, the entire output is poisoned.
This does not take into account that multiplying an uninitialized bit
with zero results in an initialized zero value.
This patch improves the instrumentation by underapproximating the
multiplication as bitwise AND, with correct handling of ANDing with
zero. This is an underapproximation (no false positives) because it is
ignoring the carry from multiplication.
The horizontal add step is modeled precisely.
It also applies the handler to the AVX512 equivalents:
llvm.avx512.pmaddw.d.512, llvm.avx512.pmaddubs.w.512
---
.../Instrumentation/MemorySanitizer.cpp | 73 ++++++++++++--
.../X86/avx2-intrinsics-x86.ll | 45 ++++++---
.../X86/avx512bw-intrinsics-upgrade.ll | 98 +++++++++----------
.../X86/avx512bw-intrinsics.ll | 97 +++++++++---------
.../MemorySanitizer/X86/mmx-intrinsics.ll | 40 +++++---
.../X86/sse2-intrinsics-x86.ll | 15 ++-
.../i386/avx2-intrinsics-i386.ll | 47 ++++++---
.../MemorySanitizer/i386/mmx-intrinsics.ll | 40 +++++---
.../i386/sse2-intrinsics-i386.ll | 15 ++-
.../MemorySanitizer/vector_arith.ll | 30 ++++--
10 files changed, 311 insertions(+), 189 deletions(-)
diff --git a/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp b/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp
index 6e8138725375a..0bf1bc587c62a 100644
--- a/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp
+++ b/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp
@@ -2865,6 +2865,9 @@ struct MemorySanitizerVisitor : public InstVisitor<MemorySanitizerVisitor> {
}
void visitMul(BinaryOperator &I) {
+ // TODO: this can only handle zero bits that are part of statically-known
+ // constants. Consider under-approximating the multiplication as AND
+ // (which ignores the carry), and using the visitAnd() logic.
Constant *constOp0 = dyn_cast<Constant>(I.getOperand(0));
Constant *constOp1 = dyn_cast<Constant>(I.getOperand(1));
if (constOp0 && !constOp1)
@@ -3827,19 +3830,67 @@ struct MemorySanitizerVisitor : public InstVisitor<MemorySanitizerVisitor> {
}
// Instrument multiply-add intrinsic.
+ //
+ // e.g., <4 x i32> @llvm.x86.sse2.pmadd.wd(<8 x i16> %a, <8 x i16> %b)
+ // <1 x i64> @llvm.x86.mmx.pmadd.wd(<1 x i64> %a, <1 x i64> %b)
void handleVectorPmaddIntrinsic(IntrinsicInst &I,
unsigned MMXEltSizeInBits = 0) {
- Type *ResTy =
- MMXEltSizeInBits ? getMMXVectorTy(MMXEltSizeInBits * 2) : I.getType();
IRBuilder<> IRB(&I);
- auto *Shadow0 = getShadow(&I, 0);
- auto *Shadow1 = getShadow(&I, 1);
- Value *S = IRB.CreateOr(Shadow0, Shadow1);
- S = IRB.CreateBitCast(S, ResTy);
- S = IRB.CreateSExt(IRB.CreateICmpNE(S, Constant::getNullValue(ResTy)),
- ResTy);
- S = IRB.CreateBitCast(S, getShadowTy(&I));
- setShadow(&I, S);
+
+ Type *ReturnType =
+ MMXEltSizeInBits ? getMMXVectorTy(MMXEltSizeInBits * 2) : I.getType();
+ assert(isa<FixedVectorType>(ReturnType));
+
+ assert(I.arg_size() == 2);
+ [[maybe_unused]] FixedVectorType *ParamType =
+ cast<FixedVectorType>(I.getArgOperand(0)->getType());
+ assert(ParamType == I.getArgOperand(1)->getType());
+
+ if (!MMXEltSizeInBits)
+ assert(ParamType->getNumElements() ==
+ 2 * cast<FixedVectorType>(ReturnType)->getNumElements());
+
+ assert(ParamType->getPrimitiveSizeInBits() ==
+ ReturnType->getPrimitiveSizeInBits());
+
+ // Step 1: multiplication of corresponding vector elements
+ // We want to take into account the fact that multiplying zero by an
+ // uninitialized bit results in an initialized value of zero.
+ // We under-approximate multiplication using the same logic as visitAnd().
+ // This ignores the carrying that may happen during multiplication.
+ Value *S1 = getShadow(&I, 0);
+ Value *S2 = getShadow(&I, 1);
+ Value *V1 = I.getOperand(0);
+ Value *V2 = I.getOperand(1);
+
+ Value *S1S2 = IRB.CreateAnd(S1, S2);
+ Value *V1S2 = IRB.CreateAnd(V1, S2);
+ Value *S1V2 = IRB.CreateAnd(S1, V2);
+
+ // After multiplying e.g., <8 x i16> %a, <8 x i16> %b, we have
+ // <8 x i16> %ab.
+ Value *ShadowAB = IRB.CreateOr({S1S2, V1S2, S1V2});
+ // For MMX, %ab has a misleading type e.g., <1 x i64>.
+ if (MMXEltSizeInBits)
+ ShadowAB = IRB.CreateBitCast(ShadowAB, getMMXVectorTy(MMXEltSizeInBits));
+
+ // Step 2: pairwise/horizontal add
+ // Handle it similarly to handlePairwiseShadowOrIntrinsic().
+ unsigned TotalNumElems =
+ cast<FixedVectorType>(ReturnType)->getNumElements() * 2;
+ SmallVector<int, 8> EvenMask;
+ SmallVector<int, 8> OddMask;
+ for (unsigned X = 0; X < TotalNumElems - 1; X += 2) {
+ EvenMask.push_back(X);
+ OddMask.push_back(X + 1);
+ }
+ Value *EvenShadow = IRB.CreateShuffleVector(ShadowAB, EvenMask);
+ Value *OddShadow = IRB.CreateShuffleVector(ShadowAB, OddMask);
+
+ Value *OrShadow = IRB.CreateOr(EvenShadow, OddShadow);
+ OrShadow = CreateShadowCast(IRB, OrShadow, getShadowTy(&I));
+
+ setShadow(&I, OrShadow);
setOriginForNaryOp(I);
}
@@ -5378,6 +5429,8 @@ struct MemorySanitizerVisitor : public InstVisitor<MemorySanitizerVisitor> {
case Intrinsic::x86_avx2_pmadd_wd:
case Intrinsic::x86_ssse3_pmadd_ub_sw_128:
case Intrinsic::x86_avx2_pmadd_ub_sw:
+ case Intrinsic::x86_avx512_pmaddw_d_512:
+ case Intrinsic::x86_avx512_pmaddubs_w_512:
handleVectorPmaddIntrinsic(I);
break;
diff --git a/llvm/test/Instrumentation/MemorySanitizer/X86/avx2-intrinsics-x86.ll b/llvm/test/Instrumentation/MemorySanitizer/X86/avx2-intrinsics-x86.ll
index f916130fe53e5..3b38fcba35f98 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/X86/avx2-intrinsics-x86.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/X86/avx2-intrinsics-x86.ll
@@ -140,11 +140,16 @@ define <8 x i32> @test_x86_avx2_pmadd_wd(<16 x i16> %a0, <16 x i16> %a1) #0 {
; CHECK-NEXT: [[TMP1:%.*]] = load <16 x i16>, ptr @__msan_param_tls, align 8
; CHECK-NEXT: [[TMP2:%.*]] = load <16 x i16>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 32) to ptr), align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP3:%.*]] = or <16 x i16> [[TMP1]], [[TMP2]]
-; CHECK-NEXT: [[TMP4:%.*]] = bitcast <16 x i16> [[TMP3]] to <8 x i32>
-; CHECK-NEXT: [[TMP5:%.*]] = icmp ne <8 x i32> [[TMP4]], zeroinitializer
-; CHECK-NEXT: [[TMP6:%.*]] = sext <8 x i1> [[TMP5]] to <8 x i32>
-; CHECK-NEXT: [[RES:%.*]] = call <8 x i32> @llvm.x86.avx2.pmadd.wd(<16 x i16> [[A0:%.*]], <16 x i16> [[A1:%.*]])
+; CHECK-NEXT: [[TMP3:%.*]] = and <16 x i16> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP4:%.*]] = and <16 x i16> [[A0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP5:%.*]] = and <16 x i16> [[TMP1]], [[A1:%.*]]
+; CHECK-NEXT: [[TMP11:%.*]] = or <16 x i16> [[TMP3]], [[TMP4]]
+; CHECK-NEXT: [[TMP7:%.*]] = or <16 x i16> [[TMP11]], [[TMP5]]
+; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <16 x i16> [[TMP7]], <16 x i16> poison, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <16 x i16> [[TMP7]], <16 x i16> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
+; CHECK-NEXT: [[TMP10:%.*]] = or <8 x i16> [[TMP8]], [[TMP9]]
+; CHECK-NEXT: [[TMP6:%.*]] = zext <8 x i16> [[TMP10]] to <8 x i32>
+; CHECK-NEXT: [[RES:%.*]] = call <8 x i32> @llvm.x86.avx2.pmadd.wd(<16 x i16> [[A0]], <16 x i16> [[A1]])
; CHECK-NEXT: store <8 x i32> [[TMP6]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <8 x i32> [[RES]]
;
@@ -677,11 +682,16 @@ define <16 x i16> @test_x86_avx2_pmadd_ub_sw(<32 x i8> %a0, <32 x i8> %a1) #0 {
; CHECK-NEXT: [[TMP1:%.*]] = load <32 x i8>, ptr @__msan_param_tls, align 8
; CHECK-NEXT: [[TMP2:%.*]] = load <32 x i8>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 32) to ptr), align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP3:%.*]] = or <32 x i8> [[TMP1]], [[TMP2]]
-; CHECK-NEXT: [[TMP4:%.*]] = bitcast <32 x i8> [[TMP3]] to <16 x i16>
-; CHECK-NEXT: [[TMP5:%.*]] = icmp ne <16 x i16> [[TMP4]], zeroinitializer
-; CHECK-NEXT: [[TMP6:%.*]] = sext <16 x i1> [[TMP5]] to <16 x i16>
-; CHECK-NEXT: [[RES:%.*]] = call <16 x i16> @llvm.x86.avx2.pmadd.ub.sw(<32 x i8> [[A0:%.*]], <32 x i8> [[A1:%.*]])
+; CHECK-NEXT: [[TMP3:%.*]] = and <32 x i8> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP4:%.*]] = and <32 x i8> [[A0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP5:%.*]] = and <32 x i8> [[TMP1]], [[A1:%.*]]
+; CHECK-NEXT: [[TMP11:%.*]] = or <32 x i8> [[TMP3]], [[TMP4]]
+; CHECK-NEXT: [[TMP7:%.*]] = or <32 x i8> [[TMP11]], [[TMP5]]
+; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <32 x i8> [[TMP7]], <32 x i8> poison, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <32 x i8> [[TMP7]], <32 x i8> poison, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
+; CHECK-NEXT: [[TMP10:%.*]] = or <16 x i8> [[TMP8]], [[TMP9]]
+; CHECK-NEXT: [[TMP6:%.*]] = zext <16 x i8> [[TMP10]] to <16 x i16>
+; CHECK-NEXT: [[RES:%.*]] = call <16 x i16> @llvm.x86.avx2.pmadd.ub.sw(<32 x i8> [[A0]], <32 x i8> [[A1]])
; CHECK-NEXT: store <16 x i16> [[TMP6]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <16 x i16> [[RES]]
;
@@ -706,11 +716,16 @@ define <16 x i16> @test_x86_avx2_pmadd_ub_sw_load_op0(ptr %ptr, <32 x i8> %a1) #
; CHECK-NEXT: [[TMP6:%.*]] = xor i64 [[TMP5]], 87960930222080
; CHECK-NEXT: [[TMP7:%.*]] = inttoptr i64 [[TMP6]] to ptr
; CHECK-NEXT: [[_MSLD:%.*]] = load <32 x i8>, ptr [[TMP7]], align 32
-; CHECK-NEXT: [[TMP8:%.*]] = or <32 x i8> [[_MSLD]], [[TMP2]]
-; CHECK-NEXT: [[TMP9:%.*]] = bitcast <32 x i8> [[TMP8]] to <16 x i16>
-; CHECK-NEXT: [[TMP10:%.*]] = icmp ne <16 x i16> [[TMP9]], zeroinitializer
-; CHECK-NEXT: [[TMP11:%.*]] = sext <16 x i1> [[TMP10]] to <16 x i16>
-; CHECK-NEXT: [[RES:%.*]] = call <16 x i16> @llvm.x86.avx2.pmadd.ub.sw(<32 x i8> [[A0]], <32 x i8> [[A1:%.*]])
+; CHECK-NEXT: [[TMP8:%.*]] = and <32 x i8> [[_MSLD]], [[TMP2]]
+; CHECK-NEXT: [[TMP9:%.*]] = and <32 x i8> [[A0]], [[TMP2]]
+; CHECK-NEXT: [[TMP10:%.*]] = and <32 x i8> [[_MSLD]], [[A1:%.*]]
+; CHECK-NEXT: [[TMP16:%.*]] = or <32 x i8> [[TMP8]], [[TMP9]]
+; CHECK-NEXT: [[TMP12:%.*]] = or <32 x i8> [[TMP16]], [[TMP10]]
+; CHECK-NEXT: [[TMP13:%.*]] = shufflevector <32 x i8> [[TMP12]], <32 x i8> poison, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
+; CHECK-NEXT: [[TMP14:%.*]] = shufflevector <32 x i8> [[TMP12]], <32 x i8> poison, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
+; CHECK-NEXT: [[TMP15:%.*]] = or <16 x i8> [[TMP13]], [[TMP14]]
+; CHECK-NEXT: [[TMP11:%.*]] = zext <16 x i8> [[TMP15]] to <16 x i16>
+; CHECK-NEXT: [[RES:%.*]] = call <16 x i16> @llvm.x86.avx2.pmadd.ub.sw(<32 x i8> [[A0]], <32 x i8> [[A1]])
; CHECK-NEXT: store <16 x i16> [[TMP11]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <16 x i16> [[RES]]
;
diff --git a/llvm/test/Instrumentation/MemorySanitizer/X86/avx512bw-intrinsics-upgrade.ll b/llvm/test/Instrumentation/MemorySanitizer/X86/avx512bw-intrinsics-upgrade.ll
index 02df9c49a010b..54e9939ace7c3 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/X86/avx512bw-intrinsics-upgrade.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/X86/avx512bw-intrinsics-upgrade.ll
@@ -7,8 +7,6 @@
; - llvm.x86.avx512.dbpsadbw.512
; - llvm.x86.avx512.packssdw.512, llvm.x86.avx512.packsswb.512
; - llvm.x86.avx512.packusdw.512, llvm.x86.avx512.packuswb.512
-; - llvm.x86.avx512.pmaddubs.w.512
-; - llvm.x86.avx512.pmaddw.d.512
;
; Heuristically handled:
; - llvm.sadd.sat.v32i16, llvm.sadd.sat.v64i8
@@ -4930,18 +4928,17 @@ define <32 x i16> @test_int_x86_avx512_pmaddubs_w_512(<64 x i8> %x0, <64 x i8> %
; CHECK-NEXT: [[TMP1:%.*]] = load <64 x i8>, ptr @__msan_param_tls, align 8
; CHECK-NEXT: [[TMP2:%.*]] = load <64 x i8>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 64) to ptr), align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP3:%.*]] = bitcast <64 x i8> [[TMP1]] to i512
-; CHECK-NEXT: [[_MSCMP:%.*]] = icmp ne i512 [[TMP3]], 0
-; CHECK-NEXT: [[TMP4:%.*]] = bitcast <64 x i8> [[TMP2]] to i512
-; CHECK-NEXT: [[_MSCMP1:%.*]] = icmp ne i512 [[TMP4]], 0
-; CHECK-NEXT: [[_MSOR:%.*]] = or i1 [[_MSCMP]], [[_MSCMP1]]
-; CHECK-NEXT: br i1 [[_MSOR]], label [[TMP5:%.*]], label [[TMP6:%.*]], !prof [[PROF1]]
-; CHECK: 5:
-; CHECK-NEXT: call void @__msan_warning_noreturn() #[[ATTR7]]
-; CHECK-NEXT: unreachable
-; CHECK: 6:
-; CHECK-NEXT: [[TMP7:%.*]] = call <32 x i16> @llvm.x86.avx512.pmaddubs.w.512(<64 x i8> [[X0:%.*]], <64 x i8> [[X1:%.*]])
-; CHECK-NEXT: store <32 x i16> zeroinitializer, ptr @__msan_retval_tls, align 8
+; CHECK-NEXT: [[TMP3:%.*]] = and <64 x i8> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP4:%.*]] = and <64 x i8> [[X0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP5:%.*]] = and <64 x i8> [[TMP1]], [[X1:%.*]]
+; CHECK-NEXT: [[TMP6:%.*]] = or <64 x i8> [[TMP3]], [[TMP4]]
+; CHECK-NEXT: [[TMP12:%.*]] = or <64 x i8> [[TMP6]], [[TMP5]]
+; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <64 x i8> [[TMP12]], <64 x i8> poison, <32 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30, i32 32, i32 34, i32 36, i32 38, i32 40, i32 42, i32 44, i32 46, i32 48, i32 50, i32 52, i32 54, i32 56, i32 58, i32 60, i32 62>
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <64 x i8> [[TMP12]], <64 x i8> poison, <32 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31, i32 33, i32 35, i32 37, i32 39, i32 41, i32 43, i32 45, i32 47, i32 49, i32 51, i32 53, i32 55, i32 57, i32 59, i32 61, i32 63>
+; CHECK-NEXT: [[TMP10:%.*]] = or <32 x i8> [[TMP8]], [[TMP9]]
+; CHECK-NEXT: [[TMP11:%.*]] = zext <32 x i8> [[TMP10]] to <32 x i16>
+; CHECK-NEXT: [[TMP7:%.*]] = call <32 x i16> @llvm.x86.avx512.pmaddubs.w.512(<64 x i8> [[X0]], <64 x i8> [[X1]])
+; CHECK-NEXT: store <32 x i16> [[TMP11]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <32 x i16> [[TMP7]]
;
%res = call <32 x i16> @llvm.x86.avx512.mask.pmaddubs.w.512(<64 x i8> %x0, <64 x i8> %x1, <32 x i16> %x2, i32 -1)
@@ -4955,22 +4952,21 @@ define <32 x i16> @test_int_x86_avx512_mask_pmaddubs_w_512(<64 x i8> %x0, <64 x
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 192) to ptr), align 8
; CHECK-NEXT: [[TMP4:%.*]] = load <32 x i16>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 128) to ptr), align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP5:%.*]] = bitcast <64 x i8> [[TMP1]] to i512
-; CHECK-NEXT: [[_MSCMP:%.*]] = icmp ne i512 [[TMP5]], 0
-; CHECK-NEXT: [[TMP6:%.*]] = bitcast <64 x i8> [[TMP2]] to i512
-; CHECK-NEXT: [[_MSCMP1:%.*]] = icmp ne i512 [[TMP6]], 0
-; CHECK-NEXT: [[_MSOR:%.*]] = or i1 [[_MSCMP]], [[_MSCMP1]]
-; CHECK-NEXT: br i1 [[_MSOR]], label [[TMP7:%.*]], label [[TMP8:%.*]], !prof [[PROF1]]
-; CHECK: 7:
-; CHECK-NEXT: call void @__msan_warning_noreturn() #[[ATTR7]]
-; CHECK-NEXT: unreachable
-; CHECK: 8:
-; CHECK-NEXT: [[TMP9:%.*]] = call <32 x i16> @llvm.x86.avx512.pmaddubs.w.512(<64 x i8> [[X0:%.*]], <64 x i8> [[X1:%.*]])
+; CHECK-NEXT: [[TMP5:%.*]] = and <64 x i8> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP6:%.*]] = and <64 x i8> [[X0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP7:%.*]] = and <64 x i8> [[TMP1]], [[X1:%.*]]
+; CHECK-NEXT: [[TMP8:%.*]] = or <64 x i8> [[TMP5]], [[TMP6]]
+; CHECK-NEXT: [[TMP17:%.*]] = or <64 x i8> [[TMP8]], [[TMP7]]
+; CHECK-NEXT: [[TMP18:%.*]] = shufflevector <64 x i8> [[TMP17]], <64 x i8> poison, <32 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30, i32 32, i32 34, i32 36, i32 38, i32 40, i32 42, i32 44, i32 46, i32 48, i32 50, i32 52, i32 54, i32 56, i32 58, i32 60, i32 62>
+; CHECK-NEXT: [[TMP19:%.*]] = shufflevector <64 x i8> [[TMP17]], <64 x i8> poison, <32 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31, i32 33, i32 35, i32 37, i32 39, i32 41, i32 43, i32 45, i32 47, i32 49, i32 51, i32 53, i32 55, i32 57, i32 59, i32 61, i32 63>
+; CHECK-NEXT: [[TMP20:%.*]] = or <32 x i8> [[TMP18]], [[TMP19]]
+; CHECK-NEXT: [[TMP21:%.*]] = zext <32 x i8> [[TMP20]] to <32 x i16>
+; CHECK-NEXT: [[TMP9:%.*]] = call <32 x i16> @llvm.x86.avx512.pmaddubs.w.512(<64 x i8> [[X0]], <64 x i8> [[X1]])
; CHECK-NEXT: [[TMP10:%.*]] = bitcast i32 [[TMP3]] to <32 x i1>
; CHECK-NEXT: [[TMP11:%.*]] = bitcast i32 [[X3:%.*]] to <32 x i1>
-; CHECK-NEXT: [[TMP12:%.*]] = select <32 x i1> [[TMP11]], <32 x i16> zeroinitializer, <32 x i16> [[TMP4]]
+; CHECK-NEXT: [[TMP12:%.*]] = select <32 x i1> [[TMP11]], <32 x i16> [[TMP21]], <32 x i16> [[TMP4]]
; CHECK-NEXT: [[TMP13:%.*]] = xor <32 x i16> [[TMP9]], [[X2:%.*]]
-; CHECK-NEXT: [[TMP14:%.*]] = or <32 x i16> [[TMP13]], zeroinitializer
+; CHECK-NEXT: [[TMP14:%.*]] = or <32 x i16> [[TMP13]], [[TMP21]]
; CHECK-NEXT: [[TMP15:%.*]] = or <32 x i16> [[TMP14]], [[TMP4]]
; CHECK-NEXT: [[_MSPROP_SELECT:%.*]] = select <32 x i1> [[TMP10]], <32 x i16> [[TMP15]], <32 x i16> [[TMP12]]
; CHECK-NEXT: [[TMP16:%.*]] = select <32 x i1> [[TMP11]], <32 x i16> [[TMP9]], <32 x i16> [[X2]]
@@ -4988,18 +4984,17 @@ define <16 x i32> @test_int_x86_avx512_pmaddw_d_512(<32 x i16> %x0, <32 x i16> %
; CHECK-NEXT: [[TMP1:%.*]] = load <32 x i16>, ptr @__msan_param_tls, align 8
; CHECK-NEXT: [[TMP2:%.*]] = load <32 x i16>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 64) to ptr), align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP3:%.*]] = bitcast <32 x i16> [[TMP1]] to i512
-; CHECK-NEXT: [[_MSCMP:%.*]] = icmp ne i512 [[TMP3]], 0
-; CHECK-NEXT: [[TMP4:%.*]] = bitcast <32 x i16> [[TMP2]] to i512
-; CHECK-NEXT: [[_MSCMP1:%.*]] = icmp ne i512 [[TMP4]], 0
-; CHECK-NEXT: [[_MSOR:%.*]] = or i1 [[_MSCMP]], [[_MSCMP1]]
-; CHECK-NEXT: br i1 [[_MSOR]], label [[TMP5:%.*]], label [[TMP6:%.*]], !prof [[PROF1]]
-; CHECK: 5:
-; CHECK-NEXT: call void @__msan_warning_noreturn() #[[ATTR7]]
-; CHECK-NEXT: unreachable
-; CHECK: 6:
-; CHECK-NEXT: [[TMP7:%.*]] = call <16 x i32> @llvm.x86.avx512.pmaddw.d.512(<32 x i16> [[X0:%.*]], <32 x i16> [[X1:%.*]])
-; CHECK-NEXT: store <16 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
+; CHECK-NEXT: [[TMP3:%.*]] = and <32 x i16> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP4:%.*]] = and <32 x i16> [[X0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP5:%.*]] = and <32 x i16> [[TMP1]], [[X1:%.*]]
+; CHECK-NEXT: [[TMP6:%.*]] = or <32 x i16> [[TMP3]], [[TMP4]]
+; CHECK-NEXT: [[TMP12:%.*]] = or <32 x i16> [[TMP6]], [[TMP5]]
+; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <32 x i16> [[TMP12]], <32 x i16> poison, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <32 x i16> [[TMP12]], <32 x i16> poison, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
+; CHECK-NEXT: [[TMP10:%.*]] = or <16 x i16> [[TMP8]], [[TMP9]]
+; CHECK-NEXT: [[TMP11:%.*]] = zext <16 x i16> [[TMP10]] to <16 x i32>
+; CHECK-NEXT: [[TMP7:%.*]] = call <16 x i32> @llvm.x86.avx512.pmaddw.d.512(<32 x i16> [[X0]], <32 x i16> [[X1]])
+; CHECK-NEXT: store <16 x i32> [[TMP11]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <16 x i32> [[TMP7]]
;
%res = call <16 x i32> @llvm.x86.avx512.mask.pmaddw.d.512(<32 x i16> %x0, <32 x i16> %x1, <16 x i32> %x2, i16 -1)
@@ -5013,22 +5008,21 @@ define <16 x i32> @test_int_x86_avx512_mask_pmaddw_d_512(<32 x i16> %x0, <32 x i
; CHECK-NEXT: [[TMP3:%.*]] = load i16, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 192) to ptr), align 8
; CHECK-NEXT: [[TMP4:%.*]] = load <16 x i32>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 128) to ptr), align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP5:%.*]] = bitcast <32 x i16> [[TMP1]] to i512
-; CHECK-NEXT: [[_MSCMP:%.*]] = icmp ne i512 [[TMP5]], 0
-; CHECK-NEXT: [[TMP6:%.*]] = bitcast <32 x i16> [[TMP2]] to i512
-; CHECK-NEXT: [[_MSCMP1:%.*]] = icmp ne i512 [[TMP6]], 0
-; CHECK-NEXT: [[_MSOR:%.*]] = or i1 [[_MSCMP]], [[_MSCMP1]]
-; CHECK-NEXT: br i1 [[_MSOR]], label [[TMP7:%.*]], label [[TMP8:%.*]], !prof [[PROF1]]
-; CHECK: 7:
-; CHECK-NEXT: call void @__msan_warning_noreturn() #[[ATTR7]]
-; CHECK-NEXT: unreachable
-; CHECK: 8:
-; CHECK-NEXT: [[TMP9:%.*]] = call <16 x i32> @llvm.x86.avx512.pmaddw.d.512(<32 x i16> [[X0:%.*]], <32 x i16> [[X1:%.*]])
+; CHECK-NEXT: [[TMP5:%.*]] = and <32 x i16> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP6:%.*]] = and <32 x i16> [[X0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP7:%.*]] = and <32 x i16> [[TMP1]], [[X1:%.*]]
+; CHECK-NEXT: [[TMP8:%.*]] = or <32 x i16> [[TMP5]], [[TMP6]]
+; CHECK-NEXT: [[TMP17:%.*]] = or <32 x i16> [[TMP8]], [[TMP7]]
+; CHECK-NEXT: [[TMP18:%.*]] = shufflevector <32 x i16> [[TMP17]], <32 x i16> poison, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
+; CHECK-NEXT: [[TMP19:%.*]] = shufflevector <32 x i16> [[TMP17]], <32 x i16> poison, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
+; CHECK-NEXT: [[TMP20:%.*]] = or <16 x i16> [[TMP18]], [[TMP19]]
+; CHECK-NEXT: [[TMP21:%.*]] = zext <16 x i16> [[TMP20]] to <16 x i32>
+; CHECK-NEXT: [[TMP9:%.*]] = call <16 x i32> @llvm.x86.avx512.pmaddw.d.512(<32 x i16> [[X0]], <32 x i16> [[X1]])
; CHECK-NEXT: [[TMP10:%.*]] = bitcast i16 [[TMP3]] to <16 x i1>
; CHECK-NEXT: [[TMP11:%.*]] = bitcast i16 [[X3:%.*]] to <16 x i1>
-; CHECK-NEXT: [[TMP12:%.*]] = select <16 x i1> [[TMP11]], <16 x i32> zeroinitializer, <16 x i32> [[TMP4]]
+; CHECK-NEXT: [[TMP12:%.*]] = select <16 x i1> [[TMP11]], <16 x i32> [[TMP21]], <16 x i32> [[TMP4]]
; CHECK-NEXT: [[TMP13:%.*]] = xor <16 x i32> [[TMP9]], [[X2:%.*]]
-; CHECK-NEXT: [[TMP14:%.*]] = or <16 x i32> [[TMP13]], zeroinitializer
+; CHECK-NEXT: [[TMP14:%.*]] = or <16 x i32> [[TMP13]], [[TMP21]]
; CHECK-NEXT: [[TMP15:%.*]] = or <16 x i32> [[TMP14]], [[TMP4]]
; CHECK-NEXT: [[_MSPROP_SELECT:%.*]] = select <16 x i1> [[TMP10]], <16 x i32> [[TMP15]], <16 x i32> [[TMP12]]
; CHECK-NEXT: [[TMP16:%.*]] = select <16 x i1> [[TMP11]], <16 x i32> [[TMP9]], <16 x i32> [[X2]]
diff --git a/llvm/test/Instrumentation/MemorySanitizer/X86/avx512bw-intrinsics.ll b/llvm/test/Instrumentation/MemorySanitizer/X86/avx512bw-intrinsics.ll
index 78c272c7b2c5a..5ec8d054de381 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/X86/avx512bw-intrinsics.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/X86/avx512bw-intrinsics.ll
@@ -9,7 +9,6 @@
; - llvm.x86.avx512.mask.pmov.wb.mem.512
; - llvm.x86.avx512.packssdw.512, llvm.x86.avx512.packsswb.512
; - llvm.x86.avx512.packusdw.512, llvm.x86.avx512.packuswb.512
-; - llvm.x86.avx512.pmaddubs.w.512, llvm.x86.avx512.pmaddw.d.512
; - llvm.x86.avx512.psad.bw.512
;
; Heuristically handled:
@@ -2204,18 +2203,17 @@ define <32 x i16> @test_int_x86_avx512_pmaddubs_w_512(<64 x i8> %x0, <64 x i8> %
; CHECK-NEXT: [[TMP1:%.*]] = load <64 x i8>, ptr @__msan_param_tls, align 8
; CHECK-NEXT: [[TMP2:%.*]] = load <64 x i8>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 64) to ptr), align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP3:%.*]] = bitcast <64 x i8> [[TMP1]] to i512
-; CHECK-NEXT: [[_MSCMP:%.*]] = icmp ne i512 [[TMP3]], 0
-; CHECK-NEXT: [[TMP4:%.*]] = bitcast <64 x i8> [[TMP2]] to i512
-; CHECK-NEXT: [[_MSCMP1:%.*]] = icmp ne i512 [[TMP4]], 0
-; CHECK-NEXT: [[_MSOR:%.*]] = or i1 [[_MSCMP]], [[_MSCMP1]]
-; CHECK-NEXT: br i1 [[_MSOR]], label [[TMP5:%.*]], label [[TMP6:%.*]], !prof [[PROF1]]
-; CHECK: 5:
-; CHECK-NEXT: call void @__msan_warning_noreturn() #[[ATTR8]]
-; CHECK-NEXT: unreachable
-; CHECK: 6:
-; CHECK-NEXT: [[TMP7:%.*]] = call <32 x i16> @llvm.x86.avx512.pmaddubs.w.512(<64 x i8> [[X0:%.*]], <64 x i8> [[X1:%.*]])
-; CHECK-NEXT: store <32 x i16> zeroinitializer, ptr @__msan_retval_tls, align 8
+; CHECK-NEXT: [[TMP3:%.*]] = and <64 x i8> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP4:%.*]] = and <64 x i8> [[X0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP5:%.*]] = and <64 x i8> [[TMP1]], [[X1:%.*]]
+; CHECK-NEXT: [[TMP6:%.*]] = or <64 x i8> [[TMP3]], [[TMP4]]
+; CHECK-NEXT: [[TMP12:%.*]] = or <64 x i8> [[TMP6]], [[TMP5]]
+; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <64 x i8> [[TMP12]], <64 x i8> poison, <32 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30, i32 32, i32 34, i32 36, i32 38, i32 40, i32 42, i32 44, i32 46, i32 48, i32 50, i32 52, i32 54, i32 56, i32 58, i32 60, i32 62>
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <64 x i8> [[TMP12]], <64 x i8> poison, <32 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31, i32 33, i32 35, i32 37, i32 39, i32 41, i32 43, i32 45, i32 47, i32 49, i32 51, i32 53, i32 55, i32 57, i32 59, i32 61, i32 63>
+; CHECK-NEXT: [[TMP10:%.*]] = or <32 x i8> [[TMP8]], [[TMP9]]
+; CHECK-NEXT: [[TMP11:%.*]] = zext <32 x i8> [[TMP10]] to <32 x i16>
+; CHECK-NEXT: [[TMP7:%.*]] = call <32 x i16> @llvm.x86.avx512.pmaddubs.w.512(<64 x i8> [[X0]], <64 x i8> [[X1]])
+; CHECK-NEXT: store <32 x i16> [[TMP11]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <32 x i16> [[TMP7]]
;
%1 = call <32 x i16> @llvm.x86.avx512.pmaddubs.w.512(<64 x i8> %x0, <64 x i8> %x1)
@@ -2229,22 +2227,21 @@ define <32 x i16> @test_int_x86_avx512_mask_pmaddubs_w_512(<64 x i8> %x0, <64 x
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 192) to ptr), align 8
; CHECK-NEXT: [[TMP4:%.*]] = load <32 x i16>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 128) to ptr), align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP5:%.*]] = bitcast <64 x i8> [[TMP1]] to i512
-; CHECK-NEXT: [[_MSCMP:%.*]] = icmp ne i512 [[TMP5]], 0
-; CHECK-NEXT: [[TMP6:%.*]] = bitcast <64 x i8> [[TMP2]] to i512
-; CHECK-NEXT: [[_MSCMP1:%.*]] = icmp ne i512 [[TMP6]], 0
-; CHECK-NEXT: [[_MSOR:%.*]] = or i1 [[_MSCMP]], [[_MSCMP1]]
-; CHECK-NEXT: br i1 [[_MSOR]], label [[TMP7:%.*]], label [[TMP8:%.*]], !prof [[PROF1]]
-; CHECK: 7:
-; CHECK-NEXT: call void @__msan_warning_noreturn() #[[ATTR8]]
-; CHECK-NEXT: unreachable
-; CHECK: 8:
-; CHECK-NEXT: [[TMP9:%.*]] = call <32 x i16> @llvm.x86.avx512.pmaddubs.w.512(<64 x i8> [[X0:%.*]], <64 x i8> [[X1:%.*]])
+; CHECK-NEXT: [[TMP5:%.*]] = and <64 x i8> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP6:%.*]] = and <64 x i8> [[X0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP7:%.*]] = and <64 x i8> [[TMP1]], [[X1:%.*]]
+; CHECK-NEXT: [[TMP8:%.*]] = or <64 x i8> [[TMP5]], [[TMP6]]
+; CHECK-NEXT: [[TMP17:%.*]] = or <64 x i8> [[TMP8]], [[TMP7]]
+; CHECK-NEXT: [[TMP18:%.*]] = shufflevector <64 x i8> [[TMP17]], <64 x i8> poison, <32 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30, i32 32, i32 34, i32 36, i32 38, i32 40, i32 42, i32 44, i32 46, i32 48, i32 50, i32 52, i32 54, i32 56, i32 58, i32 60, i32 62>
+; CHECK-NEXT: [[TMP19:%.*]] = shufflevector <64 x i8> [[TMP17]], <64 x i8> poison, <32 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31, i32 33, i32 35, i32 37, i32 39, i32 41, i32 43, i32 45, i32 47, i32 49, i32 51, i32 53, i32 55, i32 57, i32 59, i32 61, i32 63>
+; CHECK-NEXT: [[TMP20:%.*]] = or <32 x i8> [[TMP18]], [[TMP19]]
+; CHECK-NEXT: [[TMP21:%.*]] = zext <32 x i8> [[TMP20]] to <32 x i16>
+; CHECK-NEXT: [[TMP9:%.*]] = call <32 x i16> @llvm.x86.avx512.pmaddubs.w.512(<64 x i8> [[X0]], <64 x i8> [[X1]])
; CHECK-NEXT: [[TMP10:%.*]] = bitcast i32 [[TMP3]] to <32 x i1>
; CHECK-NEXT: [[TMP11:%.*]] = bitcast i32 [[X3:%.*]] to <32 x i1>
-; CHECK-NEXT: [[TMP12:%.*]] = select <32 x i1> [[TMP11]], <32 x i16> zeroinitializer, <32 x i16> [[TMP4]]
+; CHECK-NEXT: [[TMP12:%.*]] = select <32 x i1> [[TMP11]], <32 x i16> [[TMP21]], <32 x i16> [[TMP4]]
; CHECK-NEXT: [[TMP13:%.*]] = xor <32 x i16> [[TMP9]], [[X2:%.*]]
-; CHECK-NEXT: [[TMP14:%.*]] = or <32 x i16> [[TMP13]], zeroinitializer
+; CHECK-NEXT: [[TMP14:%.*]] = or <32 x i16> [[TMP13]], [[TMP21]]
; CHECK-NEXT: [[TMP15:%.*]] = or <32 x i16> [[TMP14]], [[TMP4]]
; CHECK-NEXT: [[_MSPROP_SELECT:%.*]] = select <32 x i1> [[TMP10]], <32 x i16> [[TMP15]], <32 x i16> [[TMP12]]
; CHECK-NEXT: [[TMP16:%.*]] = select <32 x i1> [[TMP11]], <32 x i16> [[TMP9]], <32 x i16> [[X2]]
@@ -2264,18 +2261,17 @@ define <16 x i32> @test_int_x86_avx512_pmaddw_d_512(<32 x i16> %x0, <32 x i16> %
; CHECK-NEXT: [[TMP1:%.*]] = load <32 x i16>, ptr @__msan_param_tls, align 8
; CHECK-NEXT: [[TMP2:%.*]] = load <32 x i16>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 64) to ptr), align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP3:%.*]] = bitcast <32 x i16> [[TMP1]] to i512
-; CHECK-NEXT: [[_MSCMP:%.*]] = icmp ne i512 [[TMP3]], 0
-; CHECK-NEXT: [[TMP4:%.*]] = bitcast <32 x i16> [[TMP2]] to i512
-; CHECK-NEXT: [[_MSCMP1:%.*]] = icmp ne i512 [[TMP4]], 0
-; CHECK-NEXT: [[_MSOR:%.*]] = or i1 [[_MSCMP]], [[_MSCMP1]]
-; CHECK-NEXT: br i1 [[_MSOR]], label [[TMP5:%.*]], label [[TMP6:%.*]], !prof [[PROF1]]
-; CHECK: 5:
-; CHECK-NEXT: call void @__msan_warning_noreturn() #[[ATTR8]]
-; CHECK-NEXT: unreachable
-; CHECK: 6:
-; CHECK-NEXT: [[TMP7:%.*]] = call <16 x i32> @llvm.x86.avx512.pmaddw.d.512(<32 x i16> [[X0:%.*]], <32 x i16> [[X1:%.*]])
-; CHECK-NEXT: store <16 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
+; CHECK-NEXT: [[TMP3:%.*]] = and <32 x i16> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP4:%.*]] = and <32 x i16> [[X0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP5:%.*]] = and <32 x i16> [[TMP1]], [[X1:%.*]]
+; CHECK-NEXT: [[TMP6:%.*]] = or <32 x i16> [[TMP3]], [[TMP4]]
+; CHECK-NEXT: [[TMP12:%.*]] = or <32 x i16> [[TMP6]], [[TMP5]]
+; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <32 x i16> [[TMP12]], <32 x i16> poison, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <32 x i16> [[TMP12]], <32 x i16> poison, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
+; CHECK-NEXT: [[TMP10:%.*]] = or <16 x i16> [[TMP8]], [[TMP9]]
+; CHECK-NEXT: [[TMP11:%.*]] = zext <16 x i16> [[TMP10]] to <16 x i32>
+; CHECK-NEXT: [[TMP7:%.*]] = call <16 x i32> @llvm.x86.avx512.pmaddw.d.512(<32 x i16> [[X0]], <32 x i16> [[X1]])
+; CHECK-NEXT: store <16 x i32> [[TMP11]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <16 x i32> [[TMP7]]
;
%1 = call <16 x i32> @llvm.x86.avx512.pmaddw.d.512(<32 x i16> %x0, <32 x i16> %x1)
@@ -2289,22 +2285,21 @@ define <16 x i32> @test_int_x86_avx512_mask_pmaddw_d_512(<32 x i16> %x0, <32 x i
; CHECK-NEXT: [[TMP3:%.*]] = load i16, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 192) to ptr), align 8
; CHECK-NEXT: [[TMP4:%.*]] = load <16 x i32>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 128) to ptr), align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP5:%.*]] = bitcast <32 x i16> [[TMP1]] to i512
-; CHECK-NEXT: [[_MSCMP:%.*]] = icmp ne i512 [[TMP5]], 0
-; CHECK-NEXT: [[TMP6:%.*]] = bitcast <32 x i16> [[TMP2]] to i512
-; CHECK-NEXT: [[_MSCMP1:%.*]] = icmp ne i512 [[TMP6]], 0
-; CHECK-NEXT: [[_MSOR:%.*]] = or i1 [[_MSCMP]], [[_MSCMP1]]
-; CHECK-NEXT: br i1 [[_MSOR]], label [[TMP7:%.*]], label [[TMP8:%.*]], !prof [[PROF1]]
-; CHECK: 7:
-; CHECK-NEXT: call void @__msan_warning_noreturn() #[[ATTR8]]
-; CHECK-NEXT: unreachable
-; CHECK: 8:
-; CHECK-NEXT: [[TMP9:%.*]] = call <16 x i32> @llvm.x86.avx512.pmaddw.d.512(<32 x i16> [[X0:%.*]], <32 x i16> [[X1:%.*]])
+; CHECK-NEXT: [[TMP5:%.*]] = and <32 x i16> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP6:%.*]] = and <32 x i16> [[X0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP7:%.*]] = and <32 x i16> [[TMP1]], [[X1:%.*]]
+; CHECK-NEXT: [[TMP8:%.*]] = or <32 x i16> [[TMP5]], [[TMP6]]
+; CHECK-NEXT: [[TMP17:%.*]] = or <32 x i16> [[TMP8]], [[TMP7]]
+; CHECK-NEXT: [[TMP18:%.*]] = shufflevector <32 x i16> [[TMP17]], <32 x i16> poison, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
+; CHECK-NEXT: [[TMP19:%.*]] = shufflevector <32 x i16> [[TMP17]], <32 x i16> poison, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
+; CHECK-NEXT: [[TMP20:%.*]] = or <16 x i16> [[TMP18]], [[TMP19]]
+; CHECK-NEXT: [[TMP21:%.*]] = zext <16 x i16> [[TMP20]] to <16 x i32>
+; CHECK-NEXT: [[TMP9:%.*]] = call <16 x i32> @llvm.x86.avx512.pmaddw.d.512(<32 x i16> [[X0]], <32 x i16> [[X1]])
; CHECK-NEXT: [[TMP10:%.*]] = bitcast i16 [[TMP3]] to <16 x i1>
; CHECK-NEXT: [[TMP11:%.*]] = bitcast i16 [[X3:%.*]] to <16 x i1>
-; CHECK-NEXT: [[TMP12:%.*]] = select <16 x i1> [[TMP11]], <16 x i32> zeroinitializer, <16 x i32> [[TMP4]]
+; CHECK-NEXT: [[TMP12:%.*]] = select <16 x i1> [[TMP11]], <16 x i32> [[TMP21]], <16 x i32> [[TMP4]]
; CHECK-NEXT: [[TMP13:%.*]] = xor <16 x i32> [[TMP9]], [[X2:%.*]]
-; CHECK-NEXT: [[TMP14:%.*]] = or <16 x i32> [[TMP13]], zeroinitializer
+; CHECK-NEXT: [[TMP14:%.*]] = or <16 x i32> [[TMP13]], [[TMP21]]
; CHECK-NEXT: [[TMP15:%.*]] = or <16 x i32> [[TMP14]], [[TMP4]]
; CHECK-NEXT: [[_MSPROP_SELECT:%.*]] = select <16 x i1> [[TMP10]], <16 x i32> [[TMP15]], <16 x i32> [[TMP12]]
; CHECK-NEXT: [[TMP16:%.*]] = select <16 x i1> [[TMP11]], <16 x i32> [[TMP9]], <16 x i32> [[X2]]
diff --git a/llvm/test/Instrumentation/MemorySanitizer/X86/mmx-intrinsics.ll b/llvm/test/Instrumentation/MemorySanitizer/X86/mmx-intrinsics.ll
index ac3bb56719038..9717471564bc7 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/X86/mmx-intrinsics.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/X86/mmx-intrinsics.ll
@@ -1687,11 +1687,18 @@ define i64 @test49(<1 x i64> %a, <1 x i64> %b) #0 {
; CHECK-NEXT: [[MMX_VAR_I:%.*]] = bitcast <4 x i16> [[TMP1]] to <1 x i64>
; CHECK-NEXT: [[TMP7:%.*]] = bitcast <4 x i16> [[TMP16]] to <1 x i64>
; CHECK-NEXT: [[MMX_VAR1_I:%.*]] = bitcast <4 x i16> [[TMP0]] to <1 x i64>
-; CHECK-NEXT: [[TMP8:%.*]] = or <1 x i64> [[TMP6]], [[TMP7]]
-; CHECK-NEXT: [[TMP9:%.*]] = bitcast <1 x i64> [[TMP8]] to <2 x i32>
-; CHECK-NEXT: [[TMP10:%.*]] = icmp ne <2 x i32> [[TMP9]], zeroinitializer
-; CHECK-NEXT: [[TMP11:%.*]] = sext <2 x i1> [[TMP10]] to <2 x i32>
-; CHECK-NEXT: [[TMP12:%.*]] = bitcast <2 x i32> [[TMP11]] to <1 x i64>
+; CHECK-NEXT: [[TMP8:%.*]] = and <1 x i64> [[TMP6]], [[TMP7]]
+; CHECK-NEXT: [[TMP9:%.*]] = and <1 x i64> [[MMX_VAR_I]], [[TMP7]]
+; CHECK-NEXT: [[TMP10:%.*]] = and <1 x i64> [[TMP6]], [[MMX_VAR1_I]]
+; CHECK-NEXT: [[TMP11:%.*]] = or <1 x i64> [[TMP8]], [[TMP9]]
+; CHECK-NEXT: [[TMP22:%.*]] = or <1 x i64> [[TMP11]], [[TMP10]]
+; CHECK-NEXT: [[TMP23:%.*]] = bitcast <1 x i64> [[TMP22]] to <4 x i16>
+; CHECK-NEXT: [[TMP24:%.*]] = shufflevector <4 x i16> [[TMP23]], <4 x i16> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT: [[TMP25:%.*]] = shufflevector <4 x i16> [[TMP23]], <4 x i16> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT: [[TMP26:%.*]] = or <2 x i16> [[TMP24]], [[TMP25]]
+; CHECK-NEXT: [[TMP27:%.*]] = bitcast <2 x i16> [[TMP26]] to i32
+; CHECK-NEXT: [[TMP28:%.*]] = zext i32 [[TMP27]] to i64
+; CHECK-NEXT: [[TMP12:%.*]] = bitcast i64 [[TMP28]] to <1 x i64>
; CHECK-NEXT: [[TMP14:%.*]] = tail call <1 x i64> @llvm.x86.mmx.pmadd.wd(<1 x i64> [[MMX_VAR_I]], <1 x i64> [[MMX_VAR1_I]]) #[[ATTR2]]
; CHECK-NEXT: [[TMP3:%.*]] = bitcast <1 x i64> [[TMP12]] to <2 x i32>
; CHECK-NEXT: [[TMP20:%.*]] = bitcast <1 x i64> [[TMP14]] to <2 x i32>
@@ -3315,16 +3322,23 @@ define i64 @test7(<1 x i64> %a, <1 x i64> %b) #0 {
; CHECK-NEXT: [[TMP22:%.*]] = bitcast <8 x i8> [[TMP1]] to <1 x i64>
; CHECK-NEXT: [[TMP8:%.*]] = bitcast <8 x i8> [[TMP17]] to <1 x i64>
; CHECK-NEXT: [[TMP23:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
-; CHECK-NEXT: [[TMP10:%.*]] = or <1 x i64> [[TMP21]], [[TMP8]]
-; CHECK-NEXT: [[TMP11:%.*]] = bitcast <1 x i64> [[TMP10]] to <4 x i16>
-; CHECK-NEXT: [[TMP12:%.*]] = icmp ne <4 x i16> [[TMP11]], zeroinitializer
-; CHECK-NEXT: [[TMP13:%.*]] = sext <4 x i1> [[TMP12]] to <4 x i16>
-; CHECK-NEXT: [[TMP14:%.*]] = bitcast <4 x i16> [[TMP13]] to <1 x i64>
-; CHECK-NEXT: [[TMP24:%.*]] = tail call <1 x i64> @llvm.x86.ssse3.pmadd.ub.sw(<1 x i64> [[TMP22]], <1 x i64> [[TMP23]]) #[[ATTR5]]
+; CHECK-NEXT: [[TMP10:%.*]] = and <1 x i64> [[TMP21]], [[TMP8]]
+; CHECK-NEXT: [[TMP11:%.*]] = and <1 x i64> [[TMP22]], [[TMP8]]
+; CHECK-NEXT: [[TMP12:%.*]] = and <1 x i64> [[TMP21]], [[TMP23]]
+; CHECK-NEXT: [[TMP13:%.*]] = or <1 x i64> [[TMP10]], [[TMP11]]
+; CHECK-NEXT: [[TMP14:%.*]] = or <1 x i64> [[TMP13]], [[TMP12]]
; CHECK-NEXT: [[TMP16:%.*]] = bitcast <1 x i64> [[TMP14]] to <8 x i8>
+; CHECK-NEXT: [[TMP25:%.*]] = shufflevector <8 x i8> [[TMP16]], <8 x i8> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; CHECK-NEXT: [[TMP26:%.*]] = shufflevector <8 x i8> [[TMP16]], <8 x i8> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; CHECK-NEXT: [[TMP27:%.*]] = or <4 x i8> [[TMP25]], [[TMP26]]
+; CHECK-NEXT: [[TMP28:%.*]] = bitcast <4 x i8> [[TMP27]] to i32
+; CHECK-NEXT: [[TMP29:%.*]] = zext i32 [[TMP28]] to i64
+; CHECK-NEXT: [[TMP24:%.*]] = bitcast i64 [[TMP29]] to <1 x i64>
+; CHECK-NEXT: [[TMP30:%.*]] = tail call <1 x i64> @llvm.x86.ssse3.pmadd.ub.sw(<1 x i64> [[TMP22]], <1 x i64> [[TMP23]]) #[[ATTR5]]
; CHECK-NEXT: [[TMP5:%.*]] = bitcast <1 x i64> [[TMP24]] to <8 x i8>
-; CHECK-NEXT: [[TMP6:%.*]] = bitcast <8 x i8> [[TMP16]] to <1 x i64>
-; CHECK-NEXT: [[TMP19:%.*]] = bitcast <8 x i8> [[TMP5]] to <1 x i64>
+; CHECK-NEXT: [[TMP31:%.*]] = bitcast <1 x i64> [[TMP30]] to <8 x i8>
+; CHECK-NEXT: [[TMP6:%.*]] = bitcast <8 x i8> [[TMP5]] to <1 x i64>
+; CHECK-NEXT: [[TMP19:%.*]] = bitcast <8 x i8> [[TMP31]] to <1 x i64>
; CHECK-NEXT: [[TMP7:%.*]] = extractelement <1 x i64> [[TMP6]], i32 0
; CHECK-NEXT: [[TMP20:%.*]] = extractelement <1 x i64> [[TMP19]], i32 0
; CHECK-NEXT: store i64 [[TMP7]], ptr @__msan_retval_tls, align 8
diff --git a/llvm/test/Instrumentation/MemorySanitizer/X86/sse2-intrinsics-x86.ll b/llvm/test/Instrumentation/MemorySanitizer/X86/sse2-intrinsics-x86.ll
index 8f915a59db8e5..afd8c7a58b6ee 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/X86/sse2-intrinsics-x86.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/X86/sse2-intrinsics-x86.ll
@@ -762,11 +762,16 @@ define <4 x i32> @test_x86_sse2_pmadd_wd(<8 x i16> %a0, <8 x i16> %a1) #0 {
; CHECK-NEXT: [[TMP1:%.*]] = load <8 x i16>, ptr @__msan_param_tls, align 8
; CHECK-NEXT: [[TMP2:%.*]] = load <8 x i16>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 16) to ptr), align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP3:%.*]] = or <8 x i16> [[TMP1]], [[TMP2]]
-; CHECK-NEXT: [[TMP4:%.*]] = bitcast <8 x i16> [[TMP3]] to <4 x i32>
-; CHECK-NEXT: [[TMP5:%.*]] = icmp ne <4 x i32> [[TMP4]], zeroinitializer
-; CHECK-NEXT: [[TMP6:%.*]] = sext <4 x i1> [[TMP5]] to <4 x i32>
-; CHECK-NEXT: [[RES:%.*]] = call <4 x i32> @llvm.x86.sse2.pmadd.wd(<8 x i16> [[A0:%.*]], <8 x i16> [[A1:%.*]])
+; CHECK-NEXT: [[TMP3:%.*]] = and <8 x i16> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP4:%.*]] = and <8 x i16> [[A0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP5:%.*]] = and <8 x i16> [[TMP1]], [[A1:%.*]]
+; CHECK-NEXT: [[TMP11:%.*]] = or <8 x i16> [[TMP3]], [[TMP4]]
+; CHECK-NEXT: [[TMP7:%.*]] = or <8 x i16> [[TMP11]], [[TMP5]]
+; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <8 x i16> [[TMP7]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <8 x i16> [[TMP7]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; CHECK-NEXT: [[TMP10:%.*]] = or <4 x i16> [[TMP8]], [[TMP9]]
+; CHECK-NEXT: [[TMP6:%.*]] = zext <4 x i16> [[TMP10]] to <4 x i32>
+; CHECK-NEXT: [[RES:%.*]] = call <4 x i32> @llvm.x86.sse2.pmadd.wd(<8 x i16> [[A0]], <8 x i16> [[A1]])
; CHECK-NEXT: store <4 x i32> [[TMP6]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <4 x i32> [[RES]]
;
diff --git a/llvm/test/Instrumentation/MemorySanitizer/i386/avx2-intrinsics-i386.ll b/llvm/test/Instrumentation/MemorySanitizer/i386/avx2-intrinsics-i386.ll
index 5cc56baf0e0de..f92f7ef3fa4a9 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/i386/avx2-intrinsics-i386.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/i386/avx2-intrinsics-i386.ll
@@ -149,11 +149,16 @@ define <8 x i32> @test_x86_avx2_pmadd_wd(<16 x i16> %a0, <16 x i16> %a1) #0 {
; CHECK-NEXT: [[TMP2:%.*]] = load <16 x i16>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 32) to ptr), align 8
; CHECK-NEXT: [[TMP7:%.*]] = load i64, ptr @__msan_va_arg_overflow_size_tls, align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP3:%.*]] = or <16 x i16> [[TMP1]], [[TMP2]]
-; CHECK-NEXT: [[TMP4:%.*]] = bitcast <16 x i16> [[TMP3]] to <8 x i32>
-; CHECK-NEXT: [[TMP5:%.*]] = icmp ne <8 x i32> [[TMP4]], zeroinitializer
-; CHECK-NEXT: [[TMP6:%.*]] = sext <8 x i1> [[TMP5]] to <8 x i32>
-; CHECK-NEXT: [[RES:%.*]] = call <8 x i32> @llvm.x86.avx2.pmadd.wd(<16 x i16> [[A0:%.*]], <16 x i16> [[A1:%.*]])
+; CHECK-NEXT: [[TMP4:%.*]] = and <16 x i16> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP5:%.*]] = and <16 x i16> [[A0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP12:%.*]] = and <16 x i16> [[TMP1]], [[A1:%.*]]
+; CHECK-NEXT: [[TMP13:%.*]] = or <16 x i16> [[TMP4]], [[TMP5]]
+; CHECK-NEXT: [[TMP8:%.*]] = or <16 x i16> [[TMP13]], [[TMP12]]
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <16 x i16> [[TMP8]], <16 x i16> poison, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
+; CHECK-NEXT: [[TMP10:%.*]] = shufflevector <16 x i16> [[TMP8]], <16 x i16> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
+; CHECK-NEXT: [[TMP11:%.*]] = or <8 x i16> [[TMP9]], [[TMP10]]
+; CHECK-NEXT: [[TMP6:%.*]] = zext <8 x i16> [[TMP11]] to <8 x i32>
+; CHECK-NEXT: [[RES:%.*]] = call <8 x i32> @llvm.x86.avx2.pmadd.wd(<16 x i16> [[A0]], <16 x i16> [[A1]])
; CHECK-NEXT: store <8 x i32> [[TMP6]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <8 x i32> [[RES]]
;
@@ -714,11 +719,16 @@ define <16 x i16> @test_x86_avx2_pmadd_ub_sw(<32 x i8> %a0, <32 x i8> %a1) #0 {
; CHECK-NEXT: [[TMP2:%.*]] = load <32 x i8>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 32) to ptr), align 8
; CHECK-NEXT: [[TMP7:%.*]] = load i64, ptr @__msan_va_arg_overflow_size_tls, align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP3:%.*]] = or <32 x i8> [[TMP1]], [[TMP2]]
-; CHECK-NEXT: [[TMP4:%.*]] = bitcast <32 x i8> [[TMP3]] to <16 x i16>
-; CHECK-NEXT: [[TMP5:%.*]] = icmp ne <16 x i16> [[TMP4]], zeroinitializer
-; CHECK-NEXT: [[TMP6:%.*]] = sext <16 x i1> [[TMP5]] to <16 x i16>
-; CHECK-NEXT: [[RES:%.*]] = call <16 x i16> @llvm.x86.avx2.pmadd.ub.sw(<32 x i8> [[A0:%.*]], <32 x i8> [[A1:%.*]])
+; CHECK-NEXT: [[TMP4:%.*]] = and <32 x i8> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP5:%.*]] = and <32 x i8> [[A0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP12:%.*]] = and <32 x i8> [[TMP1]], [[A1:%.*]]
+; CHECK-NEXT: [[TMP13:%.*]] = or <32 x i8> [[TMP4]], [[TMP5]]
+; CHECK-NEXT: [[TMP8:%.*]] = or <32 x i8> [[TMP13]], [[TMP12]]
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <32 x i8> [[TMP8]], <32 x i8> poison, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
+; CHECK-NEXT: [[TMP10:%.*]] = shufflevector <32 x i8> [[TMP8]], <32 x i8> poison, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
+; CHECK-NEXT: [[TMP11:%.*]] = or <16 x i8> [[TMP9]], [[TMP10]]
+; CHECK-NEXT: [[TMP6:%.*]] = zext <16 x i8> [[TMP11]] to <16 x i16>
+; CHECK-NEXT: [[RES:%.*]] = call <16 x i16> @llvm.x86.avx2.pmadd.ub.sw(<32 x i8> [[A0]], <32 x i8> [[A1]])
; CHECK-NEXT: store <16 x i16> [[TMP6]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <16 x i16> [[RES]]
;
@@ -734,7 +744,7 @@ define <16 x i16> @test_x86_avx2_pmadd_ub_sw_load_op0(ptr %ptr, <32 x i8> %a1) #
; CHECK-NEXT: [[TMP3:%.*]] = load i64, ptr @__msan_va_arg_overflow_size_tls, align 8
; CHECK-NEXT: call void @llvm.donothing()
; CHECK-NEXT: [[_MSCMP:%.*]] = icmp ne i64 [[TMP1]], 0
-; CHECK-NEXT: br i1 [[_MSCMP]], label [[TMP4:%.*]], label [[TMP12:%.*]], !prof [[PROF1]]
+; CHECK-NEXT: br i1 [[_MSCMP]], label [[TMP4:%.*]], label [[TMP8:%.*]], !prof [[PROF1]]
; CHECK: 4:
; CHECK-NEXT: call void @__msan_warning_noreturn() #[[ATTR6]]
; CHECK-NEXT: unreachable
@@ -744,11 +754,16 @@ define <16 x i16> @test_x86_avx2_pmadd_ub_sw_load_op0(ptr %ptr, <32 x i8> %a1) #
; CHECK-NEXT: [[TMP6:%.*]] = and i64 [[TMP5]], -2147483649
; CHECK-NEXT: [[TMP7:%.*]] = inttoptr i64 [[TMP6]] to ptr
; CHECK-NEXT: [[_MSLD:%.*]] = load <32 x i8>, ptr [[TMP7]], align 32
-; CHECK-NEXT: [[TMP8:%.*]] = or <32 x i8> [[_MSLD]], [[TMP2]]
-; CHECK-NEXT: [[TMP9:%.*]] = bitcast <32 x i8> [[TMP8]] to <16 x i16>
-; CHECK-NEXT: [[TMP10:%.*]] = icmp ne <16 x i16> [[TMP9]], zeroinitializer
-; CHECK-NEXT: [[TMP11:%.*]] = sext <16 x i1> [[TMP10]] to <16 x i16>
-; CHECK-NEXT: [[RES:%.*]] = call <16 x i16> @llvm.x86.avx2.pmadd.ub.sw(<32 x i8> [[A0]], <32 x i8> [[A1:%.*]])
+; CHECK-NEXT: [[TMP9:%.*]] = and <32 x i8> [[_MSLD]], [[TMP2]]
+; CHECK-NEXT: [[TMP10:%.*]] = and <32 x i8> [[A0]], [[TMP2]]
+; CHECK-NEXT: [[TMP17:%.*]] = and <32 x i8> [[_MSLD]], [[A1:%.*]]
+; CHECK-NEXT: [[TMP12:%.*]] = or <32 x i8> [[TMP9]], [[TMP10]]
+; CHECK-NEXT: [[TMP13:%.*]] = or <32 x i8> [[TMP12]], [[TMP17]]
+; CHECK-NEXT: [[TMP14:%.*]] = shufflevector <32 x i8> [[TMP13]], <32 x i8> poison, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
+; CHECK-NEXT: [[TMP15:%.*]] = shufflevector <32 x i8> [[TMP13]], <32 x i8> poison, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
+; CHECK-NEXT: [[TMP16:%.*]] = or <16 x i8> [[TMP14]], [[TMP15]]
+; CHECK-NEXT: [[TMP11:%.*]] = zext <16 x i8> [[TMP16]] to <16 x i16>
+; CHECK-NEXT: [[RES:%.*]] = call <16 x i16> @llvm.x86.avx2.pmadd.ub.sw(<32 x i8> [[A0]], <32 x i8> [[A1]])
; CHECK-NEXT: store <16 x i16> [[TMP11]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <16 x i16> [[RES]]
;
diff --git a/llvm/test/Instrumentation/MemorySanitizer/i386/mmx-intrinsics.ll b/llvm/test/Instrumentation/MemorySanitizer/i386/mmx-intrinsics.ll
index 0a3efaaea149f..dc9173a8b2a18 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/i386/mmx-intrinsics.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/i386/mmx-intrinsics.ll
@@ -1730,11 +1730,18 @@ define i64 @test49(<1 x i64> %a, <1 x i64> %b) #0 {
; CHECK-NEXT: [[MMX_VAR_I:%.*]] = bitcast <4 x i16> [[TMP1]] to <1 x i64>
; CHECK-NEXT: [[TMP7:%.*]] = bitcast <4 x i16> [[TMP16]] to <1 x i64>
; CHECK-NEXT: [[MMX_VAR1_I:%.*]] = bitcast <4 x i16> [[TMP0]] to <1 x i64>
-; CHECK-NEXT: [[TMP8:%.*]] = or <1 x i64> [[TMP6]], [[TMP7]]
-; CHECK-NEXT: [[TMP9:%.*]] = bitcast <1 x i64> [[TMP8]] to <2 x i32>
-; CHECK-NEXT: [[TMP10:%.*]] = icmp ne <2 x i32> [[TMP9]], zeroinitializer
-; CHECK-NEXT: [[TMP11:%.*]] = sext <2 x i1> [[TMP10]] to <2 x i32>
-; CHECK-NEXT: [[TMP12:%.*]] = bitcast <2 x i32> [[TMP11]] to <1 x i64>
+; CHECK-NEXT: [[TMP9:%.*]] = and <1 x i64> [[TMP6]], [[TMP7]]
+; CHECK-NEXT: [[TMP10:%.*]] = and <1 x i64> [[MMX_VAR_I]], [[TMP7]]
+; CHECK-NEXT: [[TMP11:%.*]] = and <1 x i64> [[TMP6]], [[MMX_VAR1_I]]
+; CHECK-NEXT: [[TMP22:%.*]] = or <1 x i64> [[TMP9]], [[TMP10]]
+; CHECK-NEXT: [[TMP23:%.*]] = or <1 x i64> [[TMP22]], [[TMP11]]
+; CHECK-NEXT: [[TMP24:%.*]] = bitcast <1 x i64> [[TMP23]] to <4 x i16>
+; CHECK-NEXT: [[TMP25:%.*]] = shufflevector <4 x i16> [[TMP24]], <4 x i16> poison, <2 x i32> <i32 0, i32 2>
+; CHECK-NEXT: [[TMP26:%.*]] = shufflevector <4 x i16> [[TMP24]], <4 x i16> poison, <2 x i32> <i32 1, i32 3>
+; CHECK-NEXT: [[TMP27:%.*]] = or <2 x i16> [[TMP25]], [[TMP26]]
+; CHECK-NEXT: [[TMP28:%.*]] = bitcast <2 x i16> [[TMP27]] to i32
+; CHECK-NEXT: [[TMP29:%.*]] = zext i32 [[TMP28]] to i64
+; CHECK-NEXT: [[TMP12:%.*]] = bitcast i64 [[TMP29]] to <1 x i64>
; CHECK-NEXT: [[TMP14:%.*]] = tail call <1 x i64> @llvm.x86.mmx.pmadd.wd(<1 x i64> [[MMX_VAR_I]], <1 x i64> [[MMX_VAR1_I]]) #[[ATTR2]]
; CHECK-NEXT: [[TMP3:%.*]] = bitcast <1 x i64> [[TMP12]] to <2 x i32>
; CHECK-NEXT: [[TMP20:%.*]] = bitcast <1 x i64> [[TMP14]] to <2 x i32>
@@ -3401,16 +3408,23 @@ define i64 @test7(<1 x i64> %a, <1 x i64> %b) #0 {
; CHECK-NEXT: [[TMP22:%.*]] = bitcast <8 x i8> [[TMP1]] to <1 x i64>
; CHECK-NEXT: [[TMP8:%.*]] = bitcast <8 x i8> [[TMP17]] to <1 x i64>
; CHECK-NEXT: [[TMP23:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
-; CHECK-NEXT: [[TMP10:%.*]] = or <1 x i64> [[TMP21]], [[TMP8]]
-; CHECK-NEXT: [[TMP11:%.*]] = bitcast <1 x i64> [[TMP10]] to <4 x i16>
-; CHECK-NEXT: [[TMP12:%.*]] = icmp ne <4 x i16> [[TMP11]], zeroinitializer
-; CHECK-NEXT: [[TMP13:%.*]] = sext <4 x i1> [[TMP12]] to <4 x i16>
-; CHECK-NEXT: [[TMP14:%.*]] = bitcast <4 x i16> [[TMP13]] to <1 x i64>
-; CHECK-NEXT: [[TMP24:%.*]] = tail call <1 x i64> @llvm.x86.ssse3.pmadd.ub.sw(<1 x i64> [[TMP22]], <1 x i64> [[TMP23]]) #[[ATTR5]]
+; CHECK-NEXT: [[TMP11:%.*]] = and <1 x i64> [[TMP21]], [[TMP8]]
+; CHECK-NEXT: [[TMP12:%.*]] = and <1 x i64> [[TMP22]], [[TMP8]]
+; CHECK-NEXT: [[TMP13:%.*]] = and <1 x i64> [[TMP21]], [[TMP23]]
+; CHECK-NEXT: [[TMP26:%.*]] = or <1 x i64> [[TMP11]], [[TMP12]]
+; CHECK-NEXT: [[TMP14:%.*]] = or <1 x i64> [[TMP26]], [[TMP13]]
; CHECK-NEXT: [[TMP16:%.*]] = bitcast <1 x i64> [[TMP14]] to <8 x i8>
+; CHECK-NEXT: [[TMP27:%.*]] = shufflevector <8 x i8> [[TMP16]], <8 x i8> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; CHECK-NEXT: [[TMP28:%.*]] = shufflevector <8 x i8> [[TMP16]], <8 x i8> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; CHECK-NEXT: [[TMP29:%.*]] = or <4 x i8> [[TMP27]], [[TMP28]]
+; CHECK-NEXT: [[TMP30:%.*]] = bitcast <4 x i8> [[TMP29]] to i32
+; CHECK-NEXT: [[TMP31:%.*]] = zext i32 [[TMP30]] to i64
+; CHECK-NEXT: [[TMP24:%.*]] = bitcast i64 [[TMP31]] to <1 x i64>
+; CHECK-NEXT: [[TMP32:%.*]] = tail call <1 x i64> @llvm.x86.ssse3.pmadd.ub.sw(<1 x i64> [[TMP22]], <1 x i64> [[TMP23]]) #[[ATTR5]]
; CHECK-NEXT: [[TMP5:%.*]] = bitcast <1 x i64> [[TMP24]] to <8 x i8>
-; CHECK-NEXT: [[TMP6:%.*]] = bitcast <8 x i8> [[TMP16]] to <1 x i64>
-; CHECK-NEXT: [[TMP19:%.*]] = bitcast <8 x i8> [[TMP5]] to <1 x i64>
+; CHECK-NEXT: [[TMP25:%.*]] = bitcast <1 x i64> [[TMP32]] to <8 x i8>
+; CHECK-NEXT: [[TMP6:%.*]] = bitcast <8 x i8> [[TMP5]] to <1 x i64>
+; CHECK-NEXT: [[TMP19:%.*]] = bitcast <8 x i8> [[TMP25]] to <1 x i64>
; CHECK-NEXT: [[TMP7:%.*]] = extractelement <1 x i64> [[TMP6]], i32 0
; CHECK-NEXT: [[TMP20:%.*]] = extractelement <1 x i64> [[TMP19]], i32 0
; CHECK-NEXT: store i64 [[TMP7]], ptr @__msan_retval_tls, align 8
diff --git a/llvm/test/Instrumentation/MemorySanitizer/i386/sse2-intrinsics-i386.ll b/llvm/test/Instrumentation/MemorySanitizer/i386/sse2-intrinsics-i386.ll
index e771e60e2f294..d4c2d8bb6749f 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/i386/sse2-intrinsics-i386.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/i386/sse2-intrinsics-i386.ll
@@ -800,11 +800,16 @@ define <4 x i32> @test_x86_sse2_pmadd_wd(<8 x i16> %a0, <8 x i16> %a1) #0 {
; CHECK-NEXT: [[TMP2:%.*]] = load <8 x i16>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 16) to ptr), align 8
; CHECK-NEXT: [[TMP7:%.*]] = load i64, ptr @__msan_va_arg_overflow_size_tls, align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP3:%.*]] = or <8 x i16> [[TMP1]], [[TMP2]]
-; CHECK-NEXT: [[TMP4:%.*]] = bitcast <8 x i16> [[TMP3]] to <4 x i32>
-; CHECK-NEXT: [[TMP5:%.*]] = icmp ne <4 x i32> [[TMP4]], zeroinitializer
-; CHECK-NEXT: [[TMP6:%.*]] = sext <4 x i1> [[TMP5]] to <4 x i32>
-; CHECK-NEXT: [[RES:%.*]] = call <4 x i32> @llvm.x86.sse2.pmadd.wd(<8 x i16> [[A0:%.*]], <8 x i16> [[A1:%.*]])
+; CHECK-NEXT: [[TMP4:%.*]] = and <8 x i16> [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP5:%.*]] = and <8 x i16> [[A0:%.*]], [[TMP2]]
+; CHECK-NEXT: [[TMP12:%.*]] = and <8 x i16> [[TMP1]], [[A1:%.*]]
+; CHECK-NEXT: [[TMP13:%.*]] = or <8 x i16> [[TMP4]], [[TMP5]]
+; CHECK-NEXT: [[TMP8:%.*]] = or <8 x i16> [[TMP13]], [[TMP12]]
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <8 x i16> [[TMP8]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; CHECK-NEXT: [[TMP10:%.*]] = shufflevector <8 x i16> [[TMP8]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; CHECK-NEXT: [[TMP11:%.*]] = or <4 x i16> [[TMP9]], [[TMP10]]
+; CHECK-NEXT: [[TMP6:%.*]] = zext <4 x i16> [[TMP11]] to <4 x i32>
+; CHECK-NEXT: [[RES:%.*]] = call <4 x i32> @llvm.x86.sse2.pmadd.wd(<8 x i16> [[A0]], <8 x i16> [[A1]])
; CHECK-NEXT: store <4 x i32> [[TMP6]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <4 x i32> [[RES]]
;
diff --git a/llvm/test/Instrumentation/MemorySanitizer/vector_arith.ll b/llvm/test/Instrumentation/MemorySanitizer/vector_arith.ll
index d614bb85d8584..e9b5462bc85cb 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/vector_arith.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/vector_arith.ll
@@ -17,10 +17,15 @@ define <4 x i32> @Test_sse2_pmadd_wd(<8 x i16> %a, <8 x i16> %b) sanitize_memory
; CHECK-NEXT: [[TMP0:%.*]] = load <8 x i16>, ptr @__msan_param_tls, align 8
; CHECK-NEXT: [[TMP1:%.*]] = load <8 x i16>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 16) to ptr), align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP2:%.*]] = or <8 x i16> [[TMP0]], [[TMP1]]
-; CHECK-NEXT: [[TMP3:%.*]] = bitcast <8 x i16> [[TMP2]] to <4 x i32>
-; CHECK-NEXT: [[TMP4:%.*]] = icmp ne <4 x i32> [[TMP3]], zeroinitializer
-; CHECK-NEXT: [[TMP5:%.*]] = sext <4 x i1> [[TMP4]] to <4 x i32>
+; CHECK-NEXT: [[TMP2:%.*]] = and <8 x i16> [[TMP0]], [[TMP1]]
+; CHECK-NEXT: [[TMP3:%.*]] = and <8 x i16> [[A]], [[TMP1]]
+; CHECK-NEXT: [[TMP4:%.*]] = and <8 x i16> [[TMP0]], [[B]]
+; CHECK-NEXT: [[TMP10:%.*]] = or <8 x i16> [[TMP2]], [[TMP3]]
+; CHECK-NEXT: [[TMP6:%.*]] = or <8 x i16> [[TMP10]], [[TMP4]]
+; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <8 x i16> [[TMP6]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <8 x i16> [[TMP6]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; CHECK-NEXT: [[TMP9:%.*]] = or <4 x i16> [[TMP7]], [[TMP8]]
+; CHECK-NEXT: [[TMP5:%.*]] = zext <4 x i16> [[TMP9]] to <4 x i32>
; CHECK-NEXT: [[C:%.*]] = tail call <4 x i32> @llvm.x86.sse2.pmadd.wd(<8 x i16> [[A]], <8 x i16> [[B]]) #[[ATTR2:[0-9]+]]
; CHECK-NEXT: store <4 x i32> [[TMP5]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <4 x i32> [[C]]
@@ -39,11 +44,18 @@ define <1 x i64> @Test_ssse3_pmadd_ub_sw(<1 x i64> %a, <1 x i64> %b) sanitize_me
; CHECK-NEXT: [[TMP0:%.*]] = load <1 x i64>, ptr @__msan_param_tls, align 8
; CHECK-NEXT: [[TMP1:%.*]] = load <1 x i64>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 8) to ptr), align 8
; CHECK-NEXT: call void @llvm.donothing()
-; CHECK-NEXT: [[TMP2:%.*]] = or <1 x i64> [[TMP0]], [[TMP1]]
-; CHECK-NEXT: [[TMP3:%.*]] = bitcast <1 x i64> [[TMP2]] to <4 x i16>
-; CHECK-NEXT: [[TMP4:%.*]] = icmp ne <4 x i16> [[TMP3]], zeroinitializer
-; CHECK-NEXT: [[TMP5:%.*]] = sext <4 x i1> [[TMP4]] to <4 x i16>
-; CHECK-NEXT: [[TMP6:%.*]] = bitcast <4 x i16> [[TMP5]] to <1 x i64>
+; CHECK-NEXT: [[TMP2:%.*]] = and <1 x i64> [[TMP0]], [[TMP1]]
+; CHECK-NEXT: [[TMP3:%.*]] = and <1 x i64> [[A]], [[TMP1]]
+; CHECK-NEXT: [[TMP4:%.*]] = and <1 x i64> [[TMP0]], [[B]]
+; CHECK-NEXT: [[TMP5:%.*]] = or <1 x i64> [[TMP2]], [[TMP3]]
+; CHECK-NEXT: [[TMP13:%.*]] = or <1 x i64> [[TMP5]], [[TMP4]]
+; CHECK-NEXT: [[TMP7:%.*]] = bitcast <1 x i64> [[TMP13]] to <8 x i8>
+; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <8 x i8> [[TMP7]], <8 x i8> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <8 x i8> [[TMP7]], <8 x i8> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
+; CHECK-NEXT: [[TMP10:%.*]] = or <4 x i8> [[TMP8]], [[TMP9]]
+; CHECK-NEXT: [[TMP11:%.*]] = bitcast <4 x i8> [[TMP10]] to i32
+; CHECK-NEXT: [[TMP12:%.*]] = zext i32 [[TMP11]] to i64
+; CHECK-NEXT: [[TMP6:%.*]] = bitcast i64 [[TMP12]] to <1 x i64>
; CHECK-NEXT: [[C:%.*]] = tail call <1 x i64> @llvm.x86.ssse3.pmadd.ub.sw(<1 x i64> [[A]], <1 x i64> [[B]]) #[[ATTR2]]
; CHECK-NEXT: store <1 x i64> [[TMP6]], ptr @__msan_retval_tls, align 8
; CHECK-NEXT: ret <1 x i64> [[C]]
>From 2ccc07f7781f74e069b6c1ab76f0af2f657adadd Mon Sep 17 00:00:00 2001
From: Thurston Dang <thurston at google.com>
Date: Sun, 10 Aug 2025 21:00:24 +0000
Subject: [PATCH 2/2] Update comment
---
llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp b/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp
index 0bf1bc587c62a..f938b0acaa0be 100644
--- a/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp
+++ b/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp
@@ -3856,8 +3856,9 @@ struct MemorySanitizerVisitor : public InstVisitor<MemorySanitizerVisitor> {
// Step 1: multiplication of corresponding vector elements
// We want to take into account the fact that multiplying zero by an
// uninitialized bit results in an initialized value of zero.
- // We under-approximate multiplication using the same logic as visitAnd().
- // This ignores the carrying that may happen during multiplication.
+ // We under-approximate multiplication by treating it as bitwise AND; this
+ // has no false positives but substantial false negatives. We then
+ // compute the shadow using the same logic as visitAnd().
Value *S1 = getShadow(&I, 0);
Value *S2 = getShadow(&I, 1);
Value *V1 = I.getOperand(0);
More information about the llvm-commits
mailing list