1bf9a7730SZachary Turner //===-- ConstString.cpp -----------------------------------------*- C++ -*-===// 2bf9a7730SZachary Turner // 3bf9a7730SZachary Turner // The LLVM Compiler Infrastructure 4bf9a7730SZachary Turner // 5bf9a7730SZachary Turner // This file is distributed under the University of Illinois Open Source 6bf9a7730SZachary Turner // License. See LICENSE.TXT for details. 7bf9a7730SZachary Turner // 8bf9a7730SZachary Turner //===----------------------------------------------------------------------===// 9bf9a7730SZachary Turner 10bf9a7730SZachary Turner #include "lldb/Utility/ConstString.h" 11bf9a7730SZachary Turner 124479ac15SZachary Turner #include "lldb/Utility/Stream.h" 13bf9a7730SZachary Turner 14bf9a7730SZachary Turner #include "llvm/ADT/StringMap.h" 15*672d2c12SJonas Devlieghere #include "llvm/ADT/iterator.h" 16*672d2c12SJonas Devlieghere #include "llvm/Support/Allocator.h" 17*672d2c12SJonas Devlieghere #include "llvm/Support/DJB.h" 18*672d2c12SJonas Devlieghere #include "llvm/Support/FormatProviders.h" 19bf9a7730SZachary Turner #include "llvm/Support/RWMutex.h" 20c5f28e2aSKamil Rytarowski #include "llvm/Support/Threading.h" 214479ac15SZachary Turner 22*672d2c12SJonas Devlieghere #include <algorithm> 234479ac15SZachary Turner #include <array> 24*672d2c12SJonas Devlieghere #include <utility> 254479ac15SZachary Turner 26*672d2c12SJonas Devlieghere #include <inttypes.h> 27*672d2c12SJonas Devlieghere #include <stdint.h> 28*672d2c12SJonas Devlieghere #include <string.h> 29bf9a7730SZachary Turner 30bf9a7730SZachary Turner using namespace lldb_private; 31bf9a7730SZachary Turner 32bf9a7730SZachary Turner class Pool { 33bf9a7730SZachary Turner public: 34bf9a7730SZachary Turner typedef const char *StringPoolValueType; 35bf9a7730SZachary Turner typedef llvm::StringMap<StringPoolValueType, llvm::BumpPtrAllocator> 36bf9a7730SZachary Turner StringPool; 37bf9a7730SZachary Turner typedef llvm::StringMapEntry<StringPoolValueType> StringPoolEntryType; 38bf9a7730SZachary Turner 39bf9a7730SZachary Turner static StringPoolEntryType & 40bf9a7730SZachary Turner GetStringMapEntryFromKeyData(const char *keyData) { 418070bf0aSPavel Labath return StringPoolEntryType::GetStringMapEntryFromKeyData(keyData); 42bf9a7730SZachary Turner } 43bf9a7730SZachary Turner 448070bf0aSPavel Labath static size_t GetConstCStringLength(const char *ccstr) { 45bf9a7730SZachary Turner if (ccstr != nullptr) { 4605097246SAdrian Prantl // Since the entry is read only, and we derive the entry entirely from 4705097246SAdrian Prantl // the pointer, we don't need the lock. 48bf9a7730SZachary Turner const StringPoolEntryType &entry = GetStringMapEntryFromKeyData(ccstr); 49bf9a7730SZachary Turner return entry.getKey().size(); 50bf9a7730SZachary Turner } 51bf9a7730SZachary Turner return 0; 52bf9a7730SZachary Turner } 53bf9a7730SZachary Turner 54bf9a7730SZachary Turner StringPoolValueType GetMangledCounterpart(const char *ccstr) const { 55bf9a7730SZachary Turner if (ccstr != nullptr) { 56bf9a7730SZachary Turner const uint8_t h = hash(llvm::StringRef(ccstr)); 57bf9a7730SZachary Turner llvm::sys::SmartScopedReader<false> rlock(m_string_pools[h].m_mutex); 58bf9a7730SZachary Turner return GetStringMapEntryFromKeyData(ccstr).getValue(); 59bf9a7730SZachary Turner } 60bf9a7730SZachary Turner return nullptr; 61bf9a7730SZachary Turner } 62bf9a7730SZachary Turner 63bf9a7730SZachary Turner bool SetMangledCounterparts(const char *key_ccstr, const char *value_ccstr) { 64bf9a7730SZachary Turner if (key_ccstr != nullptr && value_ccstr != nullptr) { 65bf9a7730SZachary Turner { 66bf9a7730SZachary Turner const uint8_t h = hash(llvm::StringRef(key_ccstr)); 67bf9a7730SZachary Turner llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex); 68bf9a7730SZachary Turner GetStringMapEntryFromKeyData(key_ccstr).setValue(value_ccstr); 69bf9a7730SZachary Turner } 70bf9a7730SZachary Turner { 71bf9a7730SZachary Turner const uint8_t h = hash(llvm::StringRef(value_ccstr)); 72bf9a7730SZachary Turner llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex); 73bf9a7730SZachary Turner GetStringMapEntryFromKeyData(value_ccstr).setValue(key_ccstr); 74bf9a7730SZachary Turner } 75bf9a7730SZachary Turner return true; 76bf9a7730SZachary Turner } 77bf9a7730SZachary Turner return false; 78bf9a7730SZachary Turner } 79bf9a7730SZachary Turner 80bf9a7730SZachary Turner const char *GetConstCString(const char *cstr) { 81bf9a7730SZachary Turner if (cstr != nullptr) 82bf9a7730SZachary Turner return GetConstCStringWithLength(cstr, strlen(cstr)); 83bf9a7730SZachary Turner return nullptr; 84bf9a7730SZachary Turner } 85bf9a7730SZachary Turner 86bf9a7730SZachary Turner const char *GetConstCStringWithLength(const char *cstr, size_t cstr_len) { 87bf9a7730SZachary Turner if (cstr != nullptr) 88bf9a7730SZachary Turner return GetConstCStringWithStringRef(llvm::StringRef(cstr, cstr_len)); 89bf9a7730SZachary Turner return nullptr; 90bf9a7730SZachary Turner } 91bf9a7730SZachary Turner 92bf9a7730SZachary Turner const char *GetConstCStringWithStringRef(const llvm::StringRef &string_ref) { 93bf9a7730SZachary Turner if (string_ref.data()) { 94bf9a7730SZachary Turner const uint8_t h = hash(string_ref); 95bf9a7730SZachary Turner 96bf9a7730SZachary Turner { 97bf9a7730SZachary Turner llvm::sys::SmartScopedReader<false> rlock(m_string_pools[h].m_mutex); 98bf9a7730SZachary Turner auto it = m_string_pools[h].m_string_map.find(string_ref); 99bf9a7730SZachary Turner if (it != m_string_pools[h].m_string_map.end()) 100bf9a7730SZachary Turner return it->getKeyData(); 101bf9a7730SZachary Turner } 102bf9a7730SZachary Turner 103bf9a7730SZachary Turner llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex); 104bf9a7730SZachary Turner StringPoolEntryType &entry = 105bf9a7730SZachary Turner *m_string_pools[h] 106bf9a7730SZachary Turner .m_string_map.insert(std::make_pair(string_ref, nullptr)) 107bf9a7730SZachary Turner .first; 108bf9a7730SZachary Turner return entry.getKeyData(); 109bf9a7730SZachary Turner } 110bf9a7730SZachary Turner return nullptr; 111bf9a7730SZachary Turner } 112bf9a7730SZachary Turner 113bf9a7730SZachary Turner const char * 11419a357adSPavel Labath GetConstCStringAndSetMangledCounterPart(llvm::StringRef demangled, 115bf9a7730SZachary Turner const char *mangled_ccstr) { 116bf9a7730SZachary Turner const char *demangled_ccstr = nullptr; 117bf9a7730SZachary Turner 118bf9a7730SZachary Turner { 11919a357adSPavel Labath const uint8_t h = hash(demangled); 120bf9a7730SZachary Turner llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex); 121bf9a7730SZachary Turner 1222397a2b6SStefan Granitz // Make or update string pool entry with the mangled counterpart 1232397a2b6SStefan Granitz StringPool &map = m_string_pools[h].m_string_map; 1242397a2b6SStefan Granitz StringPoolEntryType &entry = *map.try_emplace(demangled).first; 1252397a2b6SStefan Granitz 1262397a2b6SStefan Granitz entry.second = mangled_ccstr; 127bf9a7730SZachary Turner 128bf9a7730SZachary Turner // Extract the const version of the demangled_cstr 129bf9a7730SZachary Turner demangled_ccstr = entry.getKeyData(); 130bf9a7730SZachary Turner } 131bf9a7730SZachary Turner 132bf9a7730SZachary Turner { 133bf9a7730SZachary Turner // Now assign the demangled const string as the counterpart of the 134bf9a7730SZachary Turner // mangled const string... 135bf9a7730SZachary Turner const uint8_t h = hash(llvm::StringRef(mangled_ccstr)); 136bf9a7730SZachary Turner llvm::sys::SmartScopedWriter<false> wlock(m_string_pools[h].m_mutex); 137bf9a7730SZachary Turner GetStringMapEntryFromKeyData(mangled_ccstr).setValue(demangled_ccstr); 138bf9a7730SZachary Turner } 139bf9a7730SZachary Turner 140bf9a7730SZachary Turner // Return the constant demangled C string 141bf9a7730SZachary Turner return demangled_ccstr; 142bf9a7730SZachary Turner } 143bf9a7730SZachary Turner 144bf9a7730SZachary Turner const char *GetConstTrimmedCStringWithLength(const char *cstr, 145bf9a7730SZachary Turner size_t cstr_len) { 146bf9a7730SZachary Turner if (cstr != nullptr) { 147bf9a7730SZachary Turner const size_t trimmed_len = std::min<size_t>(strlen(cstr), cstr_len); 148bf9a7730SZachary Turner return GetConstCStringWithLength(cstr, trimmed_len); 149bf9a7730SZachary Turner } 150bf9a7730SZachary Turner return nullptr; 151bf9a7730SZachary Turner } 152bf9a7730SZachary Turner 153bf9a7730SZachary Turner //------------------------------------------------------------------ 15405097246SAdrian Prantl // Return the size in bytes that this object and any items in its collection 15505097246SAdrian Prantl // of uniqued strings + data count values takes in memory. 156bf9a7730SZachary Turner //------------------------------------------------------------------ 157bf9a7730SZachary Turner size_t MemorySize() const { 158bf9a7730SZachary Turner size_t mem_size = sizeof(Pool); 159bf9a7730SZachary Turner for (const auto &pool : m_string_pools) { 160bf9a7730SZachary Turner llvm::sys::SmartScopedReader<false> rlock(pool.m_mutex); 161bf9a7730SZachary Turner for (const auto &entry : pool.m_string_map) 162bf9a7730SZachary Turner mem_size += sizeof(StringPoolEntryType) + entry.getKey().size(); 163bf9a7730SZachary Turner } 164bf9a7730SZachary Turner return mem_size; 165bf9a7730SZachary Turner } 166bf9a7730SZachary Turner 167bf9a7730SZachary Turner protected: 168bf9a7730SZachary Turner uint8_t hash(const llvm::StringRef &s) const { 169560ce2c7SJonas Devlieghere uint32_t h = llvm::djbHash(s); 170bf9a7730SZachary Turner return ((h >> 24) ^ (h >> 16) ^ (h >> 8) ^ h) & 0xff; 171bf9a7730SZachary Turner } 172bf9a7730SZachary Turner 173bf9a7730SZachary Turner struct PoolEntry { 174bf9a7730SZachary Turner mutable llvm::sys::SmartRWMutex<false> m_mutex; 175bf9a7730SZachary Turner StringPool m_string_map; 176bf9a7730SZachary Turner }; 177bf9a7730SZachary Turner 178bf9a7730SZachary Turner std::array<PoolEntry, 256> m_string_pools; 179bf9a7730SZachary Turner }; 180bf9a7730SZachary Turner 181bf9a7730SZachary Turner //---------------------------------------------------------------------- 18205097246SAdrian Prantl // Frameworks and dylibs aren't supposed to have global C++ initializers so we 18305097246SAdrian Prantl // hide the string pool in a static function so that it will get initialized on 18405097246SAdrian Prantl // the first call to this static function. 185bf9a7730SZachary Turner // 18605097246SAdrian Prantl // Note, for now we make the string pool a pointer to the pool, because we 18705097246SAdrian Prantl // can't guarantee that some objects won't get destroyed after the global 18805097246SAdrian Prantl // destructor chain is run, and trying to make sure no destructors touch 18905097246SAdrian Prantl // ConstStrings is difficult. So we leak the pool instead. 190bf9a7730SZachary Turner //---------------------------------------------------------------------- 191bf9a7730SZachary Turner static Pool &StringPool() { 192c5f28e2aSKamil Rytarowski static llvm::once_flag g_pool_initialization_flag; 193bf9a7730SZachary Turner static Pool *g_string_pool = nullptr; 194bf9a7730SZachary Turner 195c5f28e2aSKamil Rytarowski llvm::call_once(g_pool_initialization_flag, 196bf9a7730SZachary Turner []() { g_string_pool = new Pool(); }); 197bf9a7730SZachary Turner 198bf9a7730SZachary Turner return *g_string_pool; 199bf9a7730SZachary Turner } 200bf9a7730SZachary Turner 201bf9a7730SZachary Turner ConstString::ConstString(const char *cstr) 202bf9a7730SZachary Turner : m_string(StringPool().GetConstCString(cstr)) {} 203bf9a7730SZachary Turner 204bf9a7730SZachary Turner ConstString::ConstString(const char *cstr, size_t cstr_len) 205bf9a7730SZachary Turner : m_string(StringPool().GetConstCStringWithLength(cstr, cstr_len)) {} 206bf9a7730SZachary Turner 207bf9a7730SZachary Turner ConstString::ConstString(const llvm::StringRef &s) 208bf9a7730SZachary Turner : m_string(StringPool().GetConstCStringWithLength(s.data(), s.size())) {} 209bf9a7730SZachary Turner 210bf9a7730SZachary Turner bool ConstString::operator<(const ConstString &rhs) const { 211bf9a7730SZachary Turner if (m_string == rhs.m_string) 212bf9a7730SZachary Turner return false; 213bf9a7730SZachary Turner 2148070bf0aSPavel Labath llvm::StringRef lhs_string_ref(GetStringRef()); 2158070bf0aSPavel Labath llvm::StringRef rhs_string_ref(rhs.GetStringRef()); 216bf9a7730SZachary Turner 217bf9a7730SZachary Turner // If both have valid C strings, then return the comparison 218bf9a7730SZachary Turner if (lhs_string_ref.data() && rhs_string_ref.data()) 219bf9a7730SZachary Turner return lhs_string_ref < rhs_string_ref; 220bf9a7730SZachary Turner 221bf9a7730SZachary Turner // Else one of them was nullptr, so if LHS is nullptr then it is less than 222bf9a7730SZachary Turner return lhs_string_ref.data() == nullptr; 223bf9a7730SZachary Turner } 224bf9a7730SZachary Turner 225bf9a7730SZachary Turner Stream &lldb_private::operator<<(Stream &s, const ConstString &str) { 226bf9a7730SZachary Turner const char *cstr = str.GetCString(); 227bf9a7730SZachary Turner if (cstr != nullptr) 228bf9a7730SZachary Turner s << cstr; 229bf9a7730SZachary Turner 230bf9a7730SZachary Turner return s; 231bf9a7730SZachary Turner } 232bf9a7730SZachary Turner 233bf9a7730SZachary Turner size_t ConstString::GetLength() const { 2348070bf0aSPavel Labath return Pool::GetConstCStringLength(m_string); 235bf9a7730SZachary Turner } 236bf9a7730SZachary Turner 237bf9a7730SZachary Turner bool ConstString::Equals(const ConstString &lhs, const ConstString &rhs, 238bf9a7730SZachary Turner const bool case_sensitive) { 239bf9a7730SZachary Turner if (lhs.m_string == rhs.m_string) 240bf9a7730SZachary Turner return true; 241bf9a7730SZachary Turner 242bf9a7730SZachary Turner // Since the pointers weren't equal, and identical ConstStrings always have 24305097246SAdrian Prantl // identical pointers, the result must be false for case sensitive equality 24405097246SAdrian Prantl // test. 245bf9a7730SZachary Turner if (case_sensitive) 246bf9a7730SZachary Turner return false; 247bf9a7730SZachary Turner 248bf9a7730SZachary Turner // perform case insensitive equality test 2498070bf0aSPavel Labath llvm::StringRef lhs_string_ref(lhs.GetStringRef()); 2508070bf0aSPavel Labath llvm::StringRef rhs_string_ref(rhs.GetStringRef()); 251bf9a7730SZachary Turner return lhs_string_ref.equals_lower(rhs_string_ref); 252bf9a7730SZachary Turner } 253bf9a7730SZachary Turner 254bf9a7730SZachary Turner int ConstString::Compare(const ConstString &lhs, const ConstString &rhs, 255bf9a7730SZachary Turner const bool case_sensitive) { 256bf9a7730SZachary Turner // If the iterators are the same, this is the same string 257bf9a7730SZachary Turner const char *lhs_cstr = lhs.m_string; 258bf9a7730SZachary Turner const char *rhs_cstr = rhs.m_string; 259bf9a7730SZachary Turner if (lhs_cstr == rhs_cstr) 260bf9a7730SZachary Turner return 0; 261bf9a7730SZachary Turner if (lhs_cstr && rhs_cstr) { 2628070bf0aSPavel Labath llvm::StringRef lhs_string_ref(lhs.GetStringRef()); 2638070bf0aSPavel Labath llvm::StringRef rhs_string_ref(rhs.GetStringRef()); 264bf9a7730SZachary Turner 265bf9a7730SZachary Turner if (case_sensitive) { 266bf9a7730SZachary Turner return lhs_string_ref.compare(rhs_string_ref); 267bf9a7730SZachary Turner } else { 268bf9a7730SZachary Turner return lhs_string_ref.compare_lower(rhs_string_ref); 269bf9a7730SZachary Turner } 270bf9a7730SZachary Turner } 271bf9a7730SZachary Turner 272bf9a7730SZachary Turner if (lhs_cstr) 273bf9a7730SZachary Turner return +1; // LHS isn't nullptr but RHS is 274bf9a7730SZachary Turner else 275bf9a7730SZachary Turner return -1; // LHS is nullptr but RHS isn't 276bf9a7730SZachary Turner } 277bf9a7730SZachary Turner 278bf9a7730SZachary Turner void ConstString::Dump(Stream *s, const char *fail_value) const { 279bf9a7730SZachary Turner if (s != nullptr) { 280bf9a7730SZachary Turner const char *cstr = AsCString(fail_value); 281bf9a7730SZachary Turner if (cstr != nullptr) 282bf9a7730SZachary Turner s->PutCString(cstr); 283bf9a7730SZachary Turner } 284bf9a7730SZachary Turner } 285bf9a7730SZachary Turner 286bf9a7730SZachary Turner void ConstString::DumpDebug(Stream *s) const { 287bf9a7730SZachary Turner const char *cstr = GetCString(); 288bf9a7730SZachary Turner size_t cstr_len = GetLength(); 289bf9a7730SZachary Turner // Only print the parens if we have a non-nullptr string 290bf9a7730SZachary Turner const char *parens = cstr ? "\"" : ""; 291bf9a7730SZachary Turner s->Printf("%*p: ConstString, string = %s%s%s, length = %" PRIu64, 292bf9a7730SZachary Turner static_cast<int>(sizeof(void *) * 2), 293bf9a7730SZachary Turner static_cast<const void *>(this), parens, cstr, parens, 294bf9a7730SZachary Turner static_cast<uint64_t>(cstr_len)); 295bf9a7730SZachary Turner } 296bf9a7730SZachary Turner 297bf9a7730SZachary Turner void ConstString::SetCString(const char *cstr) { 298bf9a7730SZachary Turner m_string = StringPool().GetConstCString(cstr); 299bf9a7730SZachary Turner } 300bf9a7730SZachary Turner 301bf9a7730SZachary Turner void ConstString::SetString(const llvm::StringRef &s) { 302bf9a7730SZachary Turner m_string = StringPool().GetConstCStringWithLength(s.data(), s.size()); 303bf9a7730SZachary Turner } 304bf9a7730SZachary Turner 30519a357adSPavel Labath void ConstString::SetStringWithMangledCounterpart(llvm::StringRef demangled, 306bf9a7730SZachary Turner const ConstString &mangled) { 307bf9a7730SZachary Turner m_string = StringPool().GetConstCStringAndSetMangledCounterPart( 308bf9a7730SZachary Turner demangled, mangled.m_string); 309bf9a7730SZachary Turner } 310bf9a7730SZachary Turner 311bf9a7730SZachary Turner bool ConstString::GetMangledCounterpart(ConstString &counterpart) const { 312bf9a7730SZachary Turner counterpart.m_string = StringPool().GetMangledCounterpart(m_string); 313bf9a7730SZachary Turner return (bool)counterpart; 314bf9a7730SZachary Turner } 315bf9a7730SZachary Turner 316bf9a7730SZachary Turner void ConstString::SetCStringWithLength(const char *cstr, size_t cstr_len) { 317bf9a7730SZachary Turner m_string = StringPool().GetConstCStringWithLength(cstr, cstr_len); 318bf9a7730SZachary Turner } 319bf9a7730SZachary Turner 320bf9a7730SZachary Turner void ConstString::SetTrimmedCStringWithLength(const char *cstr, 321bf9a7730SZachary Turner size_t cstr_len) { 322bf9a7730SZachary Turner m_string = StringPool().GetConstTrimmedCStringWithLength(cstr, cstr_len); 323bf9a7730SZachary Turner } 324bf9a7730SZachary Turner 325bf9a7730SZachary Turner size_t ConstString::StaticMemorySize() { 326bf9a7730SZachary Turner // Get the size of the static string pool 327bf9a7730SZachary Turner return StringPool().MemorySize(); 328bf9a7730SZachary Turner } 3293b7e1981SPavel Labath 3303b7e1981SPavel Labath void llvm::format_provider<ConstString>::format(const ConstString &CS, 3313b7e1981SPavel Labath llvm::raw_ostream &OS, 3323b7e1981SPavel Labath llvm::StringRef Options) { 3333b7e1981SPavel Labath format_provider<StringRef>::format(CS.AsCString(), OS, Options); 3343b7e1981SPavel Labath } 335