[libc-commits] [clang] [clang-tools-extra] [libc] [llvm] [libc] Fix buggy AVX2 / AVX512 `memcmp` (PR #77081)
Guillaume Chatelet via libc-commits
libc-commits at lists.llvm.org
Thu Jan 11 02:07:13 PST 2024
https://github.com/gchatelet updated https://github.com/llvm/llvm-project/pull/77081
>From fb8dbd55aacb3a25678b8092a11dd4e562857344 Mon Sep 17 00:00:00 2001
From: Guillaume Chatelet <gchatelet at google.com>
Date: Fri, 5 Jan 2024 11:01:30 +0000
Subject: [PATCH 1/6] [libc] Fix buggy AVX2 `memcmp`
Fixes 77080.
---
libc/src/string/memory_utils/op_x86.h | 33 ++++++++++++++++++++++-----
libc/test/src/string/memcmp_test.cpp | 7 ++++++
2 files changed, 34 insertions(+), 6 deletions(-)
diff --git a/libc/src/string/memory_utils/op_x86.h b/libc/src/string/memory_utils/op_x86.h
index 1a20659c178cd1..23e6b897997e90 100644
--- a/libc/src/string/memory_utils/op_x86.h
+++ b/libc/src/string/memory_utils/op_x86.h
@@ -129,7 +129,8 @@ LIBC_INLINE __m128i bytewise_reverse(__m128i value) {
8, 9, 10, 11, 12, 13, 14, 15));
}
LIBC_INLINE uint16_t big_endian_cmp_mask(__m128i max, __m128i value) {
- return static_cast<uint16_t>(_mm_movemask_epi8(bytewise_reverse(_mm_cmpeq_epi8(max, value))));
+ return static_cast<uint16_t>(
+ _mm_movemask_epi8(bytewise_reverse(_mm_cmpeq_epi8(max, value))));
}
template <> LIBC_INLINE bool eq<__m128i>(CPtr p1, CPtr p2, size_t offset) {
const auto a = load<__m128i>(p1, offset);
@@ -181,11 +182,31 @@ LIBC_INLINE __m256i bytewise_max(__m256i a, __m256i b) {
return _mm256_max_epu8(a, b);
}
LIBC_INLINE __m256i bytewise_reverse(__m256i value) {
- return _mm256_shuffle_epi8(value,
- _mm256_set_epi8(0, 1, 2, 3, 4, 5, 6, 7, //
- 8, 9, 10, 11, 12, 13, 14, 15, //
- 16, 17, 18, 19, 20, 21, 22, 23, //
- 24, 25, 26, 27, 28, 29, 30, 31));
+ const __m256i indices = _mm256_set_epi8(0, 1, 2, 3, 4, 5, 6, 7, //
+ 8, 9, 10, 11, 12, 13, 14, 15, //
+ 16, 17, 18, 19, 20, 21, 22, 23, //
+ 24, 25, 26, 27, 28, 29, 30, 31);
+#if defined(__AVX512VBMI__) && defined(__AVX512VL__)
+ // AVX512 allows full __m256i byte permutation.
+ // ymm = ymm[31,30,29,28,27,26,25,24,23,22,21,20,19,18,17,16,
+ // 15,14,13,12,11,10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0]
+ return _mm256_permutexvar_epi8(value, indices);
+#else
+ // We can't byte-reverse __m256i in a single instruction with AVX2.
+ // '_mm256_shuffle_epi8' can only shuffle within each xmm lane
+ // leading to:
+ // ymm = ymm[15,14,13,12,11,10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0,
+ // 31,30,29,28,27,26,25,24,23,22,21,20,19,18,17,16]
+ const __m256i tmp = _mm256_shuffle_epi8(value, indices);
+ // Then we shuffle accross lanes using 64 bit values.
+ // ymm = ymm[2,3,0,1]
+ // Leading to a fully reversed vector
+ // ymm = ymm[31,30,29,28,27,26,25,24,23,22,21,20,19,18,17,16,
+ // 15,14,13,12,11,10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0]
+ // The immediate encodes the 64 bit word indices : 1, 0, 3, 2.
+ // Each index is encoded with 2 bits : 0b01'00'11'10.
+ return _mm256_permute4x64_epi64(tmp, 0b01'00'11'10);
+#endif
}
LIBC_INLINE uint32_t big_endian_cmp_mask(__m256i max, __m256i value) {
return _mm256_movemask_epi8(bytewise_reverse(_mm256_cmpeq_epi8(max, value)));
diff --git a/libc/test/src/string/memcmp_test.cpp b/libc/test/src/string/memcmp_test.cpp
index 03a0ac1c0ba655..a69257704a64a2 100644
--- a/libc/test/src/string/memcmp_test.cpp
+++ b/libc/test/src/string/memcmp_test.cpp
@@ -37,6 +37,13 @@ TEST(LlvmLibcMemcmpTest, LhsAfterRhsLexically) {
EXPECT_GT(LIBC_NAMESPACE::memcmp(lhs, rhs, 2), 0);
}
+TEST(LlvmLibcMemcmpTest, Issue77080) {
+ // https://github.com/llvm/llvm-project/issues/77080
+ constexpr char lhs[35] = "1.069cd68bbe76eb2143a3284d27ebe220";
+ constexpr char rhs[35] = "1.0500185b5d966a544e2d0fa40701b0f3";
+ EXPECT_GT(LIBC_NAMESPACE::memcmp(lhs, rhs, 34), 0);
+}
+
// Adapt CheckMemcmp signature to memcmp.
static inline int Adaptor(cpp::span<char> p1, cpp::span<char> p2, size_t size) {
return LIBC_NAMESPACE::memcmp(p1.begin(), p2.begin(), size);
>From 04891668ef388ed354d9a18969ada20d54371ce6 Mon Sep 17 00:00:00 2001
From: Guillaume Chatelet <gchatelet at google.com>
Date: Mon, 8 Jan 2024 08:51:12 +0000
Subject: [PATCH 2/6] Make test clearer
---
libc/test/src/string/memcmp_test.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/libc/test/src/string/memcmp_test.cpp b/libc/test/src/string/memcmp_test.cpp
index a69257704a64a2..ca7a5c7ce37023 100644
--- a/libc/test/src/string/memcmp_test.cpp
+++ b/libc/test/src/string/memcmp_test.cpp
@@ -41,7 +41,7 @@ TEST(LlvmLibcMemcmpTest, Issue77080) {
// https://github.com/llvm/llvm-project/issues/77080
constexpr char lhs[35] = "1.069cd68bbe76eb2143a3284d27ebe220";
constexpr char rhs[35] = "1.0500185b5d966a544e2d0fa40701b0f3";
- EXPECT_GT(LIBC_NAMESPACE::memcmp(lhs, rhs, 34), 0);
+ ASSERT_GE(LIBC_NAMESPACE::memcmp(lhs, rhs, 34), 1);
}
// Adapt CheckMemcmp signature to memcmp.
>From 0c76eb213447ad4b2eb77e040053867b8d9d5505 Mon Sep 17 00:00:00 2001
From: Guillaume Chatelet <gchatelet at google.com>
Date: Mon, 8 Jan 2024 08:51:18 +0000
Subject: [PATCH 3/6] Fix typo
---
libc/src/string/memory_utils/op_x86.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/libc/src/string/memory_utils/op_x86.h b/libc/src/string/memory_utils/op_x86.h
index 23e6b897997e90..7313f67fd647bf 100644
--- a/libc/src/string/memory_utils/op_x86.h
+++ b/libc/src/string/memory_utils/op_x86.h
@@ -198,7 +198,7 @@ LIBC_INLINE __m256i bytewise_reverse(__m256i value) {
// ymm = ymm[15,14,13,12,11,10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0,
// 31,30,29,28,27,26,25,24,23,22,21,20,19,18,17,16]
const __m256i tmp = _mm256_shuffle_epi8(value, indices);
- // Then we shuffle accross lanes using 64 bit values.
+ // Then we shuffle across lanes using 64 bit values.
// ymm = ymm[2,3,0,1]
// Leading to a fully reversed vector
// ymm = ymm[31,30,29,28,27,26,25,24,23,22,21,20,19,18,17,16,
>From 2ca47472277a349f3e28834cc3918565fc00ccea Mon Sep 17 00:00:00 2001
From: Guillaume Chatelet <gchatelet at google.com>
Date: Mon, 8 Jan 2024 16:05:45 +0000
Subject: [PATCH 4/6] Use nafi3000 suggestion and fix avx512 version as well.
---
libc/src/string/memory_utils/op_x86.h | 102 +++++++++++++++++---------
1 file changed, 66 insertions(+), 36 deletions(-)
diff --git a/libc/src/string/memory_utils/op_x86.h b/libc/src/string/memory_utils/op_x86.h
index 7313f67fd647bf..96988f8093b193 100644
--- a/libc/src/string/memory_utils/op_x86.h
+++ b/libc/src/string/memory_utils/op_x86.h
@@ -181,36 +181,42 @@ template <> LIBC_INLINE uint32_t neq<__m256i>(CPtr p1, CPtr p2, size_t offset) {
LIBC_INLINE __m256i bytewise_max(__m256i a, __m256i b) {
return _mm256_max_epu8(a, b);
}
-LIBC_INLINE __m256i bytewise_reverse(__m256i value) {
- const __m256i indices = _mm256_set_epi8(0, 1, 2, 3, 4, 5, 6, 7, //
- 8, 9, 10, 11, 12, 13, 14, 15, //
- 16, 17, 18, 19, 20, 21, 22, 23, //
- 24, 25, 26, 27, 28, 29, 30, 31);
+LIBC_INLINE uint32_t big_endian_cmp_mask(__m256i max, __m256i value) {
+ // Bytewise comparison of 'max' and 'value'.
+ const __m256i little_endian_byte_mask = _mm256_cmpeq_epi8(max, value);
+ // Because x86 is little endian, bytes in the vector must be reversed before
+ // using movemask.
#if defined(__AVX512VBMI__) && defined(__AVX512VL__)
- // AVX512 allows full __m256i byte permutation.
- // ymm = ymm[31,30,29,28,27,26,25,24,23,22,21,20,19,18,17,16,
- // 15,14,13,12,11,10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0]
- return _mm256_permutexvar_epi8(value, indices);
+ // When AVX512BMI is available we can completely reverse the vector through
+ // VPERMB __m256i _mm256_permutexvar_epi8( __m256i idx, __m256i a);
+ const __m256i big_endian_byte_mask =
+ _mm256_permutexvar_epi8(_mm256_set_epi8(0, 1, 2, 3, 4, 5, 6, 7, //
+ 8, 9, 10, 11, 12, 13, 14, 15, //
+ 16, 17, 18, 19, 20, 21, 22, 23, //
+ 24, 25, 26, 27, 28, 29, 30, 31),
+ little_endian_byte_mask);
+ // And turn the byte vector mask into an 'uint32_t' for direct scalar
+ // comparison.
+ return _mm256_movemask_epi8(big_endian_byte_mask);
#else
- // We can't byte-reverse __m256i in a single instruction with AVX2.
- // '_mm256_shuffle_epi8' can only shuffle within each xmm lane
+ // We can't byte-reverse '__m256i' in a single instruction with AVX2.
+ // '_mm256_shuffle_epi8' can only shuffle within each 16-byte lane
// leading to:
// ymm = ymm[15,14,13,12,11,10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0,
// 31,30,29,28,27,26,25,24,23,22,21,20,19,18,17,16]
- const __m256i tmp = _mm256_shuffle_epi8(value, indices);
- // Then we shuffle across lanes using 64 bit values.
- // ymm = ymm[2,3,0,1]
- // Leading to a fully reversed vector
- // ymm = ymm[31,30,29,28,27,26,25,24,23,22,21,20,19,18,17,16,
- // 15,14,13,12,11,10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0]
- // The immediate encodes the 64 bit word indices : 1, 0, 3, 2.
- // Each index is encoded with 2 bits : 0b01'00'11'10.
- return _mm256_permute4x64_epi64(tmp, 0b01'00'11'10);
+ // So we first shuffle each 16-byte lane leading to half-reversed vector mask.
+ const __m256i half_reversed = _mm256_shuffle_epi8(
+ little_endian_byte_mask, _mm256_set_epi8(0, 1, 2, 3, 4, 5, 6, 7, //
+ 8, 9, 10, 11, 12, 13, 14, 15, //
+ 0, 1, 2, 3, 4, 5, 6, 7, //
+ 8, 9, 10, 11, 12, 13, 14, 15));
+ // Then we turn the vector into an uint32_t.
+ const uint32_t half_reversed_scalar = _mm256_movemask_epi8(half_reversed);
+ // And swap the lower and upper parts. This is optimized into a single `rorx`
+ // instruction.
+ return (half_reversed_scalar << 16) | (half_reversed_scalar >> 16);
#endif
}
-LIBC_INLINE uint32_t big_endian_cmp_mask(__m256i max, __m256i value) {
- return _mm256_movemask_epi8(bytewise_reverse(_mm256_cmpeq_epi8(max, value)));
-}
template <>
LIBC_INLINE MemcmpReturnType cmp_neq<__m256i>(CPtr p1, CPtr p2, size_t offset) {
const auto a = load<__m256i>(p1, offset);
@@ -219,7 +225,7 @@ LIBC_INLINE MemcmpReturnType cmp_neq<__m256i>(CPtr p1, CPtr p2, size_t offset) {
const auto le = big_endian_cmp_mask(vmax, b);
const auto ge = big_endian_cmp_mask(vmax, a);
static_assert(cpp::is_same_v<cpp::remove_cv_t<decltype(le)>, uint32_t>);
- return cmp_uint32_t(ge, le);
+ return cmp_neq_uint64_t(ge, le);
}
#endif // __AVX2__
@@ -231,19 +237,43 @@ template <> struct cmp_is_expensive<__m512i> : cpp::true_type {};
LIBC_INLINE __m512i bytewise_max(__m512i a, __m512i b) {
return _mm512_max_epu8(a, b);
}
-LIBC_INLINE __m512i bytewise_reverse(__m512i value) {
- return _mm512_shuffle_epi8(value,
- _mm512_set_epi8(0, 1, 2, 3, 4, 5, 6, 7, //
- 8, 9, 10, 11, 12, 13, 14, 15, //
- 16, 17, 18, 19, 20, 21, 22, 23, //
- 24, 25, 26, 27, 28, 29, 30, 31, //
- 32, 33, 34, 35, 36, 37, 38, 39, //
- 40, 41, 42, 43, 44, 45, 46, 47, //
- 48, 49, 50, 51, 52, 53, 54, 55, //
- 56, 57, 58, 59, 60, 61, 62, 63));
-}
LIBC_INLINE uint64_t big_endian_cmp_mask(__m512i max, __m512i value) {
- return _mm512_cmpeq_epi8_mask(bytewise_reverse(max), bytewise_reverse(value));
+#if defined(__AVX512VBMI__)
+ // When AVX512BMI is available we can completely reverse the vector through
+ // VPERMB __m512i _mm512_permutexvar_epi8( __m512i idx, __m512i a);
+ const auto indices = _mm512_set_epi8(0, 1, 2, 3, 4, 5, 6, 7, //
+ 8, 9, 10, 11, 12, 13, 14, 15, //
+ 16, 17, 18, 19, 20, 21, 22, 23, //
+ 24, 25, 26, 27, 28, 29, 30, 31, //
+ 32, 33, 34, 35, 36, 37, 38, 39, //
+ 40, 41, 42, 43, 44, 45, 46, 47, //
+ 48, 49, 50, 51, 52, 53, 54, 55, //
+ 56, 57, 58, 59, 60, 61, 62, 63);
+ // Then we compute the mask for equal bytes.
+ return _mm512_cmpeq_epi8_mask(_mm512_permutexvar_epi8(indices, max), //
+ _mm512_permutexvar_epi8(indices, value));
+#else
+ // We can't byte-reverse '__m512i' in a single instruction with __AVX512BW__.
+ // '_mm512_shuffle_epi8' can only shuffle within each 16-byte lane.
+ // zmm = | 16 bytes | 16 bytes | 16 bytes | 16 bytes |
+ // So we only reverse groups of 8 bytes, these groups are necessarily within a
+ // 16-byte lane.
+ const __m512i indices = _mm512_set_epi8(56, 57, 58, 59, 60, 61, 62, 63, //
+ 48, 49, 50, 51, 52, 53, 54, 55, //
+ 40, 41, 42, 43, 44, 45, 46, 47, //
+ 32, 33, 34, 35, 36, 37, 38, 39, //
+ 24, 25, 26, 27, 28, 29, 30, 31, //
+ 16, 17, 18, 19, 20, 21, 22, 23, //
+ 8, 9, 10, 11, 12, 13, 14, 15, //
+ 0, 1, 2, 3, 4, 5, 6, 7);
+ // Then we compute the mask for equal bytes. In this mask the bits of each
+ // byte are already reversed but the byte themselves should be reversed, this
+ // is done by using a bswap instruction.
+ return __builtin_bswap64(
+ _mm512_cmpeq_epi8_mask(_mm512_shuffle_epi8(max, indices), //
+ _mm512_shuffle_epi8(value, indices)));
+
+#endif
}
template <> LIBC_INLINE bool eq<__m512i>(CPtr p1, CPtr p2, size_t offset) {
const auto a = load<__m512i>(p1, offset);
>From de6e7b662db682d8ea72a39205a639290180be55 Mon Sep 17 00:00:00 2001
From: Guillaume Chatelet <gchatelet at google.com>
Date: Tue, 9 Jan 2024 16:06:02 +0000
Subject: [PATCH 5/6] Disable AVX512BMI version for now
---
libc/src/string/memory_utils/op_x86.h | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/libc/src/string/memory_utils/op_x86.h b/libc/src/string/memory_utils/op_x86.h
index 96988f8093b193..6ad3fba0df032d 100644
--- a/libc/src/string/memory_utils/op_x86.h
+++ b/libc/src/string/memory_utils/op_x86.h
@@ -238,7 +238,11 @@ LIBC_INLINE __m512i bytewise_max(__m512i a, __m512i b) {
return _mm512_max_epu8(a, b);
}
LIBC_INLINE uint64_t big_endian_cmp_mask(__m512i max, __m512i value) {
-#if defined(__AVX512VBMI__)
+ // The AVX512BMI version is disabled due to bad codegen.
+ // https://github.com/llvm/llvm-project/issues/77459
+ // https://github.com/llvm/llvm-project/pull/77081
+ // TODO: Re-enable when clang version meets the fixed version.
+#if false && defined(__AVX512VBMI__)
// When AVX512BMI is available we can completely reverse the vector through
// VPERMB __m512i _mm512_permutexvar_epi8( __m512i idx, __m512i a);
const auto indices = _mm512_set_epi8(0, 1, 2, 3, 4, 5, 6, 7, //
>From a49f014e9030c1578c4143d48882d09c7c152da0 Mon Sep 17 00:00:00 2001
From: Guillaume Chatelet <gchatelet at google.com>
Date: Tue, 9 Jan 2024 16:30:28 +0000
Subject: [PATCH 6/6] Reformulate documentation
---
libc/src/string/memory_utils/op_x86.h | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/libc/src/string/memory_utils/op_x86.h b/libc/src/string/memory_utils/op_x86.h
index 6ad3fba0df032d..5ee07823c64974 100644
--- a/libc/src/string/memory_utils/op_x86.h
+++ b/libc/src/string/memory_utils/op_x86.h
@@ -259,9 +259,10 @@ LIBC_INLINE uint64_t big_endian_cmp_mask(__m512i max, __m512i value) {
#else
// We can't byte-reverse '__m512i' in a single instruction with __AVX512BW__.
// '_mm512_shuffle_epi8' can only shuffle within each 16-byte lane.
- // zmm = | 16 bytes | 16 bytes | 16 bytes | 16 bytes |
// So we only reverse groups of 8 bytes, these groups are necessarily within a
// 16-byte lane.
+ // zmm = | 16 bytes | 16 bytes | 16 bytes | 16 bytes |
+ // zmm = | <8> | <8> | <8> | <8> | <8> | <8> | <8> | <8> |
const __m512i indices = _mm512_set_epi8(56, 57, 58, 59, 60, 61, 62, 63, //
48, 49, 50, 51, 52, 53, 54, 55, //
40, 41, 42, 43, 44, 45, 46, 47, //
More information about the libc-commits
mailing list