[NFCI][msan] Precommit tests for AVX-VNNI (#153135)

The tests largely cover AVX-VNNI (Vector Neural Network Instructions):
- vpdpbusd, vpdpbusds
- vpdpwssd, vpdpwssds

AVX-VNNI-INT8:
- vpdpbssd, vpdpbssds
- vpdpbsud, vpdpbsuds
- vpdpbuud, vpdpbuuds

AVX-VNNI-INT16:
- vpdpwsud, vpdpwsuds
- vpdpwusd, vpdpwusds
- vpdpwuud, vpdpwuuds

These instructions are currently heuristically handled (by OR'ing
together the vectors). This is incorrect because:
1) multiplication by a zero should result in an initialized value 2) the
addition is horizontal (within vectors, not "vertically" between
vectors).

Future work can improve the instrumentation by applying the updated
handleVectorPmaddIntrinsic() from
#152941

Author: thurstond

llvm/test/Instrumentation/MemorySanitizer/X86/avx10_2_512ni-intrinsics.ll

@@ -0,0 +1,171 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt < %s -mattr=+avx512vnni,+avx512vl,+avxvnni -passes=msan -S | FileCheck %s
+
+; Forked from llvm/test/CodeGen/X86/avx_vnni-intrinsics.ll
+;
+; Handled strictly: (none)
+;
+; Handled heuristically:
+; - llvm.x86.avx512.vpdpbusd.128
+; - llvm.x86.avx512.vpdpbusd.256
+; - llvm.x86.avx512.vpdpbusds.128
+; - llvm.x86.avx512.vpdpbusds.256
+; - llvm.x86.avx512.vpdpwssd.128
+; - llvm.x86.avx512.vpdpwssd.256
+; - llvm.x86.avx512.vpdpwssds.128
+; - llvm.x86.avx512.vpdpwssds.256
+
+target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+declare <8 x i32> @llvm.x86.avx512.vpdpbusd.256(<8 x i32>, <8 x i32>, <8 x i32>)
+
+define <8 x i32>@test_int_x86_avx_vpdpbusd_256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2) sanitize_memory {
+; CHECK-LABEL: define <8 x i32> @test_int_x86_avx_vpdpbusd_256(
+; CHECK-SAME: <8 x i32> [[X0:%.*]], <8 x i32> [[X1:%.*]], <8 x i32> [[X2:%.*]]) #[[ATTR1:[0-9]+]] {
+; 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:    [[TMP3:%.*]] = load <8 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:    [[_MSPROP:%.*]] = or <8 x i32> [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    [[_MSPROP1:%.*]] = or <8 x i32> [[_MSPROP]], [[TMP3]]
+; CHECK-NEXT:    [[RES:%.*]] = call <8 x i32> @llvm.x86.avx512.vpdpbusd.256(<8 x i32> [[X0]], <8 x i32> [[X1]], <8 x i32> [[X2]])
+; CHECK-NEXT:    store <8 x i32> [[_MSPROP1]], ptr @__msan_retval_tls, align 8
+; CHECK-NEXT:    ret <8 x i32> [[RES]]
+;
+  %res = call <8 x i32> @llvm.x86.avx512.vpdpbusd.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2)
+  ret <8 x i32> %res
+}
+
+declare <4 x i32> @llvm.x86.avx512.vpdpbusd.128(<4 x i32>, <4 x i32>, <4 x i32>)
+
+define <4 x i32>@test_int_x86_avx_vpdpbusd_128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2) sanitize_memory {
+; CHECK-LABEL: define <4 x i32> @test_int_x86_avx_vpdpbusd_128(
+; CHECK-SAME: <4 x i32> [[X0:%.*]], <4 x i32> [[X1:%.*]], <4 x i32> [[X2:%.*]]) #[[ATTR1]] {
+; 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:    [[TMP3:%.*]] = load <4 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:    [[_MSPROP:%.*]] = or <4 x i32> [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    [[_MSPROP1:%.*]] = or <4 x i32> [[_MSPROP]], [[TMP3]]
+; CHECK-NEXT:    [[RES:%.*]] = call <4 x i32> @llvm.x86.avx512.vpdpbusd.128(<4 x i32> [[X0]], <4 x i32> [[X1]], <4 x i32> [[X2]])
+; CHECK-NEXT:    store <4 x i32> [[_MSPROP1]], ptr @__msan_retval_tls, align 8
+; CHECK-NEXT:    ret <4 x i32> [[RES]]
+;
+  %res = call <4 x i32> @llvm.x86.avx512.vpdpbusd.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2)
+  ret <4 x i32> %res
+}
+
+declare <8 x i32> @llvm.x86.avx512.vpdpbusds.256(<8 x i32>, <8 x i32>, <8 x i32>)
+
+define <8 x i32>@test_int_x86_avx_vpdpbusds_256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2) sanitize_memory {
+; CHECK-LABEL: define <8 x i32> @test_int_x86_avx_vpdpbusds_256(
+; CHECK-SAME: <8 x i32> [[X0:%.*]], <8 x i32> [[X1:%.*]], <8 x i32> [[X2:%.*]]) #[[ATTR1]] {
+; 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:    [[TMP3:%.*]] = load <8 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:    [[_MSPROP:%.*]] = or <8 x i32> [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    [[_MSPROP1:%.*]] = or <8 x i32> [[_MSPROP]], [[TMP3]]
+; CHECK-NEXT:    [[RES:%.*]] = call <8 x i32> @llvm.x86.avx512.vpdpbusds.256(<8 x i32> [[X0]], <8 x i32> [[X1]], <8 x i32> [[X2]])
+; CHECK-NEXT:    store <8 x i32> [[_MSPROP1]], ptr @__msan_retval_tls, align 8
+; CHECK-NEXT:    ret <8 x i32> [[RES]]
+;
+  %res = call <8 x i32> @llvm.x86.avx512.vpdpbusds.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2)
+  ret <8 x i32> %res
+}
+
+declare <4 x i32> @llvm.x86.avx512.vpdpbusds.128(<4 x i32>, <4 x i32>, <4 x i32>)
+
+define <4 x i32>@test_int_x86_avx_vpdpbusds_128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2) sanitize_memory {
+; CHECK-LABEL: define <4 x i32> @test_int_x86_avx_vpdpbusds_128(
+; CHECK-SAME: <4 x i32> [[X0:%.*]], <4 x i32> [[X1:%.*]], <4 x i32> [[X2:%.*]]) #[[ATTR1]] {
+; 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:    [[TMP3:%.*]] = load <4 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:    [[_MSPROP:%.*]] = or <4 x i32> [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    [[_MSPROP1:%.*]] = or <4 x i32> [[_MSPROP]], [[TMP3]]
+; CHECK-NEXT:    [[RES:%.*]] = call <4 x i32> @llvm.x86.avx512.vpdpbusds.128(<4 x i32> [[X0]], <4 x i32> [[X1]], <4 x i32> [[X2]])
+; CHECK-NEXT:    store <4 x i32> [[_MSPROP1]], ptr @__msan_retval_tls, align 8
+; CHECK-NEXT:    ret <4 x i32> [[RES]]
+;
+  %res = call <4 x i32> @llvm.x86.avx512.vpdpbusds.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2)
+  ret <4 x i32> %res
+}
+
+declare <8 x i32> @llvm.x86.avx512.vpdpwssd.256(<8 x i32>, <8 x i32>, <8 x i32>)
+
+define <8 x i32>@test_int_x86_avx_vpdpwssd_256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2) sanitize_memory {
+; CHECK-LABEL: define <8 x i32> @test_int_x86_avx_vpdpwssd_256(
+; CHECK-SAME: <8 x i32> [[X0:%.*]], <8 x i32> [[X1:%.*]], <8 x i32> [[X2:%.*]]) #[[ATTR1]] {
+; 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:    [[TMP3:%.*]] = load <8 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:    [[_MSPROP:%.*]] = or <8 x i32> [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    [[_MSPROP1:%.*]] = or <8 x i32> [[_MSPROP]], [[TMP3]]
+; CHECK-NEXT:    [[RES:%.*]] = call <8 x i32> @llvm.x86.avx512.vpdpwssd.256(<8 x i32> [[X0]], <8 x i32> [[X1]], <8 x i32> [[X2]])
+; CHECK-NEXT:    store <8 x i32> [[_MSPROP1]], ptr @__msan_retval_tls, align 8
+; CHECK-NEXT:    ret <8 x i32> [[RES]]
+;
+  %res = call <8 x i32> @llvm.x86.avx512.vpdpwssd.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2)
+  ret <8 x i32> %res
+}
+
+declare <4 x i32> @llvm.x86.avx512.vpdpwssd.128(<4 x i32>, <4 x i32>, <4 x i32>)
+
+define <4 x i32>@test_int_x86_avx_vpdpwssd_128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2) sanitize_memory {
+; CHECK-LABEL: define <4 x i32> @test_int_x86_avx_vpdpwssd_128(
+; CHECK-SAME: <4 x i32> [[X0:%.*]], <4 x i32> [[X1:%.*]], <4 x i32> [[X2:%.*]]) #[[ATTR1]] {
+; 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:    [[TMP3:%.*]] = load <4 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:    [[_MSPROP:%.*]] = or <4 x i32> [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    [[_MSPROP1:%.*]] = or <4 x i32> [[_MSPROP]], [[TMP3]]
+; CHECK-NEXT:    [[RES:%.*]] = call <4 x i32> @llvm.x86.avx512.vpdpwssd.128(<4 x i32> [[X0]], <4 x i32> [[X1]], <4 x i32> [[X2]])
+; CHECK-NEXT:    store <4 x i32> [[_MSPROP1]], ptr @__msan_retval_tls, align 8
+; CHECK-NEXT:    ret <4 x i32> [[RES]]
+;
+  %res = call <4 x i32> @llvm.x86.avx512.vpdpwssd.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2)
+  ret <4 x i32> %res
+}
+
+declare <8 x i32> @llvm.x86.avx512.vpdpwssds.256(<8 x i32>, <8 x i32>, <8 x i32>)
+
+define <8 x i32>@test_int_x86_avx_vpdpwssds_256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2) sanitize_memory {
+; CHECK-LABEL: define <8 x i32> @test_int_x86_avx_vpdpwssds_256(
+; CHECK-SAME: <8 x i32> [[X0:%.*]], <8 x i32> [[X1:%.*]], <8 x i32> [[X2:%.*]]) #[[ATTR1]] {
+; 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:    [[TMP3:%.*]] = load <8 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:    [[_MSPROP:%.*]] = or <8 x i32> [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    [[_MSPROP1:%.*]] = or <8 x i32> [[_MSPROP]], [[TMP3]]
+; CHECK-NEXT:    [[RES:%.*]] = call <8 x i32> @llvm.x86.avx512.vpdpwssds.256(<8 x i32> [[X0]], <8 x i32> [[X1]], <8 x i32> [[X2]])
+; CHECK-NEXT:    store <8 x i32> [[_MSPROP1]], ptr @__msan_retval_tls, align 8
+; CHECK-NEXT:    ret <8 x i32> [[RES]]
+;
+  %res = call <8 x i32> @llvm.x86.avx512.vpdpwssds.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2)
+  ret <8 x i32> %res
+}
+
+declare <4 x i32> @llvm.x86.avx512.vpdpwssds.128(<4 x i32>, <4 x i32>, <4 x i32>)
+
+define <4 x i32>@test_int_x86_avx_vpdpwssds_128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2) sanitize_memory {
+; CHECK-LABEL: define <4 x i32> @test_int_x86_avx_vpdpwssds_128(
+; CHECK-SAME: <4 x i32> [[X0:%.*]], <4 x i32> [[X1:%.*]], <4 x i32> [[X2:%.*]]) #[[ATTR1]] {
+; 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:    [[TMP3:%.*]] = load <4 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:    [[_MSPROP:%.*]] = or <4 x i32> [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    [[_MSPROP1:%.*]] = or <4 x i32> [[_MSPROP]], [[TMP3]]
+; CHECK-NEXT:    [[RES:%.*]] = call <4 x i32> @llvm.x86.avx512.vpdpwssds.128(<4 x i32> [[X0]], <4 x i32> [[X1]], <4 x i32> [[X2]])
+; CHECK-NEXT:    store <4 x i32> [[_MSPROP1]], ptr @__msan_retval_tls, align 8
+; CHECK-NEXT:    ret <4 x i32> [[RES]]
+;
+  %res = call <4 x i32> @llvm.x86.avx512.vpdpwssds.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2)
+  ret <4 x i32> %res
+}