Skip to content

Commit a428622

Browse files
committed
Address review comments: mask shift amount with 0xFF and fix test placement
Signed-off-by: Medha Tiwari <[email protected]>
1 parent 471c262 commit a428622

File tree

5 files changed

+32
-31
lines changed

5 files changed

+32
-31
lines changed

clang/lib/AST/ByteCode/InterpBuiltin.cpp

Lines changed: 6 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -4261,9 +4261,10 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const CallExpr *Call,
42614261
case clang::X86::BI__builtin_ia32_kshiftlidi:
42624262
return interp__builtin_elementwise_int_binop(
42634263
S, OpPC, Call, [](const APSInt &LHS, const APSInt &RHS) {
4264-
if (RHS.uge(LHS.getBitWidth()))
4264+
unsigned Amt = RHS.getZExtValue() & 0xFF;
4265+
if (Amt >= LHS.getBitWidth())
42654266
return APInt::getZero(LHS.getBitWidth());
4266-
return LHS.shl(RHS.getZExtValue());
4267+
return LHS.shl(Amt);
42674268
});
42684269

42694270
case clang::X86::BI__builtin_ia32_kshiftriqi:
@@ -4272,9 +4273,10 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const CallExpr *Call,
42724273
case clang::X86::BI__builtin_ia32_kshiftridi:
42734274
return interp__builtin_elementwise_int_binop(
42744275
S, OpPC, Call, [](const APSInt &LHS, const APSInt &RHS) {
4275-
if (RHS.uge(LHS.getBitWidth()))
4276+
unsigned Amt = RHS.getZExtValue() & 0xFF;
4277+
if (Amt >= LHS.getBitWidth())
42764278
return APInt::getZero(LHS.getBitWidth());
4277-
return LHS.lshr(RHS.getZExtValue());
4279+
return LHS.lshr(Amt);
42784280
});
42794281

42804282
case clang::X86::BI__builtin_ia32_lzcnt_u16:

clang/lib/AST/ExprConstant.cpp

Lines changed: 6 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -17059,9 +17059,10 @@ bool IntExprEvaluator::VisitBuiltinCallExpr(const CallExpr *E,
1705917059
case X86::BI__builtin_ia32_kshiftlisi:
1706017060
case X86::BI__builtin_ia32_kshiftlidi: {
1706117061
return HandleMaskBinOp([](const APSInt &LHS, const APSInt &RHS) {
17062-
if (RHS.uge(LHS.getBitWidth()))
17062+
unsigned Amt = RHS.getZExtValue() & 0xFF;
17063+
if (Amt >= LHS.getBitWidth())
1706317064
return APSInt(APInt::getZero(LHS.getBitWidth()), LHS.isUnsigned());
17064-
return APSInt(LHS.shl(RHS.getZExtValue()), LHS.isUnsigned());
17065+
return APSInt(LHS.shl(Amt), LHS.isUnsigned());
1706517066
});
1706617067
}
1706717068

@@ -17070,9 +17071,10 @@ bool IntExprEvaluator::VisitBuiltinCallExpr(const CallExpr *E,
1707017071
case X86::BI__builtin_ia32_kshiftrisi:
1707117072
case X86::BI__builtin_ia32_kshiftridi: {
1707217073
return HandleMaskBinOp([](const APSInt &LHS, const APSInt &RHS) {
17073-
if (RHS.uge(LHS.getBitWidth()))
17074+
unsigned Amt = RHS.getZExtValue() & 0xFF;
17075+
if (Amt >= LHS.getBitWidth())
1707417076
return APSInt(APInt::getZero(LHS.getBitWidth()), LHS.isUnsigned());
17075-
return APSInt(LHS.lshr(RHS.getZExtValue()), LHS.isUnsigned());
17077+
return APSInt(LHS.lshr(Amt), LHS.isUnsigned());
1707617078
});
1707717079
}
1707817080

clang/test/CodeGen/X86/avx512bw-builtins.c

Lines changed: 12 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -534,40 +534,39 @@ __mmask32 test_kshiftli_mask32(__m512i A, __m512i B, __m512i C, __m512i D) {
534534
// CHECK: [[RES:%.*]] = shufflevector <32 x i1> zeroinitializer, <32 x i1> [[VAL]], <32 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32>
535535
return _mm512_mask_cmpneq_epu16_mask(_kshiftli_mask32(_mm512_cmpneq_epu16_mask(A, B), 31), C, D);
536536
}
537+
TEST_CONSTEXPR(_kshiftli_mask32(0x00000001, 1) == 0x00000002);
538+
TEST_CONSTEXPR(_kshiftli_mask32(0x00000001, 31) == 0x80000000);
539+
TEST_CONSTEXPR(_kshiftli_mask32(0x00000001, 32) == 0x00000000);
540+
TEST_CONSTEXPR(_kshiftli_mask32(0x0000FFFF, 8) == 0x00FFFF00);
537541

538542
__mmask32 test_kshiftri_mask32(__m512i A, __m512i B, __m512i C, __m512i D) {
539543
// CHECK-LABEL: test_kshiftri_mask32
540544
// CHECK: [[VAL:%.*]] = bitcast i32 %{{.*}} to <32 x i1>
541545
// CHECK: [[RES:%.*]] = shufflevector <32 x i1> [[VAL]], <32 x i1> zeroinitializer, <32 x i32> <i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62>
542546
return _mm512_mask_cmpneq_epu16_mask(_kshiftri_mask32(_mm512_cmpneq_epu16_mask(A, B), 31), C, D);
543547
}
548+
TEST_CONSTEXPR(_kshiftri_mask32(0x80000000, 1) == 0x40000000);
549+
TEST_CONSTEXPR(_kshiftri_mask32(0x80000000, 31) == 0x00000001);
550+
TEST_CONSTEXPR(_kshiftri_mask32(0x80000000, 32) == 0x00000000);
551+
TEST_CONSTEXPR(_kshiftri_mask32(0xFFFF0000, 8) == 0x00FFFF00);
544552

545553
__mmask64 test_kshiftli_mask64(__m512i A, __m512i B, __m512i C, __m512i D) {
546554
// CHECK-LABEL: test_kshiftli_mask64
547555
// CHECK: [[VAL:%.*]] = bitcast i64 %{{.*}} to <64 x i1>
548556
// CHECK: [[RES:%.*]] = shufflevector <64 x i1> zeroinitializer, <64 x i1> [[VAL]], <64 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63, i32 64, i32 65, i32 66, i32 67, i32 68, i32 69, i32 70, i32 71, i32 72, i32 73, i32 74, i32 75, i32 76, i32 77, i32 78, i32 79, i32 80, i32 81, i32 82, i32 83, i32 84, i32 85, i32 86, i32 87, i32 88, i32 89, i32 90, i32 91, i32 92, i32 93, i32 94, i32 95>
549557
return _mm512_mask_cmpneq_epu8_mask(_kshiftli_mask64(_mm512_cmpneq_epu8_mask(A, B), 32), C, D);
550558
}
559+
TEST_CONSTEXPR(_kshiftli_mask64(0x0000000000000001ULL, 1) == 0x0000000000000002ULL);
560+
TEST_CONSTEXPR(_kshiftli_mask64(0x0000000000000001ULL, 63) == 0x8000000000000000ULL);
561+
TEST_CONSTEXPR(_kshiftli_mask64(0x0000000000000001ULL, 64) == 0x0000000000000000ULL);
562+
TEST_CONSTEXPR(_kshiftli_mask64(0x00000000FFFFFFFFULL, 16) == 0x0000FFFFFFFF0000ULL);
551563

552564
__mmask64 test_kshiftri_mask64(__m512i A, __m512i B, __m512i C, __m512i D) {
553565
// CHECK-LABEL: test_kshiftri_mask64
554566
// CHECK: [[VAL:%.*]] = bitcast i64 %{{.*}} to <64 x i1>
555567
// CHECK: [[RES:%.*]] = shufflevector <64 x i1> [[VAL]], <64 x i1> zeroinitializer, <64 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63, i32 64, i32 65, i32 66, i32 67, i32 68, i32 69, i32 70, i32 71, i32 72, i32 73, i32 74, i32 75, i32 76, i32 77, i32 78, i32 79, i32 80, i32 81, i32 82, i32 83, i32 84, i32 85, i32 86, i32 87, i32 88, i32 89, i32 90, i32 91, i32 92, i32 93, i32 94, i32 95>
556568
return _mm512_mask_cmpneq_epu8_mask(_kshiftri_mask64(_mm512_cmpneq_epu8_mask(A, B), 32), C, D);
557569
}
558-
559-
TEST_CONSTEXPR(_kshiftli_mask32(0x00000001, 1) == 0x00000002);
560-
TEST_CONSTEXPR(_kshiftli_mask32(0x00000001, 31) == 0x80000000);
561-
TEST_CONSTEXPR(_kshiftli_mask32(0x00000001, 32) == 0x00000000);
562-
TEST_CONSTEXPR(_kshiftli_mask32(0x0000FFFF, 8) == 0x00FFFF00);
563-
TEST_CONSTEXPR(_kshiftri_mask32(0x80000000, 1) == 0x40000000);
564-
TEST_CONSTEXPR(_kshiftri_mask32(0x80000000, 31) == 0x00000001);
565-
TEST_CONSTEXPR(_kshiftri_mask32(0x80000000, 32) == 0x00000000);
566-
TEST_CONSTEXPR(_kshiftri_mask32(0xFFFF0000, 8) == 0x00FFFF00);
567-
TEST_CONSTEXPR(_kshiftli_mask64(0x0000000000000001ULL, 1) == 0x0000000000000002ULL);
568-
TEST_CONSTEXPR(_kshiftli_mask64(0x0000000000000001ULL, 63) == 0x8000000000000000ULL);
569-
TEST_CONSTEXPR(_kshiftli_mask64(0x0000000000000001ULL, 64) == 0x0000000000000000ULL);
570-
TEST_CONSTEXPR(_kshiftli_mask64(0x00000000FFFFFFFFULL, 16) == 0x0000FFFFFFFF0000ULL);
571570
TEST_CONSTEXPR(_kshiftri_mask64(0x8000000000000000ULL, 1) == 0x4000000000000000ULL);
572571
TEST_CONSTEXPR(_kshiftri_mask64(0x8000000000000000ULL, 63) == 0x0000000000000001ULL);
573572
TEST_CONSTEXPR(_kshiftri_mask64(0x8000000000000000ULL, 64) == 0x0000000000000000ULL);

clang/test/CodeGen/X86/avx512dq-builtins.c

Lines changed: 4 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -364,18 +364,17 @@ __mmask8 test_kshiftli_mask8(__m512i A, __m512i B, __m512i C, __m512i D) {
364364
// CHECK: [[RES:%.*]] = shufflevector <8 x i1> zeroinitializer, <8 x i1> [[VAL]], <8 x i32> <i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13>
365365
return _mm512_mask_cmpneq_epu64_mask(_kshiftli_mask8(_mm512_cmpneq_epu64_mask(A, B), 2), C, D);
366366
}
367+
TEST_CONSTEXPR(_kshiftli_mask8(0x01, 1) == 0x02);
368+
TEST_CONSTEXPR(_kshiftli_mask8(0x01, 7) == 0x80);
369+
TEST_CONSTEXPR(_kshiftli_mask8(0x01, 8) == 0x00);
370+
TEST_CONSTEXPR(_kshiftli_mask8(0x0F, 2) == 0x3C);
367371

368372
__mmask8 test_kshiftri_mask8(__m512i A, __m512i B, __m512i C, __m512i D) {
369373
// CHECK-LABEL: test_kshiftri_mask8
370374
// CHECK: [[VAL:%.*]] = bitcast i8 %{{.*}} to <8 x i1>
371375
// CHECK: [[RES:%.*]] = shufflevector <8 x i1> [[VAL]], <8 x i1> zeroinitializer, <8 x i32> <i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9>
372376
return _mm512_mask_cmpneq_epu64_mask(_kshiftri_mask8(_mm512_cmpneq_epu64_mask(A, B), 2), C, D);
373377
}
374-
375-
TEST_CONSTEXPR(_kshiftli_mask8(0x01, 1) == 0x02);
376-
TEST_CONSTEXPR(_kshiftli_mask8(0x01, 7) == 0x80);
377-
TEST_CONSTEXPR(_kshiftli_mask8(0x01, 8) == 0x00);
378-
TEST_CONSTEXPR(_kshiftli_mask8(0x0F, 2) == 0x3C);
379378
TEST_CONSTEXPR(_kshiftri_mask8(0x80, 1) == 0x40);
380379
TEST_CONSTEXPR(_kshiftri_mask8(0x80, 7) == 0x01);
381380
TEST_CONSTEXPR(_kshiftri_mask8(0x80, 8) == 0x00);

clang/test/CodeGen/X86/avx512f-builtins.c

Lines changed: 4 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -9572,6 +9572,10 @@ __mmask16 test_kshiftli_mask16(__m512i A, __m512i B, __m512i C, __m512i D) {
95729572
// CHECK: bitcast <16 x i1> {{.*}} to i16
95739573
return _mm512_mask_cmpneq_epu32_mask(_kshiftli_mask16(_mm512_cmpneq_epu32_mask(A, B), 1), C, D);
95749574
}
9575+
TEST_CONSTEXPR(_kshiftli_mask16(0x0001, 1) == 0x0002);
9576+
TEST_CONSTEXPR(_kshiftli_mask16(0x0001, 15) == 0x8000);
9577+
TEST_CONSTEXPR(_kshiftli_mask16(0x0001, 16) == 0x0000);
9578+
TEST_CONSTEXPR(_kshiftli_mask16(0x00FF, 4) == 0x0FF0);
95759579

95769580
__mmask16 test_kshiftri_mask16(__m512i A, __m512i B, __m512i C, __m512i D) {
95779581
// CHECK-LABEL: test_kshiftri_mask16
@@ -9580,11 +9584,6 @@ __mmask16 test_kshiftri_mask16(__m512i A, __m512i B, __m512i C, __m512i D) {
95809584
// CHECK: bitcast <16 x i1> {{.*}} to i16
95819585
return _mm512_mask_cmpneq_epu32_mask(_kshiftri_mask16(_mm512_cmpneq_epu32_mask(A, B), 1), C, D);
95829586
}
9583-
9584-
TEST_CONSTEXPR(_kshiftli_mask16(0x0001, 1) == 0x0002);
9585-
TEST_CONSTEXPR(_kshiftli_mask16(0x0001, 15) == 0x8000);
9586-
TEST_CONSTEXPR(_kshiftli_mask16(0x0001, 16) == 0x0000);
9587-
TEST_CONSTEXPR(_kshiftli_mask16(0x00FF, 4) == 0x0FF0);
95889587
TEST_CONSTEXPR(_kshiftri_mask16(0x8000, 1) == 0x4000);
95899588
TEST_CONSTEXPR(_kshiftri_mask16(0x8000, 15) == 0x0001);
95909589
TEST_CONSTEXPR(_kshiftri_mask16(0x8000, 16) == 0x0000);

0 commit comments

Comments
 (0)