diff --git a/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp b/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp
index ec94dcaa2c051..5a6520e964c25 100644
--- a/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp
+++ b/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp
@@ -4297,7 +4297,7 @@ struct MemorySanitizerVisitor : public InstVisitor<MemorySanitizerVisitor> {
   void handleAVXVpermilvar(IntrinsicInst &I) {
     IRBuilder<> IRB(&I);
     Value *Shadow = getShadow(&I, 0);
-    insertShadowCheck(I.getArgOperand(1), &I);
+    maskedCheckAVXIndexShadow(IRB, I.getArgOperand(1), &I);
 
     // Shadows are integer-ish types but some intrinsics require a
     // different (e.g., floating-point) type.
diff --git a/llvm/test/Instrumentation/MemorySanitizer/X86/avx-intrinsics-x86.ll b/llvm/test/Instrumentation/MemorySanitizer/X86/avx-intrinsics-x86.ll
index c4016e749e70c..6542e56b3fefe 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/X86/avx-intrinsics-x86.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/X86/avx-intrinsics-x86.ll
@@ -950,16 +950,21 @@ define <2 x double> @test_x86_avx_vpermilvar_pd(<2 x double> %a0, <2 x i64> %a1)
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <2 x i64>, ptr @__msan_param_tls, align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <2 x i64>, 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:%.*]] = extractelement <2 x i64> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP9:%.*]] = and i64 [[TMP3]], 1
+; CHECK-NEXT:    [[TMP5:%.*]] = or i64 [[TMP3]], [[TMP9]]
+; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP7:%.*]] = and i64 [[TMP6]], 1
+; CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP6]], [[TMP7]]
 ; CHECK-NEXT:    [[A0:%.*]] = bitcast <2 x i64> [[TMP1]] to <2 x double>
 ; CHECK-NEXT:    [[RES:%.*]] = call <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double> [[A0]], <2 x i64> [[A1:%.*]])
 ; CHECK-NEXT:    [[TMP4:%.*]] = bitcast <2 x double> [[RES]] to <2 x i64>
-; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <2 x i64> [[TMP2]] to i128
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i128 [[TMP6]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP7:%.*]], label [[TMP8:%.*]], !prof [[PROF1]]
-; CHECK:       7:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i64 [[TMP8]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP12:%.*]], label [[TMP13:%.*]], !prof [[PROF1]]
+; CHECK:       12:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn()
 ; CHECK-NEXT:    unreachable
-; CHECK:       8:
+; CHECK:       13:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double> [[A2:%.*]], <2 x i64> [[A1]])
 ; CHECK-NEXT:    store <2 x i64> [[TMP4]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <2 x double> [[RES1]]
@@ -975,16 +980,27 @@ define <4 x double> @test_x86_avx_vpermilvar_pd_256(<4 x double> %a0, <4 x i64>
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <4 x i64>, ptr @__msan_param_tls, align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <4 x i64>, 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:%.*]] = extractelement <4 x i64> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP15:%.*]] = and i64 [[TMP3]], 3
+; CHECK-NEXT:    [[TMP5:%.*]] = or i64 [[TMP3]], [[TMP15]]
+; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <4 x i64> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP7:%.*]] = and i64 [[TMP6]], 3
+; CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP6]], [[TMP7]]
+; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <4 x i64> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP10:%.*]] = and i64 [[TMP9]], 3
+; CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP9]], [[TMP10]]
+; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <4 x i64> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP13:%.*]] = and i64 [[TMP12]], 3
+; CHECK-NEXT:    [[TMP14:%.*]] = or i64 [[TMP12]], [[TMP13]]
 ; CHECK-NEXT:    [[A0:%.*]] = bitcast <4 x i64> [[TMP1]] to <4 x double>
 ; CHECK-NEXT:    [[RES:%.*]] = call <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double> [[A0]], <4 x i64> [[A1:%.*]])
 ; CHECK-NEXT:    [[TMP4:%.*]] = bitcast <4 x double> [[RES]] to <4 x i64>
-; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <4 x i64> [[TMP2]] to i256
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i256 [[TMP6]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP7:%.*]], label [[TMP8:%.*]], !prof [[PROF1]]
-; CHECK:       7:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i64 [[TMP14]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP18:%.*]], label [[TMP19:%.*]], !prof [[PROF1]]
+; CHECK:       18:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn()
 ; CHECK-NEXT:    unreachable
-; CHECK:       8:
+; CHECK:       19:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double> [[A2:%.*]], <4 x i64> [[A1]])
 ; CHECK-NEXT:    store <4 x i64> [[TMP4]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <4 x double> [[RES1]]
@@ -1014,16 +1030,27 @@ define <4 x float> @test_x86_avx_vpermilvar_ps(<4 x float> %a0, <4 x i32> %a1) #
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <4 x i32>, ptr @__msan_param_tls, align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <4 x i32>, 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:%.*]] = extractelement <4 x i32> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP15:%.*]] = and i32 [[TMP3]], 3
+; CHECK-NEXT:    [[TMP5:%.*]] = or i32 [[TMP3]], [[TMP15]]
+; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <4 x i32> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP7:%.*]] = and i32 [[TMP6]], 3
+; CHECK-NEXT:    [[TMP8:%.*]] = or i32 [[TMP6]], [[TMP7]]
+; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <4 x i32> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP10:%.*]] = and i32 [[TMP9]], 3
+; CHECK-NEXT:    [[TMP11:%.*]] = or i32 [[TMP9]], [[TMP10]]
+; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <4 x i32> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP13:%.*]] = and i32 [[TMP12]], 3
+; CHECK-NEXT:    [[TMP14:%.*]] = or i32 [[TMP12]], [[TMP13]]
 ; CHECK-NEXT:    [[A0:%.*]] = bitcast <4 x i32> [[TMP1]] to <4 x float>
 ; CHECK-NEXT:    [[RES:%.*]] = call <4 x float> @llvm.x86.avx.vpermilvar.ps(<4 x float> [[A0]], <4 x i32> [[A1:%.*]])
 ; CHECK-NEXT:    [[TMP4:%.*]] = bitcast <4 x float> [[RES]] to <4 x i32>
-; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <4 x i32> [[TMP2]] to i128
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i128 [[TMP6]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP7:%.*]], label [[TMP8:%.*]], !prof [[PROF1]]
-; CHECK:       7:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i32 [[TMP14]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP18:%.*]], label [[TMP19:%.*]], !prof [[PROF1]]
+; CHECK:       18:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn()
 ; CHECK-NEXT:    unreachable
-; CHECK:       8:
+; CHECK:       19:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <4 x float> @llvm.x86.avx.vpermilvar.ps(<4 x float> [[A2:%.*]], <4 x i32> [[A1]])
 ; CHECK-NEXT:    store <4 x i32> [[TMP4]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <4 x float> [[RES1]]
@@ -1047,16 +1074,27 @@ define <4 x float> @test_x86_avx_vpermilvar_ps_load(<4 x float> %a0, ptr %a1) #0
 ; CHECK-NEXT:    [[TMP6:%.*]] = xor i64 [[TMP5]], 87960930222080
 ; CHECK-NEXT:    [[TMP7:%.*]] = inttoptr i64 [[TMP6]] to ptr
 ; CHECK-NEXT:    [[_MSLD:%.*]] = load <4 x i32>, ptr [[TMP7]], align 16
+; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <4 x i32> [[_MSLD]], i64 0
+; CHECK-NEXT:    [[TMP9:%.*]] = and i32 [[TMP8]], 3
+; CHECK-NEXT:    [[TMP20:%.*]] = or i32 [[TMP8]], [[TMP9]]
+; CHECK-NEXT:    [[TMP11:%.*]] = extractelement <4 x i32> [[_MSLD]], i64 1
+; CHECK-NEXT:    [[TMP12:%.*]] = and i32 [[TMP11]], 3
+; CHECK-NEXT:    [[TMP13:%.*]] = or i32 [[TMP11]], [[TMP12]]
+; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <4 x i32> [[_MSLD]], i64 2
+; CHECK-NEXT:    [[TMP15:%.*]] = and i32 [[TMP14]], 3
+; CHECK-NEXT:    [[TMP16:%.*]] = or i32 [[TMP14]], [[TMP15]]
+; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <4 x i32> [[_MSLD]], i64 3
+; CHECK-NEXT:    [[TMP18:%.*]] = and i32 [[TMP17]], 3
+; CHECK-NEXT:    [[TMP19:%.*]] = or i32 [[TMP17]], [[TMP18]]
 ; CHECK-NEXT:    [[A0:%.*]] = bitcast <4 x i32> [[TMP2]] to <4 x float>
 ; CHECK-NEXT:    [[RES:%.*]] = call <4 x float> @llvm.x86.avx.vpermilvar.ps(<4 x float> [[A0]], <4 x i32> [[A2]])
 ; CHECK-NEXT:    [[TMP10:%.*]] = bitcast <4 x float> [[RES]] to <4 x i32>
-; CHECK-NEXT:    [[TMP11:%.*]] = bitcast <4 x i32> [[_MSLD]] to i128
-; CHECK-NEXT:    [[_MSCMP1:%.*]] = icmp ne i128 [[TMP11]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP1]], label [[TMP12:%.*]], label [[TMP13:%.*]], !prof [[PROF1]]
-; CHECK:       12:
+; CHECK-NEXT:    [[_MSCMP1:%.*]] = icmp ne i32 [[TMP19]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP1]], label [[TMP23:%.*]], label [[TMP24:%.*]], !prof [[PROF1]]
+; CHECK:       23:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn()
 ; CHECK-NEXT:    unreachable
-; CHECK:       13:
+; CHECK:       24:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <4 x float> @llvm.x86.avx.vpermilvar.ps(<4 x float> [[A3:%.*]], <4 x i32> [[A2]])
 ; CHECK-NEXT:    store <4 x i32> [[TMP10]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <4 x float> [[RES1]]
@@ -1073,16 +1111,39 @@ define <8 x float> @test_x86_avx_vpermilvar_ps_256(<8 x float> %a0, <8 x i32> %a
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <8 x i32>, ptr @__msan_param_tls, align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <8 x i32>, 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:%.*]] = extractelement <8 x i32> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP27:%.*]] = and i32 [[TMP3]], 7
+; CHECK-NEXT:    [[TMP5:%.*]] = or i32 [[TMP3]], [[TMP27]]
+; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <8 x i32> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP7:%.*]] = and i32 [[TMP6]], 7
+; CHECK-NEXT:    [[TMP8:%.*]] = or i32 [[TMP6]], [[TMP7]]
+; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <8 x i32> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP10:%.*]] = and i32 [[TMP9]], 7
+; CHECK-NEXT:    [[TMP11:%.*]] = or i32 [[TMP9]], [[TMP10]]
+; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <8 x i32> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP13:%.*]] = and i32 [[TMP12]], 7
+; CHECK-NEXT:    [[TMP14:%.*]] = or i32 [[TMP12]], [[TMP13]]
+; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <8 x i32> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP16:%.*]] = and i32 [[TMP15]], 7
+; CHECK-NEXT:    [[TMP17:%.*]] = or i32 [[TMP15]], [[TMP16]]
+; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <8 x i32> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP19:%.*]] = and i32 [[TMP18]], 7
+; CHECK-NEXT:    [[TMP20:%.*]] = or i32 [[TMP18]], [[TMP19]]
+; CHECK-NEXT:    [[TMP21:%.*]] = extractelement <8 x i32> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP22:%.*]] = and i32 [[TMP21]], 7
+; CHECK-NEXT:    [[TMP23:%.*]] = or i32 [[TMP21]], [[TMP22]]
+; CHECK-NEXT:    [[TMP24:%.*]] = extractelement <8 x i32> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP25:%.*]] = and i32 [[TMP24]], 7
+; CHECK-NEXT:    [[TMP26:%.*]] = or i32 [[TMP24]], [[TMP25]]
 ; CHECK-NEXT:    [[A0:%.*]] = bitcast <8 x i32> [[TMP1]] to <8 x float>
 ; CHECK-NEXT:    [[RES:%.*]] = call <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<8 x float> [[A0]], <8 x i32> [[A1:%.*]])
 ; CHECK-NEXT:    [[TMP4:%.*]] = bitcast <8 x float> [[RES]] to <8 x i32>
-; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <8 x i32> [[TMP2]] to i256
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i256 [[TMP6]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP7:%.*]], label [[TMP8:%.*]], !prof [[PROF1]]
-; CHECK:       7:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i32 [[TMP26]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP30:%.*]], label [[TMP31:%.*]], !prof [[PROF1]]
+; CHECK:       30:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn()
 ; CHECK-NEXT:    unreachable
-; CHECK:       8:
+; CHECK:       31:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<8 x float> [[A2:%.*]], <8 x i32> [[A1]])
 ; CHECK-NEXT:    store <8 x i32> [[TMP4]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <8 x float> [[RES1]]
diff --git a/llvm/test/Instrumentation/MemorySanitizer/X86/avx512-intrinsics-upgrade.ll b/llvm/test/Instrumentation/MemorySanitizer/X86/avx512-intrinsics-upgrade.ll
index 2c9c6c7309a09..e7854538aa3e8 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/X86/avx512-intrinsics-upgrade.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/X86/avx512-intrinsics-upgrade.ll
@@ -8143,16 +8143,39 @@ define <8 x double>@test_int_x86_avx512_vpermilvar_pd_512(<8 x double> %x0, <8 x
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <8 x i64>, ptr @__msan_param_tls, align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <8 x i64>, 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:%.*]] = extractelement <8 x i64> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP27:%.*]] = and i64 [[TMP3]], 7
+; CHECK-NEXT:    [[TMP28:%.*]] = or i64 [[TMP3]], [[TMP27]]
+; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <8 x i64> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP29:%.*]] = and i64 [[TMP6]], 7
+; CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP6]], [[TMP29]]
+; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <8 x i64> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP10:%.*]] = and i64 [[TMP9]], 7
+; CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP9]], [[TMP10]]
+; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <8 x i64> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP13:%.*]] = and i64 [[TMP12]], 7
+; CHECK-NEXT:    [[TMP14:%.*]] = or i64 [[TMP12]], [[TMP13]]
+; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <8 x i64> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP16:%.*]] = and i64 [[TMP15]], 7
+; CHECK-NEXT:    [[TMP17:%.*]] = or i64 [[TMP15]], [[TMP16]]
+; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <8 x i64> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP19:%.*]] = and i64 [[TMP18]], 7
+; CHECK-NEXT:    [[TMP20:%.*]] = or i64 [[TMP18]], [[TMP19]]
+; CHECK-NEXT:    [[TMP21:%.*]] = extractelement <8 x i64> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP22:%.*]] = and i64 [[TMP21]], 7
+; CHECK-NEXT:    [[TMP23:%.*]] = or i64 [[TMP21]], [[TMP22]]
+; CHECK-NEXT:    [[TMP24:%.*]] = extractelement <8 x i64> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP25:%.*]] = and i64 [[TMP24]], 7
+; CHECK-NEXT:    [[TMP26:%.*]] = or i64 [[TMP24]], [[TMP25]]
 ; CHECK-NEXT:    [[X0:%.*]] = bitcast <8 x i64> [[TMP1]] to <8 x double>
 ; CHECK-NEXT:    [[TMP7:%.*]] = call <8 x double> @llvm.x86.avx512.vpermilvar.pd.512(<8 x double> [[X0]], <8 x i64> [[X1:%.*]])
 ; CHECK-NEXT:    [[TMP4:%.*]] = bitcast <8 x double> [[TMP7]] to <8 x i64>
-; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <8 x i64> [[TMP2]] to i512
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i512 [[TMP6]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP9:%.*]], label [[TMP8:%.*]], !prof [[PROF1]]
-; CHECK:       7:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i64 [[TMP26]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP30:%.*]], label [[TMP31:%.*]], !prof [[PROF1]]
+; CHECK:       30:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn() #[[ATTR8]]
 ; CHECK-NEXT:    unreachable
-; CHECK:       8:
+; CHECK:       31:
 ; CHECK-NEXT:    [[TMP5:%.*]] = call <8 x double> @llvm.x86.avx512.vpermilvar.pd.512(<8 x double> [[X2:%.*]], <8 x i64> [[X1]])
 ; CHECK-NEXT:    store <8 x i64> [[TMP4]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <8 x double> [[TMP5]]
@@ -8169,16 +8192,39 @@ define <8 x double>@test_int_x86_avx512_mask_vpermilvar_pd_512(<8 x double> %x0,
 ; CHECK-NEXT:    [[TMP3:%.*]] = load i8, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 192) to ptr), align 8
 ; CHECK-NEXT:    [[TMP4:%.*]] = load <8 x i64>, 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:%.*]] = extractelement <8 x i64> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP29:%.*]] = and i64 [[TMP5]], 7
+; CHECK-NEXT:    [[TMP30:%.*]] = or i64 [[TMP5]], [[TMP29]]
+; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <8 x i64> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP31:%.*]] = and i64 [[TMP8]], 7
+; CHECK-NEXT:    [[TMP34:%.*]] = or i64 [[TMP8]], [[TMP31]]
+; CHECK-NEXT:    [[TMP35:%.*]] = extractelement <8 x i64> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP36:%.*]] = and i64 [[TMP35]], 7
+; CHECK-NEXT:    [[TMP37:%.*]] = or i64 [[TMP35]], [[TMP36]]
+; CHECK-NEXT:    [[TMP38:%.*]] = extractelement <8 x i64> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP39:%.*]] = and i64 [[TMP38]], 7
+; CHECK-NEXT:    [[TMP40:%.*]] = or i64 [[TMP38]], [[TMP39]]
+; CHECK-NEXT:    [[TMP41:%.*]] = extractelement <8 x i64> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP42:%.*]] = and i64 [[TMP41]], 7
+; CHECK-NEXT:    [[TMP19:%.*]] = or i64 [[TMP41]], [[TMP42]]
+; CHECK-NEXT:    [[TMP20:%.*]] = extractelement <8 x i64> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP21:%.*]] = and i64 [[TMP20]], 7
+; CHECK-NEXT:    [[TMP22:%.*]] = or i64 [[TMP20]], [[TMP21]]
+; CHECK-NEXT:    [[TMP23:%.*]] = extractelement <8 x i64> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP24:%.*]] = and i64 [[TMP23]], 7
+; CHECK-NEXT:    [[TMP25:%.*]] = or i64 [[TMP23]], [[TMP24]]
+; CHECK-NEXT:    [[TMP26:%.*]] = extractelement <8 x i64> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP27:%.*]] = and i64 [[TMP26]], 7
+; CHECK-NEXT:    [[TMP28:%.*]] = or i64 [[TMP26]], [[TMP27]]
 ; CHECK-NEXT:    [[X0:%.*]] = bitcast <8 x i64> [[TMP1]] to <8 x double>
 ; CHECK-NEXT:    [[TMP9:%.*]] = call <8 x double> @llvm.x86.avx512.vpermilvar.pd.512(<8 x double> [[X0]], <8 x i64> [[X1:%.*]])
 ; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <8 x double> [[TMP9]] to <8 x i64>
-; CHECK-NEXT:    [[TMP8:%.*]] = bitcast <8 x i64> [[TMP2]] to i512
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i512 [[TMP8]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP19:%.*]], label [[TMP20:%.*]], !prof [[PROF1]]
-; CHECK:       9:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i64 [[TMP28]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP32:%.*]], label [[TMP33:%.*]], !prof [[PROF1]]
+; CHECK:       32:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn() #[[ATTR8]]
 ; CHECK-NEXT:    unreachable
-; CHECK:       10:
+; CHECK:       33:
 ; CHECK-NEXT:    [[TMP7:%.*]] = call <8 x double> @llvm.x86.avx512.vpermilvar.pd.512(<8 x double> [[X4:%.*]], <8 x i64> [[X1]])
 ; CHECK-NEXT:    [[TMP10:%.*]] = bitcast i8 [[TMP3]] to <8 x i1>
 ; CHECK-NEXT:    [[TMP11:%.*]] = bitcast i8 [[X3:%.*]] to <8 x i1>
@@ -8204,16 +8250,39 @@ define <8 x double>@test_int_x86_avx512_maskz_vpermilvar_pd_512(<8 x double> %x0
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <8 x i64>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 64) to ptr), align 8
 ; CHECK-NEXT:    [[TMP3:%.*]] = load i8, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 192) to ptr), align 8
 ; CHECK-NEXT:    call void @llvm.donothing()
+; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <8 x i64> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP28:%.*]] = and i64 [[TMP4]], 7
+; CHECK-NEXT:    [[TMP29:%.*]] = or i64 [[TMP4]], [[TMP28]]
+; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <8 x i64> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP30:%.*]] = and i64 [[TMP7]], 7
+; CHECK-NEXT:    [[TMP33:%.*]] = or i64 [[TMP7]], [[TMP30]]
+; CHECK-NEXT:    [[TMP34:%.*]] = extractelement <8 x i64> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP35:%.*]] = and i64 [[TMP34]], 7
+; CHECK-NEXT:    [[TMP36:%.*]] = or i64 [[TMP34]], [[TMP35]]
+; CHECK-NEXT:    [[TMP37:%.*]] = extractelement <8 x i64> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP38:%.*]] = and i64 [[TMP37]], 7
+; CHECK-NEXT:    [[TMP39:%.*]] = or i64 [[TMP37]], [[TMP38]]
+; CHECK-NEXT:    [[TMP40:%.*]] = extractelement <8 x i64> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP17:%.*]] = and i64 [[TMP40]], 7
+; CHECK-NEXT:    [[TMP18:%.*]] = or i64 [[TMP40]], [[TMP17]]
+; CHECK-NEXT:    [[TMP19:%.*]] = extractelement <8 x i64> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP20:%.*]] = and i64 [[TMP19]], 7
+; CHECK-NEXT:    [[TMP21:%.*]] = or i64 [[TMP19]], [[TMP20]]
+; CHECK-NEXT:    [[TMP22:%.*]] = extractelement <8 x i64> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP23:%.*]] = and i64 [[TMP22]], 7
+; CHECK-NEXT:    [[TMP24:%.*]] = or i64 [[TMP22]], [[TMP23]]
+; CHECK-NEXT:    [[TMP25:%.*]] = extractelement <8 x i64> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP26:%.*]] = and i64 [[TMP25]], 7
+; CHECK-NEXT:    [[TMP27:%.*]] = or i64 [[TMP25]], [[TMP26]]
 ; CHECK-NEXT:    [[X0:%.*]] = bitcast <8 x i64> [[TMP1]] to <8 x double>
 ; CHECK-NEXT:    [[TMP8:%.*]] = call <8 x double> @llvm.x86.avx512.vpermilvar.pd.512(<8 x double> [[X0]], <8 x i64> [[X1:%.*]])
 ; CHECK-NEXT:    [[TMP5:%.*]] = bitcast <8 x double> [[TMP8]] to <8 x i64>
-; CHECK-NEXT:    [[TMP7:%.*]] = bitcast <8 x i64> [[TMP2]] to i512
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i512 [[TMP7]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP17:%.*]], label [[TMP18:%.*]], !prof [[PROF1]]
-; CHECK:       8:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i64 [[TMP27]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP31:%.*]], label [[TMP32:%.*]], !prof [[PROF1]]
+; CHECK:       31:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn() #[[ATTR8]]
 ; CHECK-NEXT:    unreachable
-; CHECK:       9:
+; CHECK:       32:
 ; CHECK-NEXT:    [[TMP6:%.*]] = call <8 x double> @llvm.x86.avx512.vpermilvar.pd.512(<8 x double> [[X2:%.*]], <8 x i64> [[X1]])
 ; CHECK-NEXT:    [[TMP9:%.*]] = bitcast i8 [[TMP3]] to <8 x i1>
 ; CHECK-NEXT:    [[TMP10:%.*]] = bitcast i8 [[X3:%.*]] to <8 x i1>
@@ -8238,16 +8307,63 @@ define <16 x float>@test_int_x86_avx512_vpermilvar_ps_512(<16 x float> %x0, <16
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <16 x i32>, ptr @__msan_param_tls, align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <16 x i32>, 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:%.*]] = extractelement <16 x i32> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP51:%.*]] = and i32 [[TMP3]], 15
+; CHECK-NEXT:    [[TMP52:%.*]] = or i32 [[TMP3]], [[TMP51]]
+; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <16 x i32> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP53:%.*]] = and i32 [[TMP6]], 15
+; CHECK-NEXT:    [[TMP8:%.*]] = or i32 [[TMP6]], [[TMP53]]
+; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <16 x i32> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP10:%.*]] = and i32 [[TMP9]], 15
+; CHECK-NEXT:    [[TMP11:%.*]] = or i32 [[TMP9]], [[TMP10]]
+; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <16 x i32> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP13:%.*]] = and i32 [[TMP12]], 15
+; CHECK-NEXT:    [[TMP14:%.*]] = or i32 [[TMP12]], [[TMP13]]
+; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <16 x i32> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP16:%.*]] = and i32 [[TMP15]], 15
+; CHECK-NEXT:    [[TMP17:%.*]] = or i32 [[TMP15]], [[TMP16]]
+; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <16 x i32> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP19:%.*]] = and i32 [[TMP18]], 15
+; CHECK-NEXT:    [[TMP20:%.*]] = or i32 [[TMP18]], [[TMP19]]
+; CHECK-NEXT:    [[TMP21:%.*]] = extractelement <16 x i32> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP22:%.*]] = and i32 [[TMP21]], 15
+; CHECK-NEXT:    [[TMP23:%.*]] = or i32 [[TMP21]], [[TMP22]]
+; CHECK-NEXT:    [[TMP24:%.*]] = extractelement <16 x i32> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP25:%.*]] = and i32 [[TMP24]], 15
+; CHECK-NEXT:    [[TMP26:%.*]] = or i32 [[TMP24]], [[TMP25]]
+; CHECK-NEXT:    [[TMP27:%.*]] = extractelement <16 x i32> [[TMP2]], i64 8
+; CHECK-NEXT:    [[TMP28:%.*]] = and i32 [[TMP27]], 15
+; CHECK-NEXT:    [[TMP29:%.*]] = or i32 [[TMP27]], [[TMP28]]
+; CHECK-NEXT:    [[TMP30:%.*]] = extractelement <16 x i32> [[TMP2]], i64 9
+; CHECK-NEXT:    [[TMP31:%.*]] = and i32 [[TMP30]], 15
+; CHECK-NEXT:    [[TMP32:%.*]] = or i32 [[TMP30]], [[TMP31]]
+; CHECK-NEXT:    [[TMP33:%.*]] = extractelement <16 x i32> [[TMP2]], i64 10
+; CHECK-NEXT:    [[TMP34:%.*]] = and i32 [[TMP33]], 15
+; CHECK-NEXT:    [[TMP35:%.*]] = or i32 [[TMP33]], [[TMP34]]
+; CHECK-NEXT:    [[TMP36:%.*]] = extractelement <16 x i32> [[TMP2]], i64 11
+; CHECK-NEXT:    [[TMP37:%.*]] = and i32 [[TMP36]], 15
+; CHECK-NEXT:    [[TMP38:%.*]] = or i32 [[TMP36]], [[TMP37]]
+; CHECK-NEXT:    [[TMP39:%.*]] = extractelement <16 x i32> [[TMP2]], i64 12
+; CHECK-NEXT:    [[TMP40:%.*]] = and i32 [[TMP39]], 15
+; CHECK-NEXT:    [[TMP41:%.*]] = or i32 [[TMP39]], [[TMP40]]
+; CHECK-NEXT:    [[TMP42:%.*]] = extractelement <16 x i32> [[TMP2]], i64 13
+; CHECK-NEXT:    [[TMP43:%.*]] = and i32 [[TMP42]], 15
+; CHECK-NEXT:    [[TMP44:%.*]] = or i32 [[TMP42]], [[TMP43]]
+; CHECK-NEXT:    [[TMP45:%.*]] = extractelement <16 x i32> [[TMP2]], i64 14
+; CHECK-NEXT:    [[TMP46:%.*]] = and i32 [[TMP45]], 15
+; CHECK-NEXT:    [[TMP47:%.*]] = or i32 [[TMP45]], [[TMP46]]
+; CHECK-NEXT:    [[TMP48:%.*]] = extractelement <16 x i32> [[TMP2]], i64 15
+; CHECK-NEXT:    [[TMP49:%.*]] = and i32 [[TMP48]], 15
+; CHECK-NEXT:    [[TMP50:%.*]] = or i32 [[TMP48]], [[TMP49]]
 ; CHECK-NEXT:    [[X0:%.*]] = bitcast <16 x i32> [[TMP1]] to <16 x float>
 ; CHECK-NEXT:    [[TMP7:%.*]] = call <16 x float> @llvm.x86.avx512.vpermilvar.ps.512(<16 x float> [[X0]], <16 x i32> [[X1:%.*]])
 ; CHECK-NEXT:    [[TMP4:%.*]] = bitcast <16 x float> [[TMP7]] to <16 x i32>
-; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <16 x i32> [[TMP2]] to i512
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i512 [[TMP6]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP9:%.*]], label [[TMP8:%.*]], !prof [[PROF1]]
-; CHECK:       7:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i32 [[TMP50]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP54:%.*]], label [[TMP55:%.*]], !prof [[PROF1]]
+; CHECK:       54:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn() #[[ATTR8]]
 ; CHECK-NEXT:    unreachable
-; CHECK:       8:
+; CHECK:       55:
 ; CHECK-NEXT:    [[TMP5:%.*]] = call <16 x float> @llvm.x86.avx512.vpermilvar.ps.512(<16 x float> [[X2:%.*]], <16 x i32> [[X1]])
 ; CHECK-NEXT:    store <16 x i32> [[TMP4]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <16 x float> [[TMP5]]
@@ -8264,16 +8380,63 @@ define <16 x float>@test_int_x86_avx512_mask_vpermilvar_ps_512(<16 x float> %x0,
 ; 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:%.*]] = extractelement <16 x i32> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP53:%.*]] = and i32 [[TMP5]], 15
+; CHECK-NEXT:    [[TMP54:%.*]] = or i32 [[TMP5]], [[TMP53]]
+; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <16 x i32> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP55:%.*]] = and i32 [[TMP8]], 15
+; CHECK-NEXT:    [[TMP58:%.*]] = or i32 [[TMP8]], [[TMP55]]
+; CHECK-NEXT:    [[TMP59:%.*]] = extractelement <16 x i32> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP60:%.*]] = and i32 [[TMP59]], 15
+; CHECK-NEXT:    [[TMP61:%.*]] = or i32 [[TMP59]], [[TMP60]]
+; CHECK-NEXT:    [[TMP62:%.*]] = extractelement <16 x i32> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP63:%.*]] = and i32 [[TMP62]], 15
+; CHECK-NEXT:    [[TMP64:%.*]] = or i32 [[TMP62]], [[TMP63]]
+; CHECK-NEXT:    [[TMP65:%.*]] = extractelement <16 x i32> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP66:%.*]] = and i32 [[TMP65]], 15
+; CHECK-NEXT:    [[TMP19:%.*]] = or i32 [[TMP65]], [[TMP66]]
+; CHECK-NEXT:    [[TMP20:%.*]] = extractelement <16 x i32> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP21:%.*]] = and i32 [[TMP20]], 15
+; CHECK-NEXT:    [[TMP22:%.*]] = or i32 [[TMP20]], [[TMP21]]
+; CHECK-NEXT:    [[TMP23:%.*]] = extractelement <16 x i32> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP24:%.*]] = and i32 [[TMP23]], 15
+; CHECK-NEXT:    [[TMP25:%.*]] = or i32 [[TMP23]], [[TMP24]]
+; CHECK-NEXT:    [[TMP26:%.*]] = extractelement <16 x i32> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP27:%.*]] = and i32 [[TMP26]], 15
+; CHECK-NEXT:    [[TMP28:%.*]] = or i32 [[TMP26]], [[TMP27]]
+; CHECK-NEXT:    [[TMP29:%.*]] = extractelement <16 x i32> [[TMP2]], i64 8
+; CHECK-NEXT:    [[TMP30:%.*]] = and i32 [[TMP29]], 15
+; CHECK-NEXT:    [[TMP31:%.*]] = or i32 [[TMP29]], [[TMP30]]
+; CHECK-NEXT:    [[TMP32:%.*]] = extractelement <16 x i32> [[TMP2]], i64 9
+; CHECK-NEXT:    [[TMP33:%.*]] = and i32 [[TMP32]], 15
+; CHECK-NEXT:    [[TMP34:%.*]] = or i32 [[TMP32]], [[TMP33]]
+; CHECK-NEXT:    [[TMP35:%.*]] = extractelement <16 x i32> [[TMP2]], i64 10
+; CHECK-NEXT:    [[TMP36:%.*]] = and i32 [[TMP35]], 15
+; CHECK-NEXT:    [[TMP37:%.*]] = or i32 [[TMP35]], [[TMP36]]
+; CHECK-NEXT:    [[TMP38:%.*]] = extractelement <16 x i32> [[TMP2]], i64 11
+; CHECK-NEXT:    [[TMP39:%.*]] = and i32 [[TMP38]], 15
+; CHECK-NEXT:    [[TMP40:%.*]] = or i32 [[TMP38]], [[TMP39]]
+; CHECK-NEXT:    [[TMP41:%.*]] = extractelement <16 x i32> [[TMP2]], i64 12
+; CHECK-NEXT:    [[TMP42:%.*]] = and i32 [[TMP41]], 15
+; CHECK-NEXT:    [[TMP43:%.*]] = or i32 [[TMP41]], [[TMP42]]
+; CHECK-NEXT:    [[TMP44:%.*]] = extractelement <16 x i32> [[TMP2]], i64 13
+; CHECK-NEXT:    [[TMP45:%.*]] = and i32 [[TMP44]], 15
+; CHECK-NEXT:    [[TMP46:%.*]] = or i32 [[TMP44]], [[TMP45]]
+; CHECK-NEXT:    [[TMP47:%.*]] = extractelement <16 x i32> [[TMP2]], i64 14
+; CHECK-NEXT:    [[TMP48:%.*]] = and i32 [[TMP47]], 15
+; CHECK-NEXT:    [[TMP49:%.*]] = or i32 [[TMP47]], [[TMP48]]
+; CHECK-NEXT:    [[TMP50:%.*]] = extractelement <16 x i32> [[TMP2]], i64 15
+; CHECK-NEXT:    [[TMP51:%.*]] = and i32 [[TMP50]], 15
+; CHECK-NEXT:    [[TMP52:%.*]] = or i32 [[TMP50]], [[TMP51]]
 ; CHECK-NEXT:    [[X0:%.*]] = bitcast <16 x i32> [[TMP1]] to <16 x float>
 ; CHECK-NEXT:    [[TMP9:%.*]] = call <16 x float> @llvm.x86.avx512.vpermilvar.ps.512(<16 x float> [[X0]], <16 x i32> [[X1:%.*]])
 ; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <16 x float> [[TMP9]] to <16 x i32>
-; CHECK-NEXT:    [[TMP8:%.*]] = bitcast <16 x i32> [[TMP2]] to i512
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i512 [[TMP8]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP19:%.*]], label [[TMP20:%.*]], !prof [[PROF1]]
-; CHECK:       9:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i32 [[TMP52]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP56:%.*]], label [[TMP57:%.*]], !prof [[PROF1]]
+; CHECK:       56:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn() #[[ATTR8]]
 ; CHECK-NEXT:    unreachable
-; CHECK:       10:
+; CHECK:       57:
 ; CHECK-NEXT:    [[TMP7:%.*]] = call <16 x float> @llvm.x86.avx512.vpermilvar.ps.512(<16 x float> [[X4:%.*]], <16 x i32> [[X1]])
 ; CHECK-NEXT:    [[TMP10:%.*]] = bitcast i16 [[TMP3]] to <16 x i1>
 ; CHECK-NEXT:    [[TMP11:%.*]] = bitcast i16 [[X3:%.*]] to <16 x i1>
@@ -8300,16 +8463,63 @@ define <16 x float>@test_int_x86_avx512_maskz_vpermilvar_ps_512(<16 x float> %x0
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <16 x i32>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 64) to ptr), align 8
 ; CHECK-NEXT:    [[TMP3:%.*]] = load 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:    [[TMP4:%.*]] = extractelement <16 x i32> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP52:%.*]] = and i32 [[TMP4]], 15
+; CHECK-NEXT:    [[TMP53:%.*]] = or i32 [[TMP4]], [[TMP52]]
+; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <16 x i32> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP54:%.*]] = and i32 [[TMP7]], 15
+; CHECK-NEXT:    [[TMP57:%.*]] = or i32 [[TMP7]], [[TMP54]]
+; CHECK-NEXT:    [[TMP58:%.*]] = extractelement <16 x i32> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP59:%.*]] = and i32 [[TMP58]], 15
+; CHECK-NEXT:    [[TMP60:%.*]] = or i32 [[TMP58]], [[TMP59]]
+; CHECK-NEXT:    [[TMP61:%.*]] = extractelement <16 x i32> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP62:%.*]] = and i32 [[TMP61]], 15
+; CHECK-NEXT:    [[TMP63:%.*]] = or i32 [[TMP61]], [[TMP62]]
+; CHECK-NEXT:    [[TMP64:%.*]] = extractelement <16 x i32> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP17:%.*]] = and i32 [[TMP64]], 15
+; CHECK-NEXT:    [[TMP18:%.*]] = or i32 [[TMP64]], [[TMP17]]
+; CHECK-NEXT:    [[TMP19:%.*]] = extractelement <16 x i32> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP20:%.*]] = and i32 [[TMP19]], 15
+; CHECK-NEXT:    [[TMP21:%.*]] = or i32 [[TMP19]], [[TMP20]]
+; CHECK-NEXT:    [[TMP22:%.*]] = extractelement <16 x i32> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP23:%.*]] = and i32 [[TMP22]], 15
+; CHECK-NEXT:    [[TMP24:%.*]] = or i32 [[TMP22]], [[TMP23]]
+; CHECK-NEXT:    [[TMP25:%.*]] = extractelement <16 x i32> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP26:%.*]] = and i32 [[TMP25]], 15
+; CHECK-NEXT:    [[TMP27:%.*]] = or i32 [[TMP25]], [[TMP26]]
+; CHECK-NEXT:    [[TMP28:%.*]] = extractelement <16 x i32> [[TMP2]], i64 8
+; CHECK-NEXT:    [[TMP29:%.*]] = and i32 [[TMP28]], 15
+; CHECK-NEXT:    [[TMP30:%.*]] = or i32 [[TMP28]], [[TMP29]]
+; CHECK-NEXT:    [[TMP31:%.*]] = extractelement <16 x i32> [[TMP2]], i64 9
+; CHECK-NEXT:    [[TMP32:%.*]] = and i32 [[TMP31]], 15
+; CHECK-NEXT:    [[TMP33:%.*]] = or i32 [[TMP31]], [[TMP32]]
+; CHECK-NEXT:    [[TMP34:%.*]] = extractelement <16 x i32> [[TMP2]], i64 10
+; CHECK-NEXT:    [[TMP35:%.*]] = and i32 [[TMP34]], 15
+; CHECK-NEXT:    [[TMP36:%.*]] = or i32 [[TMP34]], [[TMP35]]
+; CHECK-NEXT:    [[TMP37:%.*]] = extractelement <16 x i32> [[TMP2]], i64 11
+; CHECK-NEXT:    [[TMP38:%.*]] = and i32 [[TMP37]], 15
+; CHECK-NEXT:    [[TMP39:%.*]] = or i32 [[TMP37]], [[TMP38]]
+; CHECK-NEXT:    [[TMP40:%.*]] = extractelement <16 x i32> [[TMP2]], i64 12
+; CHECK-NEXT:    [[TMP41:%.*]] = and i32 [[TMP40]], 15
+; CHECK-NEXT:    [[TMP42:%.*]] = or i32 [[TMP40]], [[TMP41]]
+; CHECK-NEXT:    [[TMP43:%.*]] = extractelement <16 x i32> [[TMP2]], i64 13
+; CHECK-NEXT:    [[TMP44:%.*]] = and i32 [[TMP43]], 15
+; CHECK-NEXT:    [[TMP45:%.*]] = or i32 [[TMP43]], [[TMP44]]
+; CHECK-NEXT:    [[TMP46:%.*]] = extractelement <16 x i32> [[TMP2]], i64 14
+; CHECK-NEXT:    [[TMP47:%.*]] = and i32 [[TMP46]], 15
+; CHECK-NEXT:    [[TMP48:%.*]] = or i32 [[TMP46]], [[TMP47]]
+; CHECK-NEXT:    [[TMP49:%.*]] = extractelement <16 x i32> [[TMP2]], i64 15
+; CHECK-NEXT:    [[TMP50:%.*]] = and i32 [[TMP49]], 15
+; CHECK-NEXT:    [[TMP51:%.*]] = or i32 [[TMP49]], [[TMP50]]
 ; CHECK-NEXT:    [[X0:%.*]] = bitcast <16 x i32> [[TMP1]] to <16 x float>
 ; CHECK-NEXT:    [[TMP8:%.*]] = call <16 x float> @llvm.x86.avx512.vpermilvar.ps.512(<16 x float> [[X0]], <16 x i32> [[X1:%.*]])
 ; CHECK-NEXT:    [[TMP5:%.*]] = bitcast <16 x float> [[TMP8]] to <16 x i32>
-; CHECK-NEXT:    [[TMP7:%.*]] = bitcast <16 x i32> [[TMP2]] to i512
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i512 [[TMP7]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP17:%.*]], label [[TMP18:%.*]], !prof [[PROF1]]
-; CHECK:       8:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i32 [[TMP51]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP55:%.*]], label [[TMP56:%.*]], !prof [[PROF1]]
+; CHECK:       55:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn() #[[ATTR8]]
 ; CHECK-NEXT:    unreachable
-; CHECK:       9:
+; CHECK:       56:
 ; CHECK-NEXT:    [[TMP6:%.*]] = call <16 x float> @llvm.x86.avx512.vpermilvar.ps.512(<16 x float> [[X2:%.*]], <16 x i32> [[X1]])
 ; CHECK-NEXT:    [[TMP9:%.*]] = bitcast i16 [[TMP3]] to <16 x i1>
 ; CHECK-NEXT:    [[TMP10:%.*]] = bitcast i16 [[X3:%.*]] to <16 x i1>
diff --git a/llvm/test/Instrumentation/MemorySanitizer/X86/avx512-intrinsics.ll b/llvm/test/Instrumentation/MemorySanitizer/X86/avx512-intrinsics.ll
index 43595dccc35bc..3f5d3ab88bbea 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/X86/avx512-intrinsics.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/X86/avx512-intrinsics.ll
@@ -9207,16 +9207,39 @@ define <8 x double>@test_int_x86_avx512_vpermilvar_pd_512(<8 x double> %x0, <8 x
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <8 x i64>, ptr @__msan_param_tls, align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <8 x i64>, 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:%.*]] = extractelement <8 x i64> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP27:%.*]] = and i64 [[TMP3]], 7
+; CHECK-NEXT:    [[TMP5:%.*]] = or i64 [[TMP3]], [[TMP27]]
+; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <8 x i64> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP7:%.*]] = and i64 [[TMP6]], 7
+; CHECK-NEXT:    [[TMP8:%.*]] = or i64 [[TMP6]], [[TMP7]]
+; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <8 x i64> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP10:%.*]] = and i64 [[TMP9]], 7
+; CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP9]], [[TMP10]]
+; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <8 x i64> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP13:%.*]] = and i64 [[TMP12]], 7
+; CHECK-NEXT:    [[TMP14:%.*]] = or i64 [[TMP12]], [[TMP13]]
+; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <8 x i64> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP16:%.*]] = and i64 [[TMP15]], 7
+; CHECK-NEXT:    [[TMP17:%.*]] = or i64 [[TMP15]], [[TMP16]]
+; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <8 x i64> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP19:%.*]] = and i64 [[TMP18]], 7
+; CHECK-NEXT:    [[TMP20:%.*]] = or i64 [[TMP18]], [[TMP19]]
+; CHECK-NEXT:    [[TMP21:%.*]] = extractelement <8 x i64> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP22:%.*]] = and i64 [[TMP21]], 7
+; CHECK-NEXT:    [[TMP23:%.*]] = or i64 [[TMP21]], [[TMP22]]
+; CHECK-NEXT:    [[TMP24:%.*]] = extractelement <8 x i64> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP25:%.*]] = and i64 [[TMP24]], 7
+; CHECK-NEXT:    [[TMP26:%.*]] = or i64 [[TMP24]], [[TMP25]]
 ; CHECK-NEXT:    [[X0:%.*]] = bitcast <8 x i64> [[TMP1]] to <8 x double>
 ; CHECK-NEXT:    [[RES:%.*]] = call <8 x double> @llvm.x86.avx512.vpermilvar.pd.512(<8 x double> [[X0]], <8 x i64> [[X1:%.*]])
 ; CHECK-NEXT:    [[TMP4:%.*]] = bitcast <8 x double> [[RES]] to <8 x i64>
-; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <8 x i64> [[TMP2]] to i512
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i512 [[TMP6]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP7:%.*]], label [[TMP8:%.*]], !prof [[PROF1]]
-; CHECK:       7:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i64 [[TMP26]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP30:%.*]], label [[TMP31:%.*]], !prof [[PROF1]]
+; CHECK:       30:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn() #[[ATTR10]]
 ; CHECK-NEXT:    unreachable
-; CHECK:       8:
+; CHECK:       31:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <8 x double> @llvm.x86.avx512.vpermilvar.pd.512(<8 x double> [[X2:%.*]], <8 x i64> [[X1]])
 ; CHECK-NEXT:    store <8 x i64> [[TMP4]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <8 x double> [[RES1]]
@@ -9232,16 +9255,39 @@ define <8 x double>@test_int_x86_avx512_vpermilvar_pd_512_mask(<8 x double> %x0,
 ; CHECK-NEXT:    [[TMP3:%.*]] = load i8, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 192) to ptr), align 8
 ; CHECK-NEXT:    [[TMP4:%.*]] = load <8 x i64>, 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:%.*]] = extractelement <8 x i64> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP29:%.*]] = and i64 [[TMP5]], 7
+; CHECK-NEXT:    [[TMP7:%.*]] = or i64 [[TMP5]], [[TMP29]]
+; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <8 x i64> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP30:%.*]] = and i64 [[TMP8]], 7
+; CHECK-NEXT:    [[TMP31:%.*]] = or i64 [[TMP8]], [[TMP30]]
+; CHECK-NEXT:    [[TMP34:%.*]] = extractelement <8 x i64> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP35:%.*]] = and i64 [[TMP34]], 7
+; CHECK-NEXT:    [[TMP36:%.*]] = or i64 [[TMP34]], [[TMP35]]
+; CHECK-NEXT:    [[TMP37:%.*]] = extractelement <8 x i64> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP38:%.*]] = and i64 [[TMP37]], 7
+; CHECK-NEXT:    [[TMP16:%.*]] = or i64 [[TMP37]], [[TMP38]]
+; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <8 x i64> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP18:%.*]] = and i64 [[TMP17]], 7
+; CHECK-NEXT:    [[TMP19:%.*]] = or i64 [[TMP17]], [[TMP18]]
+; CHECK-NEXT:    [[TMP20:%.*]] = extractelement <8 x i64> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP21:%.*]] = and i64 [[TMP20]], 7
+; CHECK-NEXT:    [[TMP22:%.*]] = or i64 [[TMP20]], [[TMP21]]
+; CHECK-NEXT:    [[TMP23:%.*]] = extractelement <8 x i64> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP24:%.*]] = and i64 [[TMP23]], 7
+; CHECK-NEXT:    [[TMP25:%.*]] = or i64 [[TMP23]], [[TMP24]]
+; CHECK-NEXT:    [[TMP26:%.*]] = extractelement <8 x i64> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP27:%.*]] = and i64 [[TMP26]], 7
+; CHECK-NEXT:    [[TMP28:%.*]] = or i64 [[TMP26]], [[TMP27]]
 ; CHECK-NEXT:    [[X0:%.*]] = bitcast <8 x i64> [[TMP1]] to <8 x double>
 ; CHECK-NEXT:    [[RES:%.*]] = call <8 x double> @llvm.x86.avx512.vpermilvar.pd.512(<8 x double> [[X0]], <8 x i64> [[X1:%.*]])
 ; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <8 x double> [[RES]] to <8 x i64>
-; CHECK-NEXT:    [[TMP8:%.*]] = bitcast <8 x i64> [[TMP2]] to i512
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i512 [[TMP8]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP16:%.*]], label [[TMP17:%.*]], !prof [[PROF1]]
-; CHECK:       9:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i64 [[TMP28]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP32:%.*]], label [[TMP33:%.*]], !prof [[PROF1]]
+; CHECK:       32:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn() #[[ATTR10]]
 ; CHECK-NEXT:    unreachable
-; CHECK:       10:
+; CHECK:       33:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <8 x double> @llvm.x86.avx512.vpermilvar.pd.512(<8 x double> [[X3:%.*]], <8 x i64> [[X1]])
 ; CHECK-NEXT:    [[TMP9:%.*]] = bitcast i8 [[TMP3]] to <8 x i1>
 ; CHECK-NEXT:    [[MASK_CAST:%.*]] = bitcast i8 [[MASK:%.*]] to <8 x i1>
@@ -9268,16 +9314,39 @@ define <8 x double>@test_int_x86_avx512_vpermilvar_pd_512_maskz(<8 x double> %x0
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <8 x i64>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 64) to ptr), align 8
 ; CHECK-NEXT:    [[TMP3:%.*]] = load i8, 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:    [[TMP4:%.*]] = extractelement <8 x i64> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP28:%.*]] = and i64 [[TMP4]], 7
+; CHECK-NEXT:    [[TMP6:%.*]] = or i64 [[TMP4]], [[TMP28]]
+; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <8 x i64> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP29:%.*]] = and i64 [[TMP7]], 7
+; CHECK-NEXT:    [[TMP30:%.*]] = or i64 [[TMP7]], [[TMP29]]
+; CHECK-NEXT:    [[TMP33:%.*]] = extractelement <8 x i64> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP34:%.*]] = and i64 [[TMP33]], 7
+; CHECK-NEXT:    [[TMP35:%.*]] = or i64 [[TMP33]], [[TMP34]]
+; CHECK-NEXT:    [[TMP36:%.*]] = extractelement <8 x i64> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP14:%.*]] = and i64 [[TMP36]], 7
+; CHECK-NEXT:    [[TMP15:%.*]] = or i64 [[TMP36]], [[TMP14]]
+; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <8 x i64> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP17:%.*]] = and i64 [[TMP16]], 7
+; CHECK-NEXT:    [[TMP18:%.*]] = or i64 [[TMP16]], [[TMP17]]
+; CHECK-NEXT:    [[TMP19:%.*]] = extractelement <8 x i64> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP20:%.*]] = and i64 [[TMP19]], 7
+; CHECK-NEXT:    [[TMP21:%.*]] = or i64 [[TMP19]], [[TMP20]]
+; CHECK-NEXT:    [[TMP22:%.*]] = extractelement <8 x i64> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP23:%.*]] = and i64 [[TMP22]], 7
+; CHECK-NEXT:    [[TMP24:%.*]] = or i64 [[TMP22]], [[TMP23]]
+; CHECK-NEXT:    [[TMP25:%.*]] = extractelement <8 x i64> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP26:%.*]] = and i64 [[TMP25]], 7
+; CHECK-NEXT:    [[TMP27:%.*]] = or i64 [[TMP25]], [[TMP26]]
 ; CHECK-NEXT:    [[X0:%.*]] = bitcast <8 x i64> [[TMP1]] to <8 x double>
 ; CHECK-NEXT:    [[RES:%.*]] = call <8 x double> @llvm.x86.avx512.vpermilvar.pd.512(<8 x double> [[X0]], <8 x i64> [[X1:%.*]])
 ; CHECK-NEXT:    [[TMP5:%.*]] = bitcast <8 x double> [[RES]] to <8 x i64>
-; CHECK-NEXT:    [[TMP7:%.*]] = bitcast <8 x i64> [[TMP2]] to i512
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i512 [[TMP7]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP14:%.*]], label [[TMP15:%.*]], !prof [[PROF1]]
-; CHECK:       8:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i64 [[TMP27]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP31:%.*]], label [[TMP32:%.*]], !prof [[PROF1]]
+; CHECK:       31:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn() #[[ATTR10]]
 ; CHECK-NEXT:    unreachable
-; CHECK:       9:
+; CHECK:       32:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <8 x double> @llvm.x86.avx512.vpermilvar.pd.512(<8 x double> [[X2:%.*]], <8 x i64> [[X1]])
 ; CHECK-NEXT:    [[TMP8:%.*]] = bitcast i8 [[TMP3]] to <8 x i1>
 ; CHECK-NEXT:    [[MASK_CAST:%.*]] = bitcast i8 [[MASK:%.*]] to <8 x i1>
@@ -9304,16 +9373,63 @@ define <16 x float>@test_int_x86_avx512_vpermilvar_ps_512(<16 x float> %x0, <16
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <16 x i32>, ptr @__msan_param_tls, align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <16 x i32>, 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:%.*]] = extractelement <16 x i32> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP51:%.*]] = and i32 [[TMP3]], 15
+; CHECK-NEXT:    [[TMP5:%.*]] = or i32 [[TMP3]], [[TMP51]]
+; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <16 x i32> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP7:%.*]] = and i32 [[TMP6]], 15
+; CHECK-NEXT:    [[TMP8:%.*]] = or i32 [[TMP6]], [[TMP7]]
+; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <16 x i32> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP10:%.*]] = and i32 [[TMP9]], 15
+; CHECK-NEXT:    [[TMP11:%.*]] = or i32 [[TMP9]], [[TMP10]]
+; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <16 x i32> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP13:%.*]] = and i32 [[TMP12]], 15
+; CHECK-NEXT:    [[TMP14:%.*]] = or i32 [[TMP12]], [[TMP13]]
+; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <16 x i32> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP16:%.*]] = and i32 [[TMP15]], 15
+; CHECK-NEXT:    [[TMP17:%.*]] = or i32 [[TMP15]], [[TMP16]]
+; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <16 x i32> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP19:%.*]] = and i32 [[TMP18]], 15
+; CHECK-NEXT:    [[TMP20:%.*]] = or i32 [[TMP18]], [[TMP19]]
+; CHECK-NEXT:    [[TMP21:%.*]] = extractelement <16 x i32> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP22:%.*]] = and i32 [[TMP21]], 15
+; CHECK-NEXT:    [[TMP23:%.*]] = or i32 [[TMP21]], [[TMP22]]
+; CHECK-NEXT:    [[TMP24:%.*]] = extractelement <16 x i32> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP25:%.*]] = and i32 [[TMP24]], 15
+; CHECK-NEXT:    [[TMP26:%.*]] = or i32 [[TMP24]], [[TMP25]]
+; CHECK-NEXT:    [[TMP27:%.*]] = extractelement <16 x i32> [[TMP2]], i64 8
+; CHECK-NEXT:    [[TMP28:%.*]] = and i32 [[TMP27]], 15
+; CHECK-NEXT:    [[TMP29:%.*]] = or i32 [[TMP27]], [[TMP28]]
+; CHECK-NEXT:    [[TMP30:%.*]] = extractelement <16 x i32> [[TMP2]], i64 9
+; CHECK-NEXT:    [[TMP31:%.*]] = and i32 [[TMP30]], 15
+; CHECK-NEXT:    [[TMP32:%.*]] = or i32 [[TMP30]], [[TMP31]]
+; CHECK-NEXT:    [[TMP33:%.*]] = extractelement <16 x i32> [[TMP2]], i64 10
+; CHECK-NEXT:    [[TMP34:%.*]] = and i32 [[TMP33]], 15
+; CHECK-NEXT:    [[TMP35:%.*]] = or i32 [[TMP33]], [[TMP34]]
+; CHECK-NEXT:    [[TMP36:%.*]] = extractelement <16 x i32> [[TMP2]], i64 11
+; CHECK-NEXT:    [[TMP37:%.*]] = and i32 [[TMP36]], 15
+; CHECK-NEXT:    [[TMP38:%.*]] = or i32 [[TMP36]], [[TMP37]]
+; CHECK-NEXT:    [[TMP39:%.*]] = extractelement <16 x i32> [[TMP2]], i64 12
+; CHECK-NEXT:    [[TMP40:%.*]] = and i32 [[TMP39]], 15
+; CHECK-NEXT:    [[TMP41:%.*]] = or i32 [[TMP39]], [[TMP40]]
+; CHECK-NEXT:    [[TMP42:%.*]] = extractelement <16 x i32> [[TMP2]], i64 13
+; CHECK-NEXT:    [[TMP43:%.*]] = and i32 [[TMP42]], 15
+; CHECK-NEXT:    [[TMP44:%.*]] = or i32 [[TMP42]], [[TMP43]]
+; CHECK-NEXT:    [[TMP45:%.*]] = extractelement <16 x i32> [[TMP2]], i64 14
+; CHECK-NEXT:    [[TMP46:%.*]] = and i32 [[TMP45]], 15
+; CHECK-NEXT:    [[TMP47:%.*]] = or i32 [[TMP45]], [[TMP46]]
+; CHECK-NEXT:    [[TMP48:%.*]] = extractelement <16 x i32> [[TMP2]], i64 15
+; CHECK-NEXT:    [[TMP49:%.*]] = and i32 [[TMP48]], 15
+; CHECK-NEXT:    [[TMP50:%.*]] = or i32 [[TMP48]], [[TMP49]]
 ; CHECK-NEXT:    [[X0:%.*]] = bitcast <16 x i32> [[TMP1]] to <16 x float>
 ; CHECK-NEXT:    [[RES:%.*]] = call <16 x float> @llvm.x86.avx512.vpermilvar.ps.512(<16 x float> [[X0]], <16 x i32> [[X1:%.*]])
 ; CHECK-NEXT:    [[TMP4:%.*]] = bitcast <16 x float> [[RES]] to <16 x i32>
-; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <16 x i32> [[TMP2]] to i512
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i512 [[TMP6]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP7:%.*]], label [[TMP8:%.*]], !prof [[PROF1]]
-; CHECK:       7:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i32 [[TMP50]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP54:%.*]], label [[TMP55:%.*]], !prof [[PROF1]]
+; CHECK:       54:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn() #[[ATTR10]]
 ; CHECK-NEXT:    unreachable
-; CHECK:       8:
+; CHECK:       55:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <16 x float> @llvm.x86.avx512.vpermilvar.ps.512(<16 x float> [[X2:%.*]], <16 x i32> [[X1]])
 ; CHECK-NEXT:    store <16 x i32> [[TMP4]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <16 x float> [[RES1]]
@@ -9329,16 +9445,63 @@ define <16 x float>@test_int_x86_avx512_vpermilvar_ps_512_mask(<16 x float> %x0,
 ; 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:%.*]] = extractelement <16 x i32> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP53:%.*]] = and i32 [[TMP5]], 15
+; CHECK-NEXT:    [[TMP7:%.*]] = or i32 [[TMP5]], [[TMP53]]
+; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <16 x i32> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP54:%.*]] = and i32 [[TMP8]], 15
+; CHECK-NEXT:    [[TMP55:%.*]] = or i32 [[TMP8]], [[TMP54]]
+; CHECK-NEXT:    [[TMP58:%.*]] = extractelement <16 x i32> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP59:%.*]] = and i32 [[TMP58]], 15
+; CHECK-NEXT:    [[TMP60:%.*]] = or i32 [[TMP58]], [[TMP59]]
+; CHECK-NEXT:    [[TMP61:%.*]] = extractelement <16 x i32> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP62:%.*]] = and i32 [[TMP61]], 15
+; CHECK-NEXT:    [[TMP16:%.*]] = or i32 [[TMP61]], [[TMP62]]
+; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <16 x i32> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP18:%.*]] = and i32 [[TMP17]], 15
+; CHECK-NEXT:    [[TMP19:%.*]] = or i32 [[TMP17]], [[TMP18]]
+; CHECK-NEXT:    [[TMP20:%.*]] = extractelement <16 x i32> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP21:%.*]] = and i32 [[TMP20]], 15
+; CHECK-NEXT:    [[TMP22:%.*]] = or i32 [[TMP20]], [[TMP21]]
+; CHECK-NEXT:    [[TMP23:%.*]] = extractelement <16 x i32> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP24:%.*]] = and i32 [[TMP23]], 15
+; CHECK-NEXT:    [[TMP25:%.*]] = or i32 [[TMP23]], [[TMP24]]
+; CHECK-NEXT:    [[TMP26:%.*]] = extractelement <16 x i32> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP27:%.*]] = and i32 [[TMP26]], 15
+; CHECK-NEXT:    [[TMP28:%.*]] = or i32 [[TMP26]], [[TMP27]]
+; CHECK-NEXT:    [[TMP29:%.*]] = extractelement <16 x i32> [[TMP2]], i64 8
+; CHECK-NEXT:    [[TMP30:%.*]] = and i32 [[TMP29]], 15
+; CHECK-NEXT:    [[TMP31:%.*]] = or i32 [[TMP29]], [[TMP30]]
+; CHECK-NEXT:    [[TMP32:%.*]] = extractelement <16 x i32> [[TMP2]], i64 9
+; CHECK-NEXT:    [[TMP33:%.*]] = and i32 [[TMP32]], 15
+; CHECK-NEXT:    [[TMP34:%.*]] = or i32 [[TMP32]], [[TMP33]]
+; CHECK-NEXT:    [[TMP35:%.*]] = extractelement <16 x i32> [[TMP2]], i64 10
+; CHECK-NEXT:    [[TMP36:%.*]] = and i32 [[TMP35]], 15
+; CHECK-NEXT:    [[TMP37:%.*]] = or i32 [[TMP35]], [[TMP36]]
+; CHECK-NEXT:    [[TMP38:%.*]] = extractelement <16 x i32> [[TMP2]], i64 11
+; CHECK-NEXT:    [[TMP39:%.*]] = and i32 [[TMP38]], 15
+; CHECK-NEXT:    [[TMP40:%.*]] = or i32 [[TMP38]], [[TMP39]]
+; CHECK-NEXT:    [[TMP41:%.*]] = extractelement <16 x i32> [[TMP2]], i64 12
+; CHECK-NEXT:    [[TMP42:%.*]] = and i32 [[TMP41]], 15
+; CHECK-NEXT:    [[TMP43:%.*]] = or i32 [[TMP41]], [[TMP42]]
+; CHECK-NEXT:    [[TMP44:%.*]] = extractelement <16 x i32> [[TMP2]], i64 13
+; CHECK-NEXT:    [[TMP45:%.*]] = and i32 [[TMP44]], 15
+; CHECK-NEXT:    [[TMP46:%.*]] = or i32 [[TMP44]], [[TMP45]]
+; CHECK-NEXT:    [[TMP47:%.*]] = extractelement <16 x i32> [[TMP2]], i64 14
+; CHECK-NEXT:    [[TMP48:%.*]] = and i32 [[TMP47]], 15
+; CHECK-NEXT:    [[TMP49:%.*]] = or i32 [[TMP47]], [[TMP48]]
+; CHECK-NEXT:    [[TMP50:%.*]] = extractelement <16 x i32> [[TMP2]], i64 15
+; CHECK-NEXT:    [[TMP51:%.*]] = and i32 [[TMP50]], 15
+; CHECK-NEXT:    [[TMP52:%.*]] = or i32 [[TMP50]], [[TMP51]]
 ; CHECK-NEXT:    [[X0:%.*]] = bitcast <16 x i32> [[TMP1]] to <16 x float>
 ; CHECK-NEXT:    [[RES:%.*]] = call <16 x float> @llvm.x86.avx512.vpermilvar.ps.512(<16 x float> [[X0]], <16 x i32> [[X1:%.*]])
 ; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <16 x float> [[RES]] to <16 x i32>
-; CHECK-NEXT:    [[TMP8:%.*]] = bitcast <16 x i32> [[TMP2]] to i512
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i512 [[TMP8]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP16:%.*]], label [[TMP17:%.*]], !prof [[PROF1]]
-; CHECK:       9:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i32 [[TMP52]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP56:%.*]], label [[TMP57:%.*]], !prof [[PROF1]]
+; CHECK:       56:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn() #[[ATTR10]]
 ; CHECK-NEXT:    unreachable
-; CHECK:       10:
+; CHECK:       57:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <16 x float> @llvm.x86.avx512.vpermilvar.ps.512(<16 x float> [[X3:%.*]], <16 x i32> [[X1]])
 ; CHECK-NEXT:    [[TMP9:%.*]] = bitcast i16 [[TMP3]] to <16 x i1>
 ; CHECK-NEXT:    [[MASK_CAST:%.*]] = bitcast i16 [[MASK:%.*]] to <16 x i1>
@@ -9365,16 +9528,63 @@ define <16 x float>@test_int_x86_avx512_vpermilvar_ps_512_maskz(<16 x float> %x0
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <16 x i32>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 64) to ptr), align 8
 ; CHECK-NEXT:    [[TMP3:%.*]] = load 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:    [[TMP4:%.*]] = extractelement <16 x i32> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP52:%.*]] = and i32 [[TMP4]], 15
+; CHECK-NEXT:    [[TMP6:%.*]] = or i32 [[TMP4]], [[TMP52]]
+; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <16 x i32> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP53:%.*]] = and i32 [[TMP7]], 15
+; CHECK-NEXT:    [[TMP54:%.*]] = or i32 [[TMP7]], [[TMP53]]
+; CHECK-NEXT:    [[TMP57:%.*]] = extractelement <16 x i32> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP58:%.*]] = and i32 [[TMP57]], 15
+; CHECK-NEXT:    [[TMP59:%.*]] = or i32 [[TMP57]], [[TMP58]]
+; CHECK-NEXT:    [[TMP60:%.*]] = extractelement <16 x i32> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP14:%.*]] = and i32 [[TMP60]], 15
+; CHECK-NEXT:    [[TMP15:%.*]] = or i32 [[TMP60]], [[TMP14]]
+; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <16 x i32> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP17:%.*]] = and i32 [[TMP16]], 15
+; CHECK-NEXT:    [[TMP18:%.*]] = or i32 [[TMP16]], [[TMP17]]
+; CHECK-NEXT:    [[TMP19:%.*]] = extractelement <16 x i32> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP20:%.*]] = and i32 [[TMP19]], 15
+; CHECK-NEXT:    [[TMP21:%.*]] = or i32 [[TMP19]], [[TMP20]]
+; CHECK-NEXT:    [[TMP22:%.*]] = extractelement <16 x i32> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP23:%.*]] = and i32 [[TMP22]], 15
+; CHECK-NEXT:    [[TMP24:%.*]] = or i32 [[TMP22]], [[TMP23]]
+; CHECK-NEXT:    [[TMP25:%.*]] = extractelement <16 x i32> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP26:%.*]] = and i32 [[TMP25]], 15
+; CHECK-NEXT:    [[TMP27:%.*]] = or i32 [[TMP25]], [[TMP26]]
+; CHECK-NEXT:    [[TMP28:%.*]] = extractelement <16 x i32> [[TMP2]], i64 8
+; CHECK-NEXT:    [[TMP29:%.*]] = and i32 [[TMP28]], 15
+; CHECK-NEXT:    [[TMP30:%.*]] = or i32 [[TMP28]], [[TMP29]]
+; CHECK-NEXT:    [[TMP31:%.*]] = extractelement <16 x i32> [[TMP2]], i64 9
+; CHECK-NEXT:    [[TMP32:%.*]] = and i32 [[TMP31]], 15
+; CHECK-NEXT:    [[TMP33:%.*]] = or i32 [[TMP31]], [[TMP32]]
+; CHECK-NEXT:    [[TMP34:%.*]] = extractelement <16 x i32> [[TMP2]], i64 10
+; CHECK-NEXT:    [[TMP35:%.*]] = and i32 [[TMP34]], 15
+; CHECK-NEXT:    [[TMP36:%.*]] = or i32 [[TMP34]], [[TMP35]]
+; CHECK-NEXT:    [[TMP37:%.*]] = extractelement <16 x i32> [[TMP2]], i64 11
+; CHECK-NEXT:    [[TMP38:%.*]] = and i32 [[TMP37]], 15
+; CHECK-NEXT:    [[TMP39:%.*]] = or i32 [[TMP37]], [[TMP38]]
+; CHECK-NEXT:    [[TMP40:%.*]] = extractelement <16 x i32> [[TMP2]], i64 12
+; CHECK-NEXT:    [[TMP41:%.*]] = and i32 [[TMP40]], 15
+; CHECK-NEXT:    [[TMP42:%.*]] = or i32 [[TMP40]], [[TMP41]]
+; CHECK-NEXT:    [[TMP43:%.*]] = extractelement <16 x i32> [[TMP2]], i64 13
+; CHECK-NEXT:    [[TMP44:%.*]] = and i32 [[TMP43]], 15
+; CHECK-NEXT:    [[TMP45:%.*]] = or i32 [[TMP43]], [[TMP44]]
+; CHECK-NEXT:    [[TMP46:%.*]] = extractelement <16 x i32> [[TMP2]], i64 14
+; CHECK-NEXT:    [[TMP47:%.*]] = and i32 [[TMP46]], 15
+; CHECK-NEXT:    [[TMP48:%.*]] = or i32 [[TMP46]], [[TMP47]]
+; CHECK-NEXT:    [[TMP49:%.*]] = extractelement <16 x i32> [[TMP2]], i64 15
+; CHECK-NEXT:    [[TMP50:%.*]] = and i32 [[TMP49]], 15
+; CHECK-NEXT:    [[TMP51:%.*]] = or i32 [[TMP49]], [[TMP50]]
 ; CHECK-NEXT:    [[X0:%.*]] = bitcast <16 x i32> [[TMP1]] to <16 x float>
 ; CHECK-NEXT:    [[RES:%.*]] = call <16 x float> @llvm.x86.avx512.vpermilvar.ps.512(<16 x float> [[X0]], <16 x i32> [[X1:%.*]])
 ; CHECK-NEXT:    [[TMP5:%.*]] = bitcast <16 x float> [[RES]] to <16 x i32>
-; CHECK-NEXT:    [[TMP7:%.*]] = bitcast <16 x i32> [[TMP2]] to i512
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i512 [[TMP7]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP14:%.*]], label [[TMP15:%.*]], !prof [[PROF1]]
-; CHECK:       8:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i32 [[TMP51]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP55:%.*]], label [[TMP56:%.*]], !prof [[PROF1]]
+; CHECK:       55:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn() #[[ATTR10]]
 ; CHECK-NEXT:    unreachable
-; CHECK:       9:
+; CHECK:       56:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <16 x float> @llvm.x86.avx512.vpermilvar.ps.512(<16 x float> [[X2:%.*]], <16 x i32> [[X1]])
 ; CHECK-NEXT:    [[TMP8:%.*]] = bitcast i16 [[TMP3]] to <16 x i1>
 ; CHECK-NEXT:    [[MASK_CAST:%.*]] = bitcast i16 [[MASK:%.*]] to <16 x i1>
diff --git a/llvm/test/Instrumentation/MemorySanitizer/i386/avx-intrinsics-i386.ll b/llvm/test/Instrumentation/MemorySanitizer/i386/avx-intrinsics-i386.ll
index 613146ff8227e..294f2cae7f7ab 100644
--- a/llvm/test/Instrumentation/MemorySanitizer/i386/avx-intrinsics-i386.ll
+++ b/llvm/test/Instrumentation/MemorySanitizer/i386/avx-intrinsics-i386.ll
@@ -990,16 +990,21 @@ define <2 x double> @test_x86_avx_vpermilvar_pd(<2 x double> %a0, <2 x i64> %a1)
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <2 x i64>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 16) to ptr), align 8
 ; CHECK-NEXT:    [[TMP5:%.*]] = load i64, ptr @__msan_va_arg_overflow_size_tls, align 8
 ; CHECK-NEXT:    call void @llvm.donothing()
+; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP10:%.*]] = and i64 [[TMP4]], 1
+; CHECK-NEXT:    [[TMP11:%.*]] = or i64 [[TMP4]], [[TMP10]]
+; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <2 x i64> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP8:%.*]] = and i64 [[TMP7]], 1
+; CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP7]], [[TMP8]]
 ; CHECK-NEXT:    [[A0:%.*]] = bitcast <2 x i64> [[TMP1]] to <2 x double>
 ; CHECK-NEXT:    [[RES:%.*]] = call <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double> [[A0]], <2 x i64> [[A1:%.*]])
 ; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <2 x double> [[RES]] to <2 x i64>
-; CHECK-NEXT:    [[TMP7:%.*]] = bitcast <2 x i64> [[TMP2]] to i128
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i128 [[TMP7]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP8:%.*]], label [[TMP9:%.*]], !prof [[PROF1]]
-; CHECK:       8:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i64 [[TMP9]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP13:%.*]], label [[TMP14:%.*]], !prof [[PROF1]]
+; CHECK:       13:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn()
 ; CHECK-NEXT:    unreachable
-; CHECK:       9:
+; CHECK:       14:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double> [[A2:%.*]], <2 x i64> [[A1]])
 ; CHECK-NEXT:    store <2 x i64> [[TMP6]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <2 x double> [[RES1]]
@@ -1016,16 +1021,27 @@ define <4 x double> @test_x86_avx_vpermilvar_pd_256(<4 x double> %a0, <4 x i64>
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <4 x i64>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 32) to ptr), align 8
 ; CHECK-NEXT:    [[TMP5:%.*]] = load i64, ptr @__msan_va_arg_overflow_size_tls, align 8
 ; CHECK-NEXT:    call void @llvm.donothing()
+; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <4 x i64> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP16:%.*]] = and i64 [[TMP4]], 3
+; CHECK-NEXT:    [[TMP17:%.*]] = or i64 [[TMP4]], [[TMP16]]
+; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <4 x i64> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP8:%.*]] = and i64 [[TMP7]], 3
+; CHECK-NEXT:    [[TMP9:%.*]] = or i64 [[TMP7]], [[TMP8]]
+; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP11:%.*]] = and i64 [[TMP10]], 3
+; CHECK-NEXT:    [[TMP12:%.*]] = or i64 [[TMP10]], [[TMP11]]
+; CHECK-NEXT:    [[TMP13:%.*]] = extractelement <4 x i64> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP14:%.*]] = and i64 [[TMP13]], 3
+; CHECK-NEXT:    [[TMP15:%.*]] = or i64 [[TMP13]], [[TMP14]]
 ; CHECK-NEXT:    [[A0:%.*]] = bitcast <4 x i64> [[TMP1]] to <4 x double>
 ; CHECK-NEXT:    [[RES:%.*]] = call <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double> [[A0]], <4 x i64> [[A1:%.*]])
 ; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <4 x double> [[RES]] to <4 x i64>
-; CHECK-NEXT:    [[TMP7:%.*]] = bitcast <4 x i64> [[TMP2]] to i256
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i256 [[TMP7]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP8:%.*]], label [[TMP9:%.*]], !prof [[PROF1]]
-; CHECK:       8:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i64 [[TMP15]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP19:%.*]], label [[TMP20:%.*]], !prof [[PROF1]]
+; CHECK:       19:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn()
 ; CHECK-NEXT:    unreachable
-; CHECK:       9:
+; CHECK:       20:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double> [[A2:%.*]], <4 x i64> [[A1]])
 ; CHECK-NEXT:    store <4 x i64> [[TMP6]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <4 x double> [[RES1]]
@@ -1057,16 +1073,27 @@ define <4 x float> @test_x86_avx_vpermilvar_ps(<4 x float> %a0, <4 x i32> %a1) #
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <4 x i32>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 16) to ptr), align 8
 ; CHECK-NEXT:    [[TMP5:%.*]] = load i64, ptr @__msan_va_arg_overflow_size_tls, align 8
 ; CHECK-NEXT:    call void @llvm.donothing()
+; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <4 x i32> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP16:%.*]] = and i32 [[TMP4]], 3
+; CHECK-NEXT:    [[TMP17:%.*]] = or i32 [[TMP4]], [[TMP16]]
+; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <4 x i32> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP8:%.*]] = and i32 [[TMP7]], 3
+; CHECK-NEXT:    [[TMP9:%.*]] = or i32 [[TMP7]], [[TMP8]]
+; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <4 x i32> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP11:%.*]] = and i32 [[TMP10]], 3
+; CHECK-NEXT:    [[TMP12:%.*]] = or i32 [[TMP10]], [[TMP11]]
+; CHECK-NEXT:    [[TMP13:%.*]] = extractelement <4 x i32> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP14:%.*]] = and i32 [[TMP13]], 3
+; CHECK-NEXT:    [[TMP15:%.*]] = or i32 [[TMP13]], [[TMP14]]
 ; CHECK-NEXT:    [[A0:%.*]] = bitcast <4 x i32> [[TMP1]] to <4 x float>
 ; CHECK-NEXT:    [[RES:%.*]] = call <4 x float> @llvm.x86.avx.vpermilvar.ps(<4 x float> [[A0]], <4 x i32> [[A1:%.*]])
 ; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <4 x float> [[RES]] to <4 x i32>
-; CHECK-NEXT:    [[TMP7:%.*]] = bitcast <4 x i32> [[TMP2]] to i128
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i128 [[TMP7]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP8:%.*]], label [[TMP9:%.*]], !prof [[PROF1]]
-; CHECK:       8:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i32 [[TMP15]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP19:%.*]], label [[TMP20:%.*]], !prof [[PROF1]]
+; CHECK:       19:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn()
 ; CHECK-NEXT:    unreachable
-; CHECK:       9:
+; CHECK:       20:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <4 x float> @llvm.x86.avx.vpermilvar.ps(<4 x float> [[A2:%.*]], <4 x i32> [[A1]])
 ; CHECK-NEXT:    store <4 x i32> [[TMP6]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <4 x float> [[RES1]]
@@ -1091,16 +1118,27 @@ define <4 x float> @test_x86_avx_vpermilvar_ps_load(<4 x float> %a0, ptr %a1) #0
 ; CHECK-NEXT:    [[TMP6:%.*]] = and i64 [[TMP5]], -2147483649
 ; CHECK-NEXT:    [[TMP7:%.*]] = inttoptr i64 [[TMP6]] to ptr
 ; CHECK-NEXT:    [[_MSLD:%.*]] = load <4 x i32>, ptr [[TMP7]], align 16
+; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <4 x i32> [[_MSLD]], i64 0
+; CHECK-NEXT:    [[TMP10:%.*]] = and i32 [[TMP9]], 3
+; CHECK-NEXT:    [[TMP21:%.*]] = or i32 [[TMP9]], [[TMP10]]
+; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <4 x i32> [[_MSLD]], i64 1
+; CHECK-NEXT:    [[TMP13:%.*]] = and i32 [[TMP12]], 3
+; CHECK-NEXT:    [[TMP14:%.*]] = or i32 [[TMP12]], [[TMP13]]
+; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <4 x i32> [[_MSLD]], i64 2
+; CHECK-NEXT:    [[TMP16:%.*]] = and i32 [[TMP15]], 3
+; CHECK-NEXT:    [[TMP17:%.*]] = or i32 [[TMP15]], [[TMP16]]
+; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <4 x i32> [[_MSLD]], i64 3
+; CHECK-NEXT:    [[TMP19:%.*]] = and i32 [[TMP18]], 3
+; CHECK-NEXT:    [[TMP20:%.*]] = or i32 [[TMP18]], [[TMP19]]
 ; CHECK-NEXT:    [[A0:%.*]] = bitcast <4 x i32> [[TMP2]] to <4 x float>
 ; CHECK-NEXT:    [[RES:%.*]] = call <4 x float> @llvm.x86.avx.vpermilvar.ps(<4 x float> [[A0]], <4 x i32> [[A2]])
 ; CHECK-NEXT:    [[TMP11:%.*]] = bitcast <4 x float> [[RES]] to <4 x i32>
-; CHECK-NEXT:    [[TMP12:%.*]] = bitcast <4 x i32> [[_MSLD]] to i128
-; CHECK-NEXT:    [[_MSCMP1:%.*]] = icmp ne i128 [[TMP12]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP1]], label [[TMP13:%.*]], label [[TMP14:%.*]], !prof [[PROF1]]
-; CHECK:       13:
+; CHECK-NEXT:    [[_MSCMP1:%.*]] = icmp ne i32 [[TMP20]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP1]], label [[TMP24:%.*]], label [[TMP25:%.*]], !prof [[PROF1]]
+; CHECK:       24:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn()
 ; CHECK-NEXT:    unreachable
-; CHECK:       14:
+; CHECK:       25:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <4 x float> @llvm.x86.avx.vpermilvar.ps(<4 x float> [[A3:%.*]], <4 x i32> [[A2]])
 ; CHECK-NEXT:    store <4 x i32> [[TMP11]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <4 x float> [[RES1]]
@@ -1118,16 +1156,39 @@ define <8 x float> @test_x86_avx_vpermilvar_ps_256(<8 x float> %a0, <8 x i32> %a
 ; CHECK-NEXT:    [[TMP2:%.*]] = load <8 x i32>, ptr inttoptr (i64 add (i64 ptrtoint (ptr @__msan_param_tls to i64), i64 32) to ptr), align 8
 ; CHECK-NEXT:    [[TMP5:%.*]] = load i64, ptr @__msan_va_arg_overflow_size_tls, align 8
 ; CHECK-NEXT:    call void @llvm.donothing()
+; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <8 x i32> [[TMP2]], i64 0
+; CHECK-NEXT:    [[TMP28:%.*]] = and i32 [[TMP4]], 7
+; CHECK-NEXT:    [[TMP29:%.*]] = or i32 [[TMP4]], [[TMP28]]
+; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <8 x i32> [[TMP2]], i64 1
+; CHECK-NEXT:    [[TMP8:%.*]] = and i32 [[TMP7]], 7
+; CHECK-NEXT:    [[TMP9:%.*]] = or i32 [[TMP7]], [[TMP8]]
+; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <8 x i32> [[TMP2]], i64 2
+; CHECK-NEXT:    [[TMP11:%.*]] = and i32 [[TMP10]], 7
+; CHECK-NEXT:    [[TMP12:%.*]] = or i32 [[TMP10]], [[TMP11]]
+; CHECK-NEXT:    [[TMP13:%.*]] = extractelement <8 x i32> [[TMP2]], i64 3
+; CHECK-NEXT:    [[TMP14:%.*]] = and i32 [[TMP13]], 7
+; CHECK-NEXT:    [[TMP15:%.*]] = or i32 [[TMP13]], [[TMP14]]
+; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <8 x i32> [[TMP2]], i64 4
+; CHECK-NEXT:    [[TMP17:%.*]] = and i32 [[TMP16]], 7
+; CHECK-NEXT:    [[TMP18:%.*]] = or i32 [[TMP16]], [[TMP17]]
+; CHECK-NEXT:    [[TMP19:%.*]] = extractelement <8 x i32> [[TMP2]], i64 5
+; CHECK-NEXT:    [[TMP20:%.*]] = and i32 [[TMP19]], 7
+; CHECK-NEXT:    [[TMP21:%.*]] = or i32 [[TMP19]], [[TMP20]]
+; CHECK-NEXT:    [[TMP22:%.*]] = extractelement <8 x i32> [[TMP2]], i64 6
+; CHECK-NEXT:    [[TMP23:%.*]] = and i32 [[TMP22]], 7
+; CHECK-NEXT:    [[TMP24:%.*]] = or i32 [[TMP22]], [[TMP23]]
+; CHECK-NEXT:    [[TMP25:%.*]] = extractelement <8 x i32> [[TMP2]], i64 7
+; CHECK-NEXT:    [[TMP26:%.*]] = and i32 [[TMP25]], 7
+; CHECK-NEXT:    [[TMP27:%.*]] = or i32 [[TMP25]], [[TMP26]]
 ; CHECK-NEXT:    [[A0:%.*]] = bitcast <8 x i32> [[TMP1]] to <8 x float>
 ; CHECK-NEXT:    [[RES:%.*]] = call <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<8 x float> [[A0]], <8 x i32> [[A1:%.*]])
 ; CHECK-NEXT:    [[TMP6:%.*]] = bitcast <8 x float> [[RES]] to <8 x i32>
-; CHECK-NEXT:    [[TMP7:%.*]] = bitcast <8 x i32> [[TMP2]] to i256
-; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i256 [[TMP7]], 0
-; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP8:%.*]], label [[TMP9:%.*]], !prof [[PROF1]]
-; CHECK:       8:
+; CHECK-NEXT:    [[_MSCMP:%.*]] = icmp ne i32 [[TMP27]], 0
+; CHECK-NEXT:    br i1 [[_MSCMP]], label [[TMP31:%.*]], label [[TMP32:%.*]], !prof [[PROF1]]
+; CHECK:       31:
 ; CHECK-NEXT:    call void @__msan_warning_noreturn()
 ; CHECK-NEXT:    unreachable
-; CHECK:       9:
+; CHECK:       32:
 ; CHECK-NEXT:    [[RES1:%.*]] = call <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<8 x float> [[A2:%.*]], <8 x i32> [[A1]])
 ; CHECK-NEXT:    store <8 x i32> [[TMP6]], ptr @__msan_retval_tls, align 8
 ; CHECK-NEXT:    ret <8 x float> [[RES1]]