[clang] [llvm-profdata] Do not create numerical strings for MD5 function names read from a Sample Profile. (PR #66164)
William Junda Huang via cfe-commits
cfe-commits at lists.llvm.org
Fri Sep 15 15:36:30 PDT 2023
https://github.com/huangjd updated https://github.com/llvm/llvm-project/pull/66164
>From b6bfc6127925a66029227b1d83a2b0ae890c1b08 Mon Sep 17 00:00:00 2001
From: William Huang <williamjhuang at google.com>
Date: Fri, 15 Sep 2023 20:20:16 +0000
Subject: [PATCH 1/3] [llvm-profdata] Remove MD5 collision check in D147740
This is the patch at https://reviews.llvm.org/D153692, migrating to
Github
After testing D147740 with multiple industrial projects with ~10 million FunctionSamples, no MD5 collision has been found.
In perfect hashing, the probability of collision for N symbols over K possible hash value is 1 - K!/((K-N)! * K^N). When N is 1 million and K is 2^64, the probability is 3*10^-8, when N is 10 million the probability is 3*10^-6, so we are probably not going to find an actual case in real world application. (However if K is 2^32, the probability of collision is almost 1, this is indeed a problem, if anyone still use a large profile on 32-bit machine, as hash_code is tied to size_t).
Furthermore, when a collision happens we can't do anything to recover it, unless using a multi-map, but that is significantly slower, which contradicts the purpose of optimizing the profile reader.
One more thing, since we have been using profiles with MD5 names, and they have to be coming from non-MD5 sources, so if hash collision is to happen, it already happened when we convert a non-MD5 profile to a MD5 one, so there's no point to check for that in the reader, and this feature can be removed.
---
llvm/include/llvm/ProfileData/SampleProf.h | 68 ++-----
.../tools/llvm-profdata/CMakeLists.txt | 1 -
.../tools/llvm-profdata/MD5CollisionTest.cpp | 166 ------------------
3 files changed, 10 insertions(+), 225 deletions(-)
delete mode 100644 llvm/unittests/tools/llvm-profdata/MD5CollisionTest.cpp
diff --git a/llvm/include/llvm/ProfileData/SampleProf.h b/llvm/include/llvm/ProfileData/SampleProf.h
index 3cbfa94896e3a24..181ddac0daf91af 100644
--- a/llvm/include/llvm/ProfileData/SampleProf.h
+++ b/llvm/include/llvm/ProfileData/SampleProf.h
@@ -1299,19 +1299,13 @@ raw_ostream &operator<<(raw_ostream &OS, const FunctionSamples &FS);
/// performance of insert and query operations especially when hash values of
/// keys are available a priori, and reduces memory usage if KeyT has a large
/// size.
-/// When performing any action, if an existing entry with a given key is found,
-/// and the interface "KeyT ValueT::getKey<KeyT>() const" to retrieve a value's
-/// original key exists, this class checks if the given key actually matches
-/// the existing entry's original key. If they do not match, this class behaves
-/// as if the entry did not exist (for insertion, this means the new value will
-/// replace the existing entry's value, as if it is newly inserted). If
-/// ValueT::getKey<KeyT>() is not available, all keys with the same hash value
-/// are considered equivalent (i.e. hash collision is silently ignored). Given
-/// such feature this class should only be used where it does not affect
-/// compilation correctness, for example, when loading a sample profile.
+/// All keys with the same hash value are considered equivalent (i.e. hash
+/// collision is silently ignored). Given such feature this class should only be
+/// used where it does not affect compilation correctness, for example, when
+/// loading a sample profile.
/// Assuming the hashing algorithm is uniform, the probability of hash collision
/// with 1,000,000 entries is
-/// (2^64)!/((2^64-1000000)!*(2^64)^1000000) ~= 3*10^-8.
+/// 1 - (2^64)!/((2^64-1000000)!*(2^64)^1000000) ~= 3*10^-8.
template <template <typename, typename, typename...> typename MapT,
typename KeyT, typename ValueT, typename... MapTArgs>
class HashKeyMap : public MapT<hash_code, ValueT, MapTArgs...> {
@@ -1325,42 +1319,12 @@ class HashKeyMap : public MapT<hash_code, ValueT, MapTArgs...> {
using iterator = typename base_type::iterator;
using const_iterator = typename base_type::const_iterator;
-private:
- // If the value type has getKey(), retrieve its original key for comparison.
- template <typename U = mapped_type,
- typename = decltype(U().template getKey<original_key_type>())>
- static bool
- CheckKeyMatch(const original_key_type &Key, const mapped_type &ExistingValue,
- original_key_type *ExistingKeyIfDifferent = nullptr) {
- const original_key_type &ExistingKey =
- ExistingValue.template getKey<original_key_type>();
- bool Result = (Key == ExistingKey);
- if (!Result && ExistingKeyIfDifferent)
- *ExistingKeyIfDifferent = ExistingKey;
- return Result;
- }
-
- // If getKey() does not exist, this overload is selected, which assumes all
- // keys with the same hash are equivalent.
- static bool CheckKeyMatch(...) { return true; }
-
-public:
template <typename... Ts>
std::pair<iterator, bool> try_emplace(const key_type &Hash,
const original_key_type &Key,
Ts &&...Args) {
assert(Hash == hash_value(Key));
- auto Ret = base_type::try_emplace(Hash, std::forward<Ts>(Args)...);
- if (!Ret.second) {
- original_key_type ExistingKey;
- if (LLVM_UNLIKELY(!CheckKeyMatch(Key, Ret.first->second, &ExistingKey))) {
- dbgs() << "MD5 collision detected: " << Key << " and " << ExistingKey
- << " has same hash value " << Hash << "\n";
- Ret.second = true;
- Ret.first->second = mapped_type(std::forward<Ts>(Args)...);
- }
- }
- return Ret;
+ return base_type::try_emplace(Hash, std::forward<Ts>(Args)...);
}
template <typename... Ts>
@@ -1382,8 +1346,7 @@ class HashKeyMap : public MapT<hash_code, ValueT, MapTArgs...> {
key_type Hash = hash_value(Key);
auto It = base_type::find(Hash);
if (It != base_type::end())
- if (LLVM_LIKELY(CheckKeyMatch(Key, It->second)))
- return It;
+ return It;
return base_type::end();
}
@@ -1391,8 +1354,7 @@ class HashKeyMap : public MapT<hash_code, ValueT, MapTArgs...> {
key_type Hash = hash_value(Key);
auto It = base_type::find(Hash);
if (It != base_type::end())
- if (LLVM_LIKELY(CheckKeyMatch(Key, It->second)))
- return It;
+ return It;
return base_type::end();
}
@@ -1433,19 +1395,9 @@ class SampleProfileMap
Ctx);
}
- // Overloaded find() to lookup a function by name. This is called by IPO
- // passes with an actual function name, and it is possible that the profile
- // reader converted function names in the profile to MD5 strings, so we need
- // to check if either representation matches.
+ // Overloaded find() to lookup a function by name.
iterator find(StringRef Fname) {
- uint64_t Hash = hashFuncName(Fname);
- auto It = base_type::find(hash_code(Hash));
- if (It != end()) {
- StringRef CtxName = It->second.getContext().getName();
- if (LLVM_LIKELY(CtxName == Fname || CtxName == std::to_string(Hash)))
- return It;
- }
- return end();
+ return base_type::find(hashFuncName(Fname));
}
size_t erase(const SampleContext &Ctx) {
diff --git a/llvm/unittests/tools/llvm-profdata/CMakeLists.txt b/llvm/unittests/tools/llvm-profdata/CMakeLists.txt
index c53baf69bbc0ef8..ad91ce36bcb5bef 100644
--- a/llvm/unittests/tools/llvm-profdata/CMakeLists.txt
+++ b/llvm/unittests/tools/llvm-profdata/CMakeLists.txt
@@ -6,7 +6,6 @@ set(LLVM_LINK_COMPONENTS
add_llvm_unittest(LLVMProfdataTests
OutputSizeLimitTest.cpp
- MD5CollisionTest.cpp
)
target_link_libraries(LLVMProfdataTests PRIVATE LLVMTestingSupport)
diff --git a/llvm/unittests/tools/llvm-profdata/MD5CollisionTest.cpp b/llvm/unittests/tools/llvm-profdata/MD5CollisionTest.cpp
deleted file mode 100644
index 8af74249188dd14..000000000000000
--- a/llvm/unittests/tools/llvm-profdata/MD5CollisionTest.cpp
+++ /dev/null
@@ -1,166 +0,0 @@
-//===- llvm/unittests/tools/llvm-profdata/MD5CollisionTest.cpp ------------===//
-//
-// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-//
-//===----------------------------------------------------------------------===//
-
-/// Test whether the MD5-key SampleProfileMap can handle collision correctly.
-/// Probability of collision is rare but not negligible since we only use the
-/// lower 64 bits of the MD5 value. A unit test is required because the function
-/// names are not printable ASCII characters.
-
-#include "llvm/ProfileData/SampleProfReader.h"
-#include "llvm/Support/VirtualFileSystem.h"
-#include "llvm/Testing/Support/Error.h"
-#include "gtest/gtest.h"
-
-/// According to https://en.wikipedia.org/wiki/MD5#Preimage_vulnerability, the
-/// MD5 of the two strings are 79054025255fb1a26e4bc422aef54eb4.
-
-// First 8 bytes of the MD5.
-const uint64_t ExpectedHash = 0xa2b15f2525400579;
-
-// clang-format off
-const uint8_t ProfileData[] = {
- 0x84, 0xe4, 0xd0, 0xb1, 0xf4, 0xc9, 0x94, 0xa8,
- 0x53, 0x67, 0x03, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x7D, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x03, 0x01, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x04, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x80, 0x01, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x05, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x90, 0x01, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x00,
-
- /// Name Table
- 0x02,
- /// String1
- 0xd1, 0x31, 0xdd, 0x02, 0xc5, 0xe6, 0xee, 0xc4,
- 0x69, 0x3d, 0x9a, 0x06, 0x98, 0xaf, 0xf9, 0x5c,
- 0x2f, 0xca, 0xb5, 0x87, 0x12, 0x46, 0x7e, 0xab,
- 0x40, 0x04, 0x58, 0x3e, 0xb8, 0xfb, 0x7f, 0x89,
- 0x55, 0xad, 0x34, 0x06, 0x09, 0xf4, 0xb3, 0x02,
- 0x83, 0xe4, 0x88, 0x83, 0x25, 0x71, 0x41, 0x5a,
- 0x08, 0x51, 0x25, 0xe8, 0xf7, 0xcd, 0xc9, 0x9f,
- 0xd9, 0x1d, 0xbd, 0xf2, 0x80, 0x37, 0x3c, 0x5b,
- 0xd8, 0x82, 0x3e, 0x31, 0x56, 0x34, 0x8f, 0x5b,
- 0xae, 0x6d, 0xac, 0xd4, 0x36, 0xc9, 0x19, 0xc6,
- 0xdd, 0x53, 0xe2, 0xb4, 0x87, 0xda, 0x03, 0xfd,
- 0x02, 0x39, 0x63, 0x06, 0xd2, 0x48, 0xcd, 0xa0,
- 0xe9, 0x9f, 0x33, 0x42, 0x0f, 0x57, 0x7e, 0xe8,
- 0xce, 0x54, 0xb6, 0x70, 0x80, 0xa8, 0x0d, 0x1e,
- 0xc6, 0x98, 0x21, 0xbc, 0xb6, 0xa8, 0x83, 0x93,
- 0x96, 0xf9, 0x65, 0x2b, 0x6f, 0xf7, 0x2a, 0x70, 0x00,
- /// String2
- 0xd1, 0x31, 0xdd, 0x02, 0xc5, 0xe6, 0xee, 0xc4,
- 0x69, 0x3d, 0x9a, 0x06, 0x98, 0xaf, 0xf9, 0x5c,
- 0x2f, 0xca, 0xb5, 0x07, 0x12, 0x46, 0x7e, 0xab,
- 0x40, 0x04, 0x58, 0x3e, 0xb8, 0xfb, 0x7f, 0x89,
- 0x55, 0xad, 0x34, 0x06, 0x09, 0xf4, 0xb3, 0x02,
- 0x83, 0xe4, 0x88, 0x83, 0x25, 0xf1, 0x41, 0x5a,
- 0x08, 0x51, 0x25, 0xe8, 0xf7, 0xcd, 0xc9, 0x9f,
- 0xd9, 0x1d, 0xbd, 0x72, 0x80, 0x37, 0x3c, 0x5b,
- 0xd8, 0x82, 0x3e, 0x31, 0x56, 0x34, 0x8f, 0x5b,
- 0xae, 0x6d, 0xac, 0xd4, 0x36, 0xc9, 0x19, 0xc6,
- 0xdd, 0x53, 0xe2, 0x34, 0x87, 0xda, 0x03, 0xfd,
- 0x02, 0x39, 0x63, 0x06, 0xd2, 0x48, 0xcd, 0xa0,
- 0xe9, 0x9f, 0x33, 0x42, 0x0f, 0x57, 0x7e, 0xe8,
- 0xce, 0x54, 0xb6, 0x70, 0x80, 0x28, 0x0d, 0x1e,
- 0xc6, 0x98, 0x21, 0xbc, 0xb6, 0xa8, 0x83, 0x93,
- 0x96, 0xf9, 0x65, 0xab, 0x6f, 0xf7, 0x2a, 0x70, 0x00,
-
- /// FuncOffsetTable
- 0x02, 0x00, 0x00, 0x01, 0x17, 0x00, 0x00, 0x00,
- 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-
- /// Samples
- /// String1:10:1
- /// 1: 5
- /// 2.3: 6
- /// 4: String2:100
- /// 1: 100
- /// String2:7:3
- /// 9: 0
- 0x01, 0x00, 0x0a, 0x02, 0x01, 0x00, 0x05, 0x00,
- 0x02, 0x03, 0x06, 0x00, 0x01, 0x04, 0x00, 0x01,
- 0x64, 0x01, 0x01, 0x00, 0x64, 0x00, 0x00,
-
- 0x03, 0x01, 0x07, 0x01, 0x09, 0x00, 0x00, 0x00,
- 0x00};
-// clang-format on
-
-using namespace llvm;
-using namespace llvm::sampleprof;
-
-TEST(MD5CollisionTest, TestCollision) {
- auto InputBuffer = MemoryBuffer::getMemBuffer(
- StringRef(reinterpret_cast<const char *>(ProfileData),
- sizeof(ProfileData)),
- "", false);
- LLVMContext Context;
- auto FileSystem = vfs::getRealFileSystem();
- auto Result = SampleProfileReader::create(InputBuffer, Context, *FileSystem);
- ASSERT_TRUE(Result);
- SampleProfileReader *Reader = Result->get();
- ASSERT_FALSE(Reader->read());
-
- std::vector<StringRef> &NameTable = *Reader->getNameTable();
- ASSERT_EQ(NameTable.size(), 2U);
- StringRef S1 = NameTable[0];
- StringRef S2 = NameTable[1];
- ASSERT_NE(S1, S2);
- ASSERT_EQ(MD5Hash(S1), ExpectedHash);
- ASSERT_EQ(MD5Hash(S2), ExpectedHash);
-
- // S2's MD5 value collides with S1, S1 is expected to be dropped when S2 is
- // inserted, as if S1 never existed.
-
- FunctionSamples ExpectedFS;
- ExpectedFS.setName(S2);
- ExpectedFS.setHeadSamples(3);
- ExpectedFS.setTotalSamples(7);
- ExpectedFS.addBodySamples(9, 0, 0);
-
- SampleProfileMap &Profiles = Reader->getProfiles();
- EXPECT_EQ(Profiles.size(), 1U);
- if (Profiles.size()) {
- auto &[Hash, FS] = *Profiles.begin();
- EXPECT_EQ(Hash, hash_code(ExpectedHash));
- EXPECT_EQ(FS, ExpectedFS);
- }
-
- // Inserting S2 again should fail, returning the existing sample unchanged.
- auto [It1, Inserted1] = Profiles.try_emplace(S2, FunctionSamples());
- EXPECT_FALSE(Inserted1);
- EXPECT_EQ(Profiles.size(), 1U);
- if (Profiles.size()) {
- auto &[Hash, FS] = *It1;
- EXPECT_EQ(Hash, hash_code(ExpectedHash));
- EXPECT_EQ(FS, ExpectedFS);
- }
-
- // Inserting S1 should success as if S2 never existed, and S2 is erased.
- FunctionSamples FS1;
- FS1.setName(S1);
- FS1.setHeadSamples(5);
- FS1.setTotalSamples(10);
- FS1.addBodySamples(1, 2, 5);
-
- auto [It2, Inserted2] = Profiles.try_emplace(S1, FS1);
- EXPECT_TRUE(Inserted2);
- EXPECT_EQ(Profiles.size(), 1U);
- if (Profiles.size()) {
- auto &[Hash, FS] = *It2;
- EXPECT_EQ(Hash, hash_code(ExpectedHash));
- EXPECT_EQ(FS, FS1);
- }
-}
>From 70fd12396b79d049b08b9bfc4a9439c7f6824e49 Mon Sep 17 00:00:00 2001
From: William Huang <williamjhuang at google.com>
Date: Fri, 15 Sep 2023 22:01:48 +0000
Subject: [PATCH 2/3] Updated comments
---
llvm/include/llvm/ProfileData/SampleProf.h | 7 +++++--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/llvm/include/llvm/ProfileData/SampleProf.h b/llvm/include/llvm/ProfileData/SampleProf.h
index 181ddac0daf91af..9b2416e39b6cc9a 100644
--- a/llvm/include/llvm/ProfileData/SampleProf.h
+++ b/llvm/include/llvm/ProfileData/SampleProf.h
@@ -1303,9 +1303,12 @@ raw_ostream &operator<<(raw_ostream &OS, const FunctionSamples &FS);
/// collision is silently ignored). Given such feature this class should only be
/// used where it does not affect compilation correctness, for example, when
/// loading a sample profile.
-/// Assuming the hashing algorithm is uniform, the probability of hash collision
-/// with 1,000,000 entries is
+/// Assuming the hashing algorithm is uniform, we use the formula
+/// 1 - Permute(n, k) / n ^ k where n is the universe size and k is number of
+/// elements chosen at random to calculate the probability of collision. With
+/// 1,000,000 entries the probability is negligible:
/// 1 - (2^64)!/((2^64-1000000)!*(2^64)^1000000) ~= 3*10^-8.
+/// Source: https://en.wikipedia.org/wiki/Birthday_problem
template <template <typename, typename, typename...> typename MapT,
typename KeyT, typename ValueT, typename... MapTArgs>
class HashKeyMap : public MapT<hash_code, ValueT, MapTArgs...> {
>From f099207037f9371267f5f853f96dd7c27fb0ada8 Mon Sep 17 00:00:00 2001
From: William Huang <williamjhuang at google.com>
Date: Fri, 15 Sep 2023 20:49:19 +0000
Subject: [PATCH 3/3] Rebase with main
---
.../include/llvm/ProfileData/ProfileFuncRef.h | 225 ++++++++++++++++++
llvm/include/llvm/ProfileData/SampleProf.h | 130 +++++-----
.../llvm/ProfileData/SampleProfReader.h | 27 +--
.../llvm/ProfileData/SampleProfWriter.h | 12 +-
.../llvm/Transforms/IPO/ProfiledCallGraph.h | 24 +-
.../Transforms/IPO/SampleContextTracker.h | 27 ++-
llvm/lib/ProfileData/SampleProf.cpp | 25 +-
llvm/lib/ProfileData/SampleProfReader.cpp | 76 +++---
llvm/lib/ProfileData/SampleProfWriter.cpp | 30 ++-
llvm/lib/Target/X86/X86InsertPrefetch.cpp | 3 +-
.../Transforms/IPO/SampleContextTracker.cpp | 64 ++---
llvm/lib/Transforms/IPO/SampleProfile.cpp | 90 ++++---
llvm/tools/llvm-profdata/llvm-profdata.cpp | 16 +-
llvm/tools/llvm-profgen/CSPreInliner.cpp | 8 +-
llvm/tools/llvm-profgen/CSPreInliner.h | 7 +-
llvm/tools/llvm-profgen/ProfileGenerator.cpp | 19 +-
llvm/tools/llvm-profgen/ProfiledBinary.cpp | 15 +-
llvm/tools/llvm-profgen/ProfiledBinary.h | 20 +-
llvm/unittests/ProfileData/SampleProfTest.cpp | 45 ++--
19 files changed, 555 insertions(+), 308 deletions(-)
create mode 100644 llvm/include/llvm/ProfileData/ProfileFuncRef.h
diff --git a/llvm/include/llvm/ProfileData/ProfileFuncRef.h b/llvm/include/llvm/ProfileData/ProfileFuncRef.h
new file mode 100644
index 000000000000000..60be6232372b8c6
--- /dev/null
+++ b/llvm/include/llvm/ProfileData/ProfileFuncRef.h
@@ -0,0 +1,225 @@
+//===--- ProfileFuncRef.h - Sample profile function name ---*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This file defines the StringRefOrHashCode class. It is to represent function
+// names in a sample profile, which can be in one of two forms - either a
+// regular string, or a 64-bit hash code.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_PROFILEDATA_PROFILEFUNCREF_H
+#define LLVM_PROFILEDATA_PROFILEFUNCREF_H
+
+#include "llvm/ADT/DenseMapInfo.h"
+#include "llvm/ADT/Hashing.h"
+#include "llvm/ADT/StringRef.h"
+#include "llvm/Support/MD5.h"
+#include "llvm/Support/raw_ostream.h"
+#include <cstdint>
+
+namespace llvm {
+namespace sampleprof {
+
+/// This class represents a function name that is read from a sample profile. It
+/// comes with two forms: a string or a hash code. For efficient storage, a
+/// sample profile may store function names as 64-bit MD5 values, so when
+/// reading the profile, this class can represnet them without converting it to
+/// a string first.
+/// When representing a hash code, we utilize the Length field to store it, and
+/// Data is set to null. When representing a string, it is same as StringRef,
+/// and can be pointer-casted as one.
+/// We disallow implicit cast to StringRef because there are too many instances
+/// that it may cause break the code, such as using it in a StringMap.
+class ProfileFuncRef {
+
+ const char *Data = nullptr;
+
+ /// Use uint64_t instead of size_t so that it can also hold a MD5 value.
+ uint64_t Length = 0;
+
+ /// Extension to memcmp to handle hash code representation. If both are hash
+ /// values, Lhs and Rhs are both null, function returns 0 (and needs an extra
+ /// comparison using getIntValue). If only one is hash code, it is considered
+ /// less than the StringRef one. Otherwise perform normal string comparison.
+ static int compareMemory(const char *Lhs, const char *Rhs, uint64_t Length) {
+ if (Lhs == Rhs)
+ return 0;
+ if (!Lhs)
+ return -1;
+ if (!Rhs)
+ return 1;
+ return ::memcmp(Lhs, Rhs, (size_t)Length);
+ }
+
+public:
+ ProfileFuncRef() = default;
+
+ /// Constructor from a StringRef.
+ explicit ProfileFuncRef(StringRef Str)
+ : Data(Str.data()), Length(Str.size()) {}
+
+ /// Constructor from a hash code.
+ explicit ProfileFuncRef(uint64_t HashCode)
+ : Data(nullptr), Length(HashCode) {
+ assert(HashCode != 0);
+ }
+
+ /// Constructor from a string. Check if Str is a number, which is generated by
+ /// converting a MD5 sample profile to a format that does not support MD5, and
+ /// if so, convert the numerical string to a hash code first. We assume that
+ /// no function name (from a profile) can be a pure number.
+ explicit ProfileFuncRef(const std::string &Str)
+ : Data(Str.data()), Length(Str.size()) {
+ // Only need to check for base 10 digits, fail faster if otherwise.
+ if (Str.length() > 0 && isdigit(Str[0]) &&
+ !StringRef(Str).getAsInteger(10, Length))
+ Data = nullptr;
+ }
+
+ /// Check for equality. Similar to StringRef::equals, but will also cover for
+ /// the case where one or both are hash codes. Comparing their int values are
+ /// sufficient. A hash code ProfileFuncName is considered not equal to a
+ /// StringRef ProfileFuncName regardless of actual contents.
+ bool equals(const ProfileFuncRef &Other) const {
+ return Length == Other.Length &&
+ compareMemory(Data, Other.Data, Length) == 0;
+ }
+
+ /// Total order comparison. If both ProfileFuncName are StringRef, this is the
+ /// same as StringRef::compare. If one of them is StringRef, it is considered
+ /// greater than the hash code ProfileFuncName. Otherwise this is the the
+ /// same as comparing their int values.
+ int compare(const ProfileFuncRef &Other) const {
+ auto Res = compareMemory(Data, Other.Data, std::min(Length, Other.Length));
+ if (Res != 0)
+ return Res;
+ if (Length == Other.Length)
+ return 0;
+ return Length < Other.Length ? -1 : 1;
+ }
+
+ /// Convert to a string, usually for output purpose.
+ std::string str() const {
+ if (Data)
+ return std::string(Data, Length);
+ if (Length != 0)
+ return std::to_string(Length);
+ return std::string();
+ }
+
+ /// Convert to StringRef, a backing buffer must be provided in case this
+ /// object represents a hash code, which will be converted to a string into
+ /// the buffer. If this object represents a StringRef, the buffer is not used.
+ StringRef stringRef(std::string &Buffer) const {
+ if (Data)
+ return StringRef(Data, Length);
+ if (Length != 0) {
+ Buffer = std::to_string(Length);
+ return Buffer;
+ }
+ return StringRef();
+ }
+
+ friend raw_ostream &operator<<(raw_ostream &OS, const ProfileFuncRef &Obj);
+
+ /// Get hash code of this object. Returns this object's hash code if it is
+ /// already representing one, otherwise returns the MD5 of its string content.
+ /// Note that it is not the same as std::hash because we want to keep the
+ /// consistency that the same sample profile function in string form or MD5
+ /// form has the same hash code.
+ uint64_t getHashCode() const {
+ if (Data)
+ return MD5Hash(StringRef(Data, Length));
+ return Length;
+ }
+
+ bool empty() const { return Length == 0; }
+
+ /// Check if this object represents a StringRef, or a hash code.
+ bool isStringRef() const { return Data != nullptr; }
+};
+
+inline bool operator==(const ProfileFuncRef &LHS, const ProfileFuncRef &RHS) {
+ return LHS.equals(RHS);
+}
+
+inline bool operator!=(const ProfileFuncRef &LHS, const ProfileFuncRef &RHS) {
+ return !LHS.equals(RHS);
+}
+
+inline bool operator<(const ProfileFuncRef &LHS, const ProfileFuncRef &RHS) {
+ return LHS.compare(RHS) < 0;
+}
+
+inline bool operator<=(const ProfileFuncRef &LHS, const ProfileFuncRef &RHS) {
+ return LHS.compare(RHS) <= 0;
+}
+
+inline bool operator>(const ProfileFuncRef &LHS, const ProfileFuncRef &RHS) {
+ return LHS.compare(RHS) > 0;
+}
+
+inline bool operator>=(const ProfileFuncRef &LHS, const ProfileFuncRef &RHS) {
+ return LHS.compare(RHS) >= 0;
+}
+
+inline raw_ostream &operator<<(raw_ostream &OS, const ProfileFuncRef &Obj) {
+ if (Obj.Data)
+ return OS << StringRef(Obj.Data, Obj.Length);
+ if (Obj.Length != 0)
+ return OS << Obj.Length;
+ return OS;
+}
+
+inline uint64_t MD5Hash(const sampleprof::ProfileFuncRef &Obj) {
+ return Obj.getHashCode();
+}
+
+inline hash_code hash_value(const sampleprof::ProfileFuncRef &Obj) {
+ return Obj.getHashCode();
+}
+
+} // end namespace sampleprof
+
+/// Template specialization for ProfileFuncName so that it can be used in LLVM
+/// map containers.
+template <> struct DenseMapInfo<sampleprof::ProfileFuncRef, void> {
+
+ static inline sampleprof::ProfileFuncRef getEmptyKey() {
+ return sampleprof::ProfileFuncRef(~0ULL);
+ }
+
+ static inline sampleprof::ProfileFuncRef getTombstoneKey() {
+ return sampleprof::ProfileFuncRef(~1ULL);
+ }
+
+ static unsigned getHashValue(const sampleprof::ProfileFuncRef &Val) {
+ return Val.getHashCode();
+ }
+
+ static bool isEqual(const sampleprof::ProfileFuncRef &LHS,
+ const sampleprof::ProfileFuncRef &RHS) {
+ return LHS == RHS;
+ }
+};
+
+} // end namespace llvm
+
+namespace std {
+
+/// Template specialization for ProfileFuncName so that it can be used in STL
+/// containers.
+template <> struct hash<llvm::sampleprof::ProfileFuncRef> {
+ size_t operator()(const llvm::sampleprof::ProfileFuncRef &Val) const {
+ return Val.getHashCode();
+ }
+};
+
+} // end namespace std
+
+#endif // LLVM_PROFILEDATA_PROFILEFUNCREF_H
diff --git a/llvm/include/llvm/ProfileData/SampleProf.h b/llvm/include/llvm/ProfileData/SampleProf.h
index 9b2416e39b6cc9a..cae38de7dfb9835 100644
--- a/llvm/include/llvm/ProfileData/SampleProf.h
+++ b/llvm/include/llvm/ProfileData/SampleProf.h
@@ -21,6 +21,7 @@
#include "llvm/ADT/StringRef.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/GlobalValue.h"
+#include "llvm/ProfileData/ProfileFuncRef.h"
#include "llvm/Support/Allocator.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/ErrorOr.h"
@@ -111,12 +112,10 @@ static inline uint64_t SPMagic(SampleProfileFormat Format = SPF_Binary) {
/// Get the proper representation of a string according to whether the
/// current Format uses MD5 to represent the string.
-static inline StringRef getRepInFormat(StringRef Name, bool UseMD5,
- std::string &GUIDBuf) {
+static inline ProfileFuncRef getRepInFormat(StringRef Name, bool UseMD5) {
if (Name.empty() || !UseMD5)
- return Name;
- GUIDBuf = std::to_string(Function::getGUID(Name));
- return GUIDBuf;
+ return ProfileFuncRef(Name);
+ return ProfileFuncRef(Function::getGUID(Name));
}
static inline uint64_t SPVersion() { return 103; }
@@ -305,6 +304,10 @@ struct LineLocation {
return LineOffset != O.LineOffset || Discriminator != O.Discriminator;
}
+ uint64_t getHashCode() const {
+ return ((uint64_t) Discriminator << 32) | LineOffset;
+ }
+
uint32_t LineOffset;
uint32_t Discriminator;
};
@@ -318,14 +321,6 @@ struct LineLocationHash {
raw_ostream &operator<<(raw_ostream &OS, const LineLocation &Loc);
-static inline uint64_t hashFuncName(StringRef F) {
- // If function name is already MD5 string, do not hash again.
- uint64_t Hash;
- if (F.getAsInteger(10, Hash))
- Hash = MD5Hash(F);
- return Hash;
-}
-
/// Representation of a single sample record.
///
/// A sample record is represented by a positive integer value, which
@@ -338,7 +333,7 @@ static inline uint64_t hashFuncName(StringRef F) {
/// will be a list of one or more functions.
class SampleRecord {
public:
- using CallTarget = std::pair<StringRef, uint64_t>;
+ using CallTarget = std::pair<ProfileFuncRef, uint64_t>;
struct CallTargetComparator {
bool operator()(const CallTarget &LHS, const CallTarget &RHS) const {
if (LHS.second != RHS.second)
@@ -349,7 +344,7 @@ class SampleRecord {
};
using SortedCallTargetSet = std::set<CallTarget, CallTargetComparator>;
- using CallTargetMap = StringMap<uint64_t>;
+ using CallTargetMap = std::unordered_map<ProfileFuncRef, uint64_t>;
SampleRecord() = default;
/// Increment the number of samples for this record by \p S.
@@ -378,7 +373,7 @@ class SampleRecord {
///
/// Sample counts accumulate using saturating arithmetic, to avoid wrapping
/// around unsigned integers.
- sampleprof_error addCalledTarget(StringRef F, uint64_t S,
+ sampleprof_error addCalledTarget(ProfileFuncRef F, uint64_t S,
uint64_t Weight = 1) {
uint64_t &TargetSamples = CallTargets[F];
bool Overflowed;
@@ -390,7 +385,7 @@ class SampleRecord {
/// Remove called function from the call target map. Return the target sample
/// count of the called function.
- uint64_t removeCalledTarget(StringRef F) {
+ uint64_t removeCalledTarget(ProfileFuncRef F) {
uint64_t Count = 0;
auto I = CallTargets.find(F);
if (I != CallTargets.end()) {
@@ -476,12 +471,12 @@ enum ContextAttributeMask {
// Represents a context frame with function name and line location
struct SampleContextFrame {
- StringRef FuncName;
+ ProfileFuncRef FuncName;
LineLocation Location;
SampleContextFrame() : Location(0, 0) {}
- SampleContextFrame(StringRef FuncName, LineLocation Location)
+ SampleContextFrame(ProfileFuncRef FuncName, LineLocation Location)
: FuncName(FuncName), Location(Location) {}
bool operator==(const SampleContextFrame &That) const {
@@ -502,6 +497,12 @@ struct SampleContextFrame {
}
return OContextStr.str();
}
+
+ uint64_t getHashCode() const {
+ uint64_t NameHash = FuncName.getHashCode();
+ uint64_t LocId = Location.getHashCode();
+ return NameHash + (LocId << 5) + LocId;
+ }
};
static inline hash_code hash_value(const SampleContextFrame &arg) {
@@ -535,6 +536,9 @@ class SampleContext {
SampleContext(StringRef Name)
: Name(Name), State(UnknownContext), Attributes(ContextNone) {}
+ SampleContext(ProfileFuncRef Name)
+ : Name(Name), State(UnknownContext), Attributes(ContextNone) {}
+
SampleContext(SampleContextFrames Context,
ContextStateMask CState = RawContext)
: Attributes(ContextNone) {
@@ -555,7 +559,7 @@ class SampleContext {
bool HasContext = ContextStr.startswith("[");
if (!HasContext) {
State = UnknownContext;
- Name = ContextStr;
+ Name = ProfileFuncRef(ContextStr);
} else {
CSNameTable.emplace_back();
SampleContextFrameVector &Context = CSNameTable.back();
@@ -572,7 +576,7 @@ class SampleContext {
ContextStr = ContextStr.substr(1, ContextStr.size() - 2);
StringRef ContextRemain = ContextStr;
StringRef ChildContext;
- StringRef CalleeName;
+ ProfileFuncRef CalleeName;
while (!ContextRemain.empty()) {
auto ContextSplit = ContextRemain.split(" @ ");
ChildContext = ContextSplit.first;
@@ -585,11 +589,12 @@ class SampleContext {
// Decode context string for a frame to get function name and location.
// `ContextStr` is in the form of `FuncName:StartLine.Discriminator`.
- static void decodeContextString(StringRef ContextStr, StringRef &FName,
+ static void decodeContextString(StringRef ContextStr,
+ ProfileFuncRef &FName,
LineLocation &LineLoc) {
// Get function name
auto EntrySplit = ContextStr.split(':');
- FName = EntrySplit.first;
+ FName = ProfileFuncRef(EntrySplit.first);
LineLoc = {0, 0};
if (!EntrySplit.second.empty()) {
@@ -616,7 +621,7 @@ class SampleContext {
void clearState(ContextStateMask S) { State &= (uint32_t)~S; }
bool hasContext() const { return State != UnknownContext; }
bool isBaseContext() const { return FullContext.size() == 1; }
- StringRef getName() const { return Name; }
+ ProfileFuncRef getName() const { return Name; }
SampleContextFrames getContextFrames() const { return FullContext; }
static std::string getContextString(SampleContextFrames Context,
@@ -641,14 +646,11 @@ class SampleContext {
uint64_t getHashCode() const {
if (hasContext())
return hash_value(getContextFrames());
-
- // For non-context function name, use its MD5 as hash value, so that it is
- // consistent with the profile map's key.
- return hashFuncName(getName());
+ return getName().getHashCode();
}
/// Set the name of the function and clear the current context.
- void setName(StringRef FunctionName) {
+ void setName(ProfileFuncRef FunctionName) {
Name = FunctionName;
FullContext = SampleContextFrames();
State = UnknownContext;
@@ -713,7 +715,7 @@ class SampleContext {
private:
/// Mangled name of the function.
- StringRef Name;
+ ProfileFuncRef Name;
// Full context including calling context and leaf function name
SampleContextFrames FullContext;
// State of the associated sample profile
@@ -736,7 +738,7 @@ class SampleProfileReaderItaniumRemapper;
using BodySampleMap = std::map<LineLocation, SampleRecord>;
// NOTE: Using a StringMap here makes parsed profiles consume around 17% more
// memory, which is *very* significant for large profiles.
-using FunctionSamplesMap = std::map<std::string, FunctionSamples, std::less<>>;
+using FunctionSamplesMap = std::map<ProfileFuncRef, FunctionSamples>;
using CallsiteSampleMap = std::map<LineLocation, FunctionSamplesMap>;
using LocToLocMap =
std::unordered_map<LineLocation, LineLocation, LineLocationHash>;
@@ -788,14 +790,16 @@ class FunctionSamples {
sampleprof_error addCalledTargetSamples(uint32_t LineOffset,
uint32_t Discriminator,
- StringRef FName, uint64_t Num,
+ ProfileFuncRef FName,
+ uint64_t Num,
uint64_t Weight = 1) {
return BodySamples[LineLocation(LineOffset, Discriminator)].addCalledTarget(
FName, Num, Weight);
}
sampleprof_error addSampleRecord(LineLocation Location,
- const SampleRecord &SampleRecord, uint64_t Weight = 1) {
+ const SampleRecord &SampleRecord,
+ uint64_t Weight = 1) {
return BodySamples[Location].merge(SampleRecord, Weight);
}
@@ -803,7 +807,7 @@ class FunctionSamples {
// the number of body samples actually decreased.
uint64_t removeCalledTargetAndBodySample(uint32_t LineOffset,
uint32_t Discriminator,
- StringRef FName) {
+ ProfileFuncRef FName) {
uint64_t Count = 0;
auto I = BodySamples.find(LineLocation(LineOffset, Discriminator));
if (I != BodySamples.end()) {
@@ -1046,16 +1050,16 @@ class FunctionSamples {
};
if (isDeclaration(SymbolMap.lookup(getFuncName()))) {
// Add to the import list only when it's defined out of module.
- S.insert(getGUID(getName()));
+ S.insert(getGUID());
}
// Import hot CallTargets, which may not be available in IR because full
// profile annotation cannot be done until backend compilation in ThinLTO.
for (const auto &BS : BodySamples)
for (const auto &TS : BS.second.getCallTargets())
- if (TS.getValue() > Threshold) {
- const Function *Callee = SymbolMap.lookup(getFuncName(TS.getKey()));
+ if (TS.second > Threshold) {
+ const Function *Callee = SymbolMap.lookup(getFuncName(TS.first));
if (isDeclaration(Callee))
- S.insert(getGUID(TS.getKey()));
+ S.insert(TS.first.getHashCode());
}
for (const auto &CS : CallsiteSamples)
for (const auto &NameFS : CS.second)
@@ -1063,10 +1067,12 @@ class FunctionSamples {
}
/// Set the name of the function.
- void setName(StringRef FunctionName) { Context.setName(FunctionName); }
+ void setName(ProfileFuncRef FunctionName) {
+ Context.setName(FunctionName);
+ }
/// Return the function name.
- StringRef getName() const { return Context.getName(); }
+ ProfileFuncRef getName() const { return Context.getName(); }
/// Return the original function name.
StringRef getFuncName() const { return getFuncName(getName()); }
@@ -1133,12 +1139,12 @@ class FunctionSamples {
/// translate \p Name in current FunctionSamples into its original name
/// by looking up in the function map GUIDToFuncNameMap.
/// If the original name doesn't exist in the map, return empty StringRef.
- StringRef getFuncName(StringRef Name) const {
+ StringRef getFuncName(ProfileFuncRef Name) const {
if (!UseMD5)
- return Name;
+ return reinterpret_cast<StringRef &>(Name);
assert(GUIDToFuncNameMap && "GUIDToFuncNameMap needs to be populated first");
- return GUIDToFuncNameMap->lookup(std::stoull(Name.data()));
+ return GUIDToFuncNameMap->lookup(Name.getHashCode());
}
/// Returns the line offset to the start line of the subprogram.
@@ -1154,7 +1160,9 @@ class FunctionSamples {
/// Returns a unique hash code for a combination of a callsite location and
/// the callee function name.
- static uint64_t getCallSiteHash(StringRef CalleeName,
+ /// Guarantee MD5 and non-MD5 representation of the same function results in
+ /// the same hash.
+ static uint64_t getCallSiteHash(ProfileFuncRef CalleeName,
const LineLocation &Callsite);
/// Get the FunctionSamples of the inline instance where DIL originates
@@ -1195,16 +1203,15 @@ class FunctionSamples {
/// all the function symbols defined or declared in current module.
DenseMap<uint64_t, StringRef> *GUIDToFuncNameMap = nullptr;
- // Assume the input \p Name is a name coming from FunctionSamples itself.
- // If UseMD5 is true, the name is already a GUID and we
- // don't want to return the GUID of GUID.
- static uint64_t getGUID(StringRef Name) {
- return UseMD5 ? std::stoull(Name.data()) : Function::getGUID(Name);
+ /// Return the GUID of the context's name. If the context is already using
+ /// MD5, don't hash it again.
+ uint64_t getGUID() const {
+ return getName().getHashCode();
}
// Find all the names in the current FunctionSamples including names in
// all the inline instances and names of call targets.
- void findAllNames(DenseSet<StringRef> &NameSet) const;
+ void findAllNames(DenseSet<ProfileFuncRef> &NameSet) const;
bool operator==(const FunctionSamples &Other) const {
return (GUIDToFuncNameMap == Other.GUIDToFuncNameMap ||
@@ -1221,9 +1228,6 @@ class FunctionSamples {
return !(*this == Other);
}
- template <typename T>
- const T &getKey() const;
-
private:
/// CFG hash value for the function.
uint64_t FunctionHash = 0;
@@ -1287,9 +1291,8 @@ class FunctionSamples {
const LocToLocMap *IRToProfileLocationMap = nullptr;
};
-template <>
-inline const SampleContext &FunctionSamples::getKey<SampleContext>() const {
- return getContext();
+static inline ProfileFuncRef getRepInFormat(StringRef Name) {
+ return getRepInFormat(Name, FunctionSamples::UseMD5);
}
raw_ostream &operator<<(raw_ostream &OS, const FunctionSamples &FS);
@@ -1398,11 +1401,6 @@ class SampleProfileMap
Ctx);
}
- // Overloaded find() to lookup a function by name.
- iterator find(StringRef Fname) {
- return base_type::find(hashFuncName(Fname));
- }
-
size_t erase(const SampleContext &Ctx) {
return HashKeyMap<std::unordered_map, SampleContext, FunctionSamples>::
erase(Ctx);
@@ -1473,7 +1471,7 @@ class ProfileConverter {
// profile.
void convertCSProfiles();
struct FrameNode {
- FrameNode(StringRef FName = StringRef(),
+ FrameNode(ProfileFuncRef FName = ProfileFuncRef(),
FunctionSamples *FSamples = nullptr,
LineLocation CallLoc = {0, 0})
: FuncName(FName), FuncSamples(FSamples), CallSiteLoc(CallLoc){};
@@ -1481,14 +1479,14 @@ class ProfileConverter {
// Map line+discriminator location to child frame
std::map<uint64_t, FrameNode> AllChildFrames;
// Function name for current frame
- StringRef FuncName;
+ ProfileFuncRef FuncName;
// Function Samples for current frame
FunctionSamples *FuncSamples;
// Callsite location in parent context
LineLocation CallSiteLoc;
FrameNode *getOrCreateChildFrame(const LineLocation &CallSite,
- StringRef CalleeName);
+ ProfileFuncRef CalleeName);
};
static void flattenProfile(SampleProfileMap &ProfileMap,
@@ -1623,7 +1621,9 @@ using namespace sampleprof;
template <> struct DenseMapInfo<SampleContext> {
static inline SampleContext getEmptyKey() { return SampleContext(); }
- static inline SampleContext getTombstoneKey() { return SampleContext("@"); }
+ static inline SampleContext getTombstoneKey() {
+ return SampleContext(ProfileFuncRef(~1ULL));
+ }
static unsigned getHashValue(const SampleContext &Val) {
return Val.getHashCode();
diff --git a/llvm/include/llvm/ProfileData/SampleProfReader.h b/llvm/include/llvm/ProfileData/SampleProfReader.h
index a5b1df3ef550b2b..df30722cb1b5bc3 100644
--- a/llvm/include/llvm/ProfileData/SampleProfReader.h
+++ b/llvm/include/llvm/ProfileData/SampleProfReader.h
@@ -409,13 +409,13 @@ class SampleProfileReader {
/// Return the samples collected for function \p F.
FunctionSamples *getSamplesFor(StringRef Fname) {
- auto It = Profiles.find(Fname);
+ auto It = Profiles.find(ProfileFuncRef(Fname));
if (It != Profiles.end())
return &It->second;
if (Remapper) {
if (auto NameInProfile = Remapper->lookUpNameInProfile(Fname)) {
- auto It = Profiles.find(*NameInProfile);
+ auto It = Profiles.find(ProfileFuncRef(*NameInProfile));
if (It != Profiles.end())
return &It->second;
}
@@ -474,7 +474,7 @@ class SampleProfileReader {
/// It includes all the names that have samples either in outline instance
/// or inline instance.
- virtual std::vector<StringRef> *getNameTable() { return nullptr; }
+ virtual std::vector<ProfileFuncRef> *getNameTable() { return nullptr; }
virtual bool dumpSectionInfo(raw_ostream &OS = dbgs()) { return false; };
/// Return whether names in the profile are all MD5 numbers.
@@ -508,10 +508,6 @@ class SampleProfileReader {
/// Memory buffer holding the profile file.
std::unique_ptr<MemoryBuffer> Buffer;
- /// Extra name buffer holding names created on demand.
- /// This should only be needed for md5 profiles.
- std::unordered_set<std::string> MD5NameBuffer;
-
/// Profile summary information.
std::unique_ptr<ProfileSummary> Summary;
@@ -595,7 +591,9 @@ class SampleProfileReaderBinary : public SampleProfileReader {
/// It includes all the names that have samples either in outline instance
/// or inline instance.
- std::vector<StringRef> *getNameTable() override { return &NameTable; }
+ std::vector<ProfileFuncRef> *getNameTable() override {
+ return &NameTable;
+ }
protected:
/// Read a numeric value of type T from the profile.
@@ -637,7 +635,7 @@ class SampleProfileReaderBinary : public SampleProfileReader {
std::error_code readNameTable();
/// Read a string indirectly via the name table. Optionally return the index.
- ErrorOr<StringRef> readStringFromTable(size_t *RetIdx = nullptr);
+ ErrorOr<ProfileFuncRef> readStringFromTable(size_t *RetIdx = nullptr);
/// Read a context indirectly via the CSNameTable. Optionally return the
/// index.
@@ -654,16 +652,7 @@ class SampleProfileReaderBinary : public SampleProfileReader {
const uint8_t *End = nullptr;
/// Function name table.
- std::vector<StringRef> NameTable;
-
- /// If MD5 is used in NameTable section, the section saves uint64_t data.
- /// The uint64_t data has to be converted to a string and then the string
- /// will be used to initialize StringRef in NameTable.
- /// Note NameTable contains StringRef so it needs another buffer to own
- /// the string data. MD5StringBuf serves as the string buffer that is
- /// referenced by NameTable (vector of StringRef). We make sure
- /// the lifetime of MD5StringBuf is not shorter than that of NameTable.
- std::vector<std::string> MD5StringBuf;
+ std::vector<ProfileFuncRef> NameTable;
/// The starting address of fixed length MD5 name table section.
const uint8_t *MD5NameMemStart = nullptr;
diff --git a/llvm/include/llvm/ProfileData/SampleProfWriter.h b/llvm/include/llvm/ProfileData/SampleProfWriter.h
index 1f19283ea1dd0a8..3eff17e5e5f0900 100644
--- a/llvm/include/llvm/ProfileData/SampleProfWriter.h
+++ b/llvm/include/llvm/ProfileData/SampleProfWriter.h
@@ -196,20 +196,20 @@ class SampleProfileWriterBinary : public SampleProfileWriter {
std::error_code writeSample(const FunctionSamples &S) override;
protected:
- virtual MapVector<StringRef, uint32_t> &getNameTable() { return NameTable; }
+ virtual MapVector<ProfileFuncRef, uint32_t> &getNameTable() { return NameTable; }
virtual std::error_code writeMagicIdent(SampleProfileFormat Format);
virtual std::error_code writeNameTable();
std::error_code writeHeader(const SampleProfileMap &ProfileMap) override;
std::error_code writeSummary();
virtual std::error_code writeContextIdx(const SampleContext &Context);
- std::error_code writeNameIdx(StringRef FName);
+ std::error_code writeNameIdx(ProfileFuncRef FName);
std::error_code writeBody(const FunctionSamples &S);
- inline void stablizeNameTable(MapVector<StringRef, uint32_t> &NameTable,
- std::set<StringRef> &V);
+ inline void stablizeNameTable(MapVector<ProfileFuncRef, uint32_t> &NameTable,
+ std::set<ProfileFuncRef> &V);
- MapVector<StringRef, uint32_t> NameTable;
+ MapVector<ProfileFuncRef, uint32_t> NameTable;
- void addName(StringRef FName);
+ void addName(ProfileFuncRef FName);
virtual void addContext(const SampleContext &Context);
void addNames(const FunctionSamples &S);
diff --git a/llvm/include/llvm/Transforms/IPO/ProfiledCallGraph.h b/llvm/include/llvm/Transforms/IPO/ProfiledCallGraph.h
index bc8360a80bc02bd..b575058614211e5 100644
--- a/llvm/include/llvm/Transforms/IPO/ProfiledCallGraph.h
+++ b/llvm/include/llvm/Transforms/IPO/ProfiledCallGraph.h
@@ -53,9 +53,10 @@ struct ProfiledCallGraphNode {
using iterator = edges::iterator;
using const_iterator = edges::const_iterator;
- ProfiledCallGraphNode(StringRef FName = StringRef()) : Name(FName) {}
+ ProfiledCallGraphNode(ProfileFuncRef FName = ProfileFuncRef()) : Name(FName)
+ {}
- StringRef Name;
+ ProfileFuncRef Name;
edges Edges;
};
@@ -85,7 +86,7 @@ class ProfiledCallGraph {
std::queue<ContextTrieNode *> Queue;
for (auto &Child : ContextTracker.getRootContext().getAllChildContext()) {
ContextTrieNode *Callee = &Child.second;
- addProfiledFunction(ContextTracker.getFuncNameFor(Callee));
+ addProfiledFunction(Callee->getFuncName());
Queue.push(Callee);
}
@@ -102,7 +103,7 @@ class ProfiledCallGraph {
// context-based one, which may in turn block context-based inlining.
for (auto &Child : Caller->getAllChildContext()) {
ContextTrieNode *Callee = &Child.second;
- addProfiledFunction(ContextTracker.getFuncNameFor(Callee));
+ addProfiledFunction(Callee->getFuncName());
Queue.push(Callee);
// Fetch edge weight from the profile.
@@ -123,8 +124,7 @@ class ProfiledCallGraph {
Weight = std::max(CallsiteCount, CalleeEntryCount);
}
- addProfiledCall(ContextTracker.getFuncNameFor(Caller),
- ContextTracker.getFuncNameFor(Callee), Weight);
+ addProfiledCall(Caller->getFuncName(), Callee->getFuncName(), Weight);
}
}
@@ -137,7 +137,7 @@ class ProfiledCallGraph {
iterator end() { return Root.Edges.end(); }
ProfiledCallGraphNode *getEntryNode() { return &Root; }
- void addProfiledFunction(StringRef Name) {
+ void addProfiledFunction(ProfileFuncRef Name) {
if (!ProfiledFunctions.count(Name)) {
// Link to synthetic root to make sure every node is reachable
// from root. This does not affect SCC order.
@@ -147,7 +147,7 @@ class ProfiledCallGraph {
}
private:
- void addProfiledCall(StringRef CallerName, StringRef CalleeName,
+ void addProfiledCall(ProfileFuncRef CallerName, ProfileFuncRef CalleeName,
uint64_t Weight = 0) {
assert(ProfiledFunctions.count(CallerName));
auto CalleeIt = ProfiledFunctions.find(CalleeName);
@@ -168,19 +168,19 @@ class ProfiledCallGraph {
}
void addProfiledCalls(const FunctionSamples &Samples) {
- addProfiledFunction(Samples.getFuncName());
+ addProfiledFunction(Samples.getName());
for (const auto &Sample : Samples.getBodySamples()) {
for (const auto &[Target, Frequency] : Sample.second.getCallTargets()) {
addProfiledFunction(Target);
- addProfiledCall(Samples.getFuncName(), Target, Frequency);
+ addProfiledCall(Samples.getName(), Target, Frequency);
}
}
for (const auto &CallsiteSamples : Samples.getCallsiteSamples()) {
for (const auto &InlinedSamples : CallsiteSamples.second) {
addProfiledFunction(InlinedSamples.first);
- addProfiledCall(Samples.getFuncName(), InlinedSamples.first,
+ addProfiledCall(Samples.getName(), InlinedSamples.first,
InlinedSamples.second.getHeadSamplesEstimate());
addProfiledCalls(InlinedSamples.second);
}
@@ -206,7 +206,7 @@ class ProfiledCallGraph {
}
ProfiledCallGraphNode Root;
- StringMap<ProfiledCallGraphNode> ProfiledFunctions;
+ std::unordered_map<ProfileFuncRef, ProfiledCallGraphNode> ProfiledFunctions;
};
} // end namespace sampleprof
diff --git a/llvm/include/llvm/Transforms/IPO/SampleContextTracker.h b/llvm/include/llvm/Transforms/IPO/SampleContextTracker.h
index 347dac1e9684db1..270dd25672b9f36 100644
--- a/llvm/include/llvm/Transforms/IPO/SampleContextTracker.h
+++ b/llvm/include/llvm/Transforms/IPO/SampleContextTracker.h
@@ -35,20 +35,20 @@ class Instruction;
class ContextTrieNode {
public:
ContextTrieNode(ContextTrieNode *Parent = nullptr,
- StringRef FName = StringRef(),
+ ProfileFuncRef FName = ProfileFuncRef(),
FunctionSamples *FSamples = nullptr,
LineLocation CallLoc = {0, 0})
: ParentContext(Parent), FuncName(FName), FuncSamples(FSamples),
CallSiteLoc(CallLoc){};
ContextTrieNode *getChildContext(const LineLocation &CallSite,
- StringRef ChildName);
+ ProfileFuncRef ChildName);
ContextTrieNode *getHottestChildContext(const LineLocation &CallSite);
ContextTrieNode *getOrCreateChildContext(const LineLocation &CallSite,
- StringRef ChildName,
+ ProfileFuncRef ChildName,
bool AllowCreate = true);
- void removeChildContext(const LineLocation &CallSite, StringRef ChildName);
+ void removeChildContext(const LineLocation &CallSite, ProfileFuncRef ChildName);
std::map<uint64_t, ContextTrieNode> &getAllChildContext();
- StringRef getFuncName() const;
+ ProfileFuncRef getFuncName() const;
FunctionSamples *getFunctionSamples() const;
void setFunctionSamples(FunctionSamples *FSamples);
std::optional<uint32_t> getFunctionSize() const;
@@ -68,7 +68,7 @@ class ContextTrieNode {
ContextTrieNode *ParentContext;
// Function name for current context
- StringRef FuncName;
+ ProfileFuncRef FuncName;
// Function Samples for current context
FunctionSamples *FuncSamples;
@@ -118,7 +118,8 @@ class SampleContextTracker {
FunctionSamples *getBaseSamplesFor(const Function &Func,
bool MergeContext = true);
// Query base profile for a given function by name.
- FunctionSamples *getBaseSamplesFor(StringRef Name, bool MergeContext = true);
+ FunctionSamples *getBaseSamplesFor(ProfileFuncRef Name,
+ bool MergeContext = true);
// Retrieve the context trie node for given profile context
ContextTrieNode *getContextFor(const SampleContext &Context);
// Get real function name for a given trie node.
@@ -129,7 +130,7 @@ class SampleContextTracker {
void markContextSamplesInlined(const FunctionSamples *InlinedSamples);
ContextTrieNode &getRootContext();
void promoteMergeContextSamplesTree(const Instruction &Inst,
- StringRef CalleeName);
+ ProfileFuncRef CalleeName);
// Create a merged conext-less profile map.
void createContextLessProfileMap(SampleProfileMap &ContextLessProfiles);
@@ -140,7 +141,7 @@ class SampleContextTracker {
return nullptr;
return I->second;
}
- StringMap<ContextSamplesTy> &getFuncToCtxtProfiles() {
+ std::unordered_map<ProfileFuncRef, ContextSamplesTy> &getFuncToCtxtProfiles() {
return FuncToCtxtProfiles;
}
@@ -189,9 +190,9 @@ class SampleContextTracker {
private:
ContextTrieNode *getContextFor(const DILocation *DIL);
ContextTrieNode *getCalleeContextFor(const DILocation *DIL,
- StringRef CalleeName);
- ContextTrieNode *getTopLevelContextNode(StringRef FName);
- ContextTrieNode &addTopLevelContextNode(StringRef FName);
+ ProfileFuncRef CalleeName);
+ ContextTrieNode *getTopLevelContextNode(ProfileFuncRef FName);
+ ContextTrieNode &addTopLevelContextNode(ProfileFuncRef FName);
ContextTrieNode &promoteMergeContextSamplesTree(ContextTrieNode &NodeToPromo);
void mergeContextNode(ContextTrieNode &FromNode, ContextTrieNode &ToNode);
ContextTrieNode &
@@ -204,7 +205,7 @@ class SampleContextTracker {
ProfileToNodeMap[FSample] = Node;
}
// Map from function name to context profiles (excluding base profile)
- StringMap<ContextSamplesTy> FuncToCtxtProfiles;
+ std::unordered_map<ProfileFuncRef, ContextSamplesTy> FuncToCtxtProfiles;
// Map from current FunctionSample to the belonged context trie.
std::unordered_map<const FunctionSamples *, ContextTrieNode *>
diff --git a/llvm/lib/ProfileData/SampleProf.cpp b/llvm/lib/ProfileData/SampleProf.cpp
index b14dc01be236245..a3de3c8e7400f0a 100644
--- a/llvm/lib/ProfileData/SampleProf.cpp
+++ b/llvm/lib/ProfileData/SampleProf.cpp
@@ -121,7 +121,7 @@ sampleprof_error SampleRecord::merge(const SampleRecord &Other,
sampleprof_error Result;
Result = addSamples(Other.getSamples(), Weight);
for (const auto &I : Other.getCallTargets()) {
- MergeResult(Result, addCalledTarget(I.first(), I.second, Weight));
+ MergeResult(Result, addCalledTarget(I.first, I.second, Weight));
}
return Result;
}
@@ -234,9 +234,9 @@ LineLocation FunctionSamples::getCallSiteIdentifier(const DILocation *DIL,
}
}
-uint64_t FunctionSamples::getCallSiteHash(StringRef CalleeName,
+uint64_t FunctionSamples::getCallSiteHash(ProfileFuncRef CalleeName,
const LineLocation &Callsite) {
- uint64_t NameHash = std::hash<std::string>{}(CalleeName.str());
+ uint64_t NameHash = CalleeName.getHashCode();
uint64_t LocId =
(((uint64_t)Callsite.LineOffset) << 32) | Callsite.Discriminator;
return NameHash + (LocId << 5) + LocId;
@@ -268,11 +268,11 @@ const FunctionSamples *FunctionSamples::findFunctionSamples(
return FS;
}
-void FunctionSamples::findAllNames(DenseSet<StringRef> &NameSet) const {
+void FunctionSamples::findAllNames(DenseSet<ProfileFuncRef> &NameSet) const {
NameSet.insert(getName());
for (const auto &BS : BodySamples)
for (const auto &TS : BS.second.getCallTargets())
- NameSet.insert(TS.getKey());
+ NameSet.insert(TS.first);
for (const auto &CS : CallsiteSamples) {
for (const auto &NameFS : CS.second) {
@@ -287,18 +287,15 @@ const FunctionSamples *FunctionSamples::findFunctionSamplesAt(
SampleProfileReaderItaniumRemapper *Remapper) const {
CalleeName = getCanonicalFnName(CalleeName);
- std::string CalleeGUID;
- CalleeName = getRepInFormat(CalleeName, UseMD5, CalleeGUID);
-
auto iter = CallsiteSamples.find(mapIRLocToProfileLoc(Loc));
if (iter == CallsiteSamples.end())
return nullptr;
- auto FS = iter->second.find(CalleeName);
+ auto FS = iter->second.find(getRepInFormat(CalleeName));
if (FS != iter->second.end())
return &FS->second;
if (Remapper) {
if (auto NameInProfile = Remapper->lookUpNameInProfile(CalleeName)) {
- auto FS = iter->second.find(*NameInProfile);
+ auto FS = iter->second.find(getRepInFormat(*NameInProfile));
if (FS != iter->second.end())
return &FS->second;
}
@@ -422,7 +419,7 @@ void ProfileSymbolList::dump(raw_ostream &OS) const {
ProfileConverter::FrameNode *
ProfileConverter::FrameNode::getOrCreateChildFrame(const LineLocation &CallSite,
- StringRef CalleeName) {
+ ProfileFuncRef CalleeName) {
uint64_t Hash = FunctionSamples::getCallSiteHash(CalleeName, CallSite);
auto It = AllChildFrames.find(Hash);
if (It != AllChildFrames.end()) {
@@ -468,7 +465,7 @@ void ProfileConverter::convertCSProfiles(ProfileConverter::FrameNode &Node) {
if (!ChildProfile)
continue;
SampleContext OrigChildContext = ChildProfile->getContext();
- hash_code OrigChildContextHash = OrigChildContext.getHashCode();
+ uint64_t OrigChildContextHash = OrigChildContext.getHashCode();
// Reset the child context to be contextless.
ChildProfile->getContext().setName(OrigChildContext.getName());
if (NodeProfile) {
@@ -484,7 +481,7 @@ void ProfileConverter::convertCSProfiles(ProfileConverter::FrameNode &Node) {
NodeProfile->removeTotalSamples(Count);
}
- hash_code NewChildProfileHash(0);
+ uint64_t NewChildProfileHash = 0;
// Separate child profile to be a standalone profile, if the current parent
// profile doesn't exist. This is a duplicating operation when the child
// profile is already incorporated into the parent which is still useful and
@@ -498,7 +495,7 @@ void ProfileConverter::convertCSProfiles(ProfileConverter::FrameNode &Node) {
ProfileMap[ChildProfile->getContext()].merge(*ChildProfile);
NewChildProfileHash = ChildProfile->getContext().getHashCode();
auto &SamplesMap = NodeProfile->functionSamplesAt(ChildNode.CallSiteLoc);
- SamplesMap[ChildProfile->getName().str()].getContext().setAttribute(
+ SamplesMap[ChildProfile->getName()].getContext().setAttribute(
ContextDuplicatedIntoBase);
}
diff --git a/llvm/lib/ProfileData/SampleProfReader.cpp b/llvm/lib/ProfileData/SampleProfReader.cpp
index 6d355cdc41840b4..86306c14a048da4 100644
--- a/llvm/lib/ProfileData/SampleProfReader.cpp
+++ b/llvm/lib/ProfileData/SampleProfReader.cpp
@@ -91,7 +91,7 @@ static void dumpFunctionProfileJson(const FunctionSamples &S,
JOS.attributeArray("calls", [&] {
for (const auto &J : CallTargets) {
JOS.object([&] {
- JOS.attribute("function", J.first);
+ JOS.attribute("function", J.first.str());
JOS.attribute("samples", J.second);
});
}
@@ -117,7 +117,7 @@ static void dumpFunctionProfileJson(const FunctionSamples &S,
};
JOS.object([&] {
- JOS.attribute("name", S.getName());
+ JOS.attribute("name", S.getName().str());
JOS.attribute("total", S.getTotalSamples());
if (TopLevel)
JOS.attribute("head", S.getHeadSamples());
@@ -392,8 +392,8 @@ std::error_code SampleProfileReaderText::readImpl() {
switch (LineTy) {
case LineType::CallSiteProfile: {
FunctionSamples &FSamples = InlineStack.back()->functionSamplesAt(
- LineLocation(LineOffset, Discriminator))[std::string(FName)];
- FSamples.setName(FName);
+ LineLocation(LineOffset, Discriminator))[ProfileFuncRef(FName)];
+ FSamples.setName(ProfileFuncRef(FName));
MergeResult(Result, FSamples.addTotalSamples(NumSamples));
InlineStack.push_back(&FSamples);
DepthMetadata = 0;
@@ -406,7 +406,8 @@ std::error_code SampleProfileReaderText::readImpl() {
FunctionSamples &FProfile = *InlineStack.back();
for (const auto &name_count : TargetCountMap) {
MergeResult(Result, FProfile.addCalledTargetSamples(
- LineOffset, Discriminator, name_count.first,
+ LineOffset, Discriminator,
+ ProfileFuncRef(name_count.first),
name_count.second));
}
MergeResult(Result, FProfile.addBodySamples(LineOffset, Discriminator,
@@ -523,24 +524,21 @@ inline ErrorOr<size_t> SampleProfileReaderBinary::readStringIndex(T &Table) {
return *Idx;
}
-ErrorOr<StringRef>
+ErrorOr<ProfileFuncRef>
SampleProfileReaderBinary::readStringFromTable(size_t *RetIdx) {
auto Idx = readStringIndex(NameTable);
if (std::error_code EC = Idx.getError())
return EC;
- // Lazy loading, if the string has not been materialized from memory storing
- // MD5 values, then it is default initialized with the null pointer. This can
- // only happen when using fixed length MD5, that bounds check is performed
- // while parsing the name table to ensure MD5NameMemStart points to an array
- // with enough MD5 entries.
- StringRef &SR = NameTable[*Idx];
- if (!SR.data()) {
+ // If using fixed length MD5, just read directly from the pointer to the name
+ // table.
+ ProfileFuncRef &SR = NameTable[*Idx];
+ if (SR.empty()) {
assert(MD5NameMemStart);
using namespace support;
uint64_t FID = endian::read<uint64_t, little, unaligned>(
MD5NameMemStart + (*Idx) * sizeof(uint64_t));
- SR = MD5StringBuf.emplace_back(std::to_string(FID));
+ SR = ProfileFuncRef(FID);
}
if (RetIdx)
*RetIdx = *Idx;
@@ -662,7 +660,7 @@ SampleProfileReaderBinary::readProfile(FunctionSamples &FProfile) {
uint32_t DiscriminatorVal = (*Discriminator) & getDiscriminatorMask();
FunctionSamples &CalleeProfile = FProfile.functionSamplesAt(
- LineLocation(*LineOffset, DiscriminatorVal))[std::string(*FName)];
+ LineLocation(*LineOffset, DiscriminatorVal))[*FName];
CalleeProfile.setName(*FName);
if (std::error_code EC = readProfile(CalleeProfile))
return EC;
@@ -902,12 +900,16 @@ std::error_code SampleProfileReaderExtBinaryBase::readFuncProfiles() {
for (const auto &NameOffset : FuncOffsetList) {
const auto &FContext = NameOffset.first;
auto FName = FContext.getName();
+ StringRef FNameString;
+ if (!useMD5()) {
+ FNameString = reinterpret_cast<StringRef &>(FName);
+ }
// For function in the current module, keep its farthest ancestor
// context. This can be used to load itself and its child and
// sibling contexts.
- if ((useMD5() && FuncGuidsToUse.count(std::stoull(FName.data()))) ||
- (!useMD5() && (FuncsToUse.count(FName) ||
- (Remapper && Remapper->exist(FName))))) {
+ if ((useMD5() && FuncGuidsToUse.count(FName.getHashCode())) ||
+ (!useMD5() && (FuncsToUse.count(FNameString) ||
+ (Remapper && Remapper->exist(FNameString))))) {
if (!CommonContext || !CommonContext->IsPrefixOf(FContext))
CommonContext = &FContext;
}
@@ -937,7 +939,8 @@ std::error_code SampleProfileReaderExtBinaryBase::readFuncProfiles() {
for (auto NameOffset : FuncOffsetList) {
SampleContext FContext(NameOffset.first);
auto FuncName = FContext.getName();
- if (!FuncsToUse.count(FuncName) && !Remapper->exist(FuncName))
+ StringRef FuncNameStr = reinterpret_cast<StringRef &>(FuncName);
+ if (!FuncsToUse.count(FuncNameStr) && !Remapper->exist(FuncNameStr))
continue;
const uint8_t *FuncProfileAddr = Start + NameOffset.second;
if (std::error_code EC = readFuncProfile(FuncProfileAddr))
@@ -1069,8 +1072,6 @@ std::error_code SampleProfileReaderBinary::readNameTable() {
// tables mixing string and MD5, all of them have to be normalized to use MD5,
// because optimization passes can only handle either type.
bool UseMD5 = useMD5();
- if (UseMD5)
- MD5StringBuf.reserve(MD5StringBuf.size() + *Size);
NameTable.clear();
NameTable.reserve(*Size);
@@ -1089,12 +1090,12 @@ std::error_code SampleProfileReaderBinary::readNameTable() {
if (std::error_code EC = Name.getError())
return EC;
if (UseMD5) {
- uint64_t FID = hashFuncName(*Name);
+ ProfileFuncRef FID(*Name);
if (!ProfileIsCS)
- MD5SampleContextTable.emplace_back(FID);
- NameTable.emplace_back(MD5StringBuf.emplace_back(std::to_string(FID)));
+ MD5SampleContextTable.emplace_back(FID.getHashCode());
+ NameTable.emplace_back(FID);
} else
- NameTable.push_back(*Name);
+ NameTable.push_back(ProfileFuncRef(*Name));
}
if (!ProfileIsCS)
MD5SampleContextStart = MD5SampleContextTable.data();
@@ -1121,7 +1122,6 @@ SampleProfileReaderExtBinaryBase::readNameTableSec(bool IsMD5,
// index has been read before by checking whether the element in the
// NameTable is empty, meanwhile readStringIndex can do the boundary
// check using the size of NameTable.
- MD5StringBuf.reserve(MD5StringBuf.size() + *Size);
NameTable.clear();
NameTable.resize(*Size);
MD5NameMemStart = Data;
@@ -1137,7 +1137,6 @@ SampleProfileReaderExtBinaryBase::readNameTableSec(bool IsMD5,
if (std::error_code EC = Size.getError())
return EC;
- MD5StringBuf.reserve(MD5StringBuf.size() + *Size);
NameTable.clear();
NameTable.reserve(*Size);
if (!ProfileIsCS)
@@ -1148,7 +1147,7 @@ SampleProfileReaderExtBinaryBase::readNameTableSec(bool IsMD5,
return EC;
if (!ProfileIsCS)
support::endian::write64le(&MD5SampleContextTable[I], *FID);
- NameTable.emplace_back(MD5StringBuf.emplace_back(std::to_string(*FID)));
+ NameTable.emplace_back(ProfileFuncRef(*FID));
}
if (!ProfileIsCS)
MD5SampleContextStart = MD5SampleContextTable.data();
@@ -1250,7 +1249,7 @@ SampleProfileReaderExtBinaryBase::readFuncMetadata(bool ProfileHasAttribute,
CalleeProfile = const_cast<FunctionSamples *>(
&FProfile->functionSamplesAt(LineLocation(
*LineOffset,
- *Discriminator))[std::string(FContext.getName())]);
+ *Discriminator))[FContext.getName()]);
}
if (std::error_code EC =
readFuncMetadata(ProfileHasAttribute, CalleeProfile))
@@ -1660,7 +1659,7 @@ std::error_code SampleProfileReaderGCC::readOneFunctionProfile(
// body, there will be identical replicated profiles for the
// original function. In this case, we simply not bother updating
// the profile of the original function.
- FProfile = &Profiles[Name];
+ FProfile = &Profiles[ProfileFuncRef(Name)];
FProfile->addHeadSamples(HeadCount);
if (FProfile->getTotalSamples() > 0)
Update = false;
@@ -1672,9 +1671,9 @@ std::error_code SampleProfileReaderGCC::readOneFunctionProfile(
uint32_t LineOffset = Offset >> 16;
uint32_t Discriminator = Offset & 0xffff;
FProfile = &CallerProfile->functionSamplesAt(
- LineLocation(LineOffset, Discriminator))[std::string(Name)];
+ LineLocation(LineOffset, Discriminator))[ProfileFuncRef(Name)];
}
- FProfile->setName(Name);
+ FProfile->setName(ProfileFuncRef(Name));
for (uint32_t I = 0; I < NumPosCounts; ++I) {
uint32_t Offset;
@@ -1730,7 +1729,8 @@ std::error_code SampleProfileReaderGCC::readOneFunctionProfile(
if (Update)
FProfile->addCalledTargetSamples(LineOffset, Discriminator,
- TargetName, TargetCount);
+ ProfileFuncRef(TargetName),
+ TargetCount);
}
}
@@ -1791,11 +1791,13 @@ void SampleProfileReaderItaniumRemapper::applyRemapping(LLVMContext &Ctx) {
// We will need to remap the entire context string.
assert(Remappings && "should be initialized while creating remapper");
for (auto &Sample : Reader.getProfiles()) {
- DenseSet<StringRef> NamesInSample;
+ DenseSet<ProfileFuncRef> NamesInSample;
Sample.second.findAllNames(NamesInSample);
- for (auto &Name : NamesInSample)
- if (auto Key = Remappings->insert(Name))
- NameMap.insert({Key, Name});
+ for (auto &Name : NamesInSample) {
+ StringRef NameStr = reinterpret_cast<StringRef &>(Name);
+ if (auto Key = Remappings->insert(NameStr))
+ NameMap.insert({Key, NameStr});
+ }
}
RemappingApplied = true;
diff --git a/llvm/lib/ProfileData/SampleProfWriter.cpp b/llvm/lib/ProfileData/SampleProfWriter.cpp
index f418dbe71057095..1d5c811331cc9bf 100644
--- a/llvm/lib/ProfileData/SampleProfWriter.cpp
+++ b/llvm/lib/ProfileData/SampleProfWriter.cpp
@@ -348,7 +348,7 @@ std::error_code SampleProfileWriterExtBinaryBase::writeNameTable() {
return SampleProfileWriterBinary::writeNameTable();
auto &OS = *OutputStream;
- std::set<StringRef> V;
+ std::set<ProfileFuncRef> V;
stablizeNameTable(NameTable, V);
// Write out the MD5 name table. We wrote unencoded MD5 so reader can
@@ -357,7 +357,7 @@ std::error_code SampleProfileWriterExtBinaryBase::writeNameTable() {
encodeULEB128(NameTable.size(), OS);
support::endian::Writer Writer(OS, support::little);
for (auto N : V)
- Writer.write(hashFuncName(N));
+ Writer.write(N.getHashCode());
return sampleprof_error::success;
}
@@ -371,10 +371,14 @@ std::error_code SampleProfileWriterExtBinaryBase::writeNameTableSection(
// If NameTable contains ".__uniq." suffix, set SecFlagUniqSuffix flag
// so compiler won't strip the suffix during profile matching after
// seeing the flag in the profile.
- for (const auto &I : NameTable) {
- if (I.first.contains(FunctionSamples::UniqSuffix)) {
- addSectionFlag(SecNameTable, SecNameTableFlags::SecFlagUniqSuffix);
- break;
+ // Original names are unavailable if using MD5, so this option has no use.
+ if (!UseMD5) {
+ std::string Dummy;
+ for (const auto &I : NameTable) {
+ if (I.first.stringRef(Dummy).contains(FunctionSamples::UniqSuffix)) {
+ addSectionFlag(SecNameTable, SecNameTableFlags::SecFlagUniqSuffix);
+ break;
+ }
}
}
@@ -632,7 +636,7 @@ SampleProfileWriterBinary::writeContextIdx(const SampleContext &Context) {
return writeNameIdx(Context.getName());
}
-std::error_code SampleProfileWriterBinary::writeNameIdx(StringRef FName) {
+std::error_code SampleProfileWriterBinary::writeNameIdx(ProfileFuncRef FName) {
auto &NTable = getNameTable();
const auto &Ret = NTable.find(FName);
if (Ret == NTable.end())
@@ -641,7 +645,7 @@ std::error_code SampleProfileWriterBinary::writeNameIdx(StringRef FName) {
return sampleprof_error::success;
}
-void SampleProfileWriterBinary::addName(StringRef FName) {
+void SampleProfileWriterBinary::addName(ProfileFuncRef FName) {
auto &NTable = getNameTable();
NTable.insert(std::make_pair(FName, 0));
}
@@ -655,7 +659,7 @@ void SampleProfileWriterBinary::addNames(const FunctionSamples &S) {
for (const auto &I : S.getBodySamples()) {
const SampleRecord &Sample = I.second;
for (const auto &J : Sample.getCallTargets())
- addName(J.first());
+ addName(J.first);
}
// Recursively add all the names for inlined callsites.
@@ -679,18 +683,18 @@ void SampleProfileWriterExtBinaryBase::addContext(
}
void SampleProfileWriterBinary::stablizeNameTable(
- MapVector<StringRef, uint32_t> &NameTable, std::set<StringRef> &V) {
+ MapVector<ProfileFuncRef, uint32_t> &NameTable, std::set<ProfileFuncRef> &V) {
// Sort the names to make NameTable deterministic.
for (const auto &I : NameTable)
V.insert(I.first);
int i = 0;
- for (const StringRef &N : V)
+ for (const ProfileFuncRef &N : V)
NameTable[N] = i++;
}
std::error_code SampleProfileWriterBinary::writeNameTable() {
auto &OS = *OutputStream;
- std::set<StringRef> V;
+ std::set<ProfileFuncRef> V;
stablizeNameTable(NameTable, V);
// Write out the name table.
@@ -836,7 +840,7 @@ std::error_code SampleProfileWriterBinary::writeBody(const FunctionSamples &S) {
encodeULEB128(Sample.getSamples(), OS);
encodeULEB128(Sample.getCallTargets().size(), OS);
for (const auto &J : Sample.getSortedCallTargets()) {
- StringRef Callee = J.first;
+ auto Callee = J.first;
uint64_t CalleeSamples = J.second;
if (std::error_code EC = writeNameIdx(Callee))
return EC;
diff --git a/llvm/lib/Target/X86/X86InsertPrefetch.cpp b/llvm/lib/Target/X86/X86InsertPrefetch.cpp
index 29ae05bf0c94673..a7d74275670d521 100644
--- a/llvm/lib/Target/X86/X86InsertPrefetch.cpp
+++ b/llvm/lib/Target/X86/X86InsertPrefetch.cpp
@@ -125,7 +125,8 @@ bool X86InsertPrefetch::findPrefetchInfo(const FunctionSamples *TopSamples,
// Convert serialized prefetch hints into PrefetchInfo objects, and populate
// the Prefetches vector.
for (const auto &S_V : *T) {
- StringRef Name = S_V.getKey();
+ std::string Buffer;
+ StringRef Name = S_V.first.stringRef(Buffer);
if (Name.consume_front(SerializedPrefetchPrefix)) {
int64_t D = static_cast<int64_t>(S_V.second);
unsigned IID = 0;
diff --git a/llvm/lib/Transforms/IPO/SampleContextTracker.cpp b/llvm/lib/Transforms/IPO/SampleContextTracker.cpp
index 79bee8362ef8a9c..7bc692e4332fe0f 100644
--- a/llvm/lib/Transforms/IPO/SampleContextTracker.cpp
+++ b/llvm/lib/Transforms/IPO/SampleContextTracker.cpp
@@ -29,7 +29,7 @@ using namespace sampleprof;
namespace llvm {
ContextTrieNode *ContextTrieNode::getChildContext(const LineLocation &CallSite,
- StringRef CalleeName) {
+ ProfileFuncRef CalleeName) {
if (CalleeName.empty())
return getHottestChildContext(CallSite);
@@ -104,7 +104,7 @@ SampleContextTracker::moveContextSamples(ContextTrieNode &ToNodeParent,
}
void ContextTrieNode::removeChildContext(const LineLocation &CallSite,
- StringRef CalleeName) {
+ ProfileFuncRef CalleeName) {
uint64_t Hash = FunctionSamples::getCallSiteHash(CalleeName, CallSite);
// Note this essentially calls dtor and destroys that child context
AllChildContext.erase(Hash);
@@ -114,7 +114,7 @@ std::map<uint64_t, ContextTrieNode> &ContextTrieNode::getAllChildContext() {
return AllChildContext;
}
-StringRef ContextTrieNode::getFuncName() const { return FuncName; }
+ProfileFuncRef ContextTrieNode::getFuncName() const { return FuncName; }
FunctionSamples *ContextTrieNode::getFunctionSamples() const {
return FuncSamples;
@@ -178,7 +178,7 @@ void ContextTrieNode::dumpTree() {
}
ContextTrieNode *ContextTrieNode::getOrCreateChildContext(
- const LineLocation &CallSite, StringRef CalleeName, bool AllowCreate) {
+ const LineLocation &CallSite, ProfileFuncRef CalleeName, bool AllowCreate) {
uint64_t Hash = FunctionSamples::getCallSiteHash(CalleeName, CallSite);
auto It = AllChildContext.find(Hash);
if (It != AllChildContext.end()) {
@@ -232,14 +232,12 @@ SampleContextTracker::getCalleeContextSamplesFor(const CallBase &Inst,
return nullptr;
CalleeName = FunctionSamples::getCanonicalFnName(CalleeName);
- // Convert real function names to MD5 names, if the input profile is
- // MD5-based.
- std::string FGUID;
- CalleeName = getRepInFormat(CalleeName, FunctionSamples::UseMD5, FGUID);
+
+ ProfileFuncRef FName = getRepInFormat(CalleeName);
// For indirect call, CalleeName will be empty, in which case the context
// profile for callee with largest total samples will be returned.
- ContextTrieNode *CalleeContext = getCalleeContextFor(DIL, CalleeName);
+ ContextTrieNode *CalleeContext = getCalleeContextFor(DIL, FName);
if (CalleeContext) {
FunctionSamples *FSamples = CalleeContext->getFunctionSamples();
LLVM_DEBUG(if (FSamples) {
@@ -305,27 +303,23 @@ SampleContextTracker::getContextSamplesFor(const SampleContext &Context) {
SampleContextTracker::ContextSamplesTy &
SampleContextTracker::getAllContextSamplesFor(const Function &Func) {
StringRef CanonName = FunctionSamples::getCanonicalFnName(Func);
- return FuncToCtxtProfiles[CanonName];
+ return FuncToCtxtProfiles[getRepInFormat(CanonName)];
}
SampleContextTracker::ContextSamplesTy &
SampleContextTracker::getAllContextSamplesFor(StringRef Name) {
- return FuncToCtxtProfiles[Name];
+ return FuncToCtxtProfiles[getRepInFormat(Name)];
}
FunctionSamples *SampleContextTracker::getBaseSamplesFor(const Function &Func,
bool MergeContext) {
StringRef CanonName = FunctionSamples::getCanonicalFnName(Func);
- return getBaseSamplesFor(CanonName, MergeContext);
+ return getBaseSamplesFor(getRepInFormat(CanonName), MergeContext);
}
-FunctionSamples *SampleContextTracker::getBaseSamplesFor(StringRef Name,
+FunctionSamples *SampleContextTracker::getBaseSamplesFor(ProfileFuncRef Name,
bool MergeContext) {
LLVM_DEBUG(dbgs() << "Getting base profile for function: " << Name << "\n");
- // Convert real function names to MD5 names, if the input profile is
- // MD5-based.
- std::string FGUID;
- Name = getRepInFormat(Name, FunctionSamples::UseMD5, FGUID);
// Base profile is top-level node (child of root node), so try to retrieve
// existing top-level node for given function first. If it exists, it could be
@@ -373,7 +367,7 @@ void SampleContextTracker::markContextSamplesInlined(
ContextTrieNode &SampleContextTracker::getRootContext() { return RootContext; }
void SampleContextTracker::promoteMergeContextSamplesTree(
- const Instruction &Inst, StringRef CalleeName) {
+ const Instruction &Inst, ProfileFuncRef CalleeName) {
LLVM_DEBUG(dbgs() << "Promoting and merging context tree for instr: \n"
<< Inst << "\n");
// Get the caller context for the call instruction, we don't use callee
@@ -457,10 +451,11 @@ SampleContextTracker::getContextString(ContextTrieNode *Node) const {
void SampleContextTracker::dump() { RootContext.dumpTree(); }
StringRef SampleContextTracker::getFuncNameFor(ContextTrieNode *Node) const {
+ std::string Dummy;
if (!FunctionSamples::UseMD5)
- return Node->getFuncName();
+ return Node->getFuncName().stringRef(Dummy);
assert(GUIDToFuncNameMap && "GUIDToFuncNameMap needs to be populated first");
- return GUIDToFuncNameMap->lookup(std::stoull(Node->getFuncName().data()));
+ return GUIDToFuncNameMap->lookup(Node->getFuncName().getHashCode());
}
ContextTrieNode *
@@ -470,7 +465,7 @@ SampleContextTracker::getContextFor(const SampleContext &Context) {
ContextTrieNode *
SampleContextTracker::getCalleeContextFor(const DILocation *DIL,
- StringRef CalleeName) {
+ ProfileFuncRef CalleeName) {
assert(DIL && "Expect non-null location");
ContextTrieNode *CallContext = getContextFor(DIL);
@@ -485,7 +480,7 @@ SampleContextTracker::getCalleeContextFor(const DILocation *DIL,
ContextTrieNode *SampleContextTracker::getContextFor(const DILocation *DIL) {
assert(DIL && "Expect non-null location");
- SmallVector<std::pair<LineLocation, StringRef>, 10> S;
+ SmallVector<std::pair<LineLocation, ProfileFuncRef>, 10> S;
// Use C++ linkage name if possible.
const DILocation *PrevDIL = DIL;
@@ -494,7 +489,8 @@ ContextTrieNode *SampleContextTracker::getContextFor(const DILocation *DIL) {
if (Name.empty())
Name = PrevDIL->getScope()->getSubprogram()->getName();
S.push_back(
- std::make_pair(FunctionSamples::getCallSiteIdentifier(DIL), Name));
+ std::make_pair(FunctionSamples::getCallSiteIdentifier(DIL),
+ getRepInFormat(Name)));
PrevDIL = DIL;
}
@@ -503,24 +499,14 @@ ContextTrieNode *SampleContextTracker::getContextFor(const DILocation *DIL) {
StringRef RootName = PrevDIL->getScope()->getSubprogram()->getLinkageName();
if (RootName.empty())
RootName = PrevDIL->getScope()->getSubprogram()->getName();
- S.push_back(std::make_pair(LineLocation(0, 0), RootName));
-
- // Convert real function names to MD5 names, if the input profile is
- // MD5-based.
- std::list<std::string> MD5Names;
- if (FunctionSamples::UseMD5) {
- for (auto &Location : S) {
- MD5Names.emplace_back();
- getRepInFormat(Location.second, FunctionSamples::UseMD5, MD5Names.back());
- Location.second = MD5Names.back();
- }
- }
+ S.push_back(std::make_pair(LineLocation(0, 0),
+ getRepInFormat(RootName)));
ContextTrieNode *ContextNode = &RootContext;
int I = S.size();
while (--I >= 0 && ContextNode) {
LineLocation &CallSite = S[I].first;
- StringRef CalleeName = S[I].second;
+ ProfileFuncRef CalleeName = S[I].second;
ContextNode = ContextNode->getChildContext(CallSite, CalleeName);
}
@@ -553,12 +539,14 @@ SampleContextTracker::getOrCreateContextPath(const SampleContext &Context,
return ContextNode;
}
-ContextTrieNode *SampleContextTracker::getTopLevelContextNode(StringRef FName) {
+ContextTrieNode *
+SampleContextTracker::getTopLevelContextNode(ProfileFuncRef FName) {
assert(!FName.empty() && "Top level node query must provide valid name");
return RootContext.getChildContext(LineLocation(0, 0), FName);
}
-ContextTrieNode &SampleContextTracker::addTopLevelContextNode(StringRef FName) {
+ContextTrieNode &
+SampleContextTracker::addTopLevelContextNode(ProfileFuncRef FName) {
assert(!getTopLevelContextNode(FName) && "Node to add must not exist");
return *RootContext.getOrCreateChildContext(LineLocation(0, 0), FName);
}
diff --git a/llvm/lib/Transforms/IPO/SampleProfile.cpp b/llvm/lib/Transforms/IPO/SampleProfile.cpp
index 67a2e167b6bf7dd..e89913e2795349d 100644
--- a/llvm/lib/Transforms/IPO/SampleProfile.cpp
+++ b/llvm/lib/Transforms/IPO/SampleProfile.cpp
@@ -424,7 +424,7 @@ struct CandidateComparer {
return LCS->getBodySamples().size() > RCS->getBodySamples().size();
// Tie breaker using GUID so we have stable/deterministic inlining order
- return LCS->getGUID(LCS->getName()) < RCS->getGUID(RCS->getName());
+ return LCS->getGUID() < RCS->getGUID();
}
};
@@ -467,7 +467,7 @@ class SampleProfileMatcher {
private:
FunctionSamples *getFlattenedSamplesFor(const Function &F) {
StringRef CanonFName = FunctionSamples::getCanonicalFnName(F);
- auto It = FlattenedProfiles.find(CanonFName);
+ auto It = FlattenedProfiles.find(ProfileFuncRef(CanonFName));
if (It != FlattenedProfiles.end())
return &It->second;
return nullptr;
@@ -475,18 +475,21 @@ class SampleProfileMatcher {
void runOnFunction(const Function &F);
void findIRAnchors(const Function &F,
std::map<LineLocation, StringRef> &IRAnchors);
- void findProfileAnchors(const FunctionSamples &FS,
- std::map<LineLocation, StringSet<>> &ProfileAnchors);
+ void findProfileAnchors(
+ const FunctionSamples &FS,
+ std::map<LineLocation, std::unordered_set<ProfileFuncRef>>
+ &ProfileAnchors);
void countMismatchedSamples(const FunctionSamples &FS);
void countProfileMismatches(
const Function &F, const FunctionSamples &FS,
const std::map<LineLocation, StringRef> &IRAnchors,
- const std::map<LineLocation, StringSet<>> &ProfileAnchors);
+ const std::map<LineLocation, std::unordered_set<ProfileFuncRef>>
+ &ProfileAnchors);
void countProfileCallsiteMismatches(
const FunctionSamples &FS,
const std::map<LineLocation, StringRef> &IRAnchors,
- const std::map<LineLocation, StringSet<>> &ProfileAnchors,
-
+ const std::map<LineLocation, std::unordered_set<ProfileFuncRef>>
+ &ProfileAnchors,
uint64_t &FuncMismatchedCallsites, uint64_t &FuncProfiledCallsites);
LocToLocMap &getIRToProfileLocationMap(const Function &F) {
auto Ret = FuncMappings.try_emplace(
@@ -497,7 +500,8 @@ class SampleProfileMatcher {
void distributeIRToProfileLocationMap(FunctionSamples &FS);
void runStaleProfileMatching(
const Function &F, const std::map<LineLocation, StringRef> &IRAnchors,
- const std::map<LineLocation, StringSet<>> &ProfileAnchors,
+ const std::map<LineLocation, std::unordered_set<ProfileFuncRef>>
+ &ProfileAnchors,
LocToLocMap &IRToProfileLocationMap);
};
@@ -615,7 +619,7 @@ class SampleProfileLoader final : public SampleProfileLoaderBaseImpl<Function> {
// All the Names used in FunctionSamples including outline function
// names, inline instance names and call target names.
- StringSet<> NamesInProfile;
+ std::unordered_set<ProfileFuncRef> NamesInProfile;
// For symbol in profile symbol list, whether to regard their profiles
// to be accurate. It is mainly decided by existance of profile symbol
@@ -760,8 +764,7 @@ SampleProfileLoader::findIndirectCallFunctionSamples(
assert(L && R && "Expect non-null FunctionSamples");
if (L->getHeadSamplesEstimate() != R->getHeadSamplesEstimate())
return L->getHeadSamplesEstimate() > R->getHeadSamplesEstimate();
- return FunctionSamples::getGUID(L->getName()) <
- FunctionSamples::getGUID(R->getName());
+ return L->getGUID() < R->getGUID();
};
if (FunctionSamples::ProfileIsCS) {
@@ -1081,7 +1084,7 @@ void SampleProfileLoader::findExternalInlineCandidate(
// just add the direct GUID and move on
if (!Samples) {
InlinedGUIDs.insert(
- FunctionSamples::getGUID(CB->getCalledFunction()->getName()));
+ Function::getGUID(CB->getCalledFunction()->getName()));
return;
}
// Otherwise, drop the threshold to import everything that we can
@@ -1127,17 +1130,17 @@ void SampleProfileLoader::findExternalInlineCandidate(
Function *Func = SymbolMap.lookup(Name);
// Add to the import list only when it's defined out of module.
if (!Func || Func->isDeclaration())
- InlinedGUIDs.insert(FunctionSamples::getGUID(CalleeSample->getName()));
+ InlinedGUIDs.insert(CalleeSample->getGUID());
// Import hot CallTargets, which may not be available in IR because full
// profile annotation cannot be done until backend compilation in ThinLTO.
for (const auto &BS : CalleeSample->getBodySamples())
for (const auto &TS : BS.second.getCallTargets())
- if (TS.getValue() > Threshold) {
- StringRef CalleeName = CalleeSample->getFuncName(TS.getKey());
+ if (TS.second > Threshold) {
+ StringRef CalleeName = CalleeSample->getFuncName(TS.first);
const Function *Callee = SymbolMap.lookup(CalleeName);
if (!Callee || Callee->isDeclaration())
- InlinedGUIDs.insert(FunctionSamples::getGUID(TS.getKey()));
+ InlinedGUIDs.insert(TS.first.getHashCode());
}
// Import hot child context profile associted with callees. Note that this
@@ -1644,7 +1647,7 @@ GetSortedValueDataFromCallTargets(const SampleRecord::CallTargetMap &M) {
SmallVector<InstrProfValueData, 2> R;
for (const auto &I : SampleRecord::SortCallTargets(M)) {
R.emplace_back(
- InstrProfValueData{FunctionSamples::getGUID(I.first), I.second});
+ InstrProfValueData{I.first.getHashCode(), I.second});
}
return R;
}
@@ -1871,7 +1874,8 @@ SampleProfileLoader::buildProfiledCallGraph(Module &M) {
for (Function &F : M) {
if (F.isDeclaration() || !F.hasFnAttribute("use-sample-profile"))
continue;
- ProfiledCG->addProfiledFunction(FunctionSamples::getCanonicalFnName(F));
+ ProfiledCG->addProfiledFunction(
+ getRepInFormat(FunctionSamples::getCanonicalFnName(F)));
}
return ProfiledCG;
@@ -1962,7 +1966,10 @@ SampleProfileLoader::buildFunctionOrder(Module &M, LazyCallGraph &CG) {
Range = *SI;
}
for (auto *Node : Range) {
- Function *F = SymbolMap.lookup(Node->Name);
+ Function *F = SymbolMap.lookup(
+ FunctionSamples::UseMD5
+ ? GUIDToFuncNameMap.lookup(Node->Name.getHashCode())
+ : reinterpret_cast<StringRef &>(Node->Name));
if (F && !F->isDeclaration() && F->hasFnAttribute("use-sample-profile"))
FunctionOrderList.push_back(F);
}
@@ -2177,7 +2184,7 @@ void SampleProfileMatcher::findIRAnchors(
}
void SampleProfileMatcher::countMismatchedSamples(const FunctionSamples &FS) {
- const auto *FuncDesc = ProbeManager->getDesc(FS.getName());
+ const auto *FuncDesc = ProbeManager->getDesc(FS.getGUID());
// Skip the function that is external or renamed.
if (!FuncDesc)
return;
@@ -2194,7 +2201,8 @@ void SampleProfileMatcher::countMismatchedSamples(const FunctionSamples &FS) {
void SampleProfileMatcher::countProfileMismatches(
const Function &F, const FunctionSamples &FS,
const std::map<LineLocation, StringRef> &IRAnchors,
- const std::map<LineLocation, StringSet<>> &ProfileAnchors) {
+ const std::map<LineLocation, std::unordered_set<ProfileFuncRef>>
+ &ProfileAnchors) {
[[maybe_unused]] bool IsFuncHashMismatch = false;
if (FunctionSamples::ProfileIsProbeBased) {
TotalFuncHashSamples += FS.getTotalSamples();
@@ -2228,7 +2236,8 @@ void SampleProfileMatcher::countProfileMismatches(
void SampleProfileMatcher::countProfileCallsiteMismatches(
const FunctionSamples &FS,
const std::map<LineLocation, StringRef> &IRAnchors,
- const std::map<LineLocation, StringSet<>> &ProfileAnchors,
+ const std::map<LineLocation, std::unordered_set<ProfileFuncRef>>
+ &ProfileAnchors,
uint64_t &FuncMismatchedCallsites, uint64_t &FuncProfiledCallsites) {
// Check if there are any callsites in the profile that does not match to any
@@ -2263,7 +2272,7 @@ void SampleProfileMatcher::countProfileCallsiteMismatches(
// reported as mismatching.
if (IRCalleeName == UnknownIndirectCallee)
CallsiteIsMatched = true;
- else if (Callees.size() == 1 && Callees.count(IRCalleeName))
+ else if (Callees.size() == 1 && Callees.count(getRepInFormat(IRCalleeName)))
CallsiteIsMatched = true;
FuncProfiledCallsites++;
@@ -2275,9 +2284,8 @@ void SampleProfileMatcher::countProfileCallsiteMismatches(
}
}
-void SampleProfileMatcher::findProfileAnchors(
- const FunctionSamples &FS,
- std::map<LineLocation, StringSet<>> &ProfileAnchors) {
+void SampleProfileMatcher::findProfileAnchors(const FunctionSamples &FS,
+ std::map<LineLocation, std::unordered_set<ProfileFuncRef>> &ProfileAnchors) {
auto isInvalidLineOffset = [](uint32_t LineOffset) {
return LineOffset & 0x8000;
};
@@ -2287,8 +2295,9 @@ void SampleProfileMatcher::findProfileAnchors(
if (isInvalidLineOffset(Loc.LineOffset))
continue;
for (const auto &I : I.second.getCallTargets()) {
- auto Ret = ProfileAnchors.try_emplace(Loc, StringSet<>());
- Ret.first->second.insert(I.first());
+ auto Ret = ProfileAnchors.try_emplace(Loc,
+ std::unordered_set<ProfileFuncRef>());
+ Ret.first->second.insert(I.first);
}
}
@@ -2298,7 +2307,8 @@ void SampleProfileMatcher::findProfileAnchors(
continue;
const auto &CalleeMap = I.second;
for (const auto &I : CalleeMap) {
- auto Ret = ProfileAnchors.try_emplace(Loc, StringSet<>());
+ auto Ret = ProfileAnchors.try_emplace(Loc,
+ std::unordered_set<ProfileFuncRef>());
Ret.first->second.insert(I.first);
}
}
@@ -2322,21 +2332,24 @@ void SampleProfileMatcher::findProfileAnchors(
// [1, 2, 3(foo), 4, 7, 8(bar), 9]
// The output mapping: [2->3, 3->4, 5->7, 6->8, 7->9].
void SampleProfileMatcher::runStaleProfileMatching(
- const Function &F, const std::map<LineLocation, StringRef> &IRAnchors,
- const std::map<LineLocation, StringSet<>> &ProfileAnchors,
+ const Function &F,
+ const std::map<LineLocation, StringRef> &IRAnchors,
+ const std::map<LineLocation, std::unordered_set<ProfileFuncRef>>
+ &ProfileAnchors,
LocToLocMap &IRToProfileLocationMap) {
LLVM_DEBUG(dbgs() << "Run stale profile matching for " << F.getName()
<< "\n");
assert(IRToProfileLocationMap.empty() &&
"Run stale profile matching only once per function");
- StringMap<std::set<LineLocation>> CalleeToCallsitesMap;
+ std::unordered_map<ProfileFuncRef, std::set<LineLocation>>
+ CalleeToCallsitesMap;
for (const auto &I : ProfileAnchors) {
const auto &Loc = I.first;
const auto &Callees = I.second;
// Filter out possible indirect calls, use direct callee name as anchor.
if (Callees.size() == 1) {
- StringRef CalleeName = Callees.begin()->first();
+ ProfileFuncRef CalleeName = *Callees.begin();
const auto &Candidates = CalleeToCallsitesMap.try_emplace(
CalleeName, std::set<LineLocation>());
Candidates.first->second.insert(Loc);
@@ -2355,11 +2368,12 @@ void SampleProfileMatcher::runStaleProfileMatching(
for (const auto &IR : IRAnchors) {
const auto &Loc = IR.first;
- StringRef CalleeName = IR.second;
+ auto CalleeName = IR.second;
bool IsMatchedAnchor = false;
// Match the anchor location in lexical order.
if (!CalleeName.empty()) {
- auto CandidateAnchors = CalleeToCallsitesMap.find(CalleeName);
+ auto CandidateAnchors = CalleeToCallsitesMap.find(
+ getRepInFormat(CalleeName));
if (CandidateAnchors != CalleeToCallsitesMap.end() &&
!CandidateAnchors->second.empty()) {
auto CI = CandidateAnchors->second.begin();
@@ -2420,7 +2434,7 @@ void SampleProfileMatcher::runOnFunction(const Function &F) {
findIRAnchors(F, IRAnchors);
// Anchors for profile. It's a map from callsite location to a set of callee
// name.
- std::map<LineLocation, StringSet<>> ProfileAnchors;
+ std::map<LineLocation, std::unordered_set<ProfileFuncRef>> ProfileAnchors;
findProfileAnchors(*FSFlattened, ProfileAnchors);
// Detect profile mismatch for profile staleness metrics report.
@@ -2499,7 +2513,7 @@ void SampleProfileMatcher::runOnModule() {
void SampleProfileMatcher::distributeIRToProfileLocationMap(
FunctionSamples &FS) {
- const auto ProfileMappings = FuncMappings.find(FS.getName());
+ const auto ProfileMappings = FuncMappings.find(FS.getFuncName());
if (ProfileMappings != FuncMappings.end()) {
FS.setIRToProfileLocationMap(&(ProfileMappings->second));
}
@@ -2625,7 +2639,7 @@ bool SampleProfileLoader::runOnFunction(Function &F, ModuleAnalysisManager *AM)
// but not cold accumulatively...), so the outline function showing up as
// cold in sampled binary will actually not be cold after current build.
StringRef CanonName = FunctionSamples::getCanonicalFnName(F);
- if (NamesInProfile.count(CanonName))
+ if (NamesInProfile.count(getRepInFormat(CanonName)))
initialEntryCount = -1;
}
diff --git a/llvm/tools/llvm-profdata/llvm-profdata.cpp b/llvm/tools/llvm-profdata/llvm-profdata.cpp
index fdb2c1405f1237f..32718482296f46b 100644
--- a/llvm/tools/llvm-profdata/llvm-profdata.cpp
+++ b/llvm/tools/llvm-profdata/llvm-profdata.cpp
@@ -199,6 +199,12 @@ class SymbolRemapper {
StringRef New = RemappingTable.lookup(Name);
return New.empty() ? Name : New;
}
+
+ ProfileFuncRef operator()(ProfileFuncRef Name) {
+ std::string Buffer;
+ StringRef New = RemappingTable.lookup(Name.stringRef(Buffer));
+ return New.empty() ? Name : ProfileFuncRef(New);
+ }
};
}
@@ -716,7 +722,8 @@ adjustInstrProfile(std::unique_ptr<WriterContext> &WC,
auto BuildMaxSampleMapImpl = [&](const FunctionSamples &FS,
const StringRef &RootName,
auto &BuildImpl) -> void {
- const StringRef &Name = FS.getName();
+ std::string Buffer;
+ const StringRef &Name = FS.getName().stringRef(Buffer);
const StringRef *NewRootName = &RootName;
uint64_t EntrySample = FS.getHeadSamplesEstimate();
uint64_t MaxBodySample = FS.getMaxCountInside(/* SkipCallSite*/ true);
@@ -770,7 +777,8 @@ adjustInstrProfile(std::unique_ptr<WriterContext> &WC,
for (auto &PD : Reader->getProfiles()) {
sampleprof::FunctionSamples &FS = PD.second;
- BuildMaxSampleMap(FS, FS.getName());
+ std::string Buffer;
+ BuildMaxSampleMap(FS, FS.getName().stringRef(Buffer));
}
ProfileSummary InstrPS = *IPBuilder.getSummary();
@@ -896,7 +904,7 @@ remapSamples(const sampleprof::FunctionSamples &Samples,
for (const auto &Target : BodySample.second.getCallTargets()) {
Result.addCalledTargetSamples(BodySample.first.LineOffset,
MaskedDiscriminator,
- Remapper(Target.first()), Target.second);
+ Remapper(Target.first), Target.second);
}
}
for (const auto &CallsiteSamples : Samples.getCallsiteSamples()) {
@@ -906,7 +914,7 @@ remapSamples(const sampleprof::FunctionSamples &Samples,
sampleprof::FunctionSamples Remapped =
remapSamples(Callsite.second, Remapper, Error);
MergeResult(Error,
- Target[std::string(Remapped.getName())].merge(Remapped));
+ Target[Remapped.getName()].merge(Remapped));
}
}
return Result;
diff --git a/llvm/tools/llvm-profgen/CSPreInliner.cpp b/llvm/tools/llvm-profgen/CSPreInliner.cpp
index 9421118a3bb1b50..01796f41d188a02 100644
--- a/llvm/tools/llvm-profgen/CSPreInliner.cpp
+++ b/llvm/tools/llvm-profgen/CSPreInliner.cpp
@@ -80,8 +80,8 @@ CSPreInliner::CSPreInliner(SampleContextTracker &Tracker,
ProfileInlineLimitMax = 50000;
}
-std::vector<StringRef> CSPreInliner::buildTopDownOrder() {
- std::vector<StringRef> Order;
+std::vector<ProfileFuncRef> CSPreInliner::buildTopDownOrder() {
+ std::vector<ProfileFuncRef> Order;
// Trim cold edges to get a more stable call graph. This allows for a more
// stable top-down order which in turns helps the stablity of the generated
// profile from run to run.
@@ -202,7 +202,7 @@ bool CSPreInliner::shouldInline(ProfiledInlineCandidate &Candidate) {
return (Candidate.SizeCost < SampleThreshold);
}
-void CSPreInliner::processFunction(const StringRef Name) {
+void CSPreInliner::processFunction(const ProfileFuncRef Name) {
FunctionSamples *FSamples = ContextTracker.getBaseSamplesFor(Name);
if (!FSamples)
return;
@@ -303,7 +303,7 @@ void CSPreInliner::run() {
// It also helps better compress context profile to control profile
// size, as we now only need context profile for functions going to
// be inlined.
- for (StringRef FuncName : buildTopDownOrder()) {
+ for (ProfileFuncRef FuncName : buildTopDownOrder()) {
processFunction(FuncName);
}
diff --git a/llvm/tools/llvm-profgen/CSPreInliner.h b/llvm/tools/llvm-profgen/CSPreInliner.h
index 4d848aafdab914d..85e297cddfe8544 100644
--- a/llvm/tools/llvm-profgen/CSPreInliner.h
+++ b/llvm/tools/llvm-profgen/CSPreInliner.h
@@ -57,8 +57,7 @@ struct ProfiledCandidateComparer {
// Tie breaker using GUID so we have stable/deterministic inlining order
assert(LHS.CalleeSamples && RHS.CalleeSamples &&
"Expect non-null FunctionSamples");
- return LHS.CalleeSamples->getGUID(LHS.CalleeSamples->getName()) <
- RHS.CalleeSamples->getGUID(RHS.CalleeSamples->getName());
+ return LHS.CalleeSamples->getGUID() < RHS.CalleeSamples->getGUID();
}
};
@@ -81,8 +80,8 @@ class CSPreInliner {
private:
bool getInlineCandidates(ProfiledCandidateQueue &CQueue,
const FunctionSamples *FCallerContextSamples);
- std::vector<StringRef> buildTopDownOrder();
- void processFunction(StringRef Name);
+ std::vector<ProfileFuncRef> buildTopDownOrder();
+ void processFunction(ProfileFuncRef Name);
bool shouldInline(ProfiledInlineCandidate &Candidate);
uint32_t getFuncSize(const ContextTrieNode *ContextNode);
bool UseContextCost;
diff --git a/llvm/tools/llvm-profgen/ProfileGenerator.cpp b/llvm/tools/llvm-profgen/ProfileGenerator.cpp
index d307ab465676296..f5f2bc0d800e83a 100644
--- a/llvm/tools/llvm-profgen/ProfileGenerator.cpp
+++ b/llvm/tools/llvm-profgen/ProfileGenerator.cpp
@@ -586,20 +586,21 @@ void ProfileGenerator::populateBoundarySamplesWithProbesForAllFunctions(
FunctionProfile.addCalledTargetSamples(
FrameVec.back().Location.LineOffset,
FrameVec.back().Location.Discriminator,
- CalleeName, Count);
+ ProfileFuncRef(CalleeName), Count);
}
}
}
FunctionSamples &ProfileGenerator::getLeafProfileAndAddTotalSamples(
const SampleContextFrameVector &FrameVec, uint64_t Count) {
+ std::string Buffer;
// Get top level profile
FunctionSamples *FunctionProfile =
- &getTopLevelFunctionProfile(FrameVec[0].FuncName);
+ &getTopLevelFunctionProfile(FrameVec[0].FuncName.stringRef(Buffer));
FunctionProfile->addTotalSamples(Count);
if (Binary->usePseudoProbes()) {
const auto *FuncDesc = Binary->getFuncDescForGUID(
- Function::getGUID(FunctionProfile->getName()));
+ FunctionProfile->getName().getHashCode());
FunctionProfile->setFunctionHash(FuncDesc->FuncHash);
}
@@ -619,7 +620,7 @@ FunctionSamples &ProfileGenerator::getLeafProfileAndAddTotalSamples(
FunctionProfile->addTotalSamples(Count);
if (Binary->usePseudoProbes()) {
const auto *FuncDesc = Binary->getFuncDescForGUID(
- Function::getGUID(FunctionProfile->getName()));
+ FunctionProfile->getName().getHashCode());
FunctionProfile->setFunctionHash(FuncDesc->FuncHash);
}
}
@@ -716,7 +717,7 @@ void ProfileGenerator::populateBoundarySamplesForAllFunctions(
FunctionProfile.addCalledTargetSamples(
FrameVec.back().Location.LineOffset,
getBaseDiscriminator(FrameVec.back().Location.Discriminator),
- CalleeName, Count);
+ ProfileFuncRef(CalleeName), Count);
}
// Add head samples for callee.
FunctionSamples &CalleeProfile = getTopLevelFunctionProfile(CalleeName);
@@ -899,7 +900,8 @@ void CSProfileGenerator::populateBoundarySamplesForFunction(
if (LeafLoc) {
CallerNode->getFunctionSamples()->addCalledTargetSamples(
LeafLoc->Location.LineOffset,
- getBaseDiscriminator(LeafLoc->Location.Discriminator), CalleeName,
+ getBaseDiscriminator(LeafLoc->Location.Discriminator),
+ ProfileFuncRef(CalleeName),
Count);
// Record head sample for called target(callee)
CalleeCallSite = LeafLoc->Location;
@@ -907,7 +909,8 @@ void CSProfileGenerator::populateBoundarySamplesForFunction(
}
ContextTrieNode *CalleeNode =
- CallerNode->getOrCreateChildContext(CalleeCallSite, CalleeName);
+ CallerNode->getOrCreateChildContext(CalleeCallSite,
+ ProfileFuncRef(CalleeName));
FunctionSamples *CalleeProfile = getOrCreateFunctionSamples(CalleeNode);
CalleeProfile->addHeadSamples(Count);
}
@@ -1212,7 +1215,7 @@ void CSProfileGenerator::populateBoundarySamplesWithProbes(
continue;
FunctionProfile.addCalledTargetSamples(CallProbe->getIndex(),
CallProbe->getDiscriminator(),
- CalleeName, Count);
+ ProfileFuncRef(CalleeName), Count);
}
}
diff --git a/llvm/tools/llvm-profgen/ProfiledBinary.cpp b/llvm/tools/llvm-profgen/ProfiledBinary.cpp
index 5fb51d3ac9eba78..95cb1f91bd68fe0 100644
--- a/llvm/tools/llvm-profgen/ProfiledBinary.cpp
+++ b/llvm/tools/llvm-profgen/ProfiledBinary.cpp
@@ -77,7 +77,7 @@ void BinarySizeContextTracker::addInstructionForContext(
ContextTrieNode *CurNode = &RootContext;
bool IsLeaf = true;
for (const auto &Callsite : reverse(Context)) {
- StringRef CallerName = Callsite.FuncName;
+ ProfileFuncRef CallerName = Callsite.FuncName;
LineLocation CallsiteLoc = IsLeaf ? LineLocation(0, 0) : Callsite.Location;
CurNode = CurNode->getOrCreateChildContext(CallsiteLoc, CallerName);
IsLeaf = false;
@@ -145,7 +145,8 @@ void BinarySizeContextTracker::trackInlineesOptimizedAway(
StringRef CallerName = ProbeFrame.first;
LineLocation CallsiteLoc(ProbeFrame.second, 0);
SizeContext =
- SizeContext->getOrCreateChildContext(CallsiteLoc, CallerName);
+ SizeContext->getOrCreateChildContext(CallsiteLoc,
+ ProfileFuncRef(CallerName));
}
// Add 0 size to make known.
SizeContext->addFunctionSize(0);
@@ -831,6 +832,14 @@ void ProfiledBinary::loadSymbolsFromDWARF(ObjectFile &Obj) {
if (BinaryFunctions.empty())
WithColor::warning() << "Loading of DWARF info completed, but no binary "
"functions have been retrieved.\n";
+
+
+ // Populate the hash binary function map for MD5 function name lookup. This
+ // is done after BinaryFunctions are finalized.
+ for (auto &BinaryFunction : BinaryFunctions) {
+ HashBinaryFunctions[MD5Hash(StringRef(BinaryFunction.first))] =
+ &BinaryFunction.second;
+ }
}
void ProfiledBinary::populateSymbolListFromDWARF(
@@ -882,7 +891,7 @@ SampleContextFrameVector ProfiledBinary::symbolize(const InstructionPointer &IP,
LineLocation Line(LineOffset, Discriminator);
auto It = NameStrings.insert(FunctionName.str());
- CallStack.emplace_back(*It.first, Line);
+ CallStack.emplace_back(ProfileFuncRef(*It.first), Line);
}
return CallStack;
diff --git a/llvm/tools/llvm-profgen/ProfiledBinary.h b/llvm/tools/llvm-profgen/ProfiledBinary.h
index a6d78c661cc1c31..298ec2781b38727 100644
--- a/llvm/tools/llvm-profgen/ProfiledBinary.h
+++ b/llvm/tools/llvm-profgen/ProfiledBinary.h
@@ -220,6 +220,10 @@ class ProfiledBinary {
// A map of mapping function name to BinaryFunction info.
std::unordered_map<std::string, BinaryFunction> BinaryFunctions;
+ // Lookup BinaryFunctions using the function name's MD5 hash. Needed if the
+ // profile is using MD5.
+ std::unordered_map<uint64_t, BinaryFunction *> HashBinaryFunctions;
+
// A list of binary functions that have samples.
std::unordered_set<const BinaryFunction *> ProfiledFunctions;
@@ -477,11 +481,17 @@ class ProfiledBinary {
ProfiledFunctions = Funcs;
}
- BinaryFunction *getBinaryFunction(StringRef FName) {
- auto I = BinaryFunctions.find(FName.str());
- if (I == BinaryFunctions.end())
+ BinaryFunction *getBinaryFunction(ProfileFuncRef FName) {
+ if (FName.isStringRef()) {
+ auto I = BinaryFunctions.find(FName.str());
+ if (I == BinaryFunctions.end())
+ return nullptr;
+ return &I->second;
+ }
+ auto I = HashBinaryFunctions.find(FName.getHashCode());
+ if (I == HashBinaryFunctions.end())
return nullptr;
- return &I->second;
+ return I->second;
}
uint32_t getFuncSizeForContext(const ContextTrieNode *ContextNode) {
@@ -556,7 +566,7 @@ class ProfiledBinary {
InlineContextStack.clear();
continue;
}
- InlineContextStack.emplace_back(Callsite.first,
+ InlineContextStack.emplace_back(ProfileFuncRef(Callsite.first),
LineLocation(Callsite.second, 0));
}
}
diff --git a/llvm/unittests/ProfileData/SampleProfTest.cpp b/llvm/unittests/ProfileData/SampleProfTest.cpp
index e01025440e4a56a..a13869d901bc2eb 100644
--- a/llvm/unittests/ProfileData/SampleProfTest.cpp
+++ b/llvm/unittests/ProfileData/SampleProfTest.cpp
@@ -145,7 +145,7 @@ struct SampleProfTest : ::testing::Test {
StringRef FooName("_Z3fooi");
FunctionSamples FooSamples;
- FooSamples.setName(FooName);
+ FooSamples.setName(ProfileFuncRef(FooName));
FooSamples.addTotalSamples(7711);
FooSamples.addHeadSamples(610);
FooSamples.addBodySamples(1, 0, 610);
@@ -155,43 +155,43 @@ struct SampleProfTest : ::testing::Test {
FooSamples.addBodySamples(10, 0, 605);
// Add inline instance with name "_Z3gooi".
- StringRef GooName("_Z3gooi");
+ ProfileFuncRef GooName(StringRef("_Z3gooi"));
auto &GooSamples =
- FooSamples.functionSamplesAt(LineLocation(7, 0))[GooName.str()];
+ FooSamples.functionSamplesAt(LineLocation(7, 0))[GooName];
GooSamples.setName(GooName);
GooSamples.addTotalSamples(502);
GooSamples.addBodySamples(3, 0, 502);
// Add inline instance with name "_Z3hooi".
- StringRef HooName("_Z3hooi");
+ ProfileFuncRef HooName(StringRef("_Z3hooi"));
auto &HooSamples =
- GooSamples.functionSamplesAt(LineLocation(9, 0))[HooName.str()];
+ GooSamples.functionSamplesAt(LineLocation(9, 0))[HooName];
HooSamples.setName(HooName);
HooSamples.addTotalSamples(317);
HooSamples.addBodySamples(4, 0, 317);
StringRef BarName("_Z3bari");
FunctionSamples BarSamples;
- BarSamples.setName(BarName);
+ BarSamples.setName(ProfileFuncRef(BarName));
BarSamples.addTotalSamples(20301);
BarSamples.addHeadSamples(1437);
BarSamples.addBodySamples(1, 0, 1437);
// Test how reader/writer handles unmangled names.
StringRef MconstructName("_M_construct<char *>");
StringRef StringviewName("string_view<std::allocator<char> >");
- BarSamples.addCalledTargetSamples(1, 0, MconstructName, 1000);
- BarSamples.addCalledTargetSamples(1, 0, StringviewName, 437);
+ BarSamples.addCalledTargetSamples(1, 0, ProfileFuncRef(MconstructName), 1000);
+ BarSamples.addCalledTargetSamples(1, 0, ProfileFuncRef(StringviewName), 437);
StringRef BazName("_Z3bazi");
FunctionSamples BazSamples;
- BazSamples.setName(BazName);
+ BazSamples.setName(ProfileFuncRef(BazName));
BazSamples.addTotalSamples(12557);
BazSamples.addHeadSamples(1257);
BazSamples.addBodySamples(1, 0, 12557);
StringRef BooName("_Z3booi");
FunctionSamples BooSamples;
- BooSamples.setName(BooName);
+ BooSamples.setName(ProfileFuncRef(BooName));
BooSamples.addTotalSamples(1232);
BooSamples.addHeadSamples(1);
BooSamples.addBodySamples(1, 0, 1232);
@@ -210,8 +210,8 @@ struct SampleProfTest : ::testing::Test {
if (Remap) {
FooName = "_Z4fauxi";
BarName = "_Z3barl";
- GooName = "_Z3gool";
- HooName = "_Z3hool";
+ GooName = ProfileFuncRef(StringRef("_Z3gool"));
+ HooName = ProfileFuncRef(StringRef("_Z3hool"));
}
M.getOrInsertFunction(FooName, fn_type);
@@ -245,16 +245,17 @@ struct SampleProfTest : ::testing::Test {
FunctionSamples *ReadFooSamples = Reader->getSamplesFor(FooName);
ASSERT_TRUE(ReadFooSamples != nullptr);
if (!UseMD5) {
- ASSERT_EQ("_Z3fooi", ReadFooSamples->getName());
+ ASSERT_EQ("_Z3fooi", ReadFooSamples->getName().str());
}
ASSERT_EQ(7711u, ReadFooSamples->getTotalSamples());
ASSERT_EQ(610u, ReadFooSamples->getHeadSamples());
+ std::string Buffer;
// Try to find a FunctionSamples with GooName at given callsites containing
// inline instance for GooName. Test the correct FunctionSamples can be
// found with Remapper support.
const FunctionSamples *ReadGooSamples =
- ReadFooSamples->findFunctionSamplesAt(LineLocation(7, 0), GooName,
+ ReadFooSamples->findFunctionSamplesAt(LineLocation(7, 0), GooName.stringRef(Buffer),
Reader->getRemapper());
ASSERT_TRUE(ReadGooSamples != nullptr);
ASSERT_EQ(502u, ReadGooSamples->getTotalSamples());
@@ -263,7 +264,7 @@ struct SampleProfTest : ::testing::Test {
// no inline instance for GooName. Test no FunctionSamples will be
// found with Remapper support.
const FunctionSamples *ReadGooSamplesAgain =
- ReadFooSamples->findFunctionSamplesAt(LineLocation(9, 0), GooName,
+ ReadFooSamples->findFunctionSamplesAt(LineLocation(9, 0), GooName.stringRef(Buffer),
Reader->getRemapper());
ASSERT_TRUE(ReadGooSamplesAgain == nullptr);
@@ -272,7 +273,7 @@ struct SampleProfTest : ::testing::Test {
// inline instance for HooName. Test the correct FunctionSamples can be
// found with Remapper support.
const FunctionSamples *ReadHooSamples =
- ReadGooSamples->findFunctionSamplesAt(LineLocation(9, 0), HooName,
+ ReadGooSamples->findFunctionSamplesAt(LineLocation(9, 0), HooName.stringRef(Buffer),
Reader->getRemapper());
ASSERT_TRUE(ReadHooSamples != nullptr);
ASSERT_EQ(317u, ReadHooSamples->getTotalSamples());
@@ -280,7 +281,7 @@ struct SampleProfTest : ::testing::Test {
FunctionSamples *ReadBarSamples = Reader->getSamplesFor(BarName);
ASSERT_TRUE(ReadBarSamples != nullptr);
if (!UseMD5) {
- ASSERT_EQ("_Z3bari", ReadBarSamples->getName());
+ ASSERT_EQ("_Z3bari", ReadBarSamples->getName().str());
}
ASSERT_EQ(20301u, ReadBarSamples->getTotalSamples());
ASSERT_EQ(1437u, ReadBarSamples->getHeadSamples());
@@ -305,12 +306,8 @@ struct SampleProfTest : ::testing::Test {
ASSERT_TRUE(ReadBooSamples != nullptr);
ASSERT_EQ(1232u, ReadBooSamples->getTotalSamples());
- std::string MconstructGUID;
- StringRef MconstructRep =
- getRepInFormat(MconstructName, UseMD5, MconstructGUID);
- std::string StringviewGUID;
- StringRef StringviewRep =
- getRepInFormat(StringviewName, UseMD5, StringviewGUID);
+ ProfileFuncRef MconstructRep = getRepInFormat(MconstructName, UseMD5);
+ ProfileFuncRef StringviewRep = getRepInFormat(StringviewName, UseMD5);
ASSERT_EQ(1000u, CTMap.get()[MconstructRep]);
ASSERT_EQ(437u, CTMap.get()[StringviewRep]);
@@ -333,7 +330,7 @@ struct SampleProfTest : ::testing::Test {
uint64_t TotalSamples, uint64_t HeadSamples) {
StringRef Name(Fname);
FunctionSamples FcnSamples;
- FcnSamples.setName(Name);
+ FcnSamples.setName(ProfileFuncRef(Name));
FcnSamples.addTotalSamples(TotalSamples);
FcnSamples.addHeadSamples(HeadSamples);
FcnSamples.addBodySamples(1, 0, HeadSamples);
More information about the cfe-commits
mailing list