diff options
Diffstat (limited to 'llvm')
46 files changed, 16413 insertions, 778 deletions
diff --git a/llvm/include/llvm/ExecutionEngine/Orc/Shared/SymbolFilter.h b/llvm/include/llvm/ExecutionEngine/Orc/Shared/SymbolFilter.h new file mode 100644 index 0000000..5170893 --- /dev/null +++ b/llvm/include/llvm/ExecutionEngine/Orc/Shared/SymbolFilter.h @@ -0,0 +1,173 @@ +//===- SymbolFilter.h - Utilities for Symbol Filtering ---------*- 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 +// +//===----------------------------------------------------------------------===// + +#ifndef LLVM_EXECUTIONENGINE_ORC_SHARED_SYMBOLFILTER_H +#define LLVM_EXECUTIONENGINE_ORC_SHARED_SYMBOLFILTER_H + +#include "llvm/ExecutionEngine/Orc/Shared/SimplePackedSerialization.h" + +#include <cmath> +#include <type_traits> +#include <vector> + +namespace llvm { +namespace orc { + +namespace shared { +using SPSBloomFilter = + SPSTuple<bool, uint32_t, uint32_t, uint32_t, SPSSequence<uint64_t>>; +} + +class BloomFilter { +public: + using HashFunc = std::function<uint32_t(StringRef)>; + + BloomFilter() = default; + BloomFilter(BloomFilter &&) noexcept = default; + BloomFilter &operator=(BloomFilter &&) noexcept = default; + BloomFilter(const BloomFilter &) = delete; + BloomFilter &operator=(const BloomFilter &) = delete; + + BloomFilter(uint32_t SymbolCount, float FalsePositiveRate, HashFunc hashFn) + : HashFn(std::move(hashFn)) { + initialize(SymbolCount, FalsePositiveRate); + } + bool isInitialized() const { return Initialized; } + + void add(StringRef Sym) { + assert(Initialized); + addHash(HashFn(Sym)); + } + + bool mayContain(StringRef Sym) const { + return !isEmpty() && testHash(HashFn(Sym)); + } + + bool isEmpty() const { return SymbolCount == 0; } + +private: + friend class shared::SPSSerializationTraits<shared::SPSBloomFilter, + BloomFilter>; + static constexpr uint32_t BitsPerEntry = 64; + + bool Initialized = false; + uint32_t SymbolCount = 0; + uint32_t BloomSize = 0; + uint32_t BloomShift = 0; + std::vector<uint64_t> BloomTable; + HashFunc HashFn; + + void initialize(uint32_t SymCount, float FalsePositiveRate) { + assert(SymCount > 0); + SymbolCount = SymCount; + Initialized = true; + + float ln2 = std::log(2.0f); + float M = -1.0f * SymbolCount * std::log(FalsePositiveRate) / (ln2 * ln2); + BloomSize = static_cast<uint32_t>(std::ceil(M / BitsPerEntry)); + BloomShift = std::min(6u, log2ceil(SymbolCount)); + BloomTable.resize(BloomSize, 0); + } + + void addHash(uint32_t Hash) { + uint32_t Hash2 = Hash >> BloomShift; + uint32_t N = (Hash / BitsPerEntry) % BloomSize; + uint64_t Mask = + (1ULL << (Hash % BitsPerEntry)) | (1ULL << (Hash2 % BitsPerEntry)); + BloomTable[N] |= Mask; + } + + bool testHash(uint32_t Hash) const { + uint32_t Hash2 = Hash >> BloomShift; + uint32_t N = (Hash / BitsPerEntry) % BloomSize; + uint64_t Mask = + (1ULL << (Hash % BitsPerEntry)) | (1ULL << (Hash2 % BitsPerEntry)); + return (BloomTable[N] & Mask) == Mask; + } + + static constexpr uint32_t log2ceil(uint32_t V) { + return V <= 1 ? 0 : 32 - countl_zero(V - 1); + } +}; + +class BloomFilterBuilder { +public: + using HashFunc = BloomFilter::HashFunc; + + BloomFilterBuilder() = default; + + BloomFilterBuilder &setFalsePositiveRate(float Rate) { + assert(Rate > 0.0f && Rate < 1.0f); + FalsePositiveRate = Rate; + return *this; + } + + BloomFilterBuilder &setHashFunction(HashFunc Fn) { + HashFn = std::move(Fn); + return *this; + } + + BloomFilter build(ArrayRef<StringRef> Symbols) const { + assert(!Symbols.empty() && "Cannot build filter from empty symbol list."); + BloomFilter F(static_cast<uint32_t>(Symbols.size()), FalsePositiveRate, + HashFn); + for (const auto &Sym : Symbols) + F.add(Sym); + + return F; + } + +private: + float FalsePositiveRate = 0.02f; + HashFunc HashFn = [](StringRef S) -> uint32_t { + uint32_t H = 5381; + for (char C : S) + H = ((H << 5) + H) + static_cast<uint8_t>(C); // H * 33 + C + return H; + }; +}; + +namespace shared { + +template <> class SPSSerializationTraits<SPSBloomFilter, BloomFilter> { +public: + static size_t size(const BloomFilter &Filter) { + return SPSBloomFilter::AsArgList::size( + Filter.Initialized, Filter.SymbolCount, Filter.BloomSize, + Filter.BloomShift, Filter.BloomTable); + } + + static bool serialize(SPSOutputBuffer &OB, const BloomFilter &Filter) { + return SPSBloomFilter::AsArgList::serialize( + OB, Filter.Initialized, Filter.SymbolCount, Filter.BloomSize, + Filter.BloomShift, Filter.BloomTable); + } + + static bool deserialize(SPSInputBuffer &IB, BloomFilter &Filter) { + bool IsInitialized; + uint32_t SymbolCount = 0, BloomSize = 0, BloomShift = 0; + std::vector<uint64_t> BloomTable; + + if (!SPSBloomFilter::AsArgList::deserialize( + IB, IsInitialized, SymbolCount, BloomSize, BloomShift, BloomTable)) + return false; + + Filter.Initialized = IsInitialized; + Filter.SymbolCount = SymbolCount; + Filter.BloomSize = BloomSize; + Filter.BloomShift = BloomShift; + Filter.BloomTable = std::move(BloomTable); + + return true; + } +}; + +} // end namespace shared +} // end namespace orc +} // end namespace llvm +#endif // LLVM_EXECUTIONENGINE_ORC_SHARED_SYMBOLFILTER_H diff --git a/llvm/include/llvm/ExecutionEngine/Orc/TargetProcess/LibraryResolver.h b/llvm/include/llvm/ExecutionEngine/Orc/TargetProcess/LibraryResolver.h new file mode 100644 index 0000000..9182995 --- /dev/null +++ b/llvm/include/llvm/ExecutionEngine/Orc/TargetProcess/LibraryResolver.h @@ -0,0 +1,511 @@ +//===- LibraryResolver.h - Automatic Library Symbol Resolution -*- 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 provides support for automatically searching symbols across +// dynamic libraries that have not yet been loaded. +// +//===----------------------------------------------------------------------===// + +#ifndef LLVM_EXECUTIONENGINE_ORC_TARGETPROCESS_LIBRARYRESOLVER_H +#define LLVM_EXECUTIONENGINE_ORC_TARGETPROCESS_LIBRARYRESOLVER_H + +#include "llvm/ADT/FunctionExtras.h" +#include "llvm/ExecutionEngine/Orc/Shared/SymbolFilter.h" +#include "llvm/ExecutionEngine/Orc/TargetProcess/LibraryScanner.h" +#include "llvm/Support/Path.h" + +#include <atomic> +#include <shared_mutex> +#include <unordered_map> + +namespace llvm { +namespace orc { + +/// Manages library metadata and state for symbol resolution. +/// +/// Tracks libraries by load state and kind (user/system), and stores +/// associated Bloom filters and hash maps to speed up symbol lookups. +/// Thread-safe for concurrent access. +class LibraryManager { +public: + enum class LibState : uint8_t { Unloaded = 0, Loaded = 1, Queried = 2 }; + + class LibraryInfo { + public: + LibraryInfo(const LibraryInfo &) = delete; + LibraryInfo &operator=(const LibraryInfo &) = delete; + + LibraryInfo(std::string FilePath, LibState S, PathType K, + std::optional<BloomFilter> Filter = std::nullopt) + : FilePath(std::move(FilePath)), S(S), K(K), Filter(std::move(Filter)) { + } + + StringRef getBasePath() const { return sys::path::parent_path(FilePath); } + StringRef getFileName() const { return sys::path::filename(FilePath); } + + std::string getFullPath() const { return FilePath; } + + void setFilter(BloomFilter F) { + std::lock_guard<std::shared_mutex> Lock(Mtx); + if (Filter) + return; + Filter.emplace(std::move(F)); + } + + void ensureFilterBuilt(const BloomFilterBuilder &FB, + ArrayRef<StringRef> Symbols) { + std::lock_guard<std::shared_mutex> Lock(Mtx); + if (Filter) + return; + Filter.emplace(FB.build(Symbols)); + } + + bool mayContain(StringRef Symbol) const { + assert(hasFilter()); + std::shared_lock<std::shared_mutex> Lock(Mtx); + return Filter->mayContain(Symbol); + } + + bool hasFilter() const { + std::shared_lock<std::shared_mutex> Lock(Mtx); + return Filter.has_value(); + } + + LibState getState() const { return S.load(); } + PathType getKind() const { return K; } + + void setState(LibState s) { S.store(s); } + + bool operator==(const LibraryInfo &other) const { + return FilePath == other.FilePath; + } + + private: + std::string FilePath; + std::atomic<LibState> S; + PathType K; + std::optional<BloomFilter> Filter; + mutable std::shared_mutex Mtx; + }; + + /// A read-only view of libraries filtered by state and kind. + /// + /// Lets you loop over only the libraries in a map that match a given State + /// and PathType. + class FilteredView { + public: + using Map = StringMap<std::shared_ptr<LibraryInfo>>; + using Iterator = typename Map::const_iterator; + class FilterIterator { + public: + FilterIterator(Iterator it_, Iterator end_, LibState S, PathType K) + : it(it_), end(end_), S(S), K(K) { + advance(); + } + + bool operator!=(const FilterIterator &other) const { + return it != other.it; + } + + const std::shared_ptr<LibraryInfo> &operator*() const { + return it->second; + } + + FilterIterator &operator++() { + ++it; + advance(); + return *this; + } + + private: + void advance() { + for (; it != end; ++it) + if (it->second->getState() == S && it->second->getKind() == K) + break; + } + Iterator it; + Iterator end; + LibState S; + PathType K; + }; + FilteredView(Iterator begin, Iterator end, LibState s, PathType k) + : mapBegin(begin), mapEnd(end), state(s), kind(k) {} + + FilterIterator begin() const { + return FilterIterator(mapBegin, mapEnd, state, kind); + } + + FilterIterator end() const { + return FilterIterator(mapEnd, mapEnd, state, kind); + } + + private: + Iterator mapBegin; + Iterator mapEnd; + LibState state; + PathType kind; + }; + +private: + StringMap<std::shared_ptr<LibraryInfo>> Libraries; + mutable std::shared_mutex Mtx; + +public: + using LibraryVisitor = std::function<bool(const LibraryInfo &)>; + + LibraryManager() = default; + ~LibraryManager() = default; + + bool addLibrary(std::string Path, PathType Kind, + std::optional<BloomFilter> Filter = std::nullopt) { + std::unique_lock<std::shared_mutex> Lock(Mtx); + if (Libraries.count(Path) > 0) + return false; + Libraries.insert({std::move(Path), + std::make_shared<LibraryInfo>(Path, LibState::Unloaded, + Kind, std::move(Filter))}); + return true; + } + + bool hasLibrary(StringRef Path) const { + std::shared_lock<std::shared_mutex> Lock(Mtx); + if (Libraries.count(Path) > 0) + return true; + return false; + } + + void removeLibrary(StringRef Path) { + std::unique_lock<std::shared_mutex> Lock(Mtx); + auto I = Libraries.find(Path); + if (I == Libraries.end()) + return; + Libraries.erase(I); + } + + void markLoaded(StringRef Path) { + std::unique_lock<std::shared_mutex> Lock(Mtx); + if (auto It = Libraries.find(Path); It != Libraries.end()) + It->second->setState(LibState::Loaded); + } + + void markQueried(StringRef Path) { + std::unique_lock<std::shared_mutex> Lock(Mtx); + if (auto It = Libraries.find(Path); It != Libraries.end()) + It->second->setState(LibState::Queried); + } + + std::shared_ptr<LibraryInfo> getLibrary(StringRef Path) { + std::shared_lock<std::shared_mutex> Lock(Mtx); + if (auto It = Libraries.find(Path); It != Libraries.end()) + return It->second; + return nullptr; + } + + FilteredView getView(LibState S, PathType K) const { + std::shared_lock<std::shared_mutex> Lock(Mtx); + return FilteredView(Libraries.begin(), Libraries.end(), S, K); + } + + void forEachLibrary(const LibraryVisitor &visitor) const { + std::unique_lock<std::shared_mutex> Lock(Mtx); + for (const auto &[_, entry] : Libraries) { + if (!visitor(*entry)) + break; + } + } + + bool isLoaded(StringRef Path) const { + std::unique_lock<std::shared_mutex> Lock(Mtx); + if (auto It = Libraries.find(Path.str()); It != Libraries.end()) + return It->second->getState() == LibState::Loaded; + return false; + } + + bool isQueried(StringRef Path) const { + std::unique_lock<std::shared_mutex> Lock(Mtx); + if (auto It = Libraries.find(Path.str()); It != Libraries.end()) + return It->second->getState() == LibState::Queried; + return false; + } + + void clear() { + std::unique_lock<std::shared_mutex> Lock(Mtx); + Libraries.clear(); + } +}; + +using LibraryInfo = LibraryManager::LibraryInfo; + +struct SearchPlanEntry { + LibraryManager::LibState State; // Loaded, Queried, Unloaded + PathType Type; // User, System +}; + +struct SearchPolicy { + std::vector<SearchPlanEntry> Plan; + + static SearchPolicy defaultPlan() { + return {{{LibraryManager::LibState::Loaded, PathType::User}, + {LibraryManager::LibState::Queried, PathType::User}, + {LibraryManager::LibState::Unloaded, PathType::User}, + {LibraryManager::LibState::Loaded, PathType::System}, + {LibraryManager::LibState::Queried, PathType::System}, + {LibraryManager::LibState::Unloaded, PathType::System}}}; + } +}; + +struct SymbolEnumeratorOptions { + enum Filter : uint32_t { + None = 0, + IgnoreUndefined = 1 << 0, + IgnoreWeak = 1 << 1, + IgnoreIndirect = 1 << 2, + IgnoreHidden = 1 << 3, + IgnoreNonGlobal = 1 << 4 + }; + + static SymbolEnumeratorOptions defaultOptions() { + return {Filter::IgnoreUndefined | Filter::IgnoreWeak | + Filter::IgnoreIndirect}; + } + uint32_t FilterFlags = Filter::None; +}; + +struct SearchConfig { + SearchPolicy Policy; + SymbolEnumeratorOptions Options; + + SearchConfig() + : Policy(SearchPolicy::defaultPlan()), // default plan + Options(SymbolEnumeratorOptions::defaultOptions()) {} +}; + +/// Scans libraries and resolves Symbols across user and system paths. +/// +/// Supports symbol enumeration and filtering via SymbolEnumerator, and tracks +/// symbol resolution results through SymbolQuery. Thread-safe and uses +/// LibraryScanHelper for efficient path resolution and caching. +class LibraryResolver { + friend class LibraryResolutionDriver; + +public: + class SymbolEnumerator { + public: + enum class EnumerateResult { Continue, Stop, Error }; + + using OnEachSymbolFn = std::function<EnumerateResult(StringRef Sym)>; + + static bool enumerateSymbols(StringRef Path, OnEachSymbolFn OnEach, + const SymbolEnumeratorOptions &Opts); + }; + + /// Tracks a set of symbols and the libraries where they are resolved. + /// + /// SymbolQuery is used to keep track of which symbols have been resolved + /// to which libraries. It supports concurrent read/write access using a + /// shared mutex, allowing multiple readers or a single writer at a time. + class SymbolQuery { + public: + /// Holds the result for a single symbol. + struct Result { + std::string Name; + std::string ResolvedLibPath; + }; + + private: + mutable std::shared_mutex Mtx; + StringMap<Result> Results; + std::atomic<size_t> ResolvedCount = 0; + + public: + explicit SymbolQuery(const std::vector<std::string> &Symbols) { + for (const auto &s : Symbols) { + if (!Results.contains(s)) + Results.insert({s, Result{s, ""}}); + } + } + + SmallVector<StringRef> getUnresolvedSymbols() const { + SmallVector<StringRef> Unresolved; + std::shared_lock<std::shared_mutex> Lock(Mtx); + for (const auto &[name, res] : Results) { + if (res.ResolvedLibPath.empty()) + Unresolved.push_back(name); + } + return Unresolved; + } + + void resolve(StringRef Sym, const std::string &LibPath) { + std::unique_lock<std::shared_mutex> Lock(Mtx); + auto It = Results.find(Sym); + if (It != Results.end() && It->second.ResolvedLibPath.empty()) { + It->second.ResolvedLibPath = LibPath; + ResolvedCount.fetch_add(1, std::memory_order_relaxed); + } + } + + bool allResolved() const { + return ResolvedCount.load(std::memory_order_relaxed) == Results.size(); + } + + bool hasUnresolved() const { + return ResolvedCount.load(std::memory_order_relaxed) < Results.size(); + } + + std::optional<StringRef> getResolvedLib(StringRef Sym) const { + std::shared_lock<std::shared_mutex> Lock(Mtx); + auto It = Results.find(Sym); + if (It != Results.end() && !It->second.ResolvedLibPath.empty()) + return StringRef(It->second.ResolvedLibPath); + return std::nullopt; + } + + bool isResolved(StringRef Sym) const { + std::shared_lock<std::shared_mutex> Lock(Mtx); + auto It = Results.find(Sym.str()); + return It != Results.end() && !It->second.ResolvedLibPath.empty(); + } + + std::vector<const Result *> getAllResults() const { + std::shared_lock<std::shared_mutex> Lock(Mtx); + std::vector<const Result *> Out; + Out.reserve(Results.size()); + for (const auto &[_, res] : Results) + Out.push_back(&res); + return Out; + } + }; + + struct Setup { + std::vector<std::string> BasePaths; + std::shared_ptr<LibraryPathCache> Cache; + std::shared_ptr<PathResolver> PResolver; + + size_t ScanBatchSize = 0; + + LibraryScanner::ShouldScanFn ShouldScanCall = [](StringRef) { + return true; + }; + + BloomFilterBuilder FilterBuilder = BloomFilterBuilder(); + + static Setup + create(std::vector<std::string> BasePaths, + std::shared_ptr<LibraryPathCache> existingCache = nullptr, + std::shared_ptr<PathResolver> existingResolver = nullptr, + LibraryScanner::ShouldScanFn customShouldScan = nullptr) { + Setup S; + S.BasePaths = std::move(BasePaths); + + S.Cache = + existingCache ? existingCache : std::make_shared<LibraryPathCache>(); + + S.PResolver = existingResolver ? existingResolver + : std::make_shared<PathResolver>(S.Cache); + + if (customShouldScan) + S.ShouldScanCall = std::move(customShouldScan); + + return S; + } + }; + + LibraryResolver() = delete; + explicit LibraryResolver(const Setup &S); + ~LibraryResolver() = default; + + using OnSearchComplete = unique_function<void(SymbolQuery &)>; + + void dump() { + int i = 0; + LibMgr.forEachLibrary([&](const LibraryInfo &Lib) -> bool { + dbgs() << ++i << ". Library Path : " << Lib.getFullPath() << " -> \n\t\t:" + << " ({Type : (" + << (Lib.getKind() == PathType::User ? "User" : "System") + << ") }, { State : " + << (Lib.getState() == LibraryManager::LibState::Loaded + ? "Loaded" + : "Unloaded") + << "})\n"; + return true; + }); + } + + void searchSymbolsInLibraries(std::vector<std::string> &SymList, + OnSearchComplete OnComplete, + const SearchConfig &Config = SearchConfig()); + +private: + bool scanLibrariesIfNeeded(PathType K, size_t BatchSize = 0); + void resolveSymbolsInLibrary(LibraryInfo &Lib, SymbolQuery &Q, + const SymbolEnumeratorOptions &Opts); + bool + symbolExistsInLibrary(const LibraryInfo &Lib, StringRef Sym, + std::vector<std::string> *MatchedSymbols = nullptr); + + bool symbolExistsInLibrary(const LibraryInfo &Lib, StringRef SymName, + std::vector<std::string> *AllSymbols, + const SymbolEnumeratorOptions &Opts); + + std::shared_ptr<LibraryPathCache> LibPathCache; + std::shared_ptr<PathResolver> LibPathResolver; + LibraryScanHelper ScanHelper; + BloomFilterBuilder FB; + LibraryManager LibMgr; + LibraryScanner::ShouldScanFn ShouldScanCall; + size_t scanBatchSize; +}; + +using SymbolEnumerator = LibraryResolver::SymbolEnumerator; +using SymbolQuery = LibraryResolver::SymbolQuery; +using EnumerateResult = SymbolEnumerator::EnumerateResult; + +class LibraryResolutionDriver { +public: + static std::unique_ptr<LibraryResolutionDriver> + create(const LibraryResolver::Setup &S); + + void addScanPath(const std::string &Path, PathType Kind); + bool markLibraryLoaded(StringRef Path); + bool markLibraryUnLoaded(StringRef Path); + bool isLibraryLoaded(StringRef Path) const { + return LR->LibMgr.isLoaded(Path); + } + + void resetAll() { + LR->LibMgr.clear(); + LR->ScanHelper.resetToScan(); + LR->LibPathCache->clear(); + } + + void scanAll(size_t BatchSize = 0) { + LR->scanLibrariesIfNeeded(PathType::User, BatchSize); + LR->scanLibrariesIfNeeded(PathType::System, BatchSize); + } + + void scan(PathType PK, size_t BatchSize = 0) { + LR->scanLibrariesIfNeeded(PK, BatchSize); + } + + void resolveSymbols(std::vector<std::string> Symbols, + LibraryResolver::OnSearchComplete OnCompletion, + const SearchConfig &Config = SearchConfig()); + + ~LibraryResolutionDriver() = default; + +private: + LibraryResolutionDriver(std::unique_ptr<LibraryResolver> L) + : LR(std::move(L)) {} + + std::unique_ptr<LibraryResolver> LR; +}; + +} // end namespace orc +} // end namespace llvm + +#endif // LLVM_EXECUTIONENGINE_ORC_TARGETPROCESS_LIBRARYRESOLVER_H diff --git a/llvm/include/llvm/ExecutionEngine/Orc/TargetProcess/LibraryScanner.h b/llvm/include/llvm/ExecutionEngine/Orc/TargetProcess/LibraryScanner.h new file mode 100644 index 0000000..d1c2013 --- /dev/null +++ b/llvm/include/llvm/ExecutionEngine/Orc/TargetProcess/LibraryScanner.h @@ -0,0 +1,474 @@ +//===- LibraryScanner.h - Scanner for Shared Libraries ---------*- 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 provides functionality for scanning dynamic (shared) libraries. +// +//===----------------------------------------------------------------------===// + +#ifndef LLVM_EXECUTIONENGINE_ORC_TARGETPROCESS_LIBRARYSCANNER_H +#define LLVM_EXECUTIONENGINE_ORC_TARGETPROCESS_LIBRARYSCANNER_H + +#include "llvm/ADT/FunctionExtras.h" +#include "llvm/ADT/SmallVector.h" +#include "llvm/ADT/StringRef.h" +#include "llvm/ADT/StringSet.h" +#include "llvm/Object/ObjectFile.h" +#include "llvm/Support/Allocator.h" +#include "llvm/Support/Error.h" +#include "llvm/Support/StringSaver.h" + +#include <atomic> +#include <mutex> +#include <queue> +#include <shared_mutex> +#include <string> +#include <unordered_map> +#include <unordered_set> + +namespace llvm { +namespace orc { + +class LibraryManager; + +class LibraryPathCache { + friend class PathResolver; + +public: + LibraryPathCache() = default; + + void clear(bool isRealPathCache = false) { + std::unique_lock<std::shared_mutex> lock(Mtx); + Seen.clear(); + if (isRealPathCache) { + RealPathCache.clear(); +#ifndef _WIN32 + ReadlinkCache.clear(); + LstatCache.clear(); +#endif + } + } + + void markSeen(const std::string &CanonPath) { + std::unique_lock<std::shared_mutex> lock(Mtx); + Seen.insert(CanonPath); + } + + bool hasSeen(StringRef CanonPath) const { + std::shared_lock<std::shared_mutex> lock(Mtx); + return Seen.contains(CanonPath); + } + + bool hasSeenOrMark(StringRef CanonPath) { + std::string s = CanonPath.str(); + { + std::shared_lock<std::shared_mutex> lock(Mtx); + if (Seen.contains(s)) + return true; + } + { + std::unique_lock<std::shared_mutex> lock(Mtx); + Seen.insert(s); + } + return false; + } + +private: + mutable std::shared_mutex Mtx; + + struct PathInfo { + std::string canonicalPath; + std::error_code ErrnoCode; + }; + + void insert_realpath(StringRef Path, const PathInfo &Info) { + std::unique_lock<std::shared_mutex> lock(Mtx); + RealPathCache.insert({Path, Info}); + } + + std::optional<PathInfo> read_realpath(StringRef Path) const { + std::shared_lock<std::shared_mutex> lock(Mtx); + auto It = RealPathCache.find(Path); + if (It != RealPathCache.end()) + return It->second; + + return std::nullopt; + } + + StringSet<> Seen; + StringMap<PathInfo> RealPathCache; + +#ifndef _WIN32 + StringMap<std::string> ReadlinkCache; + StringMap<mode_t> LstatCache; + + void insert_link(StringRef Path, const std::string &s) { + std::unique_lock<std::shared_mutex> lock(Mtx); + ReadlinkCache.insert({Path, s}); + } + + std::optional<std::string> read_link(StringRef Path) const { + std::shared_lock<std::shared_mutex> lock(Mtx); + auto It = ReadlinkCache.find(Path); + if (It != ReadlinkCache.end()) + return It->second; + + return std::nullopt; + } + + void insert_lstat(StringRef Path, mode_t m) { + std::unique_lock<std::shared_mutex> lock(Mtx); + LstatCache.insert({Path, m}); + } + + std::optional<mode_t> read_lstat(StringRef Path) const { + std::shared_lock<std::shared_mutex> lock(Mtx); + auto It = LstatCache.find(Path); + if (It != LstatCache.end()) + return It->second; + + return std::nullopt; + } + +#endif +}; + +/// Resolves file system paths with optional caching of results. +/// +/// Supports lstat, readlink, and realpath operations. Can resolve paths +/// relative to a base and handle symbolic links. Caches results to reduce +/// repeated system calls when enabled. +class PathResolver { +private: + std::shared_ptr<LibraryPathCache> LibPathCache; + +public: + PathResolver(std::shared_ptr<LibraryPathCache> cache) + : LibPathCache(std::move(cache)) {} + + std::optional<std::string> resolve(StringRef Path, std::error_code &ec) { + return realpathCached(Path, ec); + } +#ifndef _WIN32 + mode_t lstatCached(StringRef Path); + std::optional<std::string> readlinkCached(StringRef Path); +#endif + std::optional<std::string> realpathCached(StringRef Path, std::error_code &ec, + StringRef base = "", + bool baseIsResolved = false, + long symloopLevel = 40); +}; + +/// Performs placeholder substitution in dynamic library paths. +/// +/// Configures known placeholders (like @loader_path) and replaces them +/// in input paths with their resolved values. +class DylibSubstitutor { +public: + void configure(StringRef loaderPath); + + std::string substitute(StringRef input) const { + for (const auto &[ph, value] : Placeholders) { + if (input.starts_with_insensitive(ph)) + return (Twine(value) + input.drop_front(ph.size())).str(); + } + return input.str(); + } + +private: + StringMap<std::string> Placeholders; +}; + +/// Validates and normalizes dynamic library paths. +/// +/// Uses a `PathResolver` to resolve paths to their canonical form and +/// checks whether they point to valid shared libraries. +class DylibPathValidator { +public: + DylibPathValidator(PathResolver &PR) : LibPathResolver(PR) {} + + static bool isSharedLibrary(StringRef Path); + + std::optional<std::string> normalize(StringRef Path) const { + std::error_code ec; + auto real = LibPathResolver.resolve(Path, ec); + if (!real || ec) + return std::nullopt; + + return real; + } + + /// Validate the given path as a shared library. + std::optional<std::string> validate(StringRef Path) const { + auto realOpt = normalize(Path); + if (!realOpt) + return std::nullopt; + + if (!isSharedLibrary(*realOpt)) + return std::nullopt; + + return realOpt; + } + +private: + PathResolver &LibPathResolver; +}; + +enum class SearchPathType { + RPath, + UsrOrSys, + RunPath, +}; + +struct SearchPathConfig { + ArrayRef<StringRef> Paths; + SearchPathType type; +}; + +class SearchPathResolver { +public: + SearchPathResolver(const SearchPathConfig &Cfg, + StringRef PlaceholderPrefix = "") + : Kind(Cfg.type), PlaceholderPrefix(PlaceholderPrefix) { + for (auto &path : Cfg.Paths) + Paths.emplace_back(path.str()); + } + + std::optional<std::string> resolve(StringRef libStem, + const DylibSubstitutor &Subst, + DylibPathValidator &Validator) const; + SearchPathType searchPathType() const { return Kind; } + +private: + std::vector<std::string> Paths; + SearchPathType Kind; + std::string PlaceholderPrefix; +}; + +class DylibResolverImpl { +public: + DylibResolverImpl(DylibSubstitutor Substitutor, DylibPathValidator &Validator, + std::vector<SearchPathResolver> Resolvers) + : Substitutor(std::move(Substitutor)), Validator(Validator), + Resolvers(std::move(Resolvers)) {} + + std::optional<std::string> resolve(StringRef Stem, + bool VariateLibStem = false) const; + +private: + std::optional<std::string> tryWithExtensions(StringRef libstem) const; + + DylibSubstitutor Substitutor; + DylibPathValidator &Validator; + std::vector<SearchPathResolver> Resolvers; +}; + +class DylibResolver { +public: + DylibResolver(DylibPathValidator &Validator) : Validator(Validator) {} + + void configure(StringRef loaderPath, + ArrayRef<SearchPathConfig> SearchPathCfg) { + DylibSubstitutor Substitutor; + Substitutor.configure(loaderPath); + + std::vector<SearchPathResolver> Resolvers; + for (const auto &cfg : SearchPathCfg) { + Resolvers.emplace_back(cfg, + cfg.type == SearchPathType::RPath ? "@rpath" : ""); + } + + impl_ = std::make_unique<DylibResolverImpl>( + std::move(Substitutor), Validator, std::move(Resolvers)); + } + + std::optional<std::string> resolve(StringRef libStem, + bool VariateLibStem = false) const { + if (!impl_) + return std::nullopt; + return impl_->resolve(libStem, VariateLibStem); + } + + static std::string resolvelinkerFlag(StringRef libStem, + StringRef loaderPath) { + DylibSubstitutor Substitutor; + Substitutor.configure(loaderPath); + return Substitutor.substitute(libStem); + } + +private: + DylibPathValidator &Validator; + std::unique_ptr<DylibResolverImpl> impl_; +}; + +enum class PathType : uint8_t { User, System, Unknown }; + +enum class ScanState : uint8_t { NotScanned, Scanning, Scanned }; + +struct LibrarySearchPath { + std::string BasePath; // Canonical base directory path + PathType Kind; // User or System + std::atomic<ScanState> State; + + LibrarySearchPath(std::string Base, PathType K) + : BasePath(std::move(Base)), Kind(K), State(ScanState::NotScanned) {} +}; + +/// Scans and tracks libraries for symbol resolution. +/// +/// Maintains a list of library paths to scan, caches scanned units, +/// and resolves paths canonically for consistent tracking. +class LibraryScanHelper { +public: + explicit LibraryScanHelper(const std::vector<std::string> &SPaths, + std::shared_ptr<LibraryPathCache> LibPathCache, + std::shared_ptr<PathResolver> LibPathResolver) + : LibPathCache(std::move(LibPathCache)), + LibPathResolver(std::move(LibPathResolver)) { + DEBUG_WITH_TYPE( + "orc", dbgs() << "LibraryScanHelper::LibraryScanHelper: base paths : " + << SPaths.size() << "\n";); + for (const auto &p : SPaths) + addBasePath(p); + } + + void + addBasePath(const std::string &P, + PathType Kind = + PathType::Unknown); // Add a canonical directory for scanning + std::vector<std::shared_ptr<LibrarySearchPath>> + getNextBatch(PathType Kind, size_t batchSize); + + bool leftToScan(PathType K) const; + void resetToScan(); + + bool isTrackedBasePath(StringRef P) const; + std::vector<std::shared_ptr<LibrarySearchPath>> getAllUnits() const; + + SmallVector<StringRef> getSearchPaths() const { + SmallVector<StringRef> SearchPaths; + for (const auto &[_, SP] : LibSearchPaths) + SearchPaths.push_back(SP->BasePath); + return SearchPaths; + } + + PathResolver &getPathResolver() const { return *LibPathResolver; } + + LibraryPathCache &getCache() const { return *LibPathCache; } + + bool hasSeenOrMark(StringRef P) const { + return LibPathCache->hasSeenOrMark(P); + } + + std::optional<std::string> resolve(StringRef P, std::error_code &ec) const { + return LibPathResolver->resolve(P.str(), ec); + } + +private: + std::string resolveCanonical(StringRef P, std::error_code &ec) const; + PathType classifyKind(StringRef P) const; + + mutable std::shared_mutex Mtx; + std::shared_ptr<LibraryPathCache> LibPathCache; + std::shared_ptr<PathResolver> LibPathResolver; + + StringMap<std::shared_ptr<LibrarySearchPath>> + LibSearchPaths; // key: canonical path + std::deque<StringRef> UnscannedUsr; + std::deque<StringRef> UnscannedSys; +}; + +/// Loads an object file and provides access to it. +/// +/// Owns the underlying `ObjectFile` and ensures it is valid. +/// Any errors encountered during construction are stored and +/// returned when attempting to access the file. +class ObjectFileLoader { +public: + /// Construct an object file loader from the given path. + explicit ObjectFileLoader(StringRef Path) { + auto ObjOrErr = loadObjectFileWithOwnership(Path); + if (ObjOrErr) + Obj = std::move(*ObjOrErr); + else { + consumeError(std::move(Err)); + Err = ObjOrErr.takeError(); + } + } + + ObjectFileLoader(const ObjectFileLoader &) = delete; + ObjectFileLoader &operator=(const ObjectFileLoader &) = delete; + + ObjectFileLoader(ObjectFileLoader &&) = default; + ObjectFileLoader &operator=(ObjectFileLoader &&) = default; + + /// Get the loaded object file, or return an error if loading failed. + Expected<object::ObjectFile &> getObjectFile() { + if (Err) + return std::move(Err); + return *Obj.getBinary(); + } + + static bool isArchitectureCompatible(const object::ObjectFile &Obj); + +private: + object::OwningBinary<object::ObjectFile> Obj; + Error Err = Error::success(); + + static Expected<object::OwningBinary<object::ObjectFile>> + loadObjectFileWithOwnership(StringRef FilePath); +}; + +/// Scans libraries, resolves dependencies, and registers them. +class LibraryScanner { +public: + using ShouldScanFn = std::function<bool(StringRef)>; + + LibraryScanner( + LibraryScanHelper &H, LibraryManager &LibMgr, + ShouldScanFn ShouldScanCall = [](StringRef path) { return true; }) + : ScanHelper(H), LibMgr(LibMgr), + ShouldScanCall(std::move(ShouldScanCall)) {} + + void scanNext(PathType Kind, size_t batchSize = 1); + + /// Dependency info for a library. + struct LibraryDepsInfo { + llvm::BumpPtrAllocator Alloc; + llvm::StringSaver Saver{Alloc}; + + SmallVector<StringRef, 2> rpath; + SmallVector<StringRef, 2> runPath; + SmallVector<StringRef, 4> deps; + bool isPIE = false; + + void addRPath(StringRef s) { rpath.push_back(Saver.save(s)); } + + void addRunPath(StringRef s) { runPath.push_back(Saver.save(s)); } + + void addDep(StringRef s) { deps.push_back(Saver.save(s)); } + }; + +private: + LibraryScanHelper &ScanHelper; + LibraryManager &LibMgr; + ShouldScanFn ShouldScanCall; + + std::optional<std::string> shouldScan(StringRef FilePath); + Expected<LibraryDepsInfo> extractDeps(StringRef FilePath); + + void handleLibrary(StringRef P, PathType K, int level = 1); + + void scanBaseDir(std::shared_ptr<LibrarySearchPath> U); +}; + +using LibraryDepsInfo = LibraryScanner::LibraryDepsInfo; + +} // end namespace orc +} // end namespace llvm + +#endif // LLVM_EXECUTIONENGINE_ORC_TARGETPROCESS_LIBRARYSCANNER_H diff --git a/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp b/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp index bf1abfe..58983cb 100644 --- a/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp +++ b/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp @@ -1172,6 +1172,12 @@ bool DAGTypeLegalizer::SoftenFloatOperand(SDNode *N, unsigned OpNo) { case ISD::FAKE_USE: Res = SoftenFloatOp_FAKE_USE(N); break; + case ISD::STACKMAP: + Res = SoftenFloatOp_STACKMAP(N, OpNo); + break; + case ISD::PATCHPOINT: + Res = SoftenFloatOp_PATCHPOINT(N, OpNo); + break; } // If the result is null, the sub-method took care of registering results etc. @@ -1512,6 +1518,20 @@ SDValue DAGTypeLegalizer::SoftenFloatOp_FAKE_USE(SDNode *N) { N->getOperand(0), Op1); } +SDValue DAGTypeLegalizer::SoftenFloatOp_STACKMAP(SDNode *N, unsigned OpNo) { + assert(OpNo > 1); // Because the first two arguments are guaranteed legal. + SmallVector<SDValue> NewOps(N->ops()); + NewOps[OpNo] = GetSoftenedFloat(NewOps[OpNo]); + return SDValue(DAG.UpdateNodeOperands(N, NewOps), 0); +} + +SDValue DAGTypeLegalizer::SoftenFloatOp_PATCHPOINT(SDNode *N, unsigned OpNo) { + assert(OpNo >= 7); + SmallVector<SDValue> NewOps(N->ops()); + NewOps[OpNo] = GetSoftenedFloat(NewOps[OpNo]); + return SDValue(DAG.UpdateNodeOperands(N, NewOps), 0); +} + //===----------------------------------------------------------------------===// // Float Result Expansion //===----------------------------------------------------------------------===// diff --git a/llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.h b/llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.h index 9656a30..ede522e 100644 --- a/llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.h +++ b/llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.h @@ -658,6 +658,8 @@ private: SDValue SoftenFloatOp_ATOMIC_STORE(SDNode *N, unsigned OpNo); SDValue SoftenFloatOp_FCOPYSIGN(SDNode *N); SDValue SoftenFloatOp_FAKE_USE(SDNode *N); + SDValue SoftenFloatOp_STACKMAP(SDNode *N, unsigned OpNo); + SDValue SoftenFloatOp_PATCHPOINT(SDNode *N, unsigned OpNo); //===--------------------------------------------------------------------===// // Float Expansion Support: LegalizeFloatTypes.cpp diff --git a/llvm/lib/DebugInfo/CodeView/LazyRandomTypeCollection.cpp b/llvm/lib/DebugInfo/CodeView/LazyRandomTypeCollection.cpp index 6c23ba8..23ab534 100644 --- a/llvm/lib/DebugInfo/CodeView/LazyRandomTypeCollection.cpp +++ b/llvm/lib/DebugInfo/CodeView/LazyRandomTypeCollection.cpp @@ -102,7 +102,8 @@ std::optional<CVType> LazyRandomTypeCollection::tryGetType(TypeIndex Index) { return std::nullopt; } - assert(contains(Index)); + if (!contains(Index)) + return std::nullopt; return Records[Index.toArrayIndex()].Type; } diff --git a/llvm/lib/ExecutionEngine/Orc/TargetProcess/CMakeLists.txt b/llvm/lib/ExecutionEngine/Orc/TargetProcess/CMakeLists.txt index 9275586..ca8192b 100644 --- a/llvm/lib/ExecutionEngine/Orc/TargetProcess/CMakeLists.txt +++ b/llvm/lib/ExecutionEngine/Orc/TargetProcess/CMakeLists.txt @@ -16,9 +16,11 @@ add_llvm_component_library(LLVMOrcTargetProcess ExecutorSharedMemoryMapperService.cpp DefaultHostBootstrapValues.cpp ExecutorResolver.cpp + LibraryResolver.cpp JITLoaderGDB.cpp JITLoaderPerf.cpp JITLoaderVTune.cpp + LibraryScanner.cpp OrcRTBootstrap.cpp RegisterEHFrames.cpp SimpleExecutorDylibManager.cpp @@ -36,6 +38,8 @@ add_llvm_component_library(LLVMOrcTargetProcess LINK_COMPONENTS ${intel_jit_profiling} + BinaryFormat + Object OrcShared Support TargetParser diff --git a/llvm/lib/ExecutionEngine/Orc/TargetProcess/LibraryResolver.cpp b/llvm/lib/ExecutionEngine/Orc/TargetProcess/LibraryResolver.cpp new file mode 100644 index 0000000..35da82a --- /dev/null +++ b/llvm/lib/ExecutionEngine/Orc/TargetProcess/LibraryResolver.cpp @@ -0,0 +1,370 @@ +//===- LibraryResolver.cpp - Library Resolution of Unresolved Symbols ---===// +// +// 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 +// +//===----------------------------------------------------------------------===// +// +// Library resolution impl for unresolved symbols +// +//===----------------------------------------------------------------------===// + +#include "llvm/ExecutionEngine/Orc/TargetProcess/LibraryResolver.h" +#include "llvm/ExecutionEngine/Orc/TargetProcess/LibraryScanner.h" + +#include "llvm/ADT/StringSet.h" + +#include "llvm/BinaryFormat/MachO.h" +#include "llvm/Object/COFF.h" +#include "llvm/Object/ELF.h" +#include "llvm/Object/ELFObjectFile.h" +#include "llvm/Object/MachO.h" +#include "llvm/Object/ObjectFile.h" +#include "llvm/Support/Error.h" + +#include <mutex> +#include <thread> + +#define DEBUG_TYPE "orc-resolver" + +namespace llvm::orc { + +LibraryResolver::LibraryResolver(const LibraryResolver::Setup &S) + : LibPathCache(S.Cache ? S.Cache : std::make_shared<LibraryPathCache>()), + LibPathResolver(S.PResolver + ? S.PResolver + : std::make_shared<PathResolver>(LibPathCache)), + ScanHelper(S.BasePaths, LibPathCache, LibPathResolver), + FB(S.FilterBuilder), LibMgr(), + ShouldScanCall(S.ShouldScanCall ? S.ShouldScanCall + : [](StringRef) -> bool { return true; }), + scanBatchSize(S.ScanBatchSize) { + + if (ScanHelper.getAllUnits().empty()) { + LLVM_DEBUG(dbgs() << "Warning: No base paths provided for scanning.\n"); + } +} + +std::unique_ptr<LibraryResolutionDriver> +LibraryResolutionDriver::create(const LibraryResolver::Setup &S) { + auto LR = std::make_unique<LibraryResolver>(S); + return std::unique_ptr<LibraryResolutionDriver>( + new LibraryResolutionDriver(std::move(LR))); +} + +void LibraryResolutionDriver::addScanPath(const std::string &Path, PathType K) { + LR->ScanHelper.addBasePath(Path, K); +} + +bool LibraryResolutionDriver::markLibraryLoaded(StringRef Path) { + auto Lib = LR->LibMgr.getLibrary(Path); + if (!Lib) + return false; + + Lib->setState(LibraryManager::LibState::Loaded); + + return true; +} + +bool LibraryResolutionDriver::markLibraryUnLoaded(StringRef Path) { + auto Lib = LR->LibMgr.getLibrary(Path); + if (!Lib) + return false; + + Lib->setState(LibraryManager::LibState::Unloaded); + + return true; +} + +void LibraryResolutionDriver::resolveSymbols( + std::vector<std::string> Syms, + LibraryResolver::OnSearchComplete OnCompletion, + const SearchConfig &Config) { + LR->searchSymbolsInLibraries(Syms, std::move(OnCompletion), Config); +} + +static bool shouldIgnoreSymbol(const object::SymbolRef &Sym, + uint32_t IgnoreFlags) { + Expected<uint32_t> FlagsOrErr = Sym.getFlags(); + if (!FlagsOrErr) { + consumeError(FlagsOrErr.takeError()); + return true; + } + + uint32_t Flags = *FlagsOrErr; + + using Filter = SymbolEnumeratorOptions; + if ((IgnoreFlags & Filter::IgnoreUndefined) && + (Flags & object::SymbolRef::SF_Undefined)) + return true; + if ((IgnoreFlags & Filter::IgnoreIndirect) && + (Flags & object::SymbolRef::SF_Indirect)) + return true; + if ((IgnoreFlags & Filter::IgnoreWeak) && + (Flags & object::SymbolRef::SF_Weak)) + return true; + + return false; +} + +bool SymbolEnumerator::enumerateSymbols(StringRef Path, OnEachSymbolFn OnEach, + const SymbolEnumeratorOptions &Opts) { + if (Path.empty()) + return false; + + ObjectFileLoader ObjLoader(Path); + + auto ObjOrErr = ObjLoader.getObjectFile(); + if (!ObjOrErr) { + std::string ErrMsg; + handleAllErrors(ObjOrErr.takeError(), + [&](const ErrorInfoBase &EIB) { ErrMsg = EIB.message(); }); + LLVM_DEBUG(dbgs() << "Failed loading object file: " << Path + << "\nError: " << ErrMsg << "\n"); + return false; + } + + object::ObjectFile *Obj = &ObjOrErr.get(); + + auto processSymbolRange = + [&](object::ObjectFile::symbol_iterator_range Range) -> EnumerateResult { + for (const auto &Sym : Range) { + if (shouldIgnoreSymbol(Sym, Opts.FilterFlags)) + continue; + + auto NameOrErr = Sym.getName(); + if (!NameOrErr) { + consumeError(NameOrErr.takeError()); + continue; + } + + StringRef Name = *NameOrErr; + if (Name.empty()) + continue; + + EnumerateResult Res = OnEach(Name); + if (Res != EnumerateResult::Continue) + return Res; + } + return EnumerateResult::Continue; + }; + + EnumerateResult Res = processSymbolRange(Obj->symbols()); + if (Res != EnumerateResult::Continue) + return Res == EnumerateResult::Stop; + + if (Obj->isELF()) { + const auto *ElfObj = cast<object::ELFObjectFileBase>(Obj); + Res = processSymbolRange(ElfObj->getDynamicSymbolIterators()); + if (Res != EnumerateResult::Continue) + return Res == EnumerateResult::Stop; + } else if (Obj->isCOFF()) { + const auto *CoffObj = cast<object::COFFObjectFile>(Obj); + for (auto I = CoffObj->export_directory_begin(), + E = CoffObj->export_directory_end(); + I != E; ++I) { + StringRef Name; + if (I->getSymbolName(Name)) + continue; + if (Name.empty()) + continue; + + EnumerateResult Res = OnEach(Name); + if (Res != EnumerateResult::Continue) + return Res == EnumerateResult::Stop; + } + } else if (Obj->isMachO()) { + } + + return true; +} + +class SymbolSearchContext { +public: + SymbolSearchContext(SymbolQuery &Q) : Q(Q) {} + + bool hasSearched(LibraryInfo *Lib) const { return Searched.count(Lib); } + + void markSearched(LibraryInfo *Lib) { Searched.insert(Lib); } + + inline bool allResolved() const { return Q.allResolved(); } + + SymbolQuery &query() { return Q; } + +private: + SymbolQuery &Q; + DenseSet<LibraryInfo *> Searched; +}; + +void LibraryResolver::resolveSymbolsInLibrary( + LibraryInfo &Lib, SymbolQuery &UnresolvedSymbols, + const SymbolEnumeratorOptions &Opts) { + LLVM_DEBUG(dbgs() << "Checking unresolved symbols " + << " in library : " << Lib.getFileName() << "\n";); + StringSet<> DiscoveredSymbols; + + if (!UnresolvedSymbols.hasUnresolved()) { + LLVM_DEBUG(dbgs() << "Skipping library: " << Lib.getFullPath() + << " — unresolved symbols exist.\n";); + return; + } + + bool HasEnumerated = false; + auto enumerateSymbolsIfNeeded = [&]() { + if (HasEnumerated) + return; + + HasEnumerated = true; + + LLVM_DEBUG(dbgs() << "Enumerating symbols in library: " << Lib.getFullPath() + << "\n";); + SymbolEnumerator::enumerateSymbols( + Lib.getFullPath(), + [&](StringRef sym) { + DiscoveredSymbols.insert(sym); + return EnumerateResult::Continue; + }, + Opts); + + if (DiscoveredSymbols.empty()) { + LLVM_DEBUG(dbgs() << " No symbols and remove library : " + << Lib.getFullPath() << "\n";); + LibMgr.removeLibrary(Lib.getFullPath()); + return; + } + }; + + if (!Lib.hasFilter()) { + LLVM_DEBUG(dbgs() << "Building filter for library: " << Lib.getFullPath() + << "\n";); + enumerateSymbolsIfNeeded(); + SmallVector<StringRef> SymbolVec; + SymbolVec.reserve(DiscoveredSymbols.size()); + for (const auto &KV : DiscoveredSymbols) + SymbolVec.push_back(KV.first()); + + Lib.ensureFilterBuilt(FB, SymbolVec); + LLVM_DEBUG({ + dbgs() << "DiscoveredSymbols : " << DiscoveredSymbols.size() << "\n"; + for (const auto &KV : DiscoveredSymbols) + dbgs() << "DiscoveredSymbols : " << KV.first() << "\n"; + }); + } + + const auto &Unresolved = UnresolvedSymbols.getUnresolvedSymbols(); + bool HadAnySym = false; + LLVM_DEBUG(dbgs() << "Total unresolved symbols : " << Unresolved.size() + << "\n";); + for (const auto &Sym : Unresolved) { + if (Lib.mayContain(Sym)) { + LLVM_DEBUG(dbgs() << "Checking symbol '" << Sym + << "' in library: " << Lib.getFullPath() << "\n";); + enumerateSymbolsIfNeeded(); + if (DiscoveredSymbols.count(Sym) > 0) { + LLVM_DEBUG(dbgs() << " Resolved symbol: " << Sym + << " in library: " << Lib.getFullPath() << "\n";); + UnresolvedSymbols.resolve(Sym, Lib.getFullPath()); + HadAnySym = true; + } + } + } + + using LibraryState = LibraryManager::LibState; + if (HadAnySym && Lib.getState() != LibraryState::Loaded) + Lib.setState(LibraryState::Queried); +} + +void LibraryResolver::searchSymbolsInLibraries( + std::vector<std::string> &SymbolList, OnSearchComplete OnComplete, + const SearchConfig &Config) { + SymbolQuery Q(SymbolList); + + using LibraryState = LibraryManager::LibState; + using LibraryType = PathType; + auto tryResolveFrom = [&](LibraryState S, LibraryType K) { + LLVM_DEBUG(dbgs() << "Trying resolve from state=" << static_cast<int>(S) + << " type=" << static_cast<int>(K) << "\n";); + + SymbolSearchContext Ctx(Q); + while (!Ctx.allResolved()) { + + for (auto &Lib : LibMgr.getView(S, K)) { + if (Ctx.hasSearched(Lib.get())) + continue; + + // can use Async here? + resolveSymbolsInLibrary(*Lib, Ctx.query(), Config.Options); + Ctx.markSearched(Lib.get()); + + if (Ctx.allResolved()) + return; + } + + if (Ctx.allResolved()) + return; + + if (!scanLibrariesIfNeeded(K, scanBatchSize)) + break; // no more new libs to scan + } + }; + + for (const auto &[St, Ty] : Config.Policy.Plan) { + tryResolveFrom(St, Ty); + if (Q.allResolved()) + break; + } + + // done: + LLVM_DEBUG({ + dbgs() << "Search complete.\n"; + for (const auto &r : Q.getAllResults()) + dbgs() << "Resolved Symbol:" << r->Name << " -> " << r->ResolvedLibPath + << "\n"; + }); + + OnComplete(Q); +} + +bool LibraryResolver::scanLibrariesIfNeeded(PathType PK, size_t BatchSize) { + LLVM_DEBUG(dbgs() << "LibraryResolver::scanLibrariesIfNeeded: Scanning for " + << (PK == PathType::User ? "User" : "System") + << " libraries\n";); + if (!ScanHelper.leftToScan(PK)) + return false; + + LibraryScanner Scanner(ScanHelper, LibMgr, ShouldScanCall); + Scanner.scanNext(PK, BatchSize); + return true; +} + +bool LibraryResolver::symbolExistsInLibrary(const LibraryInfo &Lib, + StringRef SymName, + std::vector<std::string> *AllSyms) { + SymbolEnumeratorOptions Opts; + return symbolExistsInLibrary(Lib, SymName, AllSyms, Opts); +} + +bool LibraryResolver::symbolExistsInLibrary( + const LibraryInfo &Lib, StringRef SymName, + std::vector<std::string> *AllSyms, const SymbolEnumeratorOptions &Opts) { + bool Found = false; + + SymbolEnumerator::enumerateSymbols( + Lib.getFullPath(), + [&](StringRef Sym) { + if (AllSyms) + AllSyms->emplace_back(Sym.str()); + + if (Sym == SymName) { + Found = true; + } + + return EnumerateResult::Continue; + }, + Opts); + + return Found; +} + +} // end namespace llvm::orc diff --git a/llvm/lib/ExecutionEngine/Orc/TargetProcess/LibraryScanner.cpp b/llvm/lib/ExecutionEngine/Orc/TargetProcess/LibraryScanner.cpp new file mode 100644 index 0000000..d93f686 --- /dev/null +++ b/llvm/lib/ExecutionEngine/Orc/TargetProcess/LibraryScanner.cpp @@ -0,0 +1,1161 @@ +//===- LibraryScanner.cpp - Provide Library Scanning Implementation ----===// +// +// 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 +// +//===----------------------------------------------------------------------===// + +#include "llvm/ExecutionEngine/Orc/TargetProcess/LibraryScanner.h" +#include "llvm/ExecutionEngine/Orc/TargetProcess/LibraryResolver.h" + +#include "llvm/ADT/StringExtras.h" +#include "llvm/Object/COFF.h" +#include "llvm/Object/ELF.h" +#include "llvm/Object/ELFObjectFile.h" +#include "llvm/Object/ELFTypes.h" +#include "llvm/Object/MachO.h" +#include "llvm/Object/MachOUniversal.h" +#include "llvm/Object/ObjectFile.h" +#include "llvm/Support/Error.h" +#include "llvm/Support/FileSystem.h" +#include "llvm/Support/MemoryBuffer.h" +#include "llvm/Support/Path.h" +#include "llvm/Support/Program.h" +#include "llvm/TargetParser/Host.h" +#include "llvm/TargetParser/Triple.h" + +#ifdef LLVM_ON_UNIX +#include <sys/stat.h> +#include <unistd.h> +#endif // LLVM_ON_UNIX + +#ifdef __APPLE__ +#include <sys/stat.h> +#undef LC_LOAD_DYLIB +#undef LC_RPATH +#endif // __APPLE__ + +#define DEBUG_TYPE "orc-scanner" + +namespace llvm::orc { + +void handleError(Error Err, StringRef context = "") { + consumeError(handleErrors(std::move(Err), [&](const ErrorInfoBase &EIB) { + dbgs() << "LLVM Error"; + if (!context.empty()) + dbgs() << " [" << context << "]"; + dbgs() << ": " << EIB.message() << "\n"; + })); +} + +bool ObjectFileLoader::isArchitectureCompatible(const object::ObjectFile &Obj) { + Triple HostTriple(sys::getDefaultTargetTriple()); + Triple ObjTriple = Obj.makeTriple(); + + LLVM_DEBUG({ + dbgs() << "Host triple: " << HostTriple.str() + << ", Object triple: " << ObjTriple.str() << "\n"; + }); + + if (ObjTriple.getArch() != Triple::UnknownArch && + HostTriple.getArch() != ObjTriple.getArch()) + return false; + + if (ObjTriple.getOS() != Triple::UnknownOS && + HostTriple.getOS() != ObjTriple.getOS()) + return false; + + if (ObjTriple.getEnvironment() != Triple::UnknownEnvironment && + HostTriple.getEnvironment() != Triple::UnknownEnvironment && + HostTriple.getEnvironment() != ObjTriple.getEnvironment()) + return false; + + return true; +} + +Expected<object::OwningBinary<object::ObjectFile>> +ObjectFileLoader::loadObjectFileWithOwnership(StringRef FilePath) { + LLVM_DEBUG(dbgs() << "ObjectFileLoader: Attempting to open file " << FilePath + << "\n";); + auto BinOrErr = object::createBinary(FilePath); + if (!BinOrErr) { + LLVM_DEBUG(dbgs() << "ObjectFileLoader: Failed to open file " << FilePath + << "\n";); + return BinOrErr.takeError(); + } + + LLVM_DEBUG(dbgs() << "ObjectFileLoader: Successfully opened file " << FilePath + << "\n";); + + auto OwningBin = BinOrErr->takeBinary(); + object::Binary *Bin = OwningBin.first.get(); + + if (Bin->isArchive()) { + LLVM_DEBUG(dbgs() << "ObjectFileLoader: File is an archive, not supported: " + << FilePath << "\n";); + return createStringError(std::errc::invalid_argument, + "Archive files are not supported: %s", + FilePath.str().c_str()); + } + +#if defined(__APPLE__) + if (auto *UB = dyn_cast<object::MachOUniversalBinary>(Bin)) { + LLVM_DEBUG(dbgs() << "ObjectFileLoader: Detected Mach-O universal binary: " + << FilePath << "\n";); + for (auto ObjForArch : UB->objects()) { + auto ObjOrErr = ObjForArch.getAsObjectFile(); + if (!ObjOrErr) { + LLVM_DEBUG( + dbgs() + << "ObjectFileLoader: Skipping invalid architecture slice\n";); + + consumeError(ObjOrErr.takeError()); + continue; + } + + std::unique_ptr<object::ObjectFile> Obj = std::move(ObjOrErr.get()); + if (isArchitectureCompatible(*Obj)) { + LLVM_DEBUG( + dbgs() << "ObjectFileLoader: Found compatible object slice\n";); + + return object::OwningBinary<object::ObjectFile>( + std::move(Obj), std::move(OwningBin.second)); + + } else { + LLVM_DEBUG(dbgs() << "ObjectFileLoader: Incompatible architecture " + "slice skipped\n";); + } + } + LLVM_DEBUG(dbgs() << "ObjectFileLoader: No compatible slices found in " + "universal binary\n";); + return createStringError(inconvertibleErrorCode(), + "No compatible object found in fat binary: %s", + FilePath.str().c_str()); + } +#endif + + auto ObjOrErr = + object::ObjectFile::createObjectFile(Bin->getMemoryBufferRef()); + if (!ObjOrErr) { + LLVM_DEBUG(dbgs() << "ObjectFileLoader: Failed to create object file\n";); + return ObjOrErr.takeError(); + } + LLVM_DEBUG(dbgs() << "ObjectFileLoader: Detected object file\n";); + + std::unique_ptr<object::ObjectFile> Obj = std::move(*ObjOrErr); + if (!isArchitectureCompatible(*Obj)) { + LLVM_DEBUG(dbgs() << "ObjectFileLoader: Incompatible architecture: " + << FilePath << "\n";); + return createStringError(inconvertibleErrorCode(), + "Incompatible object file: %s", + FilePath.str().c_str()); + } + + LLVM_DEBUG(dbgs() << "ObjectFileLoader: Object file is compatible\n";); + + return object::OwningBinary<object::ObjectFile>(std::move(Obj), + std::move(OwningBin.second)); +} + +template <class ELFT> +bool isELFSharedLibrary(const object::ELFFile<ELFT> &ELFObj) { + if (ELFObj.getHeader().e_type != ELF::ET_DYN) + return false; + + auto PHOrErr = ELFObj.program_headers(); + if (!PHOrErr) { + consumeError(PHOrErr.takeError()); + return true; + } + + for (auto Phdr : *PHOrErr) { + if (Phdr.p_type == ELF::PT_INTERP) + return false; + } + + return true; +} + +bool isSharedLibraryObject(object::ObjectFile &Obj) { + if (Obj.isELF()) { + if (auto *ELF32LE = dyn_cast<object::ELF32LEObjectFile>(&Obj)) + return isELFSharedLibrary(ELF32LE->getELFFile()); + if (auto *ELF64LE = dyn_cast<object::ELF64LEObjectFile>(&Obj)) + return isELFSharedLibrary(ELF64LE->getELFFile()); + if (auto *ELF32BE = dyn_cast<object::ELF32BEObjectFile>(&Obj)) + return isELFSharedLibrary(ELF32BE->getELFFile()); + if (auto *ELF64BE = dyn_cast<object::ELF64BEObjectFile>(&Obj)) + return isELFSharedLibrary(ELF64BE->getELFFile()); + } else if (Obj.isMachO()) { + const object::MachOObjectFile *MachO = + dyn_cast<object::MachOObjectFile>(&Obj); + if (!MachO) { + LLVM_DEBUG(dbgs() << "Failed to cast to MachOObjectFile.\n";); + return false; + } + LLVM_DEBUG({ + bool Result = + MachO->getHeader().filetype == MachO::HeaderFileType::MH_DYLIB; + dbgs() << "Mach-O filetype: " << MachO->getHeader().filetype + << " (MH_DYLIB == " << MachO::HeaderFileType::MH_DYLIB + << "), shared: " << Result << "\n"; + }); + + return MachO->getHeader().filetype == MachO::HeaderFileType::MH_DYLIB; + } else if (Obj.isCOFF()) { + const object::COFFObjectFile *coff = dyn_cast<object::COFFObjectFile>(&Obj); + if (!coff) + return false; + return coff->getCharacteristics() & COFF::IMAGE_FILE_DLL; + } else { + LLVM_DEBUG(dbgs() << "Binary is not an ObjectFile.\n";); + } + + return false; +} + +bool DylibPathValidator::isSharedLibrary(StringRef Path) { + LLVM_DEBUG(dbgs() << "Checking if path is a shared library: " << Path + << "\n";); + + auto FileType = sys::fs::get_file_type(Path, /*Follow*/ true); + if (FileType != sys::fs::file_type::regular_file) { + LLVM_DEBUG(dbgs() << "File type is not a regular file for path: " << Path + << "\n";); + return false; + } + + file_magic MagicCode; + identify_magic(Path, MagicCode); + + // Skip archives. + if (MagicCode == file_magic::archive) + return false; + + // Universal binary handling. +#if defined(__APPLE__) + if (MagicCode == file_magic::macho_universal_binary) { + ObjectFileLoader ObjLoader(Path); + auto ObjOrErr = ObjLoader.getObjectFile(); + if (!ObjOrErr) { + consumeError(ObjOrErr.takeError()); + return false; + } + return isSharedLibraryObject(ObjOrErr.get()); + } +#endif + + // Object file inspection for PE/COFF, ELF, and Mach-O + bool NeedsObjectInspection = +#if defined(_WIN32) + (MagicCode == file_magic::pecoff_executable); +#elif defined(__APPLE__) + (MagicCode == file_magic::macho_fixed_virtual_memory_shared_lib || + MagicCode == file_magic::macho_dynamically_linked_shared_lib || + MagicCode == file_magic::macho_dynamically_linked_shared_lib_stub); +#elif defined(LLVM_ON_UNIX) +#ifdef __CYGWIN__ + (MagicCode == file_magic::pecoff_executable); +#else + (MagicCode == file_magic::elf_shared_object); +#endif +#else +#error "Unsupported platform." +#endif + + if (NeedsObjectInspection) { + ObjectFileLoader ObjLoader(Path); + auto ObjOrErr = ObjLoader.getObjectFile(); + if (!ObjOrErr) { + consumeError(ObjOrErr.takeError()); + return false; + } + return isSharedLibraryObject(ObjOrErr.get()); + } + + LLVM_DEBUG(dbgs() << "Path is not identified as a shared library: " << Path + << "\n";); + return false; +} + +void DylibSubstitutor::configure(StringRef LoaderPath) { + SmallString<512> ExecPath(sys::fs::getMainExecutable(nullptr, nullptr)); + sys::path::remove_filename(ExecPath); + + SmallString<512> LoaderDir; + if (LoaderPath.empty()) { + LoaderDir = ExecPath; + } else { + LoaderDir = LoaderPath.str(); + if (!sys::fs::is_directory(LoaderPath)) + sys::path::remove_filename(LoaderDir); + } + +#ifdef __APPLE__ + Placeholders["@loader_path"] = std::string(LoaderDir); + Placeholders["@executable_path"] = std::string(ExecPath); +#else + Placeholders["$origin"] = std::string(LoaderDir); +#endif +} + +std::optional<std::string> +SearchPathResolver::resolve(StringRef Stem, const DylibSubstitutor &Subst, + DylibPathValidator &Validator) const { + for (const auto &SP : Paths) { + std::string Base = Subst.substitute(SP); + + SmallString<512> FullPath(Base); + if (!PlaceholderPrefix.empty() && + Stem.starts_with_insensitive(PlaceholderPrefix)) + FullPath.append(Stem.drop_front(PlaceholderPrefix.size())); + else + sys::path::append(FullPath, Stem); + + LLVM_DEBUG(dbgs() << "SearchPathResolver::resolve FullPath = " << FullPath + << "\n";); + + if (auto Valid = Validator.validate(FullPath.str())) + return Valid; + } + + return std::nullopt; +} + +std::optional<std::string> +DylibResolverImpl::tryWithExtensions(StringRef LibStem) const { + LLVM_DEBUG(dbgs() << "tryWithExtensions: baseName = " << LibStem << "\n";); + SmallVector<SmallString<256>, 8> Candidates; + + // Add extensions by platform +#if defined(__APPLE__) + Candidates.emplace_back(LibStem); + Candidates.back() += ".dylib"; +#elif defined(_WIN32) + Candidates.emplace_back(LibStem); + Candidates.back() += ".dll"; +#else + Candidates.emplace_back(LibStem); + Candidates.back() += ".so"; +#endif + + // Optionally try "lib" prefix if not already there + StringRef FileName = sys::path::filename(LibStem); + StringRef Base = sys::path::parent_path(LibStem); + if (!FileName.starts_with("lib")) { + SmallString<256> WithPrefix(Base); + if (!WithPrefix.empty()) + sys::path::append(WithPrefix, ""); // ensure separator if needed + WithPrefix += "lib"; + WithPrefix += FileName; + +#if defined(__APPLE__) + WithPrefix += ".dylib"; +#elif defined(_WIN32) + WithPrefix += ".dll"; +#else + WithPrefix += ".so"; +#endif + + Candidates.push_back(std::move(WithPrefix)); + } + + LLVM_DEBUG({ + dbgs() << " Candidates to try:\n"; + for (const auto &C : Candidates) + dbgs() << " " << C << "\n"; + }); + + // Try all variants using tryAllPaths + for (const auto &Name : Candidates) { + + LLVM_DEBUG(dbgs() << " Trying candidate: " << Name << "\n";); + + for (const auto &R : Resolvers) { + if (auto Res = R.resolve(Name, Substitutor, Validator)) + return Res; + } + } + + LLVM_DEBUG(dbgs() << " -> No candidate Resolved.\n";); + + return std::nullopt; +} + +std::optional<std::string> +DylibResolverImpl::resolve(StringRef LibStem, bool VariateLibStem) const { + LLVM_DEBUG(dbgs() << "Resolving library stem: " << LibStem << "\n";); + + // If it is an absolute path, don't try iterate over the paths. + if (sys::path::is_absolute(LibStem)) { + LLVM_DEBUG(dbgs() << " -> Absolute path detected.\n";); + return Validator.validate(LibStem); + } + + if (!LibStem.starts_with_insensitive("@rpath")) { + if (auto norm = Validator.validate(Substitutor.substitute(LibStem))) { + LLVM_DEBUG(dbgs() << " -> Resolved after substitution: " << *norm + << "\n";); + + return norm; + } + } + + for (const auto &R : Resolvers) { + LLVM_DEBUG(dbgs() << " -> Resolving via search path ... \n";); + if (auto Result = R.resolve(LibStem, Substitutor, Validator)) { + LLVM_DEBUG(dbgs() << " -> Resolved via search path: " << *Result + << "\n";); + + return Result; + } + } + + // Expand libStem with paths, extensions, etc. + // std::string foundName; + if (VariateLibStem) { + LLVM_DEBUG(dbgs() << " -> Trying with extensions...\n";); + + if (auto Norm = tryWithExtensions(LibStem)) { + LLVM_DEBUG(dbgs() << " -> Resolved via tryWithExtensions: " << *Norm + << "\n";); + + return Norm; + } + } + + LLVM_DEBUG(dbgs() << " -> Could not resolve: " << LibStem << "\n";); + + return std::nullopt; +} + +#ifndef _WIN32 +mode_t PathResolver::lstatCached(StringRef Path) { + // If already cached - retun cached result + if (auto Cache = LibPathCache->read_lstat(Path)) + return *Cache; + + // Not cached: perform lstat and store + struct stat buf{}; + mode_t st_mode = (lstat(Path.str().c_str(), &buf) == -1) ? 0 : buf.st_mode; + + LibPathCache->insert_lstat(Path, st_mode); + + return st_mode; +} + +std::optional<std::string> PathResolver::readlinkCached(StringRef Path) { + // If already cached - retun cached result + if (auto Cache = LibPathCache->read_link(Path)) + return Cache; + + // If result not in cache - call system function and cache result + char buf[PATH_MAX]; + ssize_t len; + if ((len = readlink(Path.str().c_str(), buf, sizeof(buf))) != -1) { + buf[len] = '\0'; + std::string s(buf); + LibPathCache->insert_link(Path, s); + return s; + } + return std::nullopt; +} + +void createComponent(StringRef Path, StringRef BasePath, bool BaseIsResolved, + SmallVector<StringRef, 16> &Component) { + StringRef Separator = sys::path::get_separator(); + if (!BaseIsResolved) { + if (Path[0] == '~' && + (Path.size() == 1 || sys::path::is_separator(Path[1]))) { + static SmallString<128> HomeP; + if (HomeP.str().empty()) + sys::path::home_directory(HomeP); + StringRef(HomeP).split(Component, Separator, /*MaxSplit*/ -1, + /*KeepEmpty*/ false); + } else if (BasePath.empty()) { + static SmallString<256> CurrentPath; + if (CurrentPath.str().empty()) + sys::fs::current_path(CurrentPath); + StringRef(CurrentPath) + .split(Component, Separator, /*MaxSplit*/ -1, /*KeepEmpty*/ false); + } else { + BasePath.split(Component, Separator, /*MaxSplit*/ -1, + /*KeepEmpty*/ false); + } + } + + Path.split(Component, Separator, /*MaxSplit*/ -1, /*KeepEmpty*/ false); +} + +void normalizePathSegments(SmallVector<StringRef, 16> &PathParts) { + SmallVector<StringRef, 16> NormalizedPath; + for (auto &Part : PathParts) { + if (Part == ".") { + continue; + } else if (Part == "..") { + if (!NormalizedPath.empty() && NormalizedPath.back() != "..") { + NormalizedPath.pop_back(); + } else { + NormalizedPath.push_back(".."); + } + } else { + NormalizedPath.push_back(Part); + } + } + PathParts.swap(NormalizedPath); +} +#endif + +std::optional<std::string> PathResolver::realpathCached(StringRef Path, + std::error_code &EC, + StringRef Base, + bool BaseIsResolved, + long SymLoopLevel) { + EC.clear(); + + if (Path.empty()) { + EC = std::make_error_code(std::errc::no_such_file_or_directory); + LLVM_DEBUG(dbgs() << "PathResolver::realpathCached: Empty path\n";); + + return std::nullopt; + } + + if (SymLoopLevel <= 0) { + EC = std::make_error_code(std::errc::too_many_symbolic_link_levels); + LLVM_DEBUG( + dbgs() << "PathResolver::realpathCached: Too many Symlink levels: " + << Path << "\n";); + + return std::nullopt; + } + + // If already cached - retun cached result + bool isRelative = sys::path::is_relative(Path); + if (!isRelative) { + if (auto Cached = LibPathCache->read_realpath(Path)) { + EC = Cached->ErrnoCode; + if (EC) { + LLVM_DEBUG(dbgs() << "PathResolver::realpathCached: Cached (error) for " + << Path << "\n";); + } else { + LLVM_DEBUG( + dbgs() << "PathResolver::realpathCached: Cached (success) for " + << Path << " => " << Cached->canonicalPath << "\n";); + } + return Cached->canonicalPath.empty() + ? std::nullopt + : std::make_optional(Cached->canonicalPath); + } + } + + LLVM_DEBUG(dbgs() << "PathResolver::realpathCached: Resolving path: " << Path + << "\n";); + + // If result not in cache - call system function and cache result + + StringRef Separator(sys::path::get_separator()); + SmallString<256> Resolved(Separator); +#ifndef _WIN32 + SmallVector<StringRef, 16> Components; + + if (isRelative) { + if (BaseIsResolved) { + Resolved.assign(Base); + LLVM_DEBUG(dbgs() << " Using Resolved base: " << Base << "\n";); + } + createComponent(Path, Base, BaseIsResolved, Components); + } else { + Path.split(Components, Separator, /*MaxSplit*/ -1, /*KeepEmpty*/ false); + } + + normalizePathSegments(Components); + LLVM_DEBUG({ + for (auto &C : Components) + dbgs() << " " << C << " "; + + dbgs() << "\n"; + }); + + // Handle path list items + for (const auto &Component : Components) { + if (Component == ".") + continue; + if (Component == "..") { + // collapse "a/b/../c" to "a/c" + size_t S = Resolved.rfind(Separator); + if (S != llvm::StringRef::npos) + Resolved.resize(S); + if (Resolved.empty()) + Resolved = Separator; + continue; + } + + size_t oldSize = Resolved.size(); + sys::path::append(Resolved, Component); + const char *ResolvedPath = Resolved.c_str(); + LLVM_DEBUG(dbgs() << " Processing Component: " << Component << " => " + << ResolvedPath << "\n";); + mode_t st_mode = lstatCached(ResolvedPath); + + if (S_ISLNK(st_mode)) { + LLVM_DEBUG(dbgs() << " Found symlink: " << ResolvedPath << "\n";); + + auto SymlinkOpt = readlinkCached(ResolvedPath); + if (!SymlinkOpt) { + EC = std::make_error_code(std::errc::no_such_file_or_directory); + LibPathCache->insert_realpath(Path, LibraryPathCache::PathInfo{"", EC}); + LLVM_DEBUG(dbgs() << " Failed to read symlink: " << ResolvedPath + << "\n";); + + return std::nullopt; + } + + StringRef Symlink = *SymlinkOpt; + LLVM_DEBUG(dbgs() << " Symlink points to: " << Symlink << "\n";); + + std::string resolvedBase = ""; + if (sys::path::is_relative(Symlink)) { + Resolved.resize(oldSize); + resolvedBase = Resolved.str().str(); + } + + auto RealSymlink = + realpathCached(Symlink, EC, resolvedBase, + /*BaseIsResolved=*/true, SymLoopLevel - 1); + if (!RealSymlink) { + LibPathCache->insert_realpath(Path, LibraryPathCache::PathInfo{"", EC}); + LLVM_DEBUG(dbgs() << " Failed to resolve symlink target: " << Symlink + << "\n";); + + return std::nullopt; + } + + Resolved.assign(*RealSymlink); + LLVM_DEBUG(dbgs() << " Symlink Resolved to: " << Resolved << "\n";); + + } else if (st_mode == 0) { + EC = std::make_error_code(std::errc::no_such_file_or_directory); + LibPathCache->insert_realpath(Path, LibraryPathCache::PathInfo{"", EC}); + LLVM_DEBUG(dbgs() << " Component does not exist: " << ResolvedPath + << "\n";); + + return std::nullopt; + } + } +#else + EC = sys::fs::real_path(Path, Resolved); // Windows fallback +#endif + + std::string Canonical = Resolved.str().str(); + { + LibPathCache->insert_realpath(Path, LibraryPathCache::PathInfo{ + Canonical, + std::error_code() // success + }); + } + LLVM_DEBUG(dbgs() << "PathResolver::realpathCached: Final Resolved: " << Path + << " => " << Canonical << "\n";); + return Canonical; +} + +void LibraryScanHelper::addBasePath(const std::string &Path, PathType K) { + std::error_code EC; + std::string Canon = resolveCanonical(Path, EC); + if (EC) { + LLVM_DEBUG( + dbgs() + << "LibraryScanHelper::addBasePath: Failed to canonicalize path: " + << Path << "\n";); + return; + } + std::unique_lock<std::shared_mutex> Lock(Mtx); + if (LibSearchPaths.count(Canon)) { + LLVM_DEBUG(dbgs() << "LibraryScanHelper::addBasePath: Already added: " + << Canon << "\n";); + return; + } + K = K == PathType::Unknown ? classifyKind(Canon) : K; + auto SP = std::make_shared<LibrarySearchPath>(Canon, K); + LibSearchPaths[Canon] = SP; + + if (K == PathType::User) { + LLVM_DEBUG(dbgs() << "LibraryScanHelper::addBasePath: Added User path: " + << Canon << "\n";); + UnscannedUsr.push_back(StringRef(SP->BasePath)); + } else { + LLVM_DEBUG(dbgs() << "LibraryScanHelper::addBasePath: Added System path: " + << Canon << "\n";); + UnscannedSys.push_back(StringRef(SP->BasePath)); + } +} + +std::vector<std::shared_ptr<LibrarySearchPath>> +LibraryScanHelper::getNextBatch(PathType K, size_t BatchSize) { + std::vector<std::shared_ptr<LibrarySearchPath>> Result; + auto &Queue = (K == PathType::User) ? UnscannedUsr : UnscannedSys; + + std::unique_lock<std::shared_mutex> Lock(Mtx); + + while (!Queue.empty() && (BatchSize == 0 || Result.size() < BatchSize)) { + StringRef Base = Queue.front(); + auto It = LibSearchPaths.find(Base); + if (It != LibSearchPaths.end()) { + auto &SP = It->second; + ScanState Expected = ScanState::NotScanned; + if (SP->State.compare_exchange_strong(Expected, ScanState::Scanning)) { + Result.push_back(SP); + } + } + Queue.pop_front(); + } + + return Result; +} + +bool LibraryScanHelper::isTrackedBasePath(StringRef Path) const { + std::error_code EC; + std::string Canon = resolveCanonical(Path, EC); + if (EC) + return false; + + std::shared_lock<std::shared_mutex> Lock(Mtx); + return LibSearchPaths.count(Canon) > 0; +} + +bool LibraryScanHelper::leftToScan(PathType K) const { + std::shared_lock<std::shared_mutex> Lock(Mtx); + for (const auto &KV : LibSearchPaths) { + const auto &SP = KV.second; + if (SP->Kind == K && SP->State == ScanState::NotScanned) + return true; + } + return false; +} + +void LibraryScanHelper::resetToScan() { + std::shared_lock<std::shared_mutex> Lock(Mtx); + + for (auto &[_, SP] : LibSearchPaths) { + ScanState Expected = ScanState::Scanned; + + if (!SP->State.compare_exchange_strong(Expected, ScanState::NotScanned)) + continue; + + auto &TargetList = + (SP->Kind == PathType::User) ? UnscannedUsr : UnscannedSys; + TargetList.emplace_back(SP->BasePath); + } +} + +std::vector<std::shared_ptr<LibrarySearchPath>> +LibraryScanHelper::getAllUnits() const { + std::shared_lock<std::shared_mutex> Lock(Mtx); + std::vector<std::shared_ptr<LibrarySearchPath>> Result; + Result.reserve(LibSearchPaths.size()); + for (const auto &[_, SP] : LibSearchPaths) { + Result.push_back(SP); + } + return Result; +} + +std::string LibraryScanHelper::resolveCanonical(StringRef Path, + std::error_code &EC) const { + auto Canon = LibPathResolver->resolve(Path, EC); + return EC ? Path.str() : *Canon; +} + +PathType LibraryScanHelper::classifyKind(StringRef Path) const { + // Detect home directory + const char *Home = getenv("HOME"); + if (Home && Path.find(Home) == 0) + return PathType::User; + + static const std::array<std::string, 5> UserPrefixes = { + "/usr/local", // often used by users for manual installs + "/opt/homebrew", // common on macOS + "/opt/local", // MacPorts + "/home", // Linux home dirs + "/Users", // macOS user dirs + }; + + for (const auto &Prefix : UserPrefixes) { + if (Path.find(Prefix) == 0) + return PathType::User; + } + + return PathType::System; +} + +Expected<LibraryDepsInfo> parseMachODeps(const object::MachOObjectFile &Obj) { + LibraryDepsInfo Libdeps; + LLVM_DEBUG(dbgs() << "Parsing Mach-O dependencies...\n";); + for (const auto &Command : Obj.load_commands()) { + switch (Command.C.cmd) { + case MachO::LC_LOAD_DYLIB: { + MachO::dylib_command dylibCmd = Obj.getDylibIDLoadCommand(Command); + const char *name = Command.Ptr + dylibCmd.dylib.name; + Libdeps.addDep(name); + LLVM_DEBUG(dbgs() << " Found LC_LOAD_DYLIB: " << name << "\n";); + } break; + case MachO::LC_LOAD_WEAK_DYLIB: + case MachO::LC_REEXPORT_DYLIB: + case MachO::LC_LOAD_UPWARD_DYLIB: + case MachO::LC_LAZY_LOAD_DYLIB: + break; + case MachO::LC_RPATH: { + // Extract RPATH + MachO::rpath_command rpathCmd = Obj.getRpathCommand(Command); + const char *rpath = Command.Ptr + rpathCmd.path; + LLVM_DEBUG(dbgs() << " Found LC_RPATH: " << rpath << "\n";); + + SmallVector<StringRef, 4> RawPaths; + SplitString(StringRef(rpath), RawPaths, + sys::EnvPathSeparator == ':' ? ":" : ";"); + + for (const auto &raw : RawPaths) { + Libdeps.addRPath(raw.str()); // Convert to std::string + LLVM_DEBUG(dbgs() << " Parsed RPATH entry: " << raw << "\n";); + } + break; + } + } + } + + return Expected<LibraryDepsInfo>(std::move(Libdeps)); +} + +template <class ELFT> +static Expected<StringRef> getDynamicStrTab(const object::ELFFile<ELFT> &Elf) { + auto DynamicEntriesOrError = Elf.dynamicEntries(); + if (!DynamicEntriesOrError) + return DynamicEntriesOrError.takeError(); + + for (const typename ELFT::Dyn &Dyn : *DynamicEntriesOrError) { + if (Dyn.d_tag == ELF::DT_STRTAB) { + auto MappedAddrOrError = Elf.toMappedAddr(Dyn.getPtr()); + if (!MappedAddrOrError) + return MappedAddrOrError.takeError(); + return StringRef(reinterpret_cast<const char *>(*MappedAddrOrError)); + } + } + + // If the dynamic segment is not present, we fall back on the sections. + auto SectionsOrError = Elf.sections(); + if (!SectionsOrError) + return SectionsOrError.takeError(); + + for (const typename ELFT::Shdr &Sec : *SectionsOrError) { + if (Sec.sh_type == ELF::SHT_DYNSYM) + return Elf.getStringTableForSymtab(Sec); + } + + return make_error<StringError>("dynamic string table not found", + inconvertibleErrorCode()); +} + +template <typename ELFT> +Expected<LibraryDepsInfo> parseELF(const object::ELFFile<ELFT> &Elf) { + LibraryDepsInfo Deps; + Expected<StringRef> StrTabOrErr = getDynamicStrTab(Elf); + if (!StrTabOrErr) + return StrTabOrErr.takeError(); + + const char *Data = StrTabOrErr->data(); + + auto DynamicEntriesOrError = Elf.dynamicEntries(); + if (!DynamicEntriesOrError) { + return DynamicEntriesOrError.takeError(); + } + + for (const typename ELFT::Dyn &Dyn : *DynamicEntriesOrError) { + switch (Dyn.d_tag) { + case ELF::DT_NEEDED: + Deps.addDep(Data + Dyn.d_un.d_val); + break; + case ELF::DT_RPATH: { + SmallVector<StringRef, 4> RawPaths; + SplitString(Data + Dyn.d_un.d_val, RawPaths, + sys::EnvPathSeparator == ':' ? ":" : ";"); + for (const auto &raw : RawPaths) + Deps.addRPath(raw.str()); + break; + } + case ELF::DT_RUNPATH: { + SmallVector<StringRef, 4> RawPaths; + SplitString(Data + Dyn.d_un.d_val, RawPaths, + sys::EnvPathSeparator == ':' ? ":" : ";"); + for (const auto &raw : RawPaths) + Deps.addRunPath(raw.str()); + break; + } + case ELF::DT_FLAGS_1: + // Check if this is not a pie executable. + if (Dyn.d_un.d_val & ELF::DF_1_PIE) + Deps.isPIE = true; + break; + // (Dyn.d_tag == ELF::DT_NULL) continue; + // (Dyn.d_tag == ELF::DT_AUXILIARY || Dyn.d_tag == ELF::DT_FILTER) + default: + break; + } + } + + return Expected<LibraryDepsInfo>(std::move(Deps)); +} + +Expected<LibraryDepsInfo> parseELFDeps(const object::ELFObjectFileBase &Obj) { + using namespace object; + LLVM_DEBUG(dbgs() << "parseELFDeps: Detected ELF object\n";); + if (const auto *ELF = dyn_cast<ELF32LEObjectFile>(&Obj)) + return parseELF(ELF->getELFFile()); + else if (const auto *ELF = dyn_cast<ELF32BEObjectFile>(&Obj)) + return parseELF(ELF->getELFFile()); + else if (const auto *ELF = dyn_cast<ELF64LEObjectFile>(&Obj)) + return parseELF(ELF->getELFFile()); + else if (const auto *ELF = dyn_cast<ELF64BEObjectFile>(&Obj)) + return parseELF(ELF->getELFFile()); + + LLVM_DEBUG(dbgs() << "parseELFDeps: Unknown ELF format\n";); + return createStringError(std::errc::not_supported, "Unknown ELF format"); +} + +Expected<LibraryDepsInfo> LibraryScanner::extractDeps(StringRef FilePath) { + LLVM_DEBUG(dbgs() << "extractDeps: Attempting to open file " << FilePath + << "\n";); + + ObjectFileLoader ObjLoader(FilePath); + auto ObjOrErr = ObjLoader.getObjectFile(); + if (!ObjOrErr) { + LLVM_DEBUG(dbgs() << "extractDeps: Failed to open " << FilePath << "\n";); + return ObjOrErr.takeError(); + } + + object::ObjectFile *Obj = &ObjOrErr.get(); + + if (auto *elfObj = dyn_cast<object::ELFObjectFileBase>(Obj)) { + LLVM_DEBUG(dbgs() << "extractDeps: File " << FilePath + << " is an ELF object\n";); + + return parseELFDeps(*elfObj); + } + + if (auto *macho = dyn_cast<object::MachOObjectFile>(Obj)) { + LLVM_DEBUG(dbgs() << "extractDeps: File " << FilePath + << " is a Mach-O object\n";); + return parseMachODeps(*macho); + } + + if (Obj->isCOFF()) { + // TODO: COFF support + return LibraryDepsInfo(); + } + + LLVM_DEBUG(dbgs() << "extractDeps: Unsupported binary format for file " + << FilePath << "\n";); + return createStringError(inconvertibleErrorCode(), + "Unsupported binary format: %s", + FilePath.str().c_str()); +} + +std::optional<std::string> LibraryScanner::shouldScan(StringRef FilePath) { + std::error_code EC; + + LLVM_DEBUG(dbgs() << "[shouldScan] Checking: " << FilePath << "\n";); + + // [1] Check file existence early + if (!sys::fs::exists(FilePath)) { + LLVM_DEBUG(dbgs() << " -> Skipped: file does not exist.\n";); + + return std::nullopt; + } + + // [2] Resolve to canonical path + auto CanonicalPathOpt = ScanHelper.resolve(FilePath, EC); + if (EC || !CanonicalPathOpt) { + LLVM_DEBUG(dbgs() << " -> Skipped: failed to resolve path (EC=" + << EC.message() << ").\n";); + + return std::nullopt; + } + + const std::string &CanonicalPath = *CanonicalPathOpt; + LLVM_DEBUG(dbgs() << " -> Canonical path: " << CanonicalPath << "\n"); + + // [3] Check if it's a directory — skip directories + if (sys::fs::is_directory(CanonicalPath)) { + LLVM_DEBUG(dbgs() << " -> Skipped: path is a directory.\n";); + + return std::nullopt; + } + + // [4] Skip if it's not a shared library. + if (!DylibPathValidator::isSharedLibrary(CanonicalPath)) { + LLVM_DEBUG(dbgs() << " -> Skipped: not a shared library.\n";); + return std::nullopt; + } + + // [5] Skip if we've already seen this path (via cache) + if (ScanHelper.hasSeenOrMark(CanonicalPath)) { + LLVM_DEBUG(dbgs() << " -> Skipped: already seen.\n";); + + return std::nullopt; + } + + // [6] Already tracked in LibraryManager? + if (LibMgr.hasLibrary(CanonicalPath)) { + LLVM_DEBUG(dbgs() << " -> Skipped: already tracked by LibraryManager.\n";); + + return std::nullopt; + } + + // [7] Run user-defined hook (default: always true) + if (!ShouldScanCall(CanonicalPath)) { + LLVM_DEBUG(dbgs() << " -> Skipped: user-defined hook rejected.\n";); + + return std::nullopt; + } + + LLVM_DEBUG(dbgs() << " -> Accepted: ready to scan " << CanonicalPath + << "\n";); + return CanonicalPath; +} + +void LibraryScanner::handleLibrary(StringRef FilePath, PathType K, int level) { + LLVM_DEBUG(dbgs() << "LibraryScanner::handleLibrary: Scanning: " << FilePath + << ", level=" << level << "\n";); + auto CanonPathOpt = shouldScan(FilePath); + if (!CanonPathOpt) { + LLVM_DEBUG(dbgs() << " Skipped (shouldScan returned false): " << FilePath + << "\n";); + + return; + } + const std::string CanonicalPath = *CanonPathOpt; + + auto DepsOrErr = extractDeps(CanonicalPath); + if (!DepsOrErr) { + LLVM_DEBUG(dbgs() << " Failed to extract deps for: " << CanonicalPath + << "\n";); + handleError(DepsOrErr.takeError()); + return; + } + + LibraryDepsInfo &Deps = *DepsOrErr; + + LLVM_DEBUG({ + dbgs() << " Found deps : \n"; + for (const auto &dep : Deps.deps) + dbgs() << " : " << dep << "\n"; + dbgs() << " Found @rpath : " << Deps.rpath.size() << "\n"; + for (const auto &r : Deps.rpath) + dbgs() << " : " << r << "\n"; + dbgs() << " Found @runpath : \n"; + for (const auto &r : Deps.runPath) + dbgs() << " : " << r << "\n"; + }); + + if (Deps.isPIE && level == 0) { + LLVM_DEBUG(dbgs() << " Skipped PIE executable at top level: " + << CanonicalPath << "\n";); + + return; + } + + bool Added = LibMgr.addLibrary(CanonicalPath, K); + if (!Added) { + LLVM_DEBUG(dbgs() << " Already added: " << CanonicalPath << "\n";); + return; + } + + // Heuristic 1: No RPATH/RUNPATH, skip deps + if (Deps.rpath.empty() && Deps.runPath.empty()) { + LLVM_DEBUG( + dbgs() << "LibraryScanner::handleLibrary: Skipping deps (Heuristic1): " + << CanonicalPath << "\n";); + return; + } + + // Heuristic 2: All RPATH and RUNPATH already tracked + auto allTracked = [&](const auto &Paths) { + LLVM_DEBUG(dbgs() << " Checking : " << Paths.size() << "\n";); + return std::all_of(Paths.begin(), Paths.end(), [&](StringRef P) { + LLVM_DEBUG(dbgs() << " Checking isTrackedBasePath : " << P << "\n";); + return ScanHelper.isTrackedBasePath( + DylibResolver::resolvelinkerFlag(P, CanonicalPath)); + }); + }; + + if (allTracked(Deps.rpath) && allTracked(Deps.runPath)) { + LLVM_DEBUG( + dbgs() << "LibraryScanner::handleLibrary: Skipping deps (Heuristic2): " + << CanonicalPath << "\n";); + return; + } + + DylibPathValidator Validator(ScanHelper.getPathResolver()); + DylibResolver Resolver(Validator); + Resolver.configure(CanonicalPath, + {{Deps.rpath, SearchPathType::RPath}, + {ScanHelper.getSearchPaths(), SearchPathType::UsrOrSys}, + {Deps.runPath, SearchPathType::RunPath}}); + for (StringRef Dep : Deps.deps) { + LLVM_DEBUG(dbgs() << " Resolving dep: " << Dep << "\n";); + auto DepFullOpt = Resolver.resolve(Dep); + if (!DepFullOpt) { + LLVM_DEBUG(dbgs() << " Failed to resolve dep: " << Dep << "\n";); + + continue; + } + LLVM_DEBUG(dbgs() << " Resolved dep to: " << *DepFullOpt << "\n";); + + handleLibrary(*DepFullOpt, K, level + 1); + } +} + +void LibraryScanner::scanBaseDir(std::shared_ptr<LibrarySearchPath> SP) { + if (!sys::fs::is_directory(SP->BasePath) || SP->BasePath.empty()) { + LLVM_DEBUG( + dbgs() << "LibraryScanner::scanBaseDir: Invalid or empty basePath: " + << SP->BasePath << "\n";); + return; + } + + LLVM_DEBUG(dbgs() << "LibraryScanner::scanBaseDir: Scanning directory: " + << SP->BasePath << "\n";); + std::error_code EC; + + SP->State.store(ScanState::Scanning); + + for (sys::fs::directory_iterator It(SP->BasePath, EC), end; It != end && !EC; + It.increment(EC)) { + auto Entry = *It; + if (!Entry.status()) + continue; + + auto Status = *Entry.status(); + if (sys::fs::is_regular_file(Status) || sys::fs::is_symlink_file(Status)) { + LLVM_DEBUG(dbgs() << " Found file: " << Entry.path() << "\n";); + // async support ? + handleLibrary(Entry.path(), SP->Kind); + } + } + + SP->State.store(ScanState::Scanned); +} + +void LibraryScanner::scanNext(PathType K, size_t BatchSize) { + LLVM_DEBUG(dbgs() << "LibraryScanner::scanNext: Scanning next batch of size " + << BatchSize << " for kind " + << (K == PathType::User ? "User" : "System") << "\n";); + + auto SearchPaths = ScanHelper.getNextBatch(K, BatchSize); + for (auto &SP : SearchPaths) { + LLVM_DEBUG(dbgs() << " Scanning unit with basePath: " << SP->BasePath + << "\n";); + + scanBaseDir(SP); + } +} + +} // end namespace llvm::orc diff --git a/llvm/lib/Support/Windows/Signals.inc b/llvm/lib/Support/Windows/Signals.inc index 648d6a5..da68994 100644 --- a/llvm/lib/Support/Windows/Signals.inc +++ b/llvm/lib/Support/Windows/Signals.inc @@ -421,8 +421,13 @@ bool sys::RemoveFileOnSignal(StringRef Filename, std::string *ErrMsg) { return true; } - if (FilesToRemove == NULL) + if (FilesToRemove == NULL) { FilesToRemove = new std::vector<std::string>; + std::atexit([]() { + delete FilesToRemove; + FilesToRemove = NULL; + }); + } FilesToRemove->push_back(std::string(Filename)); diff --git a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp index 6dcbced..b7fa899 100644 --- a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp +++ b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp @@ -1288,18 +1288,38 @@ void WaitcntBrackets::applyWaitcnt(InstCounterType T, unsigned Count) { } void WaitcntBrackets::applyXcnt(const AMDGPU::Waitcnt &Wait) { + // On entry to a block with multiple predescessors, there may + // be pending SMEM and VMEM events active at the same time. + // In such cases, only clear one active event at a time. + auto applyPendingXcntGroup = [this](unsigned E) { + unsigned LowerBound = getScoreLB(X_CNT); + applyWaitcnt(X_CNT, 0); + PendingEvents |= (1 << E); + setScoreLB(X_CNT, LowerBound); + }; + // Wait on XCNT is redundant if we are already waiting for a load to complete. // SMEM can return out of order, so only omit XCNT wait if we are waiting till // zero. - if (Wait.KmCnt == 0 && hasPendingEvent(SMEM_GROUP)) - return applyWaitcnt(X_CNT, 0); + if (Wait.KmCnt == 0 && hasPendingEvent(SMEM_GROUP)) { + if (hasPendingEvent(VMEM_GROUP)) + applyPendingXcntGroup(VMEM_GROUP); + else + applyWaitcnt(X_CNT, 0); + return; + } // If we have pending store we cannot optimize XCnt because we do not wait for // stores. VMEM loads retun in order, so if we only have loads XCnt is // decremented to the same number as LOADCnt. if (Wait.LoadCnt != ~0u && hasPendingEvent(VMEM_GROUP) && - !hasPendingEvent(STORE_CNT)) - return applyWaitcnt(X_CNT, std::min(Wait.XCnt, Wait.LoadCnt)); + !hasPendingEvent(STORE_CNT)) { + if (hasPendingEvent(SMEM_GROUP)) + applyPendingXcntGroup(SMEM_GROUP); + else + applyWaitcnt(X_CNT, std::min(Wait.XCnt, Wait.LoadCnt)); + return; + } applyWaitcnt(X_CNT, Wait.XCnt); } diff --git a/llvm/lib/Target/CSKY/CSKYISelLowering.cpp b/llvm/lib/Target/CSKY/CSKYISelLowering.cpp index e5b4f6e..ab4ee55 100644 --- a/llvm/lib/Target/CSKY/CSKYISelLowering.cpp +++ b/llvm/lib/Target/CSKY/CSKYISelLowering.cpp @@ -889,7 +889,7 @@ CSKYTargetLowering::getRegForInlineAsmConstraint(const TargetRegisterInfo *TRI, .Cases("{t9}", "{bsp}", CSKY::R25) .Case("{r26}", CSKY::R26) .Case("{r27}", CSKY::R27) - .Cases("{gb}", "{rgb}", "{rdb}", CSKY::R28) + .Cases({"{gb}", "{rgb}", "{rdb}"}, CSKY::R28) .Cases("{tb}", "{rtb}", CSKY::R29) .Case("{svbr}", CSKY::R30) .Case("{tls}", CSKY::R31) diff --git a/llvm/lib/Target/M68k/MCTargetDesc/M68kAsmBackend.cpp b/llvm/lib/Target/M68k/MCTargetDesc/M68kAsmBackend.cpp index fe83dc6..51bafe4 100644 --- a/llvm/lib/Target/M68k/MCTargetDesc/M68kAsmBackend.cpp +++ b/llvm/lib/Target/M68k/MCTargetDesc/M68kAsmBackend.cpp @@ -49,7 +49,7 @@ public: M68kAsmBackend(const Target &T, const MCSubtargetInfo &STI) : MCAsmBackend(llvm::endianness::big), Allows32BitBranch(llvm::StringSwitch<bool>(STI.getCPU()) - .CasesLower("m68020", "m68030", "m68040", true) + .CasesLower({"m68020", "m68030", "m68040"}, true) .Default(false)) {} void applyFixup(const MCFragment &, const MCFixup &, const MCValue &, diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp index 007074c..133406b 100644 --- a/llvm/lib/Target/X86/X86ISelLowering.cpp +++ b/llvm/lib/Target/X86/X86ISelLowering.cpp @@ -22861,6 +22861,13 @@ static SDValue combineVectorSizedSetCCEquality(EVT VT, SDValue X, SDValue Y, if (!OpVT.isScalarInteger() || OpSize < 128) return SDValue(); + // Don't do this if we're not supposed to use the FPU. + bool NoImplicitFloatOps = + DAG.getMachineFunction().getFunction().hasFnAttribute( + Attribute::NoImplicitFloat); + if (Subtarget.useSoftFloat() || NoImplicitFloatOps) + return SDValue(); + // Ignore a comparison with zero because that gets special treatment in // EmitTest(). But make an exception for the special case of a pair of // logically-combined vector-sized operands compared to zero. This pattern may @@ -22883,13 +22890,9 @@ static SDValue combineVectorSizedSetCCEquality(EVT VT, SDValue X, SDValue Y, // Use XOR (plus OR) and PTEST after SSE4.1 for 128/256-bit operands. // Use PCMPNEQ (plus OR) and KORTEST for 512-bit operands. // Otherwise use PCMPEQ (plus AND) and mask testing. - bool NoImplicitFloatOps = - DAG.getMachineFunction().getFunction().hasFnAttribute( - Attribute::NoImplicitFloat); - if (!Subtarget.useSoftFloat() && !NoImplicitFloatOps && - ((OpSize == 128 && Subtarget.hasSSE2()) || - (OpSize == 256 && Subtarget.hasAVX()) || - (OpSize == 512 && Subtarget.useAVX512Regs()))) { + if ((OpSize == 128 && Subtarget.hasSSE2()) || + (OpSize == 256 && Subtarget.hasAVX()) || + (OpSize == 512 && Subtarget.useAVX512Regs())) { bool HasPT = Subtarget.hasSSE41(); // PTEST and MOVMSK are slow on Knights Landing and Knights Mill and widened @@ -53344,105 +53347,6 @@ static SDValue combineMaskedStore(SDNode *N, SelectionDAG &DAG, return SDValue(); } -// Look for a RMW operation that only touches one bit of a larger than legal -// type and fold it to a BTC/BTR/BTS or bit insertion pattern acting on a single -// i32 sub value. -static SDValue narrowBitOpRMW(StoreSDNode *St, const SDLoc &DL, - SelectionDAG &DAG, - const X86Subtarget &Subtarget) { - using namespace SDPatternMatch; - - // Only handle normal stores and its chain was a matching normal load. - auto *Ld = dyn_cast<LoadSDNode>(St->getChain()); - if (!ISD::isNormalStore(St) || !St->isSimple() || !Ld || - !ISD::isNormalLoad(Ld) || !Ld->isSimple() || - Ld->getBasePtr() != St->getBasePtr() || - Ld->getOffset() != St->getOffset()) - return SDValue(); - - SDValue LoadVal(Ld, 0); - SDValue StoredVal = St->getValue(); - EVT VT = StoredVal.getValueType(); - - // Only narrow larger than legal scalar integers. - if (!VT.isScalarInteger() || - VT.getSizeInBits() <= (Subtarget.is64Bit() ? 64 : 32)) - return SDValue(); - - // BTR: X & ~(1 << ShAmt) - // BTS: X | (1 << ShAmt) - // BTC: X ^ (1 << ShAmt) - // - // BitInsert: (X & ~(1 << ShAmt)) | (InsertBit << ShAmt) - SDValue InsertBit, ShAmt; - if (!StoredVal.hasOneUse() || - !(sd_match(StoredVal, m_And(m_Specific(LoadVal), - m_Not(m_Shl(m_One(), m_Value(ShAmt))))) || - sd_match(StoredVal, - m_Or(m_Specific(LoadVal), m_Shl(m_One(), m_Value(ShAmt)))) || - sd_match(StoredVal, - m_Xor(m_Specific(LoadVal), m_Shl(m_One(), m_Value(ShAmt)))) || - sd_match(StoredVal, - m_Or(m_And(m_Specific(LoadVal), - m_Not(m_Shl(m_One(), m_Value(ShAmt)))), - m_Shl(m_Value(InsertBit), m_Deferred(ShAmt)))))) - return SDValue(); - - // Ensure the shift amount is in bounds. - KnownBits KnownAmt = DAG.computeKnownBits(ShAmt); - if (KnownAmt.getMaxValue().uge(VT.getSizeInBits())) - return SDValue(); - - // If we're inserting a bit then it must be the LSB. - if (InsertBit) { - KnownBits KnownInsert = DAG.computeKnownBits(InsertBit); - if (KnownInsert.countMinLeadingZeros() < (VT.getSizeInBits() - 1)) - return SDValue(); - } - - // Split the shift into an alignment shift that moves the active i32 block to - // the bottom bits for truncation and a modulo shift that can act on the i32. - EVT AmtVT = ShAmt.getValueType(); - SDValue AlignAmt = DAG.getNode(ISD::AND, DL, AmtVT, ShAmt, - DAG.getSignedConstant(-32LL, DL, AmtVT)); - SDValue ModuloAmt = - DAG.getNode(ISD::AND, DL, AmtVT, ShAmt, DAG.getConstant(31, DL, AmtVT)); - ModuloAmt = DAG.getZExtOrTrunc(ModuloAmt, DL, MVT::i8); - - // Compute the byte offset for the i32 block that is changed by the RMW. - // combineTruncate will adjust the load for us in a similar way. - EVT PtrVT = St->getBasePtr().getValueType(); - SDValue PtrBitOfs = DAG.getZExtOrTrunc(AlignAmt, DL, PtrVT); - SDValue PtrByteOfs = DAG.getNode(ISD::SRL, DL, PtrVT, PtrBitOfs, - DAG.getShiftAmountConstant(3, PtrVT, DL)); - SDValue NewPtr = DAG.getMemBasePlusOffset(St->getBasePtr(), PtrByteOfs, DL, - SDNodeFlags::NoUnsignedWrap); - - // Reconstruct the BTC/BTR/BTS pattern for the i32 block and store. - SDValue X = DAG.getNode(ISD::SRL, DL, VT, LoadVal, AlignAmt); - X = DAG.getNode(ISD::TRUNCATE, DL, MVT::i32, X); - - SDValue Mask = DAG.getNode(ISD::SHL, DL, MVT::i32, - DAG.getConstant(1, DL, MVT::i32), ModuloAmt); - - SDValue Res; - if (InsertBit) { - SDValue BitMask = - DAG.getNode(ISD::SHL, DL, MVT::i32, - DAG.getZExtOrTrunc(InsertBit, DL, MVT::i32), ModuloAmt); - Res = - DAG.getNode(ISD::AND, DL, MVT::i32, X, DAG.getNOT(DL, Mask, MVT::i32)); - Res = DAG.getNode(ISD::OR, DL, MVT::i32, Res, BitMask); - } else { - if (StoredVal.getOpcode() == ISD::AND) - Mask = DAG.getNOT(DL, Mask, MVT::i32); - Res = DAG.getNode(StoredVal.getOpcode(), DL, MVT::i32, X, Mask); - } - - return DAG.getStore(St->getChain(), DL, Res, NewPtr, St->getPointerInfo(), - Align(), St->getMemOperand()->getFlags()); -} - static SDValue combineStore(SDNode *N, SelectionDAG &DAG, TargetLowering::DAGCombinerInfo &DCI, const X86Subtarget &Subtarget) { @@ -53669,9 +53573,6 @@ static SDValue combineStore(SDNode *N, SelectionDAG &DAG, } } - if (SDValue R = narrowBitOpRMW(St, dl, DAG, Subtarget)) - return R; - // Convert store(cmov(load(p), x, CC), p) to cstore(x, p, CC) // store(cmov(x, load(p), CC), p) to cstore(x, p, InvertCC) if ((VT == MVT::i16 || VT == MVT::i32 || VT == MVT::i64) && @@ -54604,9 +54505,8 @@ static SDValue combineTruncate(SDNode *N, SelectionDAG &DAG, // truncation, see if we can convert the shift into a pointer offset instead. // Limit this to normal (non-ext) scalar integer loads. if (SrcVT.isScalarInteger() && Src.getOpcode() == ISD::SRL && - Src.hasOneUse() && ISD::isNormalLoad(Src.getOperand(0).getNode()) && - (Src.getOperand(0).hasOneUse() || - !DAG.getTargetLoweringInfo().isOperationLegal(ISD::LOAD, SrcVT))) { + Src.hasOneUse() && Src.getOperand(0).hasOneUse() && + ISD::isNormalLoad(Src.getOperand(0).getNode())) { auto *Ld = cast<LoadSDNode>(Src.getOperand(0)); if (Ld->isSimple() && VT.isByteSized() && isPowerOf2_64(VT.getSizeInBits())) { @@ -56406,7 +56306,6 @@ static SDValue combineAVX512SetCCToKMOV(EVT VT, SDValue Op0, ISD::CondCode CC, static SDValue combineSetCC(SDNode *N, SelectionDAG &DAG, TargetLowering::DAGCombinerInfo &DCI, const X86Subtarget &Subtarget) { - using namespace SDPatternMatch; const ISD::CondCode CC = cast<CondCodeSDNode>(N->getOperand(2))->get(); const SDValue LHS = N->getOperand(0); const SDValue RHS = N->getOperand(1); @@ -56465,37 +56364,6 @@ static SDValue combineSetCC(SDNode *N, SelectionDAG &DAG, if (SDValue AndN = MatchAndCmpEq(RHS, LHS)) return DAG.getSetCC(DL, VT, AndN, DAG.getConstant(0, DL, OpVT), CC); - // If we're performing a bit test on a larger than legal type, attempt - // to (aligned) shift down the value to the bottom 32-bits and then - // perform the bittest on the i32 value. - // ICMP_ZERO(AND(X,SHL(1,IDX))) - // --> ICMP_ZERO(AND(TRUNC(SRL(X,AND(IDX,-32))),SHL(1,AND(IDX,31)))) - if (isNullConstant(RHS) && - OpVT.getScalarSizeInBits() > (Subtarget.is64Bit() ? 64 : 32)) { - SDValue X, ShAmt; - if (sd_match(LHS, m_OneUse(m_And(m_Value(X), - m_Shl(m_One(), m_Value(ShAmt)))))) { - // Only attempt this if the shift amount is known to be in bounds. - KnownBits KnownAmt = DAG.computeKnownBits(ShAmt); - if (KnownAmt.getMaxValue().ult(OpVT.getScalarSizeInBits())) { - EVT AmtVT = ShAmt.getValueType(); - SDValue AlignAmt = - DAG.getNode(ISD::AND, DL, AmtVT, ShAmt, - DAG.getSignedConstant(-32LL, DL, AmtVT)); - SDValue ModuloAmt = DAG.getNode(ISD::AND, DL, AmtVT, ShAmt, - DAG.getConstant(31, DL, AmtVT)); - SDValue Mask = DAG.getNode( - ISD::SHL, DL, MVT::i32, DAG.getConstant(1, DL, MVT::i32), - DAG.getZExtOrTrunc(ModuloAmt, DL, MVT::i8)); - X = DAG.getNode(ISD::SRL, DL, OpVT, X, AlignAmt); - X = DAG.getNode(ISD::TRUNCATE, DL, MVT::i32, X); - X = DAG.getNode(ISD::AND, DL, MVT::i32, X, Mask); - return DAG.getSetCC(DL, VT, X, DAG.getConstant(0, DL, MVT::i32), - CC); - } - } - } - // cmpeq(trunc(x),C) --> cmpeq(x,C) // cmpne(trunc(x),C) --> cmpne(x,C) // iff x upper bits are zero. diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h b/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h index 3fed003..5298728 100644 --- a/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h +++ b/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h @@ -280,8 +280,7 @@ public: VPValue *createElementCount(Type *Ty, ElementCount EC) { VPlan &Plan = *getInsertBlock()->getPlan(); - VPValue *RuntimeEC = - Plan.getOrAddLiveIn(ConstantInt::get(Ty, EC.getKnownMinValue())); + VPValue *RuntimeEC = Plan.getConstantInt(Ty, EC.getKnownMinValue()); if (EC.isScalable()) { VPValue *VScale = createNaryOp(VPInstruction::VScale, {}, Ty); RuntimeEC = EC.getKnownMinValue() == 1 diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp index 25bf49d..e5c3f17 100644 --- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp +++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp @@ -7752,8 +7752,7 @@ VPWidenRecipe *VPRecipeBuilder::tryToWiden(Instruction *I, if (CM.isPredicatedInst(I)) { SmallVector<VPValue *> Ops(Operands); VPValue *Mask = getBlockInMask(Builder.getInsertBlock()); - VPValue *One = - Plan.getOrAddLiveIn(ConstantInt::get(I->getType(), 1u, false)); + VPValue *One = Plan.getConstantInt(I->getType(), 1u); auto *SafeRHS = Builder.createSelect(Mask, Ops[1], One, I->getDebugLoc()); Ops[1] = SafeRHS; return new VPWidenRecipe(*I, Ops); @@ -7806,11 +7805,10 @@ VPWidenRecipe *VPRecipeBuilder::tryToWiden(Instruction *I, } case Instruction::ExtractValue: { SmallVector<VPValue *> NewOps(Operands); - Type *I32Ty = IntegerType::getInt32Ty(I->getContext()); auto *EVI = cast<ExtractValueInst>(I); assert(EVI->getNumIndices() == 1 && "Expected one extractvalue index"); unsigned Idx = EVI->getIndices()[0]; - NewOps.push_back(Plan.getOrAddLiveIn(ConstantInt::get(I32Ty, Idx, false))); + NewOps.push_back(Plan.getConstantInt(32, Idx)); return new VPWidenRecipe(*I, NewOps); } }; @@ -8179,8 +8177,7 @@ VPRecipeBuilder::tryToCreatePartialReduction(Instruction *Reduction, "Expected an ADD or SUB operation for predicated partial " "reductions (because the neutral element in the mask is zero)!"); Cond = getBlockInMask(Builder.getInsertBlock()); - VPValue *Zero = - Plan.getOrAddLiveIn(ConstantInt::get(Reduction->getType(), 0)); + VPValue *Zero = Plan.getConstantInt(Reduction->getType(), 0); BinOp = Builder.createSelect(Cond, BinOp, Zero, Reduction->getDebugLoc()); } return new VPPartialReductionRecipe(ReductionOpcode, Accumulator, BinOp, Cond, @@ -8643,7 +8640,7 @@ void LoopVectorizationPlanner::adjustRecipesForReductions( } else if (PhiR->isInLoop() && Kind == RecurKind::AddChainWithSubs && CurrentLinkI->getOpcode() == Instruction::Sub) { Type *PhiTy = PhiR->getUnderlyingValue()->getType(); - auto *Zero = Plan->getOrAddLiveIn(ConstantInt::get(PhiTy, 0)); + auto *Zero = Plan->getConstantInt(PhiTy, 0); VPWidenRecipe *Sub = new VPWidenRecipe( Instruction::Sub, {Zero, CurrentLink->getOperand(1)}, {}, VPIRMetadata(), CurrentLinkI->getDebugLoc()); @@ -8857,8 +8854,7 @@ void LoopVectorizationPlanner::adjustRecipesForReductions( ToDelete.push_back(Select); // Convert the reduction phi to operate on bools. - PhiR->setOperand(0, Plan->getOrAddLiveIn(ConstantInt::getFalse( - OrigLoop->getHeader()->getContext()))); + PhiR->setOperand(0, Plan->getFalse()); continue; } @@ -8880,9 +8876,7 @@ void LoopVectorizationPlanner::adjustRecipesForReductions( unsigned ScaleFactor = RecipeBuilder.getScalingForReduction(RdxDesc.getLoopExitInstr()) .value_or(1); - Type *I32Ty = IntegerType::getInt32Ty(PhiTy->getContext()); - auto *ScaleFactorVPV = - Plan->getOrAddLiveIn(ConstantInt::get(I32Ty, ScaleFactor)); + auto *ScaleFactorVPV = Plan->getConstantInt(32, ScaleFactor); VPValue *StartV = PHBuilder.createNaryOp( VPInstruction::ReductionStartVector, {PhiR->getStartValue(), Iden, ScaleFactorVPV}, diff --git a/llvm/lib/Transforms/Vectorize/VPlan.h b/llvm/lib/Transforms/Vectorize/VPlan.h index 1504acf..08c9c15 100644 --- a/llvm/lib/Transforms/Vectorize/VPlan.h +++ b/llvm/lib/Transforms/Vectorize/VPlan.h @@ -4393,15 +4393,25 @@ public: } /// Return a VPValue wrapping i1 true. - VPValue *getTrue() { - LLVMContext &Ctx = getContext(); - return getOrAddLiveIn(ConstantInt::getTrue(Ctx)); - } + VPValue *getTrue() { return getConstantInt(1, 1); } /// Return a VPValue wrapping i1 false. - VPValue *getFalse() { - LLVMContext &Ctx = getContext(); - return getOrAddLiveIn(ConstantInt::getFalse(Ctx)); + VPValue *getFalse() { return getConstantInt(1, 0); } + + /// Return a VPValue wrapping a ConstantInt with the given type and value. + VPValue *getConstantInt(Type *Ty, uint64_t Val, bool IsSigned = false) { + return getOrAddLiveIn(ConstantInt::get(Ty, Val, IsSigned)); + } + + /// Return a VPValue wrapping a ConstantInt with the given bitwidth and value. + VPValue *getConstantInt(unsigned BitWidth, uint64_t Val, + bool IsSigned = false) { + return getConstantInt(APInt(BitWidth, Val, IsSigned)); + } + + /// Return a VPValue wrapping a ConstantInt with the given APInt value. + VPValue *getConstantInt(const APInt &Val) { + return getOrAddLiveIn(ConstantInt::get(getContext(), Val)); } /// Return the live-in VPValue for \p V, if there is one or nullptr otherwise. diff --git a/llvm/lib/Transforms/Vectorize/VPlanConstruction.cpp b/llvm/lib/Transforms/Vectorize/VPlanConstruction.cpp index 65688a3..1a66d20 100644 --- a/llvm/lib/Transforms/Vectorize/VPlanConstruction.cpp +++ b/llvm/lib/Transforms/Vectorize/VPlanConstruction.cpp @@ -612,8 +612,7 @@ void VPlanTransforms::addMiddleCheck(VPlan &Plan, if (!RequiresScalarEpilogueCheck) Cmp = Plan.getFalse(); else if (TailFolded) - Cmp = Plan.getOrAddLiveIn( - ConstantInt::getTrue(IntegerType::getInt1Ty(Plan.getContext()))); + Cmp = Plan.getTrue(); else Cmp = Builder.createICmp(CmpInst::ICMP_EQ, Plan.getTripCount(), &Plan.getVectorTripCount(), LatchDL, "cmp.n"); @@ -712,8 +711,8 @@ void VPlanTransforms::addMinimumIterationCheck( // additional overflow check is required before entering the vector loop. // Get the maximum unsigned value for the type. - VPValue *MaxUIntTripCount = Plan.getOrAddLiveIn(ConstantInt::get( - TripCountTy, cast<IntegerType>(TripCountTy)->getMask())); + VPValue *MaxUIntTripCount = + Plan.getConstantInt(cast<IntegerType>(TripCountTy)->getMask()); VPValue *DistanceToMax = Builder.createNaryOp( Instruction::Sub, {MaxUIntTripCount, TripCountVPV}, DebugLoc::getUnknown()); diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp index d491d56..6a8231b 100644 --- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp +++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp @@ -699,8 +699,7 @@ static void legalizeAndOptimizeInductions(VPlan &Plan) { continue; const InductionDescriptor &ID = PtrIV->getInductionDescriptor(); - VPValue *StartV = - Plan.getOrAddLiveIn(ConstantInt::get(ID.getStep()->getType(), 0)); + VPValue *StartV = Plan.getConstantInt(ID.getStep()->getType(), 0); VPValue *StepV = PtrIV->getOperand(1); VPScalarIVStepsRecipe *Steps = createScalarIVSteps( Plan, InductionDescriptor::IK_IntInduction, Instruction::Add, nullptr, @@ -836,7 +835,7 @@ static VPValue *optimizeEarlyExitInductionUser(VPlan &Plan, // changed it means the exit is using the incremented value, so we need to // add the step. if (Incoming != WideIV) { - VPValue *One = Plan.getOrAddLiveIn(ConstantInt::get(CanonicalIVType, 1)); + VPValue *One = Plan.getConstantInt(CanonicalIVType, 1); EndValue = B.createNaryOp(Instruction::Add, {EndValue, One}, DL); } @@ -882,7 +881,7 @@ static VPValue *optimizeLatchExitInductionUser( return B.createNaryOp(Instruction::Sub, {EndValue, Step}, {}, "ind.escape"); if (ScalarTy->isPointerTy()) { Type *StepTy = TypeInfo.inferScalarType(Step); - auto *Zero = Plan.getOrAddLiveIn(ConstantInt::get(StepTy, 0)); + auto *Zero = Plan.getConstantInt(StepTy, 0); return B.createPtrAdd(EndValue, B.createNaryOp(Instruction::Sub, {Zero, Step}), DebugLoc::getUnknown(), "ind.escape"); @@ -1574,9 +1573,9 @@ static bool optimizeVectorInductionWidthForTCAndVFUF(VPlan &Plan, continue; // Update IV operands and comparison bound to use new narrower type. - auto *NewStart = Plan.getOrAddLiveIn(ConstantInt::get(NewIVTy, 0)); + auto *NewStart = Plan.getConstantInt(NewIVTy, 0); WideIV->setStartValue(NewStart); - auto *NewStep = Plan.getOrAddLiveIn(ConstantInt::get(NewIVTy, 1)); + auto *NewStep = Plan.getConstantInt(NewIVTy, 1); WideIV->setStepValue(NewStep); auto *NewBTC = new VPWidenCastRecipe( @@ -1695,8 +1694,7 @@ static bool tryToReplaceALMWithWideALM(VPlan &Plan, ElementCount VF, // When using wide lane masks, the return type of the get.active.lane.mask // intrinsic is VF x UF (last operand). - VPValue *ALMMultiplier = - Plan.getOrAddLiveIn(ConstantInt::get(IntegerType::getInt64Ty(Ctx), UF)); + VPValue *ALMMultiplier = Plan.getConstantInt(64, UF); EntryALM->setOperand(2, ALMMultiplier); LoopALM->setOperand(2, ALMMultiplier); @@ -2403,7 +2401,7 @@ static VPActiveLaneMaskPHIRecipe *addVPLaneMaskPhiAndUpdateExitBranch( // Create the active lane mask instruction in the VPlan preheader. VPValue *ALMMultiplier = - Plan.getOrAddLiveIn(ConstantInt::get(TopRegion->getCanonicalIVType(), 1)); + Plan.getConstantInt(TopRegion->getCanonicalIVType(), 1); auto *EntryALM = Builder.createNaryOp(VPInstruction::ActiveLaneMask, {EntryIncrement, TC, ALMMultiplier}, DL, "active.lane.mask.entry"); @@ -2790,8 +2788,7 @@ void VPlanTransforms::addExplicitVectorLength( if (MaxSafeElements) { // Support for MaxSafeDist for correct loop emission. - VPValue *AVLSafe = - Plan.getOrAddLiveIn(ConstantInt::get(CanIVTy, *MaxSafeElements)); + VPValue *AVLSafe = Plan.getConstantInt(CanIVTy, *MaxSafeElements); VPValue *Cmp = Builder.createICmp(ICmpInst::ICMP_ULT, AVL, AVLSafe); AVL = Builder.createSelect(Cmp, AVL, AVLSafe, DebugLoc::getUnknown(), "safe_avl"); @@ -2904,9 +2901,8 @@ void VPlanTransforms::canonicalizeEVLLoops(VPlan &Plan) { Type *AVLTy = VPTypeAnalysis(Plan).inferScalarType(AVLNext); VPBuilder Builder(LatchExitingBr); - VPValue *Cmp = - Builder.createICmp(CmpInst::ICMP_EQ, AVLNext, - Plan.getOrAddLiveIn(ConstantInt::getNullValue(AVLTy))); + VPValue *Cmp = Builder.createICmp(CmpInst::ICMP_EQ, AVLNext, + Plan.getConstantInt(AVLTy, 0)); Builder.createNaryOp(VPInstruction::BranchOnCond, Cmp); LatchExitingBr->eraseFromParent(); } @@ -2930,8 +2926,7 @@ void VPlanTransforms::replaceSymbolicStrides( // Only handle constant strides for now. continue; - auto *CI = - Plan.getOrAddLiveIn(ConstantInt::get(Stride->getType(), *StrideConst)); + auto *CI = Plan.getConstantInt(*StrideConst); if (VPValue *StrideVPV = Plan.getLiveIn(StrideV)) StrideVPV->replaceUsesWithIf(CI, CanUseVersionedStride); @@ -2946,7 +2941,7 @@ void VPlanTransforms::replaceSymbolicStrides( unsigned BW = U->getType()->getScalarSizeInBits(); APInt C = isa<SExtInst>(U) ? StrideConst->sext(BW) : StrideConst->zext(BW); - VPValue *CI = Plan.getOrAddLiveIn(ConstantInt::get(U->getType(), C)); + VPValue *CI = Plan.getConstantInt(C); StrideVPV->replaceUsesWithIf(CI, CanUseVersionedStride); } RewriteMap[StrideV] = PSE.getSCEV(StrideV); @@ -3125,8 +3120,7 @@ void VPlanTransforms::createInterleaveGroups( DL.getTypeAllocSize(getLoadStoreType(IRInsertPos)) * IG->getIndex(IRInsertPos), /*IsSigned=*/true); - VPValue *OffsetVPV = - Plan.getOrAddLiveIn(ConstantInt::get(Plan.getContext(), -Offset)); + VPValue *OffsetVPV = Plan.getConstantInt(-Offset); VPBuilder B(InsertPos); Addr = B.createNoWrapPtrAdd(InsertPos->getAddr(), OffsetVPV, NW); } @@ -3867,8 +3861,7 @@ void VPlanTransforms::materializeBackedgeTakenCount(VPlan &Plan, VPBuilder Builder(VectorPH, VectorPH->begin()); auto *TCTy = VPTypeAnalysis(Plan).inferScalarType(Plan.getTripCount()); auto *TCMO = Builder.createNaryOp( - Instruction::Sub, - {Plan.getTripCount(), Plan.getOrAddLiveIn(ConstantInt::get(TCTy, 1))}, + Instruction::Sub, {Plan.getTripCount(), Plan.getConstantInt(TCTy, 1)}, DebugLoc::getCompilerGenerated(), "trip.count.minus.1"); BTC->replaceAllUsesWith(TCMO); } @@ -3993,9 +3986,8 @@ void VPlanTransforms::materializeVectorTripCount(VPlan &Plan, if (TailByMasking) { TC = Builder.createNaryOp( Instruction::Add, - {TC, Builder.createNaryOp( - Instruction::Sub, - {Step, Plan.getOrAddLiveIn(ConstantInt::get(TCTy, 1))})}, + {TC, Builder.createNaryOp(Instruction::Sub, + {Step, Plan.getConstantInt(TCTy, 1)})}, DebugLoc::getCompilerGenerated(), "n.rnd.up"); } @@ -4017,8 +4009,8 @@ void VPlanTransforms::materializeVectorTripCount(VPlan &Plan, if (RequiresScalarEpilogue) { assert(!TailByMasking && "requiring scalar epilogue is not supported with fail folding"); - VPValue *IsZero = Builder.createICmp( - CmpInst::ICMP_EQ, R, Plan.getOrAddLiveIn(ConstantInt::get(TCTy, 0))); + VPValue *IsZero = + Builder.createICmp(CmpInst::ICMP_EQ, R, Plan.getConstantInt(TCTy, 0)); R = Builder.createSelect(IsZero, Step, R); } @@ -4056,7 +4048,7 @@ void VPlanTransforms::materializeVFAndVFxUF(VPlan &Plan, VPBasicBlock *VectorPH, } VF.replaceAllUsesWith(RuntimeVF); - VPValue *UF = Plan.getOrAddLiveIn(ConstantInt::get(TCTy, Plan.getUF())); + VPValue *UF = Plan.getConstantInt(TCTy, Plan.getUF()); VPValue *MulByUF = Builder.createNaryOp(Instruction::Mul, {RuntimeVF, UF}); VFxUF.replaceAllUsesWith(MulByUF); } @@ -4346,7 +4338,7 @@ void VPlanTransforms::narrowInterleaveGroups(VPlan &Plan, ElementCount VF, } else { Inc->setOperand(1, UF); Plan.getVF().replaceAllUsesWith( - Plan.getOrAddLiveIn(ConstantInt::get(CanIV->getScalarType(), 1))); + Plan.getConstantInt(CanIV->getScalarType(), 1)); } removeDeadRecipes(Plan); } diff --git a/llvm/lib/Transforms/Vectorize/VPlanUnroll.cpp b/llvm/lib/Transforms/Vectorize/VPlanUnroll.cpp index f15113c..d6a0028 100644 --- a/llvm/lib/Transforms/Vectorize/VPlanUnroll.cpp +++ b/llvm/lib/Transforms/Vectorize/VPlanUnroll.cpp @@ -68,9 +68,9 @@ class UnrollState { void unrollWidenInductionByUF(VPWidenInductionRecipe *IV, VPBasicBlock::iterator InsertPtForPhi); - VPValue *getConstantVPV(unsigned Part) { + VPValue *getConstantInt(unsigned Part) { Type *CanIVIntTy = Plan.getVectorLoopRegion()->getCanonicalIVType(); - return Plan.getOrAddLiveIn(ConstantInt::get(CanIVIntTy, Part)); + return Plan.getConstantInt(CanIVIntTy, Part); } public: @@ -137,7 +137,7 @@ void UnrollState::unrollReplicateRegionByUF(VPRegionBlock *VPR) { for (const auto &[PartIR, Part0R] : zip(*PartIVPBB, *Part0VPBB)) { remapOperands(&PartIR, Part); if (auto *ScalarIVSteps = dyn_cast<VPScalarIVStepsRecipe>(&PartIR)) { - ScalarIVSteps->addOperand(getConstantVPV(Part)); + ScalarIVSteps->addOperand(getConstantInt(Part)); } addRecipeForPart(&Part0R, &PartIR, Part); @@ -249,7 +249,7 @@ void UnrollState::unrollHeaderPHIByUF(VPHeaderPHIRecipe *R, for (unsigned Part = 1; Part != UF; ++Part) VPV2Parts[VPI][Part - 1] = StartV; } - Copy->addOperand(getConstantVPV(Part)); + Copy->addOperand(getConstantInt(Part)); } else { assert(isa<VPActiveLaneMaskPHIRecipe>(R) && "unexpected header phi recipe not needing unrolled part"); @@ -318,7 +318,7 @@ void UnrollState::unrollRecipeByUF(VPRecipeBase &R) { VPVectorPointerRecipe, VPVectorEndPointerRecipe>(Copy) || match(Copy, m_VPInstruction<VPInstruction::CanonicalIVIncrementForPart>())) - Copy->addOperand(getConstantVPV(Part)); + Copy->addOperand(getConstantInt(Part)); if (isa<VPVectorPointerRecipe, VPVectorEndPointerRecipe>(R)) Copy->setOperand(0, R.getOperand(0)); @@ -474,8 +474,7 @@ cloneForLane(VPlan &Plan, VPBuilder &Builder, Type *IdxTy, if (LaneDefs != Def2LaneDefs.end()) return LaneDefs->second[Lane.getKnownLane()]; - VPValue *Idx = - Plan.getOrAddLiveIn(ConstantInt::get(IdxTy, Lane.getKnownLane())); + VPValue *Idx = Plan.getConstantInt(IdxTy, Lane.getKnownLane()); return Builder.createNaryOp(Instruction::ExtractElement, {Op, Idx}); } @@ -509,8 +508,7 @@ cloneForLane(VPlan &Plan, VPBuilder &Builder, Type *IdxTy, cast<VPInstruction>(Op)->getOperand(Lane.getKnownLane())); continue; } - VPValue *Idx = - Plan.getOrAddLiveIn(ConstantInt::get(IdxTy, Lane.getKnownLane())); + VPValue *Idx = Plan.getConstantInt(IdxTy, Lane.getKnownLane()); VPValue *Ext = Builder.createNaryOp(Instruction::ExtractElement, {Op, Idx}); NewOps.push_back(Ext); } diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.abs.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.abs.ll index 02d0e52..6facdfd 100644 --- a/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.abs.ll +++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.abs.ll @@ -104,109 +104,110 @@ define amdgpu_cs <4 x i32> @abs_sgpr_v4i32(<4 x i32> inreg %arg) { ret <4 x i32> %res } -define amdgpu_cs i16 @abs_vgpr_i16(i16 %arg) { +define i16 @abs_vgpr_i16(i16 %arg) { ; GFX6-LABEL: abs_vgpr_i16: ; GFX6: ; %bb.0: +; GFX6-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX6-NEXT: v_bfe_i32 v0, v0, 0, 16 ; GFX6-NEXT: v_sub_i32_e32 v1, vcc, 0, v0 ; GFX6-NEXT: v_max_i32_e32 v0, v0, v1 -; GFX6-NEXT: v_readfirstlane_b32 s0, v0 -; GFX6-NEXT: ; return to shader part epilog +; GFX6-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: abs_vgpr_i16: ; GFX8: ; %bb.0: +; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_sub_u16_e32 v1, 0, v0 ; GFX8-NEXT: v_max_i16_e32 v0, v0, v1 -; GFX8-NEXT: v_readfirstlane_b32 s0, v0 -; GFX8-NEXT: ; return to shader part epilog +; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: abs_vgpr_i16: ; GFX10: ; %bb.0: +; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: v_sub_nc_u16 v1, 0, v0 ; GFX10-NEXT: v_max_i16 v0, v0, v1 -; GFX10-NEXT: v_readfirstlane_b32 s0, v0 -; GFX10-NEXT: ; return to shader part epilog +; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX1250-LABEL: abs_vgpr_i16: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_sub_nc_u16 v1, 0, v0 -; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-NEXT: v_max_i16 v0, v0, v1 -; GFX1250-NEXT: v_readfirstlane_b32 s0, v0 -; GFX1250-NEXT: ; return to shader part epilog +; GFX1250-NEXT: s_set_pc_i64 s[30:31] %res = call i16 @llvm.abs.i16(i16 %arg, i1 false) ret i16 %res } -define amdgpu_cs i32 @abs_vgpr_i32(i32 %arg) { +define i32 @abs_vgpr_i32(i32 %arg) { ; GFX6-LABEL: abs_vgpr_i32: ; GFX6: ; %bb.0: +; GFX6-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX6-NEXT: v_sub_i32_e32 v1, vcc, 0, v0 ; GFX6-NEXT: v_max_i32_e32 v0, v0, v1 -; GFX6-NEXT: v_readfirstlane_b32 s0, v0 -; GFX6-NEXT: ; return to shader part epilog +; GFX6-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: abs_vgpr_i32: ; GFX8: ; %bb.0: +; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_sub_u32_e32 v1, vcc, 0, v0 ; GFX8-NEXT: v_max_i32_e32 v0, v0, v1 -; GFX8-NEXT: v_readfirstlane_b32 s0, v0 -; GFX8-NEXT: ; return to shader part epilog +; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: abs_vgpr_i32: ; GFX10: ; %bb.0: +; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: v_sub_nc_u32_e32 v1, 0, v0 ; GFX10-NEXT: v_max_i32_e32 v0, v0, v1 -; GFX10-NEXT: v_readfirstlane_b32 s0, v0 -; GFX10-NEXT: ; return to shader part epilog +; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX1250-LABEL: abs_vgpr_i32: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_sub_nc_u32_e32 v1, 0, v0 -; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-NEXT: v_max_i32_e32 v0, v0, v1 -; GFX1250-NEXT: v_readfirstlane_b32 s0, v0 -; GFX1250-NEXT: ; return to shader part epilog +; GFX1250-NEXT: s_set_pc_i64 s[30:31] %res = call i32 @llvm.abs.i32(i32 %arg, i1 false) ret i32 %res } -define amdgpu_cs i64 @abs_vgpr_i64(i64 %arg) { +define i64 @abs_vgpr_i64(i64 %arg) { ; GFX6-LABEL: abs_vgpr_i64: ; GFX6: ; %bb.0: +; GFX6-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX6-NEXT: v_ashrrev_i32_e32 v2, 31, v1 ; GFX6-NEXT: v_add_i32_e32 v0, vcc, v0, v2 ; GFX6-NEXT: v_addc_u32_e32 v1, vcc, v1, v2, vcc ; GFX6-NEXT: v_xor_b32_e32 v0, v0, v2 ; GFX6-NEXT: v_xor_b32_e32 v1, v1, v2 -; GFX6-NEXT: v_readfirstlane_b32 s0, v0 -; GFX6-NEXT: v_readfirstlane_b32 s1, v1 -; GFX6-NEXT: ; return to shader part epilog +; GFX6-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: abs_vgpr_i64: ; GFX8: ; %bb.0: +; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_ashrrev_i32_e32 v2, 31, v1 ; GFX8-NEXT: v_add_u32_e32 v0, vcc, v0, v2 ; GFX8-NEXT: v_addc_u32_e32 v1, vcc, v1, v2, vcc ; GFX8-NEXT: v_xor_b32_e32 v0, v0, v2 ; GFX8-NEXT: v_xor_b32_e32 v1, v1, v2 -; GFX8-NEXT: v_readfirstlane_b32 s0, v0 -; GFX8-NEXT: v_readfirstlane_b32 s1, v1 -; GFX8-NEXT: ; return to shader part epilog +; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: abs_vgpr_i64: ; GFX10: ; %bb.0: +; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: v_ashrrev_i32_e32 v2, 31, v1 ; GFX10-NEXT: v_add_co_u32 v0, vcc_lo, v0, v2 ; GFX10-NEXT: v_add_co_ci_u32_e32 v1, vcc_lo, v1, v2, vcc_lo ; GFX10-NEXT: v_xor_b32_e32 v0, v0, v2 ; GFX10-NEXT: v_xor_b32_e32 v1, v1, v2 -; GFX10-NEXT: v_readfirstlane_b32 s0, v0 -; GFX10-NEXT: v_readfirstlane_b32 s1, v1 -; GFX10-NEXT: ; return to shader part epilog +; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX1250-LABEL: abs_vgpr_i64: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_ashrrev_i32_e32 v2, 31, v1 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) ; GFX1250-NEXT: v_mov_b32_e32 v3, v2 @@ -214,17 +215,15 @@ define amdgpu_cs i64 @abs_vgpr_i64(i64 %arg) { ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) ; GFX1250-NEXT: v_xor_b32_e32 v0, v0, v2 ; GFX1250-NEXT: v_xor_b32_e32 v1, v1, v2 -; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) -; GFX1250-NEXT: v_readfirstlane_b32 s0, v0 -; GFX1250-NEXT: v_readfirstlane_b32 s1, v1 -; GFX1250-NEXT: ; return to shader part epilog +; GFX1250-NEXT: s_set_pc_i64 s[30:31] %res = call i64 @llvm.abs.i64(i64 %arg, i1 false) ret i64 %res } -define amdgpu_cs <4 x i32> @abs_vgpr_v4i32(<4 x i32> %arg) { +define <4 x i32> @abs_vgpr_v4i32(<4 x i32> %arg) { ; GFX6-LABEL: abs_vgpr_v4i32: ; GFX6: ; %bb.0: +; GFX6-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX6-NEXT: v_sub_i32_e32 v4, vcc, 0, v0 ; GFX6-NEXT: v_max_i32_e32 v0, v0, v4 ; GFX6-NEXT: v_sub_i32_e32 v4, vcc, 0, v1 @@ -233,14 +232,11 @@ define amdgpu_cs <4 x i32> @abs_vgpr_v4i32(<4 x i32> %arg) { ; GFX6-NEXT: v_max_i32_e32 v2, v2, v4 ; GFX6-NEXT: v_sub_i32_e32 v4, vcc, 0, v3 ; GFX6-NEXT: v_max_i32_e32 v3, v3, v4 -; GFX6-NEXT: v_readfirstlane_b32 s0, v0 -; GFX6-NEXT: v_readfirstlane_b32 s1, v1 -; GFX6-NEXT: v_readfirstlane_b32 s2, v2 -; GFX6-NEXT: v_readfirstlane_b32 s3, v3 -; GFX6-NEXT: ; return to shader part epilog +; GFX6-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: abs_vgpr_v4i32: ; GFX8: ; %bb.0: +; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_sub_u32_e32 v4, vcc, 0, v0 ; GFX8-NEXT: v_max_i32_e32 v0, v0, v4 ; GFX8-NEXT: v_sub_u32_e32 v4, vcc, 0, v1 @@ -249,14 +245,11 @@ define amdgpu_cs <4 x i32> @abs_vgpr_v4i32(<4 x i32> %arg) { ; GFX8-NEXT: v_max_i32_e32 v2, v2, v4 ; GFX8-NEXT: v_sub_u32_e32 v4, vcc, 0, v3 ; GFX8-NEXT: v_max_i32_e32 v3, v3, v4 -; GFX8-NEXT: v_readfirstlane_b32 s0, v0 -; GFX8-NEXT: v_readfirstlane_b32 s1, v1 -; GFX8-NEXT: v_readfirstlane_b32 s2, v2 -; GFX8-NEXT: v_readfirstlane_b32 s3, v3 -; GFX8-NEXT: ; return to shader part epilog +; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: abs_vgpr_v4i32: ; GFX10: ; %bb.0: +; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: v_sub_nc_u32_e32 v4, 0, v0 ; GFX10-NEXT: v_sub_nc_u32_e32 v5, 0, v1 ; GFX10-NEXT: v_sub_nc_u32_e32 v6, 0, v2 @@ -265,14 +258,12 @@ define amdgpu_cs <4 x i32> @abs_vgpr_v4i32(<4 x i32> %arg) { ; GFX10-NEXT: v_max_i32_e32 v1, v1, v5 ; GFX10-NEXT: v_max_i32_e32 v2, v2, v6 ; GFX10-NEXT: v_max_i32_e32 v3, v3, v7 -; GFX10-NEXT: v_readfirstlane_b32 s0, v0 -; GFX10-NEXT: v_readfirstlane_b32 s1, v1 -; GFX10-NEXT: v_readfirstlane_b32 s2, v2 -; GFX10-NEXT: v_readfirstlane_b32 s3, v3 -; GFX10-NEXT: ; return to shader part epilog +; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX1250-LABEL: abs_vgpr_v4i32: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_dual_sub_nc_u32 v4, 0, v0 :: v_dual_sub_nc_u32 v5, 0, v1 ; GFX1250-NEXT: v_dual_sub_nc_u32 v6, 0, v2 :: v_dual_sub_nc_u32 v7, 0, v3 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) @@ -281,13 +272,7 @@ define amdgpu_cs <4 x i32> @abs_vgpr_v4i32(<4 x i32> %arg) { ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_4) ; GFX1250-NEXT: v_max_i32_e32 v2, v2, v6 ; GFX1250-NEXT: v_max_i32_e32 v3, v3, v7 -; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4) -; GFX1250-NEXT: v_readfirstlane_b32 s0, v0 -; GFX1250-NEXT: v_readfirstlane_b32 s1, v1 -; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4) -; GFX1250-NEXT: v_readfirstlane_b32 s2, v2 -; GFX1250-NEXT: v_readfirstlane_b32 s3, v3 -; GFX1250-NEXT: ; return to shader part epilog +; GFX1250-NEXT: s_set_pc_i64 s[30:31] %res = call <4 x i32> @llvm.abs.v4i32(<4 x i32> %arg, i1 false) ret <4 x i32> %res } @@ -304,44 +289,43 @@ define amdgpu_cs <2 x i8> @abs_sgpr_v2i8(<2 x i8> inreg %arg) { ret <2 x i8> %res } -define amdgpu_cs <2 x i8> @abs_vgpr_v2i8(<2 x i8> %arg) { +define <2 x i8> @abs_vgpr_v2i8(<2 x i8> %arg) { ; GFX6-LABEL: abs_vgpr_v2i8: ; GFX6: ; %bb.0: +; GFX6-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX6-NEXT: v_bfe_i32 v0, v0, 0, 8 ; GFX6-NEXT: v_sub_i32_e32 v2, vcc, 0, v0 ; GFX6-NEXT: v_bfe_i32 v1, v1, 0, 8 ; GFX6-NEXT: v_max_i32_e32 v0, v0, v2 ; GFX6-NEXT: v_sub_i32_e32 v2, vcc, 0, v1 ; GFX6-NEXT: v_max_i32_e32 v1, v1, v2 -; GFX6-NEXT: v_readfirstlane_b32 s0, v0 -; GFX6-NEXT: v_readfirstlane_b32 s1, v1 -; GFX6-NEXT: ; return to shader part epilog +; GFX6-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: abs_vgpr_v2i8: ; GFX8: ; %bb.0: +; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_mov_b32_e32 v2, 0 ; GFX8-NEXT: v_sub_u16_sdwa v3, v2, sext(v0) dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0 ; GFX8-NEXT: v_sub_u16_sdwa v2, v2, sext(v1) dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0 ; GFX8-NEXT: v_max_i16_sdwa v0, sext(v0), v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD ; GFX8-NEXT: v_max_i16_sdwa v1, sext(v1), v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD -; GFX8-NEXT: v_readfirstlane_b32 s0, v0 -; GFX8-NEXT: v_readfirstlane_b32 s1, v1 -; GFX8-NEXT: ; return to shader part epilog +; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: abs_vgpr_v2i8: ; GFX10: ; %bb.0: +; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: v_bfe_i32 v0, v0, 0, 8 ; GFX10-NEXT: v_bfe_i32 v1, v1, 0, 8 ; GFX10-NEXT: v_sub_nc_u16 v2, 0, v0 ; GFX10-NEXT: v_sub_nc_u16 v3, 0, v1 ; GFX10-NEXT: v_max_i16 v0, v0, v2 ; GFX10-NEXT: v_max_i16 v1, v1, v3 -; GFX10-NEXT: v_readfirstlane_b32 s0, v0 -; GFX10-NEXT: v_readfirstlane_b32 s1, v1 -; GFX10-NEXT: ; return to shader part epilog +; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX1250-LABEL: abs_vgpr_v2i8: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_bfe_i32 v0, v0, 0, 8 ; GFX1250-NEXT: v_bfe_i32 v1, v1, 0, 8 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) @@ -350,10 +334,7 @@ define amdgpu_cs <2 x i8> @abs_vgpr_v2i8(<2 x i8> %arg) { ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) ; GFX1250-NEXT: v_max_i16 v0, v0, v2 ; GFX1250-NEXT: v_max_i16 v1, v1, v3 -; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) -; GFX1250-NEXT: v_readfirstlane_b32 s0, v0 -; GFX1250-NEXT: v_readfirstlane_b32 s1, v1 -; GFX1250-NEXT: ; return to shader part epilog +; GFX1250-NEXT: s_set_pc_i64 s[30:31] %res = call <2 x i8> @llvm.abs.v2i8(<2 x i8> %arg, i1 false) ret <2 x i8> %res } @@ -372,9 +353,10 @@ define amdgpu_cs <3 x i8> @abs_sgpr_v3i8(<3 x i8> inreg %arg) { ret <3 x i8> %res } -define amdgpu_cs <3 x i8> @abs_vgpr_v3i8(<3 x i8> %arg) { +define <3 x i8> @abs_vgpr_v3i8(<3 x i8> %arg) { ; GFX6-LABEL: abs_vgpr_v3i8: ; GFX6: ; %bb.0: +; GFX6-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX6-NEXT: v_bfe_i32 v0, v0, 0, 8 ; GFX6-NEXT: v_sub_i32_e32 v3, vcc, 0, v0 ; GFX6-NEXT: v_bfe_i32 v1, v1, 0, 8 @@ -384,13 +366,11 @@ define amdgpu_cs <3 x i8> @abs_vgpr_v3i8(<3 x i8> %arg) { ; GFX6-NEXT: v_max_i32_e32 v1, v1, v3 ; GFX6-NEXT: v_sub_i32_e32 v3, vcc, 0, v2 ; GFX6-NEXT: v_max_i32_e32 v2, v2, v3 -; GFX6-NEXT: v_readfirstlane_b32 s0, v0 -; GFX6-NEXT: v_readfirstlane_b32 s1, v1 -; GFX6-NEXT: v_readfirstlane_b32 s2, v2 -; GFX6-NEXT: ; return to shader part epilog +; GFX6-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: abs_vgpr_v3i8: ; GFX8: ; %bb.0: +; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_mov_b32_e32 v3, 0 ; GFX8-NEXT: v_sub_u16_sdwa v4, v3, sext(v0) dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0 ; GFX8-NEXT: v_max_i16_sdwa v0, sext(v0), v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD @@ -398,13 +378,11 @@ define amdgpu_cs <3 x i8> @abs_vgpr_v3i8(<3 x i8> %arg) { ; GFX8-NEXT: v_sub_u16_sdwa v3, v3, sext(v2) dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0 ; GFX8-NEXT: v_max_i16_sdwa v1, sext(v1), v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD ; GFX8-NEXT: v_max_i16_sdwa v2, sext(v2), v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD -; GFX8-NEXT: v_readfirstlane_b32 s0, v0 -; GFX8-NEXT: v_readfirstlane_b32 s1, v1 -; GFX8-NEXT: v_readfirstlane_b32 s2, v2 -; GFX8-NEXT: ; return to shader part epilog +; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: abs_vgpr_v3i8: ; GFX10: ; %bb.0: +; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: v_bfe_i32 v0, v0, 0, 8 ; GFX10-NEXT: v_bfe_i32 v1, v1, 0, 8 ; GFX10-NEXT: v_bfe_i32 v2, v2, 0, 8 @@ -414,13 +392,12 @@ define amdgpu_cs <3 x i8> @abs_vgpr_v3i8(<3 x i8> %arg) { ; GFX10-NEXT: v_max_i16 v0, v0, v3 ; GFX10-NEXT: v_max_i16 v1, v1, v4 ; GFX10-NEXT: v_max_i16 v2, v2, v5 -; GFX10-NEXT: v_readfirstlane_b32 s0, v0 -; GFX10-NEXT: v_readfirstlane_b32 s1, v1 -; GFX10-NEXT: v_readfirstlane_b32 s2, v2 -; GFX10-NEXT: ; return to shader part epilog +; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX1250-LABEL: abs_vgpr_v3i8: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_bfe_i32 v0, v0, 0, 8 ; GFX1250-NEXT: v_bfe_i32 v1, v1, 0, 8 ; GFX1250-NEXT: v_bfe_i32 v2, v2, 0, 8 @@ -433,12 +410,7 @@ define amdgpu_cs <3 x i8> @abs_vgpr_v3i8(<3 x i8> %arg) { ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) ; GFX1250-NEXT: v_max_i16 v1, v1, v4 ; GFX1250-NEXT: v_max_i16 v2, v2, v5 -; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) -; GFX1250-NEXT: v_readfirstlane_b32 s0, v0 -; GFX1250-NEXT: v_readfirstlane_b32 s1, v1 -; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_3) -; GFX1250-NEXT: v_readfirstlane_b32 s2, v2 -; GFX1250-NEXT: ; return to shader part epilog +; GFX1250-NEXT: s_set_pc_i64 s[30:31] %res = call <3 x i8> @llvm.abs.v3i8(<3 x i8> %arg, i1 false) ret <3 x i8> %res } @@ -485,44 +457,44 @@ define amdgpu_cs <2 x i16> @abs_sgpr_v2i16(<2 x i16> inreg %arg) { ret <2 x i16> %res } -define amdgpu_cs <2 x i16> @abs_vgpr_v2i16(<2 x i16> %arg) { +define <2 x i16> @abs_vgpr_v2i16(<2 x i16> %arg) { ; GFX6-LABEL: abs_vgpr_v2i16: ; GFX6: ; %bb.0: +; GFX6-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX6-NEXT: v_bfe_i32 v0, v0, 0, 16 ; GFX6-NEXT: v_sub_i32_e32 v2, vcc, 0, v0 ; GFX6-NEXT: v_bfe_i32 v1, v1, 0, 16 ; GFX6-NEXT: v_max_i32_e32 v0, v0, v2 ; GFX6-NEXT: v_sub_i32_e32 v2, vcc, 0, v1 ; GFX6-NEXT: v_max_i32_e32 v1, v1, v2 -; GFX6-NEXT: v_readfirstlane_b32 s0, v0 -; GFX6-NEXT: v_readfirstlane_b32 s1, v1 -; GFX6-NEXT: ; return to shader part epilog +; GFX6-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: abs_vgpr_v2i16: ; GFX8: ; %bb.0: +; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_mov_b32_e32 v2, 0 ; GFX8-NEXT: v_sub_u16_e32 v1, 0, v0 ; GFX8-NEXT: v_sub_u16_sdwa v2, v2, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1 ; GFX8-NEXT: v_max_i16_e32 v1, v0, v1 ; GFX8-NEXT: v_max_i16_sdwa v0, v0, v2 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD ; GFX8-NEXT: v_or_b32_e32 v0, v1, v0 -; GFX8-NEXT: v_readfirstlane_b32 s0, v0 -; GFX8-NEXT: ; return to shader part epilog +; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: abs_vgpr_v2i16: ; GFX10: ; %bb.0: +; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: v_pk_sub_i16 v1, 0, v0 ; GFX10-NEXT: v_pk_max_i16 v0, v0, v1 -; GFX10-NEXT: v_readfirstlane_b32 s0, v0 -; GFX10-NEXT: ; return to shader part epilog +; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX1250-LABEL: abs_vgpr_v2i16: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_pk_sub_i16 v1, 0, v0 -; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) +; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_1) ; GFX1250-NEXT: v_pk_max_i16 v0, v0, v1 -; GFX1250-NEXT: v_readfirstlane_b32 s0, v0 -; GFX1250-NEXT: ; return to shader part epilog +; GFX1250-NEXT: s_set_pc_i64 s[30:31] %res = call <2 x i16> @llvm.abs.v2i16(<2 x i16> %arg, i1 false) ret <2 x i16> %res } @@ -576,9 +548,10 @@ define amdgpu_cs <3 x i16> @abs_sgpr_v3i16(<3 x i16> inreg %arg) { ret <3 x i16> %res } -define amdgpu_cs <3 x i16> @abs_vgpr_v3i16(<3 x i16> %arg) { +define <3 x i16> @abs_vgpr_v3i16(<3 x i16> %arg) { ; GFX6-LABEL: abs_vgpr_v3i16: ; GFX6: ; %bb.0: +; GFX6-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX6-NEXT: v_bfe_i32 v0, v0, 0, 16 ; GFX6-NEXT: v_sub_i32_e32 v3, vcc, 0, v0 ; GFX6-NEXT: v_bfe_i32 v1, v1, 0, 16 @@ -588,13 +561,11 @@ define amdgpu_cs <3 x i16> @abs_vgpr_v3i16(<3 x i16> %arg) { ; GFX6-NEXT: v_max_i32_e32 v1, v1, v3 ; GFX6-NEXT: v_sub_i32_e32 v3, vcc, 0, v2 ; GFX6-NEXT: v_max_i32_e32 v2, v2, v3 -; GFX6-NEXT: v_readfirstlane_b32 s0, v0 -; GFX6-NEXT: v_readfirstlane_b32 s1, v1 -; GFX6-NEXT: v_readfirstlane_b32 s2, v2 -; GFX6-NEXT: ; return to shader part epilog +; GFX6-NEXT: s_setpc_b64 s[30:31] ; ; GFX8-LABEL: abs_vgpr_v3i16: ; GFX8: ; %bb.0: +; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX8-NEXT: v_mov_b32_e32 v3, 0 ; GFX8-NEXT: v_sub_u16_e32 v2, 0, v0 ; GFX8-NEXT: v_sub_u16_sdwa v3, v3, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1 @@ -603,31 +574,27 @@ define amdgpu_cs <3 x i16> @abs_vgpr_v3i16(<3 x i16> %arg) { ; GFX8-NEXT: v_max_i16_sdwa v0, v0, v3 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD ; GFX8-NEXT: v_or_b32_e32 v0, v2, v0 ; GFX8-NEXT: v_max_i16_e32 v1, v1, v4 -; GFX8-NEXT: v_readfirstlane_b32 s0, v0 -; GFX8-NEXT: v_readfirstlane_b32 s1, v1 -; GFX8-NEXT: ; return to shader part epilog +; GFX8-NEXT: s_setpc_b64 s[30:31] ; ; GFX10-LABEL: abs_vgpr_v3i16: ; GFX10: ; %bb.0: +; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0) ; GFX10-NEXT: v_pk_sub_i16 v2, 0, v0 ; GFX10-NEXT: v_sub_nc_u16 v3, 0, v1 ; GFX10-NEXT: v_pk_max_i16 v0, v0, v2 ; GFX10-NEXT: v_max_i16 v1, v1, v3 -; GFX10-NEXT: v_readfirstlane_b32 s0, v0 -; GFX10-NEXT: v_readfirstlane_b32 s1, v1 -; GFX10-NEXT: ; return to shader part epilog +; GFX10-NEXT: s_setpc_b64 s[30:31] ; ; GFX1250-LABEL: abs_vgpr_v3i16: ; GFX1250: ; %bb.0: +; GFX1250-NEXT: s_wait_loadcnt_dscnt 0x0 +; GFX1250-NEXT: s_wait_kmcnt 0x0 ; GFX1250-NEXT: v_pk_sub_i16 v2, 0, v0 ; GFX1250-NEXT: v_sub_nc_u16 v3, 0, v1 ; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) ; GFX1250-NEXT: v_pk_max_i16 v0, v0, v2 ; GFX1250-NEXT: v_max_i16 v1, v1, v3 -; GFX1250-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) -; GFX1250-NEXT: v_readfirstlane_b32 s0, v0 -; GFX1250-NEXT: v_readfirstlane_b32 s1, v1 -; GFX1250-NEXT: ; return to shader part epilog +; GFX1250-NEXT: s_set_pc_i64 s[30:31] %res = call <3 x i16> @llvm.abs.v3i16(<3 x i16> %arg, i1 false) ret <3 x i16> %res } diff --git a/llvm/test/CodeGen/AMDGPU/wait-xcnt.mir b/llvm/test/CodeGen/AMDGPU/wait-xcnt.mir index 1b8e126..a1381ec 100644 --- a/llvm/test/CodeGen/AMDGPU/wait-xcnt.mir +++ b/llvm/test/CodeGen/AMDGPU/wait-xcnt.mir @@ -945,7 +945,6 @@ body: | $vgpr0 = V_MOV_B32_e32 0, implicit $exec ... -# FIXME: Missing S_WAIT_XCNT before overwriting vgpr0. --- name: wait_kmcnt_with_outstanding_vmem_2 tracksRegLiveness: true @@ -971,6 +970,7 @@ body: | ; GCN-NEXT: {{ $}} ; GCN-NEXT: S_WAIT_KMCNT 0 ; GCN-NEXT: $sgpr2 = S_MOV_B32 $sgpr2 + ; GCN-NEXT: S_WAIT_XCNT 0 ; GCN-NEXT: $vgpr0 = V_MOV_B32_e32 0, implicit $exec bb.0: liveins: $vgpr0_vgpr1, $sgpr0_sgpr1, $scc @@ -986,6 +986,180 @@ body: | ... --- +name: wait_kmcnt_and_wait_loadcnt +tracksRegLiveness: true +machineFunctionInfo: + isEntryFunction: true +body: | + ; GCN-LABEL: name: wait_kmcnt_and_wait_loadcnt + ; GCN: bb.0: + ; GCN-NEXT: successors: %bb.2(0x40000000), %bb.1(0x40000000) + ; GCN-NEXT: liveins: $vgpr0_vgpr1, $sgpr0_sgpr1, $scc + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: $sgpr2 = S_LOAD_DWORD_IMM $sgpr0_sgpr1, 0, 0 + ; GCN-NEXT: S_CBRANCH_SCC1 %bb.2, implicit $scc + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: bb.1: + ; GCN-NEXT: successors: %bb.2(0x80000000) + ; GCN-NEXT: liveins: $vgpr0_vgpr1, $sgpr2 + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: $vgpr2 = GLOBAL_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: bb.2: + ; GCN-NEXT: liveins: $sgpr2 + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: S_WAIT_KMCNT 0 + ; GCN-NEXT: $sgpr2 = S_MOV_B32 $sgpr2 + ; GCN-NEXT: S_WAIT_LOADCNT 0 + ; GCN-NEXT: $vgpr2 = V_MOV_B32_e32 0, implicit $exec + bb.0: + liveins: $vgpr0_vgpr1, $sgpr0_sgpr1, $scc + $sgpr2 = S_LOAD_DWORD_IMM $sgpr0_sgpr1, 0, 0 + S_CBRANCH_SCC1 %bb.2, implicit $scc + bb.1: + liveins: $vgpr0_vgpr1, $sgpr2 + $vgpr2 = GLOBAL_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec + bb.2: + liveins: $sgpr2 + $sgpr2 = S_MOV_B32 $sgpr2 + $vgpr2 = V_MOV_B32_e32 0, implicit $exec +... + +--- +name: implicit_handling_of_pending_vmem_group +tracksRegLiveness: true +machineFunctionInfo: + isEntryFunction: true +body: | + ; GCN-LABEL: name: implicit_handling_of_pending_vmem_group + ; GCN: bb.0: + ; GCN-NEXT: successors: %bb.2(0x40000000), %bb.1(0x40000000) + ; GCN-NEXT: liveins: $vgpr0_vgpr1, $sgpr0_sgpr1, $scc + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: $sgpr2 = S_LOAD_DWORD_IMM $sgpr0_sgpr1, 0, 0 + ; GCN-NEXT: S_CBRANCH_SCC1 %bb.2, implicit $scc + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: bb.1: + ; GCN-NEXT: successors: %bb.2(0x80000000) + ; GCN-NEXT: liveins: $vgpr0_vgpr1, $sgpr2 + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: $vgpr2 = GLOBAL_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: bb.2: + ; GCN-NEXT: liveins: $sgpr0_sgpr1, $sgpr2 + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: S_WAIT_KMCNT 0 + ; GCN-NEXT: $sgpr2 = S_MOV_B32 $sgpr2 + ; GCN-NEXT: $sgpr2 = S_LOAD_DWORD_IMM $sgpr0_sgpr1, 0, 0 + ; GCN-NEXT: $vgpr0 = V_MOV_B32_e32 0, implicit $exec + ; GCN-NEXT: S_WAIT_XCNT 0 + ; GCN-NEXT: $sgpr0 = S_MOV_B32 $sgpr0 + bb.0: + liveins: $vgpr0_vgpr1, $sgpr0_sgpr1, $scc + $sgpr2 = S_LOAD_DWORD_IMM $sgpr0_sgpr1, 0, 0 + S_CBRANCH_SCC1 %bb.2, implicit $scc + bb.1: + liveins: $vgpr0_vgpr1, $sgpr2 + $vgpr2 = GLOBAL_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec + bb.2: + liveins: $sgpr0_sgpr1, $sgpr2 + $sgpr2 = S_MOV_B32 $sgpr2 + $sgpr2 = S_LOAD_DWORD_IMM $sgpr0_sgpr1, 0, 0 + $vgpr0 = V_MOV_B32_e32 0, implicit $exec + $sgpr0 = S_MOV_B32 $sgpr0 +... + +--- +name: pending_vmem_event_between_block +tracksRegLiveness: true +machineFunctionInfo: + isEntryFunction: true +body: | + ; GCN-LABEL: name: pending_vmem_event_between_block + ; GCN: bb.0: + ; GCN-NEXT: successors: %bb.2(0x40000000), %bb.1(0x40000000) + ; GCN-NEXT: liveins: $vgpr0_vgpr1, $sgpr0_sgpr1, $scc + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: $sgpr2 = S_LOAD_DWORD_IMM $sgpr0_sgpr1, 0, 0 + ; GCN-NEXT: S_CBRANCH_SCC1 %bb.2, implicit $scc + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: bb.1: + ; GCN-NEXT: successors: %bb.2(0x80000000) + ; GCN-NEXT: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3, $sgpr2 + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: $vgpr4 = GLOBAL_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec + ; GCN-NEXT: $vgpr5 = GLOBAL_LOAD_DWORD $vgpr2_vgpr3, 0, 0, implicit $exec + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: bb.2: + ; GCN-NEXT: liveins: $sgpr0_sgpr1, $sgpr2, $vgpr2 + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: S_WAIT_KMCNT 0 + ; GCN-NEXT: $sgpr2 = S_MOV_B32 $sgpr2 + ; GCN-NEXT: S_WAIT_XCNT 1 + ; GCN-NEXT: $vgpr1 = V_MOV_B32_e32 0, implicit $exec + ; GCN-NEXT: S_WAIT_XCNT 0 + ; GCN-NEXT: $vgpr2 = V_MOV_B32_e32 0, implicit $exec + ; GCN-NEXT: $sgpr0 = S_MOV_B32 $sgpr0 + bb.0: + liveins: $vgpr0_vgpr1, $sgpr0_sgpr1, $scc + $sgpr2 = S_LOAD_DWORD_IMM $sgpr0_sgpr1, 0, 0 + S_CBRANCH_SCC1 %bb.2, implicit $scc + bb.1: + liveins: $vgpr0_vgpr1, $vgpr2_vgpr3, $sgpr2 + $vgpr4 = GLOBAL_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec + $vgpr5 = GLOBAL_LOAD_DWORD $vgpr2_vgpr3, 0, 0, implicit $exec + bb.2: + liveins: $sgpr0_sgpr1, $sgpr2, $vgpr2 + $sgpr2 = S_MOV_B32 $sgpr2 + $vgpr1 = V_MOV_B32_e32 0, implicit $exec + $vgpr2 = V_MOV_B32_e32 0, implicit $exec + $sgpr0 = S_MOV_B32 $sgpr0 +... + +--- +name: flushing_vmem_cnt_on_block_entry +tracksRegLiveness: true +machineFunctionInfo: + isEntryFunction: true +body: | + ; GCN-LABEL: name: flushing_vmem_cnt_on_block_entry + ; GCN: bb.0: + ; GCN-NEXT: successors: %bb.2(0x40000000), %bb.1(0x40000000) + ; GCN-NEXT: liveins: $vgpr0_vgpr1, $sgpr0_sgpr1, $scc + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: $sgpr2 = S_LOAD_DWORD_IMM $sgpr0_sgpr1, 0, 0 + ; GCN-NEXT: S_CBRANCH_SCC1 %bb.2, implicit $scc + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: bb.1: + ; GCN-NEXT: successors: %bb.2(0x80000000) + ; GCN-NEXT: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3, $sgpr2 + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: $vgpr4 = GLOBAL_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec + ; GCN-NEXT: $vgpr5 = GLOBAL_LOAD_DWORD $vgpr2_vgpr3, 0, 0, implicit $exec + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: bb.2: + ; GCN-NEXT: liveins: $sgpr0_sgpr1, $sgpr2, $vgpr2 + ; GCN-NEXT: {{ $}} + ; GCN-NEXT: S_WAIT_XCNT 0 + ; GCN-NEXT: $vgpr1 = V_MOV_B32_e32 0, implicit $exec + ; GCN-NEXT: $vgpr2 = V_MOV_B32_e32 0, implicit $exec + ; GCN-NEXT: $sgpr0 = S_MOV_B32 $sgpr0 + bb.0: + liveins: $vgpr0_vgpr1, $sgpr0_sgpr1, $scc + $sgpr2 = S_LOAD_DWORD_IMM $sgpr0_sgpr1, 0, 0 + S_CBRANCH_SCC1 %bb.2, implicit $scc + bb.1: + liveins: $vgpr0_vgpr1, $vgpr2_vgpr3, $sgpr2 + $vgpr4 = GLOBAL_LOAD_DWORD $vgpr0_vgpr1, 0, 0, implicit $exec + $vgpr5 = GLOBAL_LOAD_DWORD $vgpr2_vgpr3, 0, 0, implicit $exec + bb.2: + liveins: $sgpr0_sgpr1, $sgpr2, $vgpr2 + $vgpr1 = V_MOV_B32_e32 0, implicit $exec + $vgpr2 = V_MOV_B32_e32 0, implicit $exec + $sgpr0 = S_MOV_B32 $sgpr0 +... + +--- name: wait_loadcnt_with_outstanding_smem tracksRegLiveness: true machineFunctionInfo: diff --git a/llvm/test/CodeGen/LoongArch/lasx/ctpop-ctlz.ll b/llvm/test/CodeGen/LoongArch/lasx/ctpop-ctlz.ll index ba2118f..b3155c9 100644 --- a/llvm/test/CodeGen/LoongArch/lasx/ctpop-ctlz.ll +++ b/llvm/test/CodeGen/LoongArch/lasx/ctpop-ctlz.ll @@ -106,6 +106,69 @@ define void @ctlz_v4i64(ptr %src, ptr %dst) nounwind { ret void } +define void @not_ctlz_v32i8(ptr %src, ptr %dst) nounwind { +; CHECK-LABEL: not_ctlz_v32i8: +; CHECK: # %bb.0: +; CHECK-NEXT: xvld $xr0, $a0, 0 +; CHECK-NEXT: xvxori.b $xr0, $xr0, 255 +; CHECK-NEXT: xvclz.b $xr0, $xr0 +; CHECK-NEXT: xvst $xr0, $a1, 0 +; CHECK-NEXT: ret + %v = load <32 x i8>, ptr %src + %neg = xor <32 x i8> %v, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1> + %res = call <32 x i8> @llvm.ctlz.v32i8(<32 x i8> %neg, i1 false) + store <32 x i8> %res, ptr %dst + ret void +} + +define void @not_ctlz_v16i16(ptr %src, ptr %dst) nounwind { +; CHECK-LABEL: not_ctlz_v16i16: +; CHECK: # %bb.0: +; CHECK-NEXT: xvld $xr0, $a0, 0 +; CHECK-NEXT: xvrepli.b $xr1, -1 +; CHECK-NEXT: xvxor.v $xr0, $xr0, $xr1 +; CHECK-NEXT: xvclz.h $xr0, $xr0 +; CHECK-NEXT: xvst $xr0, $a1, 0 +; CHECK-NEXT: ret + %v = load <16 x i16>, ptr %src + %neg = xor <16 x i16> %v, <i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1> + %res = call <16 x i16> @llvm.ctlz.v16i16(<16 x i16> %neg, i1 false) + store <16 x i16> %res, ptr %dst + ret void +} + +define void @not_ctlz_v8i32(ptr %src, ptr %dst) nounwind { +; CHECK-LABEL: not_ctlz_v8i32: +; CHECK: # %bb.0: +; CHECK-NEXT: xvld $xr0, $a0, 0 +; CHECK-NEXT: xvrepli.b $xr1, -1 +; CHECK-NEXT: xvxor.v $xr0, $xr0, $xr1 +; CHECK-NEXT: xvclz.w $xr0, $xr0 +; CHECK-NEXT: xvst $xr0, $a1, 0 +; CHECK-NEXT: ret + %v = load <8 x i32>, ptr %src + %neg = xor <8 x i32> %v, <i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1> + %res = call <8 x i32> @llvm.ctlz.v8i32(<8 x i32> %neg, i1 false) + store <8 x i32> %res, ptr %dst + ret void +} + +define void @not_ctlz_v4i64(ptr %src, ptr %dst) nounwind { +; CHECK-LABEL: not_ctlz_v4i64: +; CHECK: # %bb.0: +; CHECK-NEXT: xvld $xr0, $a0, 0 +; CHECK-NEXT: xvrepli.b $xr1, -1 +; CHECK-NEXT: xvxor.v $xr0, $xr0, $xr1 +; CHECK-NEXT: xvclz.d $xr0, $xr0 +; CHECK-NEXT: xvst $xr0, $a1, 0 +; CHECK-NEXT: ret + %v = load <4 x i64>, ptr %src + %neg = xor <4 x i64> %v, <i64 -1, i64 -1, i64 -1, i64 -1> + %res = call <4 x i64> @llvm.ctlz.v4i64(<4 x i64> %neg, i1 false) + store <4 x i64> %res, ptr %dst + ret void +} + declare <32 x i8> @llvm.ctpop.v32i8(<32 x i8>) declare <16 x i16> @llvm.ctpop.v16i16(<16 x i16>) declare <8 x i32> @llvm.ctpop.v8i32(<8 x i32>) diff --git a/llvm/test/CodeGen/LoongArch/lsx/ctpop-ctlz.ll b/llvm/test/CodeGen/LoongArch/lsx/ctpop-ctlz.ll index a9a38e8..6ac7d51 100644 --- a/llvm/test/CodeGen/LoongArch/lsx/ctpop-ctlz.ll +++ b/llvm/test/CodeGen/LoongArch/lsx/ctpop-ctlz.ll @@ -106,6 +106,69 @@ define void @ctlz_v2i64(ptr %src, ptr %dst) nounwind { ret void } +define void @not_ctlz_v16i8(ptr %src, ptr %dst) nounwind { +; CHECK-LABEL: not_ctlz_v16i8: +; CHECK: # %bb.0: +; CHECK-NEXT: vld $vr0, $a0, 0 +; CHECK-NEXT: vxori.b $vr0, $vr0, 255 +; CHECK-NEXT: vclz.b $vr0, $vr0 +; CHECK-NEXT: vst $vr0, $a1, 0 +; CHECK-NEXT: ret + %v = load <16 x i8>, ptr %src + %neg = xor <16 x i8> %v, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1> + %res = call <16 x i8> @llvm.ctlz.v16i8(<16 x i8> %neg, i1 false) + store <16 x i8> %res, ptr %dst + ret void +} + +define void @not_ctlz_v8i16(ptr %src, ptr %dst) nounwind { +; CHECK-LABEL: not_ctlz_v8i16: +; CHECK: # %bb.0: +; CHECK-NEXT: vld $vr0, $a0, 0 +; CHECK-NEXT: vrepli.b $vr1, -1 +; CHECK-NEXT: vxor.v $vr0, $vr0, $vr1 +; CHECK-NEXT: vclz.h $vr0, $vr0 +; CHECK-NEXT: vst $vr0, $a1, 0 +; CHECK-NEXT: ret + %v = load <8 x i16>, ptr %src + %neg = xor <8 x i16> %v, <i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1> + %res = call <8 x i16> @llvm.ctlz.v8i16(<8 x i16> %neg, i1 false) + store <8 x i16> %res, ptr %dst + ret void +} + +define void @not_ctlz_v4i32(ptr %src, ptr %dst) nounwind { +; CHECK-LABEL: not_ctlz_v4i32: +; CHECK: # %bb.0: +; CHECK-NEXT: vld $vr0, $a0, 0 +; CHECK-NEXT: vrepli.b $vr1, -1 +; CHECK-NEXT: vxor.v $vr0, $vr0, $vr1 +; CHECK-NEXT: vclz.w $vr0, $vr0 +; CHECK-NEXT: vst $vr0, $a1, 0 +; CHECK-NEXT: ret + %v = load <4 x i32>, ptr %src + %neg = xor <4 x i32> %v, <i32 -1, i32 -1, i32 -1, i32 -1> + %res = call <4 x i32> @llvm.ctlz.v4i32(<4 x i32> %neg, i1 false) + store <4 x i32> %res, ptr %dst + ret void +} + +define void @not_ctlz_v2i64(ptr %src, ptr %dst) nounwind { +; CHECK-LABEL: not_ctlz_v2i64: +; CHECK: # %bb.0: +; CHECK-NEXT: vld $vr0, $a0, 0 +; CHECK-NEXT: vrepli.b $vr1, -1 +; CHECK-NEXT: vxor.v $vr0, $vr0, $vr1 +; CHECK-NEXT: vclz.d $vr0, $vr0 +; CHECK-NEXT: vst $vr0, $a1, 0 +; CHECK-NEXT: ret + %v = load <2 x i64>, ptr %src + %neg = xor <2 x i64> %v, <i64 -1, i64 -1> + %res = call <2 x i64> @llvm.ctlz.v2i64(<2 x i64> %neg, i1 false) + store <2 x i64> %res, ptr %dst + ret void +} + declare <16 x i8> @llvm.ctpop.v16i8(<16 x i8>) declare <8 x i16> @llvm.ctpop.v8i16(<8 x i16>) declare <4 x i32> @llvm.ctpop.v4i32(<4 x i32>) diff --git a/llvm/test/CodeGen/LoongArch/sink-fold-addi.ll b/llvm/test/CodeGen/LoongArch/sink-fold-addi.ll new file mode 100644 index 0000000..9a806a1 --- /dev/null +++ b/llvm/test/CodeGen/LoongArch/sink-fold-addi.ll @@ -0,0 +1,758 @@ +; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6 +; RUN: llc --mtriple=loongarch32 --mattr=+32s,+lasx --verify-machineinstrs < %s \ +; RUN: | FileCheck --check-prefix=LA32 %s +; RUN: llc --mtriple=loongarch64 --mattr=+lasx --verify-machineinstrs < %s \ +; RUN: | FileCheck --check-prefix=LA64 %s + +%struct.S = type { i64, i64, i8 } +%struct.F = type { float, double, float } +%struct.V = type { <4 x i32>, <4 x i32>, <16 x i16> } + +define void @sink_fold_i64(i64 %k, i64 %n, ptr %a) nounwind { +; LA32-LABEL: sink_fold_i64: +; LA32: # %bb.0: # %entry +; LA32-NEXT: addi.w $sp, $sp, -48 +; LA32-NEXT: st.w $ra, $sp, 44 # 4-byte Folded Spill +; LA32-NEXT: st.w $fp, $sp, 40 # 4-byte Folded Spill +; LA32-NEXT: st.w $s0, $sp, 36 # 4-byte Folded Spill +; LA32-NEXT: st.w $s1, $sp, 32 # 4-byte Folded Spill +; LA32-NEXT: st.w $s2, $sp, 28 # 4-byte Folded Spill +; LA32-NEXT: st.w $s3, $sp, 24 # 4-byte Folded Spill +; LA32-NEXT: st.w $s4, $sp, 20 # 4-byte Folded Spill +; LA32-NEXT: st.w $s5, $sp, 16 # 4-byte Folded Spill +; LA32-NEXT: st.w $s6, $sp, 12 # 4-byte Folded Spill +; LA32-NEXT: move $s0, $a3 +; LA32-NEXT: move $s1, $a2 +; LA32-NEXT: slli.w $a1, $a0, 4 +; LA32-NEXT: alsl.w $a0, $a0, $a1, 3 +; LA32-NEXT: add.w $a0, $a4, $a0 +; LA32-NEXT: sltui $a1, $a3, 1 +; LA32-NEXT: slti $a2, $a3, 0 +; LA32-NEXT: masknez $a2, $a2, $a1 +; LA32-NEXT: sltui $a3, $s1, 1 +; LA32-NEXT: maskeqz $a1, $a3, $a1 +; LA32-NEXT: or $a1, $a1, $a2 +; LA32-NEXT: addi.w $s2, $a0, 8 +; LA32-NEXT: bnez $a1, .LBB0_3 +; LA32-NEXT: # %bb.1: # %for.body.preheader +; LA32-NEXT: move $fp, $a4 +; LA32-NEXT: move $s4, $zero +; LA32-NEXT: move $s5, $zero +; LA32-NEXT: move $s3, $zero +; LA32-NEXT: move $s6, $zero +; LA32-NEXT: .p2align 4, , 16 +; LA32-NEXT: .LBB0_2: # %for.body +; LA32-NEXT: # =>This Inner Loop Header: Depth=1 +; LA32-NEXT: move $a0, $fp +; LA32-NEXT: bl f +; LA32-NEXT: ld.w $a0, $s2, 4 +; LA32-NEXT: ld.w $a1, $s2, 0 +; LA32-NEXT: add.w $a0, $a0, $s6 +; LA32-NEXT: add.w $s3, $a1, $s3 +; LA32-NEXT: sltu $a1, $s3, $a1 +; LA32-NEXT: addi.w $s4, $s4, 1 +; LA32-NEXT: sltui $a2, $s4, 1 +; LA32-NEXT: add.w $s5, $s5, $a2 +; LA32-NEXT: xor $a2, $s4, $s1 +; LA32-NEXT: xor $a3, $s5, $s0 +; LA32-NEXT: or $a2, $a2, $a3 +; LA32-NEXT: add.w $s6, $a0, $a1 +; LA32-NEXT: bnez $a2, .LBB0_2 +; LA32-NEXT: b .LBB0_4 +; LA32-NEXT: .LBB0_3: +; LA32-NEXT: move $s3, $zero +; LA32-NEXT: move $s6, $zero +; LA32-NEXT: .LBB0_4: # %for.cond.cleanup +; LA32-NEXT: st.w $s3, $s2, 0 +; LA32-NEXT: st.w $s6, $s2, 4 +; LA32-NEXT: ld.w $s6, $sp, 12 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s5, $sp, 16 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s4, $sp, 20 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s3, $sp, 24 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s2, $sp, 28 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s1, $sp, 32 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s0, $sp, 36 # 4-byte Folded Reload +; LA32-NEXT: ld.w $fp, $sp, 40 # 4-byte Folded Reload +; LA32-NEXT: ld.w $ra, $sp, 44 # 4-byte Folded Reload +; LA32-NEXT: addi.w $sp, $sp, 48 +; LA32-NEXT: ret +; +; LA64-LABEL: sink_fold_i64: +; LA64: # %bb.0: # %entry +; LA64-NEXT: addi.d $sp, $sp, -48 +; LA64-NEXT: st.d $ra, $sp, 40 # 8-byte Folded Spill +; LA64-NEXT: st.d $fp, $sp, 32 # 8-byte Folded Spill +; LA64-NEXT: st.d $s0, $sp, 24 # 8-byte Folded Spill +; LA64-NEXT: st.d $s1, $sp, 16 # 8-byte Folded Spill +; LA64-NEXT: st.d $s2, $sp, 8 # 8-byte Folded Spill +; LA64-NEXT: move $s0, $a1 +; LA64-NEXT: slli.d $a1, $a0, 4 +; LA64-NEXT: alsl.d $a0, $a0, $a1, 3 +; LA64-NEXT: add.d $a0, $a2, $a0 +; LA64-NEXT: addi.d $s1, $a0, 8 +; LA64-NEXT: blez $s0, .LBB0_3 +; LA64-NEXT: # %bb.1: # %for.body.preheader +; LA64-NEXT: move $fp, $a2 +; LA64-NEXT: move $s2, $zero +; LA64-NEXT: .p2align 4, , 16 +; LA64-NEXT: .LBB0_2: # %for.body +; LA64-NEXT: # =>This Inner Loop Header: Depth=1 +; LA64-NEXT: move $a0, $fp +; LA64-NEXT: pcaddu18i $ra, %call36(f) +; LA64-NEXT: jirl $ra, $ra, 0 +; LA64-NEXT: ld.d $a0, $s1, 0 +; LA64-NEXT: addi.d $s0, $s0, -1 +; LA64-NEXT: add.d $s2, $a0, $s2 +; LA64-NEXT: bnez $s0, .LBB0_2 +; LA64-NEXT: b .LBB0_4 +; LA64-NEXT: .LBB0_3: +; LA64-NEXT: move $s2, $zero +; LA64-NEXT: .LBB0_4: # %for.cond.cleanup +; LA64-NEXT: st.d $s2, $s1, 0 +; LA64-NEXT: ld.d $s2, $sp, 8 # 8-byte Folded Reload +; LA64-NEXT: ld.d $s1, $sp, 16 # 8-byte Folded Reload +; LA64-NEXT: ld.d $s0, $sp, 24 # 8-byte Folded Reload +; LA64-NEXT: ld.d $fp, $sp, 32 # 8-byte Folded Reload +; LA64-NEXT: ld.d $ra, $sp, 40 # 8-byte Folded Reload +; LA64-NEXT: addi.d $sp, $sp, 48 +; LA64-NEXT: ret +entry: + %y = getelementptr inbounds %struct.S, ptr %a, i64 %k, i32 1 + %cmp4 = icmp sgt i64 %n, 0 + br i1 %cmp4, label %for.body, label %for.cond.cleanup + +for.body: ; preds = %entry, %for.body + %i.06 = phi i64 [ 0, %entry ], [ %inc, %for.body ] + %s.05 = phi i64 [ 0, %entry ], [ %add, %for.body ] + call void @f(ptr %a) + %0 = load i64, ptr %y + %add = add nsw i64 %0, %s.05 + %inc = add nuw nsw i64 %i.06, 1 + %exitcond.not = icmp eq i64 %inc, %n + br i1 %exitcond.not, label %for.cond.cleanup, label %for.body + +for.cond.cleanup: ; preds = %for.body, %entry + %s.0.lcssa = phi i64 [ 0, %entry ], [ %add, %for.body ] + store i64 %s.0.lcssa, ptr %y + ret void +} + +define void @sink_fold_f32(i64 %k, i64 %n, ptr %a) nounwind { +; LA32-LABEL: sink_fold_f32: +; LA32: # %bb.0: # %entry +; LA32-NEXT: addi.w $sp, $sp, -48 +; LA32-NEXT: st.w $ra, $sp, 44 # 4-byte Folded Spill +; LA32-NEXT: st.w $fp, $sp, 40 # 4-byte Folded Spill +; LA32-NEXT: st.w $s0, $sp, 36 # 4-byte Folded Spill +; LA32-NEXT: st.w $s1, $sp, 32 # 4-byte Folded Spill +; LA32-NEXT: st.w $s2, $sp, 28 # 4-byte Folded Spill +; LA32-NEXT: st.w $s3, $sp, 24 # 4-byte Folded Spill +; LA32-NEXT: st.w $s4, $sp, 20 # 4-byte Folded Spill +; LA32-NEXT: fst.d $fs0, $sp, 8 # 8-byte Folded Spill +; LA32-NEXT: move $s0, $a3 +; LA32-NEXT: move $s1, $a2 +; LA32-NEXT: slli.w $a1, $a0, 4 +; LA32-NEXT: alsl.w $a0, $a0, $a1, 3 +; LA32-NEXT: add.w $a0, $a4, $a0 +; LA32-NEXT: sltui $a1, $a3, 1 +; LA32-NEXT: slti $a2, $a3, 0 +; LA32-NEXT: masknez $a2, $a2, $a1 +; LA32-NEXT: sltui $a3, $s1, 1 +; LA32-NEXT: maskeqz $a1, $a3, $a1 +; LA32-NEXT: or $a1, $a1, $a2 +; LA32-NEXT: addi.w $s2, $a0, 16 +; LA32-NEXT: bnez $a1, .LBB1_3 +; LA32-NEXT: # %bb.1: # %for.body.preheader +; LA32-NEXT: move $fp, $a4 +; LA32-NEXT: move $s3, $zero +; LA32-NEXT: move $s4, $zero +; LA32-NEXT: movgr2fr.w $fs0, $zero +; LA32-NEXT: .p2align 4, , 16 +; LA32-NEXT: .LBB1_2: # %for.body +; LA32-NEXT: # =>This Inner Loop Header: Depth=1 +; LA32-NEXT: move $a0, $fp +; LA32-NEXT: bl f +; LA32-NEXT: fld.s $fa0, $s2, 0 +; LA32-NEXT: addi.w $s3, $s3, 1 +; LA32-NEXT: sltui $a0, $s3, 1 +; LA32-NEXT: add.w $s4, $s4, $a0 +; LA32-NEXT: xor $a0, $s3, $s1 +; LA32-NEXT: xor $a1, $s4, $s0 +; LA32-NEXT: or $a0, $a0, $a1 +; LA32-NEXT: fadd.s $fs0, $fa0, $fs0 +; LA32-NEXT: bnez $a0, .LBB1_2 +; LA32-NEXT: b .LBB1_4 +; LA32-NEXT: .LBB1_3: +; LA32-NEXT: movgr2fr.w $fs0, $zero +; LA32-NEXT: .LBB1_4: # %for.cond.cleanup +; LA32-NEXT: fst.s $fs0, $s2, 0 +; LA32-NEXT: fld.d $fs0, $sp, 8 # 8-byte Folded Reload +; LA32-NEXT: ld.w $s4, $sp, 20 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s3, $sp, 24 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s2, $sp, 28 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s1, $sp, 32 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s0, $sp, 36 # 4-byte Folded Reload +; LA32-NEXT: ld.w $fp, $sp, 40 # 4-byte Folded Reload +; LA32-NEXT: ld.w $ra, $sp, 44 # 4-byte Folded Reload +; LA32-NEXT: addi.w $sp, $sp, 48 +; LA32-NEXT: ret +; +; LA64-LABEL: sink_fold_f32: +; LA64: # %bb.0: # %entry +; LA64-NEXT: addi.d $sp, $sp, -48 +; LA64-NEXT: st.d $ra, $sp, 40 # 8-byte Folded Spill +; LA64-NEXT: st.d $fp, $sp, 32 # 8-byte Folded Spill +; LA64-NEXT: st.d $s0, $sp, 24 # 8-byte Folded Spill +; LA64-NEXT: st.d $s1, $sp, 16 # 8-byte Folded Spill +; LA64-NEXT: fst.d $fs0, $sp, 8 # 8-byte Folded Spill +; LA64-NEXT: move $s0, $a1 +; LA64-NEXT: slli.d $a1, $a0, 4 +; LA64-NEXT: alsl.d $a0, $a0, $a1, 3 +; LA64-NEXT: add.d $a0, $a2, $a0 +; LA64-NEXT: addi.d $s1, $a0, 16 +; LA64-NEXT: blez $s0, .LBB1_3 +; LA64-NEXT: # %bb.1: # %for.body.preheader +; LA64-NEXT: move $fp, $a2 +; LA64-NEXT: movgr2fr.w $fs0, $zero +; LA64-NEXT: .p2align 4, , 16 +; LA64-NEXT: .LBB1_2: # %for.body +; LA64-NEXT: # =>This Inner Loop Header: Depth=1 +; LA64-NEXT: move $a0, $fp +; LA64-NEXT: pcaddu18i $ra, %call36(f) +; LA64-NEXT: jirl $ra, $ra, 0 +; LA64-NEXT: fld.s $fa0, $s1, 0 +; LA64-NEXT: addi.d $s0, $s0, -1 +; LA64-NEXT: fadd.s $fs0, $fa0, $fs0 +; LA64-NEXT: bnez $s0, .LBB1_2 +; LA64-NEXT: b .LBB1_4 +; LA64-NEXT: .LBB1_3: +; LA64-NEXT: movgr2fr.w $fs0, $zero +; LA64-NEXT: .LBB1_4: # %for.cond.cleanup +; LA64-NEXT: fst.s $fs0, $s1, 0 +; LA64-NEXT: fld.d $fs0, $sp, 8 # 8-byte Folded Reload +; LA64-NEXT: ld.d $s1, $sp, 16 # 8-byte Folded Reload +; LA64-NEXT: ld.d $s0, $sp, 24 # 8-byte Folded Reload +; LA64-NEXT: ld.d $fp, $sp, 32 # 8-byte Folded Reload +; LA64-NEXT: ld.d $ra, $sp, 40 # 8-byte Folded Reload +; LA64-NEXT: addi.d $sp, $sp, 48 +; LA64-NEXT: ret +entry: + %y = getelementptr inbounds %struct.F, ptr %a, i64 %k, i32 2 + %cmp4 = icmp sgt i64 %n, 0 + br i1 %cmp4, label %for.body, label %for.cond.cleanup + +for.body: ; preds = %entry, %for.body + %i.06 = phi i64 [ 0, %entry ], [ %inc, %for.body ] + %s.05 = phi float [ 0.0, %entry ], [ %add, %for.body ] + call void @f(ptr %a) + %0 = load float, ptr %y + %add = fadd float %0, %s.05 + %inc = add nuw nsw i64 %i.06, 1 + %exitcond.not = icmp eq i64 %inc, %n + br i1 %exitcond.not, label %for.cond.cleanup, label %for.body + +for.cond.cleanup: ; preds = %for.body, %entry + %s.0.lcssa = phi float [ 0.0, %entry ], [ %add, %for.body ] + store float %s.0.lcssa, ptr %y + ret void +} + +define void @sink_fold_v4i32(i64 %k, i64 %n, ptr %a) nounwind { +; LA32-LABEL: sink_fold_v4i32: +; LA32: # %bb.0: # %entry +; LA32-NEXT: addi.w $sp, $sp, -48 +; LA32-NEXT: st.w $ra, $sp, 44 # 4-byte Folded Spill +; LA32-NEXT: st.w $fp, $sp, 40 # 4-byte Folded Spill +; LA32-NEXT: st.w $s0, $sp, 36 # 4-byte Folded Spill +; LA32-NEXT: st.w $s1, $sp, 32 # 4-byte Folded Spill +; LA32-NEXT: st.w $s2, $sp, 28 # 4-byte Folded Spill +; LA32-NEXT: st.w $s3, $sp, 24 # 4-byte Folded Spill +; LA32-NEXT: st.w $s4, $sp, 20 # 4-byte Folded Spill +; LA32-NEXT: move $s0, $a3 +; LA32-NEXT: move $s1, $a2 +; LA32-NEXT: slli.w $a0, $a0, 6 +; LA32-NEXT: add.w $a0, $a4, $a0 +; LA32-NEXT: sltui $a1, $a3, 1 +; LA32-NEXT: slti $a2, $a3, 0 +; LA32-NEXT: masknez $a2, $a2, $a1 +; LA32-NEXT: sltui $a3, $s1, 1 +; LA32-NEXT: maskeqz $a1, $a3, $a1 +; LA32-NEXT: or $a1, $a1, $a2 +; LA32-NEXT: addi.w $s2, $a0, 16 +; LA32-NEXT: bnez $a1, .LBB2_3 +; LA32-NEXT: # %bb.1: # %for.body.preheader +; LA32-NEXT: move $fp, $a4 +; LA32-NEXT: move $s3, $zero +; LA32-NEXT: move $s4, $zero +; LA32-NEXT: vrepli.b $vr0, 0 +; LA32-NEXT: .p2align 4, , 16 +; LA32-NEXT: .LBB2_2: # %for.body +; LA32-NEXT: # =>This Inner Loop Header: Depth=1 +; LA32-NEXT: vst $vr0, $sp, 0 # 16-byte Folded Spill +; LA32-NEXT: move $a0, $fp +; LA32-NEXT: bl f +; LA32-NEXT: vld $vr0, $s2, 0 +; LA32-NEXT: addi.w $s3, $s3, 1 +; LA32-NEXT: sltui $a0, $s3, 1 +; LA32-NEXT: add.w $s4, $s4, $a0 +; LA32-NEXT: xor $a0, $s3, $s1 +; LA32-NEXT: xor $a1, $s4, $s0 +; LA32-NEXT: or $a0, $a0, $a1 +; LA32-NEXT: vld $vr1, $sp, 0 # 16-byte Folded Reload +; LA32-NEXT: vadd.w $vr1, $vr0, $vr1 +; LA32-NEXT: vst $vr1, $sp, 0 # 16-byte Folded Spill +; LA32-NEXT: vld $vr0, $sp, 0 # 16-byte Folded Reload +; LA32-NEXT: bnez $a0, .LBB2_2 +; LA32-NEXT: b .LBB2_4 +; LA32-NEXT: .LBB2_3: +; LA32-NEXT: vrepli.b $vr0, 0 +; LA32-NEXT: .LBB2_4: # %for.cond.cleanup +; LA32-NEXT: vst $vr0, $s2, 0 +; LA32-NEXT: ld.w $s4, $sp, 20 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s3, $sp, 24 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s2, $sp, 28 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s1, $sp, 32 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s0, $sp, 36 # 4-byte Folded Reload +; LA32-NEXT: ld.w $fp, $sp, 40 # 4-byte Folded Reload +; LA32-NEXT: ld.w $ra, $sp, 44 # 4-byte Folded Reload +; LA32-NEXT: addi.w $sp, $sp, 48 +; LA32-NEXT: ret +; +; LA64-LABEL: sink_fold_v4i32: +; LA64: # %bb.0: # %entry +; LA64-NEXT: addi.d $sp, $sp, -48 +; LA64-NEXT: st.d $ra, $sp, 40 # 8-byte Folded Spill +; LA64-NEXT: st.d $fp, $sp, 32 # 8-byte Folded Spill +; LA64-NEXT: st.d $s0, $sp, 24 # 8-byte Folded Spill +; LA64-NEXT: st.d $s1, $sp, 16 # 8-byte Folded Spill +; LA64-NEXT: slli.d $a0, $a0, 6 +; LA64-NEXT: add.d $a0, $a2, $a0 +; LA64-NEXT: addi.d $s1, $a0, 16 +; LA64-NEXT: blez $a1, .LBB2_3 +; LA64-NEXT: # %bb.1: # %for.body.preheader +; LA64-NEXT: move $fp, $a2 +; LA64-NEXT: move $s0, $a1 +; LA64-NEXT: vrepli.b $vr0, 0 +; LA64-NEXT: .p2align 4, , 16 +; LA64-NEXT: .LBB2_2: # %for.body +; LA64-NEXT: # =>This Inner Loop Header: Depth=1 +; LA64-NEXT: vst $vr0, $sp, 0 # 16-byte Folded Spill +; LA64-NEXT: move $a0, $fp +; LA64-NEXT: pcaddu18i $ra, %call36(f) +; LA64-NEXT: jirl $ra, $ra, 0 +; LA64-NEXT: vld $vr0, $s1, 0 +; LA64-NEXT: addi.d $s0, $s0, -1 +; LA64-NEXT: vld $vr1, $sp, 0 # 16-byte Folded Reload +; LA64-NEXT: vadd.w $vr1, $vr0, $vr1 +; LA64-NEXT: vst $vr1, $sp, 0 # 16-byte Folded Spill +; LA64-NEXT: vld $vr0, $sp, 0 # 16-byte Folded Reload +; LA64-NEXT: bnez $s0, .LBB2_2 +; LA64-NEXT: b .LBB2_4 +; LA64-NEXT: .LBB2_3: +; LA64-NEXT: vrepli.b $vr0, 0 +; LA64-NEXT: .LBB2_4: # %for.cond.cleanup +; LA64-NEXT: vst $vr0, $s1, 0 +; LA64-NEXT: ld.d $s1, $sp, 16 # 8-byte Folded Reload +; LA64-NEXT: ld.d $s0, $sp, 24 # 8-byte Folded Reload +; LA64-NEXT: ld.d $fp, $sp, 32 # 8-byte Folded Reload +; LA64-NEXT: ld.d $ra, $sp, 40 # 8-byte Folded Reload +; LA64-NEXT: addi.d $sp, $sp, 48 +; LA64-NEXT: ret +entry: + %y = getelementptr inbounds %struct.V, ptr %a, i64 %k, i32 1 + %cmp = icmp sgt i64 %n, 0 + br i1 %cmp, label %for.body, label %for.cond.cleanup + +for.body: ; preds = %entry, %for.body + %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ] + %sum.0 = phi <4 x i32> [ zeroinitializer, %entry ], [ %addv, %for.body ] + call void @f(ptr %a) + %v = load <4 x i32>, ptr %y + %addv = add <4 x i32> %v, %sum.0 + %inc = add nuw nsw i64 %i.0, 1 + %exitcond = icmp eq i64 %inc, %n + br i1 %exitcond, label %for.cond.cleanup, label %for.body + +for.cond.cleanup: ; preds = %for.body, %entry + %sum.lcssa = phi <4 x i32> [ zeroinitializer, %entry ], [ %addv, %for.body ] + store <4 x i32> %sum.lcssa, ptr %y + ret void +} + +define void @sink_fold_v16i16(i64 %k, i64 %n, ptr %a) nounwind { +; LA32-LABEL: sink_fold_v16i16: +; LA32: # %bb.0: # %entry +; LA32-NEXT: addi.w $sp, $sp, -80 +; LA32-NEXT: st.w $ra, $sp, 76 # 4-byte Folded Spill +; LA32-NEXT: st.w $fp, $sp, 72 # 4-byte Folded Spill +; LA32-NEXT: st.w $s0, $sp, 68 # 4-byte Folded Spill +; LA32-NEXT: st.w $s1, $sp, 64 # 4-byte Folded Spill +; LA32-NEXT: st.w $s2, $sp, 60 # 4-byte Folded Spill +; LA32-NEXT: st.w $s3, $sp, 56 # 4-byte Folded Spill +; LA32-NEXT: st.w $s4, $sp, 52 # 4-byte Folded Spill +; LA32-NEXT: move $s0, $a3 +; LA32-NEXT: move $s1, $a2 +; LA32-NEXT: slli.w $a0, $a0, 6 +; LA32-NEXT: add.w $a0, $a4, $a0 +; LA32-NEXT: sltui $a1, $a3, 1 +; LA32-NEXT: slti $a2, $a3, 0 +; LA32-NEXT: masknez $a2, $a2, $a1 +; LA32-NEXT: sltui $a3, $s1, 1 +; LA32-NEXT: maskeqz $a1, $a3, $a1 +; LA32-NEXT: or $a1, $a1, $a2 +; LA32-NEXT: addi.w $s2, $a0, 32 +; LA32-NEXT: bnez $a1, .LBB3_3 +; LA32-NEXT: # %bb.1: # %for.body.preheader +; LA32-NEXT: move $fp, $a4 +; LA32-NEXT: move $s3, $zero +; LA32-NEXT: move $s4, $zero +; LA32-NEXT: xvrepli.b $xr0, 0 +; LA32-NEXT: .p2align 4, , 16 +; LA32-NEXT: .LBB3_2: # %for.body +; LA32-NEXT: # =>This Inner Loop Header: Depth=1 +; LA32-NEXT: xvst $xr0, $sp, 16 # 32-byte Folded Spill +; LA32-NEXT: move $a0, $fp +; LA32-NEXT: bl f +; LA32-NEXT: xvld $xr0, $s2, 0 +; LA32-NEXT: addi.w $s3, $s3, 1 +; LA32-NEXT: sltui $a0, $s3, 1 +; LA32-NEXT: add.w $s4, $s4, $a0 +; LA32-NEXT: xor $a0, $s3, $s1 +; LA32-NEXT: xor $a1, $s4, $s0 +; LA32-NEXT: or $a0, $a0, $a1 +; LA32-NEXT: xvld $xr1, $sp, 16 # 32-byte Folded Reload +; LA32-NEXT: xvadd.h $xr1, $xr0, $xr1 +; LA32-NEXT: xvst $xr1, $sp, 16 # 32-byte Folded Spill +; LA32-NEXT: xvld $xr0, $sp, 16 # 32-byte Folded Reload +; LA32-NEXT: bnez $a0, .LBB3_2 +; LA32-NEXT: b .LBB3_4 +; LA32-NEXT: .LBB3_3: +; LA32-NEXT: xvrepli.b $xr0, 0 +; LA32-NEXT: .LBB3_4: # %for.cond.cleanup +; LA32-NEXT: xvst $xr0, $s2, 0 +; LA32-NEXT: ld.w $s4, $sp, 52 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s3, $sp, 56 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s2, $sp, 60 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s1, $sp, 64 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s0, $sp, 68 # 4-byte Folded Reload +; LA32-NEXT: ld.w $fp, $sp, 72 # 4-byte Folded Reload +; LA32-NEXT: ld.w $ra, $sp, 76 # 4-byte Folded Reload +; LA32-NEXT: addi.w $sp, $sp, 80 +; LA32-NEXT: ret +; +; LA64-LABEL: sink_fold_v16i16: +; LA64: # %bb.0: # %entry +; LA64-NEXT: addi.d $sp, $sp, -80 +; LA64-NEXT: st.d $ra, $sp, 72 # 8-byte Folded Spill +; LA64-NEXT: st.d $fp, $sp, 64 # 8-byte Folded Spill +; LA64-NEXT: st.d $s0, $sp, 56 # 8-byte Folded Spill +; LA64-NEXT: st.d $s1, $sp, 48 # 8-byte Folded Spill +; LA64-NEXT: slli.d $a0, $a0, 6 +; LA64-NEXT: add.d $a0, $a2, $a0 +; LA64-NEXT: addi.d $s1, $a0, 32 +; LA64-NEXT: blez $a1, .LBB3_3 +; LA64-NEXT: # %bb.1: # %for.body.preheader +; LA64-NEXT: move $fp, $a2 +; LA64-NEXT: move $s0, $a1 +; LA64-NEXT: xvrepli.b $xr0, 0 +; LA64-NEXT: .p2align 4, , 16 +; LA64-NEXT: .LBB3_2: # %for.body +; LA64-NEXT: # =>This Inner Loop Header: Depth=1 +; LA64-NEXT: xvst $xr0, $sp, 16 # 32-byte Folded Spill +; LA64-NEXT: move $a0, $fp +; LA64-NEXT: pcaddu18i $ra, %call36(f) +; LA64-NEXT: jirl $ra, $ra, 0 +; LA64-NEXT: xvld $xr0, $s1, 0 +; LA64-NEXT: addi.d $s0, $s0, -1 +; LA64-NEXT: xvld $xr1, $sp, 16 # 32-byte Folded Reload +; LA64-NEXT: xvadd.h $xr1, $xr0, $xr1 +; LA64-NEXT: xvst $xr1, $sp, 16 # 32-byte Folded Spill +; LA64-NEXT: xvld $xr0, $sp, 16 # 32-byte Folded Reload +; LA64-NEXT: bnez $s0, .LBB3_2 +; LA64-NEXT: b .LBB3_4 +; LA64-NEXT: .LBB3_3: +; LA64-NEXT: xvrepli.b $xr0, 0 +; LA64-NEXT: .LBB3_4: # %for.cond.cleanup +; LA64-NEXT: xvst $xr0, $s1, 0 +; LA64-NEXT: ld.d $s1, $sp, 48 # 8-byte Folded Reload +; LA64-NEXT: ld.d $s0, $sp, 56 # 8-byte Folded Reload +; LA64-NEXT: ld.d $fp, $sp, 64 # 8-byte Folded Reload +; LA64-NEXT: ld.d $ra, $sp, 72 # 8-byte Folded Reload +; LA64-NEXT: addi.d $sp, $sp, 80 +; LA64-NEXT: ret +entry: + %y = getelementptr inbounds %struct.V, ptr %a, i64 %k, i32 2 + %cmp = icmp sgt i64 %n, 0 + br i1 %cmp, label %for.body, label %for.cond.cleanup + +for.body: ; preds = %entry, %for.body + %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ] + %sum.0 = phi <16 x i16> [ zeroinitializer, %entry ], [ %addv, %for.body ] + call void @f(ptr %a) + %v = load <16 x i16>, ptr %y + %addv = add <16 x i16> %v, %sum.0 + %inc = add nuw nsw i64 %i.0, 1 + %exitcond = icmp eq i64 %inc, %n + br i1 %exitcond, label %for.cond.cleanup, label %for.body + +for.cond.cleanup: ; preds = %for.body, %entry + %sum.lcssa = phi <16 x i16> [ zeroinitializer, %entry ], [ %addv, %for.body ] + store <16 x i16> %sum.lcssa, ptr %y + ret void +} + +define void @sink_fold_extracti8(i64 %k, i64 %n, ptr %a) nounwind { +; LA32-LABEL: sink_fold_extracti8: +; LA32: # %bb.0: # %entry +; LA32-NEXT: addi.w $sp, $sp, -48 +; LA32-NEXT: st.w $ra, $sp, 44 # 4-byte Folded Spill +; LA32-NEXT: st.w $fp, $sp, 40 # 4-byte Folded Spill +; LA32-NEXT: st.w $s0, $sp, 36 # 4-byte Folded Spill +; LA32-NEXT: st.w $s1, $sp, 32 # 4-byte Folded Spill +; LA32-NEXT: st.w $s2, $sp, 28 # 4-byte Folded Spill +; LA32-NEXT: st.w $s3, $sp, 24 # 4-byte Folded Spill +; LA32-NEXT: st.w $s4, $sp, 20 # 4-byte Folded Spill +; LA32-NEXT: move $s0, $a3 +; LA32-NEXT: move $s1, $a2 +; LA32-NEXT: slli.w $a1, $a0, 4 +; LA32-NEXT: alsl.w $a0, $a0, $a1, 3 +; LA32-NEXT: add.w $a0, $a4, $a0 +; LA32-NEXT: sltui $a1, $a3, 1 +; LA32-NEXT: slti $a2, $a3, 0 +; LA32-NEXT: masknez $a2, $a2, $a1 +; LA32-NEXT: sltui $a3, $s1, 1 +; LA32-NEXT: maskeqz $a1, $a3, $a1 +; LA32-NEXT: or $a1, $a1, $a2 +; LA32-NEXT: addi.w $s2, $a0, 16 +; LA32-NEXT: bnez $a1, .LBB4_3 +; LA32-NEXT: # %bb.1: # %for.body.preheader +; LA32-NEXT: move $fp, $a4 +; LA32-NEXT: move $s3, $zero +; LA32-NEXT: move $s4, $zero +; LA32-NEXT: vrepli.b $vr0, 0 +; LA32-NEXT: .p2align 4, , 16 +; LA32-NEXT: .LBB4_2: # %for.body +; LA32-NEXT: # =>This Inner Loop Header: Depth=1 +; LA32-NEXT: vst $vr0, $sp, 0 # 16-byte Folded Spill +; LA32-NEXT: move $a0, $fp +; LA32-NEXT: bl f +; LA32-NEXT: vldrepl.b $vr0, $s2, 0 +; LA32-NEXT: addi.w $s3, $s3, 1 +; LA32-NEXT: sltui $a0, $s3, 1 +; LA32-NEXT: add.w $s4, $s4, $a0 +; LA32-NEXT: xor $a0, $s3, $s1 +; LA32-NEXT: xor $a1, $s4, $s0 +; LA32-NEXT: or $a0, $a0, $a1 +; LA32-NEXT: vld $vr1, $sp, 0 # 16-byte Folded Reload +; LA32-NEXT: vadd.b $vr1, $vr0, $vr1 +; LA32-NEXT: vst $vr1, $sp, 0 # 16-byte Folded Spill +; LA32-NEXT: vld $vr0, $sp, 0 # 16-byte Folded Reload +; LA32-NEXT: bnez $a0, .LBB4_2 +; LA32-NEXT: b .LBB4_4 +; LA32-NEXT: .LBB4_3: +; LA32-NEXT: vrepli.b $vr0, 0 +; LA32-NEXT: .LBB4_4: # %for.cond.cleanup +; LA32-NEXT: vstelm.b $vr0, $s2, 0, 1 +; LA32-NEXT: ld.w $s4, $sp, 20 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s3, $sp, 24 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s2, $sp, 28 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s1, $sp, 32 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s0, $sp, 36 # 4-byte Folded Reload +; LA32-NEXT: ld.w $fp, $sp, 40 # 4-byte Folded Reload +; LA32-NEXT: ld.w $ra, $sp, 44 # 4-byte Folded Reload +; LA32-NEXT: addi.w $sp, $sp, 48 +; LA32-NEXT: ret +; +; LA64-LABEL: sink_fold_extracti8: +; LA64: # %bb.0: # %entry +; LA64-NEXT: addi.d $sp, $sp, -48 +; LA64-NEXT: st.d $ra, $sp, 40 # 8-byte Folded Spill +; LA64-NEXT: st.d $fp, $sp, 32 # 8-byte Folded Spill +; LA64-NEXT: st.d $s0, $sp, 24 # 8-byte Folded Spill +; LA64-NEXT: st.d $s1, $sp, 16 # 8-byte Folded Spill +; LA64-NEXT: move $s0, $a1 +; LA64-NEXT: slli.d $a1, $a0, 4 +; LA64-NEXT: alsl.d $a0, $a0, $a1, 3 +; LA64-NEXT: add.d $a0, $a2, $a0 +; LA64-NEXT: addi.d $s1, $a0, 16 +; LA64-NEXT: blez $s0, .LBB4_3 +; LA64-NEXT: # %bb.1: # %for.body.preheader +; LA64-NEXT: move $fp, $a2 +; LA64-NEXT: vrepli.b $vr0, 0 +; LA64-NEXT: .p2align 4, , 16 +; LA64-NEXT: .LBB4_2: # %for.body +; LA64-NEXT: # =>This Inner Loop Header: Depth=1 +; LA64-NEXT: vst $vr0, $sp, 0 # 16-byte Folded Spill +; LA64-NEXT: move $a0, $fp +; LA64-NEXT: pcaddu18i $ra, %call36(f) +; LA64-NEXT: jirl $ra, $ra, 0 +; LA64-NEXT: vldrepl.b $vr0, $s1, 0 +; LA64-NEXT: addi.d $s0, $s0, -1 +; LA64-NEXT: vld $vr1, $sp, 0 # 16-byte Folded Reload +; LA64-NEXT: vadd.b $vr1, $vr0, $vr1 +; LA64-NEXT: vst $vr1, $sp, 0 # 16-byte Folded Spill +; LA64-NEXT: vld $vr0, $sp, 0 # 16-byte Folded Reload +; LA64-NEXT: bnez $s0, .LBB4_2 +; LA64-NEXT: b .LBB4_4 +; LA64-NEXT: .LBB4_3: +; LA64-NEXT: vrepli.b $vr0, 0 +; LA64-NEXT: .LBB4_4: # %for.cond.cleanup +; LA64-NEXT: vstelm.b $vr0, $s1, 0, 1 +; LA64-NEXT: ld.d $s1, $sp, 16 # 8-byte Folded Reload +; LA64-NEXT: ld.d $s0, $sp, 24 # 8-byte Folded Reload +; LA64-NEXT: ld.d $fp, $sp, 32 # 8-byte Folded Reload +; LA64-NEXT: ld.d $ra, $sp, 40 # 8-byte Folded Reload +; LA64-NEXT: addi.d $sp, $sp, 48 +; LA64-NEXT: ret +entry: + %y = getelementptr inbounds %struct.S, ptr %a, i64 %k, i32 2 + %cmp = icmp sgt i64 %n, 0 + br i1 %cmp, label %for.body, label %for.cond.cleanup + +for.body: ; preds = %entry, %for.body + %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ] + %sum.0 = phi <16 x i8> [ zeroinitializer, %entry ], [ %addv, %for.body ] + call void @f(ptr %a) + %e = load i8, ptr %y + %ins0 = insertelement <16 x i8> poison, i8 %e, i32 0 + %v = shufflevector <16 x i8> %ins0, <16 x i8> poison, <16 x i32> zeroinitializer + %addv = add <16 x i8> %v, %sum.0 + %inc = add nuw nsw i64 %i.0, 1 + %exitcond = icmp eq i64 %inc, %n + br i1 %exitcond, label %for.cond.cleanup, label %for.body + +for.cond.cleanup: ; preds = %for.body, %entry + %sum.lcssa = phi <16 x i8> [ zeroinitializer, %entry ], [ %addv, %for.body ] + %res = extractelement <16 x i8> %sum.lcssa, i32 1 + store i8 %res, ptr %y + ret void +} + +define void @sink_fold_extractf64(i64 %k, i64 %n, ptr %a) nounwind { +; LA32-LABEL: sink_fold_extractf64: +; LA32: # %bb.0: # %entry +; LA32-NEXT: addi.w $sp, $sp, -80 +; LA32-NEXT: st.w $ra, $sp, 76 # 4-byte Folded Spill +; LA32-NEXT: st.w $fp, $sp, 72 # 4-byte Folded Spill +; LA32-NEXT: st.w $s0, $sp, 68 # 4-byte Folded Spill +; LA32-NEXT: st.w $s1, $sp, 64 # 4-byte Folded Spill +; LA32-NEXT: st.w $s2, $sp, 60 # 4-byte Folded Spill +; LA32-NEXT: st.w $s3, $sp, 56 # 4-byte Folded Spill +; LA32-NEXT: st.w $s4, $sp, 52 # 4-byte Folded Spill +; LA32-NEXT: move $s0, $a3 +; LA32-NEXT: move $s1, $a2 +; LA32-NEXT: slli.w $a1, $a0, 4 +; LA32-NEXT: alsl.w $a0, $a0, $a1, 3 +; LA32-NEXT: add.w $a0, $a4, $a0 +; LA32-NEXT: sltui $a1, $a3, 1 +; LA32-NEXT: slti $a2, $a3, 0 +; LA32-NEXT: masknez $a2, $a2, $a1 +; LA32-NEXT: sltui $a3, $s1, 1 +; LA32-NEXT: maskeqz $a1, $a3, $a1 +; LA32-NEXT: or $a1, $a1, $a2 +; LA32-NEXT: addi.w $s2, $a0, 8 +; LA32-NEXT: bnez $a1, .LBB5_3 +; LA32-NEXT: # %bb.1: # %for.body.preheader +; LA32-NEXT: move $fp, $a4 +; LA32-NEXT: move $s3, $zero +; LA32-NEXT: move $s4, $zero +; LA32-NEXT: xvrepli.b $xr0, 0 +; LA32-NEXT: .p2align 4, , 16 +; LA32-NEXT: .LBB5_2: # %for.body +; LA32-NEXT: # =>This Inner Loop Header: Depth=1 +; LA32-NEXT: xvst $xr0, $sp, 16 # 32-byte Folded Spill +; LA32-NEXT: move $a0, $fp +; LA32-NEXT: bl f +; LA32-NEXT: xvldrepl.d $xr0, $s2, 0 +; LA32-NEXT: addi.w $s3, $s3, 1 +; LA32-NEXT: sltui $a0, $s3, 1 +; LA32-NEXT: add.w $s4, $s4, $a0 +; LA32-NEXT: xor $a0, $s3, $s1 +; LA32-NEXT: xor $a1, $s4, $s0 +; LA32-NEXT: or $a0, $a0, $a1 +; LA32-NEXT: xvld $xr1, $sp, 16 # 32-byte Folded Reload +; LA32-NEXT: xvfadd.d $xr1, $xr0, $xr1 +; LA32-NEXT: xvst $xr1, $sp, 16 # 32-byte Folded Spill +; LA32-NEXT: xvld $xr0, $sp, 16 # 32-byte Folded Reload +; LA32-NEXT: bnez $a0, .LBB5_2 +; LA32-NEXT: b .LBB5_4 +; LA32-NEXT: .LBB5_3: +; LA32-NEXT: xvrepli.b $xr0, 0 +; LA32-NEXT: .LBB5_4: # %for.cond.cleanup +; LA32-NEXT: xvstelm.d $xr0, $s2, 0, 1 +; LA32-NEXT: ld.w $s4, $sp, 52 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s3, $sp, 56 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s2, $sp, 60 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s1, $sp, 64 # 4-byte Folded Reload +; LA32-NEXT: ld.w $s0, $sp, 68 # 4-byte Folded Reload +; LA32-NEXT: ld.w $fp, $sp, 72 # 4-byte Folded Reload +; LA32-NEXT: ld.w $ra, $sp, 76 # 4-byte Folded Reload +; LA32-NEXT: addi.w $sp, $sp, 80 +; LA32-NEXT: ret +; +; LA64-LABEL: sink_fold_extractf64: +; LA64: # %bb.0: # %entry +; LA64-NEXT: addi.d $sp, $sp, -80 +; LA64-NEXT: st.d $ra, $sp, 72 # 8-byte Folded Spill +; LA64-NEXT: st.d $fp, $sp, 64 # 8-byte Folded Spill +; LA64-NEXT: st.d $s0, $sp, 56 # 8-byte Folded Spill +; LA64-NEXT: st.d $s1, $sp, 48 # 8-byte Folded Spill +; LA64-NEXT: move $s0, $a1 +; LA64-NEXT: slli.d $a1, $a0, 4 +; LA64-NEXT: alsl.d $a0, $a0, $a1, 3 +; LA64-NEXT: add.d $a0, $a2, $a0 +; LA64-NEXT: addi.d $s1, $a0, 8 +; LA64-NEXT: blez $s0, .LBB5_3 +; LA64-NEXT: # %bb.1: # %for.body.preheader +; LA64-NEXT: move $fp, $a2 +; LA64-NEXT: xvrepli.b $xr0, 0 +; LA64-NEXT: .p2align 4, , 16 +; LA64-NEXT: .LBB5_2: # %for.body +; LA64-NEXT: # =>This Inner Loop Header: Depth=1 +; LA64-NEXT: xvst $xr0, $sp, 16 # 32-byte Folded Spill +; LA64-NEXT: move $a0, $fp +; LA64-NEXT: pcaddu18i $ra, %call36(f) +; LA64-NEXT: jirl $ra, $ra, 0 +; LA64-NEXT: xvldrepl.d $xr0, $s1, 0 +; LA64-NEXT: addi.d $s0, $s0, -1 +; LA64-NEXT: xvld $xr1, $sp, 16 # 32-byte Folded Reload +; LA64-NEXT: xvfadd.d $xr1, $xr0, $xr1 +; LA64-NEXT: xvst $xr1, $sp, 16 # 32-byte Folded Spill +; LA64-NEXT: xvld $xr0, $sp, 16 # 32-byte Folded Reload +; LA64-NEXT: bnez $s0, .LBB5_2 +; LA64-NEXT: b .LBB5_4 +; LA64-NEXT: .LBB5_3: +; LA64-NEXT: xvrepli.b $xr0, 0 +; LA64-NEXT: .LBB5_4: # %for.cond.cleanup +; LA64-NEXT: xvstelm.d $xr0, $s1, 0, 1 +; LA64-NEXT: ld.d $s1, $sp, 48 # 8-byte Folded Reload +; LA64-NEXT: ld.d $s0, $sp, 56 # 8-byte Folded Reload +; LA64-NEXT: ld.d $fp, $sp, 64 # 8-byte Folded Reload +; LA64-NEXT: ld.d $ra, $sp, 72 # 8-byte Folded Reload +; LA64-NEXT: addi.d $sp, $sp, 80 +; LA64-NEXT: ret +entry: + %y = getelementptr inbounds %struct.F, ptr %a, i64 %k, i32 1 + %cmp = icmp sgt i64 %n, 0 + br i1 %cmp, label %for.body, label %for.cond.cleanup + +for.body: ; preds = %entry, %for.body + %i.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ] + %sum.0 = phi <4 x double> [ zeroinitializer, %entry ], [ %addv, %for.body ] + call void @f(ptr %a) + %e = load double, ptr %y + %ins0 = insertelement <4 x double> poison, double %e, i32 0 + %v = shufflevector <4 x double> %ins0, <4 x double> poison, <4 x i32> zeroinitializer + %addv = fadd <4 x double> %v, %sum.0 + %inc = add nuw nsw i64 %i.0, 1 + %exitcond = icmp eq i64 %inc, %n + br i1 %exitcond, label %for.cond.cleanup, label %for.body + +for.cond.cleanup: ; preds = %for.body, %entry + %sum.lcssa = phi <4 x double> [ zeroinitializer, %entry ], [ %addv, %for.body ] + %res = extractelement <4 x double> %sum.lcssa, i32 1 + store double %res, ptr %y + ret void +} + +declare void @f(ptr) diff --git a/llvm/test/CodeGen/RISCV/rv64-stackmap.ll b/llvm/test/CodeGen/RISCV/rv64-stackmap.ll index 9aefa90..c50a0fb3 100644 --- a/llvm/test/CodeGen/RISCV/rv64-stackmap.ll +++ b/llvm/test/CodeGen/RISCV/rv64-stackmap.ll @@ -7,11 +7,11 @@ ; CHECK-NEXT: .byte 0 ; CHECK-NEXT: .half 0 ; Num Functions -; CHECK-NEXT: .word 12 +; CHECK-NEXT: .word 13 ; Num LargeConstants -; CHECK-NEXT: .word 2 +; CHECK-NEXT: .word 3 ; Num Callsites -; CHECK-NEXT: .word 16 +; CHECK-NEXT: .word 17 ; Functions and stack size ; CHECK-NEXT: .quad constantargs @@ -50,10 +50,14 @@ ; CHECK-NEXT: .quad needsStackRealignment ; CHECK-NEXT: .quad -1 ; CHECK-NEXT: .quad 1 +; CHECK-NEXT: .quad floats +; CHECK-NEXT: .quad 32 +; CHECK-NEXT: .quad 1 ; Num LargeConstants ; CHECK-NEXT: .quad 4294967295 ; CHECK-NEXT: .quad 4294967296 +; CHECK-NEXT: .quad 4609434218613702656 ; Constant arguments ; @@ -379,6 +383,104 @@ define void @needsStackRealignment() { } declare void @escape_values(...) +; CHECK-LABEL: .word .L{{.*}}-floats +; CHECK-NEXT: .half 0 +; Num Locations +; CHECK-NEXT: .half 12 +; Loc 0: constant float as constant integer +; CHECK-NEXT: .byte 4 +; CHECK-NEXT: .byte 0 +; CHECK-NEXT: .half 8 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .word +; Loc 1: constant double as large constant integer +; CHECK-NEXT: .byte 5 +; CHECK-NEXT: .byte 0 +; CHECK-NEXT: .half 8 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .word +; Loc 2: constant half as constant integer +; CHECK-NEXT: .byte 4 +; CHECK-NEXT: .byte 0 +; CHECK-NEXT: .half 8 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .word +; Loc 3: constant bfloat as constant integer +; CHECK-NEXT: .byte 4 +; CHECK-NEXT: .byte 0 +; CHECK-NEXT: .half 8 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .word +; Loc 4: float value in X register +; CHECK-NEXT: .byte 1 +; CHECK-NEXT: .byte 0 +; CHECK-NEXT: .half 8 +; CHECK-NEXT: .half 10 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .word +; Loc 5: double value in X register +; CHECK-NEXT: .byte 1 +; CHECK-NEXT: .byte 0 +; CHECK-NEXT: .half 8 +; CHECK-NEXT: .half 11 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .word +; Loc 6: half value in X register +; CHECK-NEXT: .byte 1 +; CHECK-NEXT: .byte 0 +; CHECK-NEXT: .half 8 +; CHECK-NEXT: .half 12 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .word +; Loc 7: bfloat value in X register +; CHECK-NEXT: .byte 1 +; CHECK-NEXT: .byte 0 +; CHECK-NEXT: .half 8 +; CHECK-NEXT: .half 13 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .word +; Loc 8: float on stack +; CHECK-NEXT: .byte 2 +; CHECK-NEXT: .byte 0 +; CHECK-NEXT: .half 8 +; CHECK-NEXT: .half 2 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .word +; Loc 9: double on stack +; CHECK-NEXT: .byte 2 +; CHECK-NEXT: .byte 0 +; CHECK-NEXT: .half 8 +; CHECK-NEXT: .half 2 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .word +; Loc 10: half on stack +; CHECK-NEXT: .byte 2 +; CHECK-NEXT: .byte 0 +; CHECK-NEXT: .half 8 +; CHECK-NEXT: .half 2 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .word +; Loc 11: bfloat on stack +; CHECK-NEXT: .byte 2 +; CHECK-NEXT: .byte 0 +; CHECK-NEXT: .half 8 +; CHECK-NEXT: .half 2 +; CHECK-NEXT: .half 0 +; CHECK-NEXT: .word +define void @floats(float %f, double %g, half %h, bfloat %i) { + %ff = alloca float + %gg = alloca double + %hh = alloca half + %ii = alloca bfloat + call void (i64, i32, ...) @llvm.experimental.stackmap(i64 888, i32 0, float 1.25, + double 1.5, half 1.5, bfloat 1.5, float %f, double %g, half %h, bfloat %i, ptr %ff, ptr %gg, ptr %hh, ptr %ii) + ret void +} + declare void @llvm.experimental.stackmap(i64, i32, ...) declare void @llvm.experimental.patchpoint.void(i64, i32, ptr, i32, ...) declare i64 @llvm.experimental.patchpoint.i64(i64, i32, ptr, i32, ...) diff --git a/llvm/test/CodeGen/X86/bittest-big-integer.ll b/llvm/test/CodeGen/X86/bittest-big-integer.ll index dffe900..8007d9d 100644 --- a/llvm/test/CodeGen/X86/bittest-big-integer.ll +++ b/llvm/test/CodeGen/X86/bittest-big-integer.ll @@ -203,14 +203,24 @@ define i1 @init_eq_i32(ptr %word, i32 %position, i1 zeroext %value) nounwind { define i1 @test_ne_i64(ptr %word, i32 %position) nounwind { ; X86-LABEL: test_ne_i64: ; X86: # %bb.0: +; X86-NEXT: pushl %esi ; X86-NEXT: movl {{[0-9]+}}(%esp), %eax -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X86-NEXT: movl %ecx, %edx -; X86-NEXT: andl $32, %edx -; X86-NEXT: shrl $3, %edx -; X86-NEXT: movl (%eax,%edx), %eax -; X86-NEXT: btl %ecx, %eax -; X86-NEXT: setb %al +; X86-NEXT: movzbl {{[0-9]+}}(%esp), %ecx +; X86-NEXT: movl $1, %edx +; X86-NEXT: xorl %esi, %esi +; X86-NEXT: shldl %cl, %edx, %esi +; X86-NEXT: shll %cl, %edx +; X86-NEXT: testb $32, %cl +; X86-NEXT: je .LBB5_2 +; X86-NEXT: # %bb.1: +; X86-NEXT: movl %edx, %esi +; X86-NEXT: xorl %edx, %edx +; X86-NEXT: .LBB5_2: +; X86-NEXT: andl 4(%eax), %esi +; X86-NEXT: andl (%eax), %edx +; X86-NEXT: orl %esi, %edx +; X86-NEXT: setne %al +; X86-NEXT: popl %esi ; X86-NEXT: retl ; ; X64-LABEL: test_ne_i64: @@ -232,20 +242,38 @@ define i1 @test_ne_i64(ptr %word, i32 %position) nounwind { define i1 @complement_ne_i64(ptr %word, i32 %position) nounwind { ; X86-LABEL: complement_ne_i64: ; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: pushl %ebx ; X86-NEXT: pushl %edi ; X86-NEXT: pushl %esi -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx ; X86-NEXT: movl {{[0-9]+}}(%esp), %edx -; X86-NEXT: movl %edx, %esi -; X86-NEXT: andl $32, %esi -; X86-NEXT: shrl $3, %esi -; X86-NEXT: movl (%ecx,%esi), %edi -; X86-NEXT: btl %edx, %edi -; X86-NEXT: setb %al -; X86-NEXT: btcl %edx, %edi -; X86-NEXT: movl %edi, (%ecx,%esi) +; X86-NEXT: movzbl {{[0-9]+}}(%esp), %ecx +; X86-NEXT: movl $1, %eax +; X86-NEXT: xorl %esi, %esi +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: shll %cl, %eax +; X86-NEXT: testb $32, %cl +; X86-NEXT: je .LBB6_2 +; X86-NEXT: # %bb.1: +; X86-NEXT: movl %eax, %esi +; X86-NEXT: xorl %eax, %eax +; X86-NEXT: .LBB6_2: +; X86-NEXT: movl (%edx), %ecx +; X86-NEXT: movl 4(%edx), %edi +; X86-NEXT: movl %edi, %ebx +; X86-NEXT: andl %esi, %ebx +; X86-NEXT: movl %ecx, %ebp +; X86-NEXT: andl %eax, %ebp +; X86-NEXT: xorl %esi, %edi +; X86-NEXT: xorl %eax, %ecx +; X86-NEXT: orl %ebx, %ebp +; X86-NEXT: setne %al +; X86-NEXT: movl %ecx, (%edx) +; X86-NEXT: movl %edi, 4(%edx) ; X86-NEXT: popl %esi ; X86-NEXT: popl %edi +; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; ; X64-LABEL: complement_ne_i64: @@ -272,20 +300,40 @@ define i1 @complement_ne_i64(ptr %word, i32 %position) nounwind { define i1 @reset_eq_i64(ptr %word, i32 %position) nounwind { ; X86-LABEL: reset_eq_i64: ; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: pushl %ebx ; X86-NEXT: pushl %edi ; X86-NEXT: pushl %esi -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx ; X86-NEXT: movl {{[0-9]+}}(%esp), %edx -; X86-NEXT: movl %edx, %esi -; X86-NEXT: andl $32, %esi -; X86-NEXT: shrl $3, %esi -; X86-NEXT: movl (%ecx,%esi), %edi -; X86-NEXT: btl %edx, %edi -; X86-NEXT: setae %al -; X86-NEXT: btrl %edx, %edi -; X86-NEXT: movl %edi, (%ecx,%esi) +; X86-NEXT: movzbl {{[0-9]+}}(%esp), %ecx +; X86-NEXT: movl $1, %esi +; X86-NEXT: xorl %edi, %edi +; X86-NEXT: shldl %cl, %esi, %edi +; X86-NEXT: shll %cl, %esi +; X86-NEXT: testb $32, %cl +; X86-NEXT: je .LBB7_2 +; X86-NEXT: # %bb.1: +; X86-NEXT: movl %esi, %edi +; X86-NEXT: xorl %esi, %esi +; X86-NEXT: .LBB7_2: +; X86-NEXT: movl (%edx), %eax +; X86-NEXT: movl 4(%edx), %ecx +; X86-NEXT: movl %ecx, %ebx +; X86-NEXT: andl %edi, %ebx +; X86-NEXT: notl %edi +; X86-NEXT: movl %eax, %ebp +; X86-NEXT: andl %esi, %ebp +; X86-NEXT: notl %esi +; X86-NEXT: andl %ecx, %edi +; X86-NEXT: andl %eax, %esi +; X86-NEXT: orl %ebx, %ebp +; X86-NEXT: sete %al +; X86-NEXT: movl %esi, (%edx) +; X86-NEXT: movl %edi, 4(%edx) ; X86-NEXT: popl %esi ; X86-NEXT: popl %edi +; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; ; X64-LABEL: reset_eq_i64: @@ -313,20 +361,38 @@ define i1 @reset_eq_i64(ptr %word, i32 %position) nounwind { define i1 @set_ne_i64(ptr %word, i32 %position) nounwind { ; X86-LABEL: set_ne_i64: ; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: pushl %ebx ; X86-NEXT: pushl %edi ; X86-NEXT: pushl %esi -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx ; X86-NEXT: movl {{[0-9]+}}(%esp), %edx -; X86-NEXT: movl %edx, %esi -; X86-NEXT: andl $32, %esi -; X86-NEXT: shrl $3, %esi -; X86-NEXT: movl (%ecx,%esi), %edi -; X86-NEXT: btl %edx, %edi -; X86-NEXT: setb %al -; X86-NEXT: btsl %edx, %edi -; X86-NEXT: movl %edi, (%ecx,%esi) +; X86-NEXT: movzbl {{[0-9]+}}(%esp), %ecx +; X86-NEXT: movl $1, %eax +; X86-NEXT: xorl %esi, %esi +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: shll %cl, %eax +; X86-NEXT: testb $32, %cl +; X86-NEXT: je .LBB8_2 +; X86-NEXT: # %bb.1: +; X86-NEXT: movl %eax, %esi +; X86-NEXT: xorl %eax, %eax +; X86-NEXT: .LBB8_2: +; X86-NEXT: movl (%edx), %ecx +; X86-NEXT: movl 4(%edx), %edi +; X86-NEXT: movl %edi, %ebx +; X86-NEXT: andl %esi, %ebx +; X86-NEXT: movl %ecx, %ebp +; X86-NEXT: andl %eax, %ebp +; X86-NEXT: orl %esi, %edi +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: orl %ebx, %ebp +; X86-NEXT: setne %al +; X86-NEXT: movl %ecx, (%edx) +; X86-NEXT: movl %edi, 4(%edx) ; X86-NEXT: popl %esi ; X86-NEXT: popl %edi +; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; ; X64-LABEL: set_ne_i64: @@ -353,26 +419,52 @@ define i1 @set_ne_i64(ptr %word, i32 %position) nounwind { define i1 @init_eq_i64(ptr %word, i32 %position, i1 zeroext %value) nounwind { ; X86-LABEL: init_eq_i64: ; X86: # %bb.0: +; X86-NEXT: pushl %ebp ; X86-NEXT: pushl %ebx ; X86-NEXT: pushl %edi ; X86-NEXT: pushl %esi -; X86-NEXT: movl {{[0-9]+}}(%esp), %edx +; X86-NEXT: movzbl {{[0-9]+}}(%esp), %ecx +; X86-NEXT: movl $1, %eax +; X86-NEXT: xorl %edx, %edx +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: shll %cl, %eax +; X86-NEXT: movzbl {{[0-9]+}}(%esp), %esi +; X86-NEXT: xorl %edi, %edi +; X86-NEXT: shldl %cl, %esi, %edi +; X86-NEXT: shll %cl, %esi +; X86-NEXT: testb $32, %cl +; X86-NEXT: je .LBB9_2 +; X86-NEXT: # %bb.1: +; X86-NEXT: movl %eax, %edx +; X86-NEXT: movl $0, %eax +; X86-NEXT: .LBB9_2: +; X86-NEXT: movl %edx, %ebx +; X86-NEXT: notl %ebx +; X86-NEXT: movl %eax, %ebp +; X86-NEXT: notl %ebp +; X86-NEXT: je .LBB9_4 +; X86-NEXT: # %bb.3: +; X86-NEXT: movl %esi, %edi +; X86-NEXT: xorl %esi, %esi +; X86-NEXT: .LBB9_4: ; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X86-NEXT: movl %ecx, %esi -; X86-NEXT: andl $32, %esi -; X86-NEXT: shrl $3, %esi -; X86-NEXT: movl (%edx,%esi), %edi -; X86-NEXT: btl %ecx, %edi -; X86-NEXT: setae %al -; X86-NEXT: btrl %ecx, %edi -; X86-NEXT: movzbl {{[0-9]+}}(%esp), %ebx -; X86-NEXT: # kill: def $cl killed $cl killed $ecx -; X86-NEXT: shll %cl, %ebx +; X86-NEXT: movl 4(%ecx), %ecx +; X86-NEXT: andl %ecx, %edx +; X86-NEXT: andl %ecx, %ebx ; X86-NEXT: orl %edi, %ebx -; X86-NEXT: movl %ebx, (%edx,%esi) +; X86-NEXT: movl {{[0-9]+}}(%esp), %edi +; X86-NEXT: movl (%edi), %ecx +; X86-NEXT: andl %ecx, %eax +; X86-NEXT: andl %ecx, %ebp +; X86-NEXT: orl %esi, %ebp +; X86-NEXT: orl %edx, %eax +; X86-NEXT: movl %ebp, (%edi) +; X86-NEXT: movl %ebx, 4(%edi) +; X86-NEXT: sete %al ; X86-NEXT: popl %esi ; X86-NEXT: popl %edi ; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; ; SSE-LABEL: init_eq_i64: @@ -424,25 +516,101 @@ define i1 @init_eq_i64(ptr %word, i32 %position, i1 zeroext %value) nounwind { define i1 @test_ne_i128(ptr %word, i32 %position) nounwind { ; X86-LABEL: test_ne_i128: ; X86: # %bb.0: -; X86-NEXT: movl {{[0-9]+}}(%esp), %eax -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X86-NEXT: movl %ecx, %edx -; X86-NEXT: andl $96, %edx -; X86-NEXT: shrl $3, %edx -; X86-NEXT: movl (%eax,%edx), %eax -; X86-NEXT: btl %ecx, %eax -; X86-NEXT: setb %al +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: pushl %ebx +; X86-NEXT: pushl %edi +; X86-NEXT: pushl %esi +; X86-NEXT: andl $-16, %esp +; X86-NEXT: subl $48, %esp +; X86-NEXT: movzbl 12(%ebp), %ecx +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $1, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, (%esp) +; X86-NEXT: movl %ecx, %eax +; X86-NEXT: shrb $3, %al +; X86-NEXT: andb $12, %al +; X86-NEXT: negb %al +; X86-NEXT: movsbl %al, %esi +; X86-NEXT: movl 24(%esp,%esi), %edi +; X86-NEXT: movl 28(%esp,%esi), %eax +; X86-NEXT: shldl %cl, %edi, %eax +; X86-NEXT: movl 16(%esp,%esi), %edx +; X86-NEXT: movl 20(%esp,%esi), %esi +; X86-NEXT: shldl %cl, %esi, %edi +; X86-NEXT: shldl %cl, %edx, %esi +; X86-NEXT: movl 8(%ebp), %ebx +; X86-NEXT: shll %cl, %edx +; X86-NEXT: andl 8(%ebx), %edi +; X86-NEXT: andl (%ebx), %edx +; X86-NEXT: orl %edi, %edx +; X86-NEXT: andl 12(%ebx), %eax +; X86-NEXT: andl 4(%ebx), %esi +; X86-NEXT: orl %eax, %esi +; X86-NEXT: orl %edx, %esi +; X86-NEXT: setne %al +; X86-NEXT: leal -12(%ebp), %esp +; X86-NEXT: popl %esi +; X86-NEXT: popl %edi +; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; -; X64-LABEL: test_ne_i128: -; X64: # %bb.0: -; X64-NEXT: movl %esi, %eax -; X64-NEXT: andl $96, %eax -; X64-NEXT: shrl $3, %eax -; X64-NEXT: movl (%rdi,%rax), %eax -; X64-NEXT: btl %esi, %eax -; X64-NEXT: setb %al -; X64-NEXT: retq +; SSE-LABEL: test_ne_i128: +; SSE: # %bb.0: +; SSE-NEXT: movl %esi, %ecx +; SSE-NEXT: movl $1, %eax +; SSE-NEXT: xorl %edx, %edx +; SSE-NEXT: shldq %cl, %rax, %rdx +; SSE-NEXT: xorl %esi, %esi +; SSE-NEXT: shlq %cl, %rax +; SSE-NEXT: testb $64, %cl +; SSE-NEXT: cmovneq %rax, %rdx +; SSE-NEXT: cmovneq %rsi, %rax +; SSE-NEXT: andq 8(%rdi), %rdx +; SSE-NEXT: andq (%rdi), %rax +; SSE-NEXT: orq %rdx, %rax +; SSE-NEXT: setne %al +; SSE-NEXT: retq +; +; AVX2-LABEL: test_ne_i128: +; AVX2: # %bb.0: +; AVX2-NEXT: movl %esi, %ecx +; AVX2-NEXT: xorl %eax, %eax +; AVX2-NEXT: movl $1, %edx +; AVX2-NEXT: xorl %esi, %esi +; AVX2-NEXT: shldq %cl, %rdx, %rsi +; AVX2-NEXT: shlxq %rcx, %rdx, %rdx +; AVX2-NEXT: testb $64, %cl +; AVX2-NEXT: cmovneq %rdx, %rsi +; AVX2-NEXT: cmovneq %rax, %rdx +; AVX2-NEXT: andq 8(%rdi), %rsi +; AVX2-NEXT: andq (%rdi), %rdx +; AVX2-NEXT: orq %rsi, %rdx +; AVX2-NEXT: setne %al +; AVX2-NEXT: retq +; +; AVX512-LABEL: test_ne_i128: +; AVX512: # %bb.0: +; AVX512-NEXT: movl %esi, %ecx +; AVX512-NEXT: movl $1, %eax +; AVX512-NEXT: xorl %edx, %edx +; AVX512-NEXT: shldq %cl, %rax, %rdx +; AVX512-NEXT: xorl %esi, %esi +; AVX512-NEXT: shlxq %rcx, %rax, %rax +; AVX512-NEXT: testb $64, %cl +; AVX512-NEXT: cmovneq %rax, %rdx +; AVX512-NEXT: cmovneq %rsi, %rax +; AVX512-NEXT: andq 8(%rdi), %rdx +; AVX512-NEXT: andq (%rdi), %rax +; AVX512-NEXT: orq %rdx, %rax +; AVX512-NEXT: setne %al +; AVX512-NEXT: retq %rem = and i32 %position, 127 %ofs = zext nneg i32 %rem to i128 %bit = shl nuw i128 1, %ofs @@ -455,33 +623,124 @@ define i1 @test_ne_i128(ptr %word, i32 %position) nounwind { define i1 @complement_ne_i128(ptr %word, i32 %position) nounwind { ; X86-LABEL: complement_ne_i128: ; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: pushl %ebx ; X86-NEXT: pushl %edi ; X86-NEXT: pushl %esi -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X86-NEXT: movl {{[0-9]+}}(%esp), %edx -; X86-NEXT: movl %edx, %esi -; X86-NEXT: andl $96, %esi -; X86-NEXT: shrl $3, %esi -; X86-NEXT: movl (%ecx,%esi), %edi -; X86-NEXT: btl %edx, %edi -; X86-NEXT: setb %al -; X86-NEXT: btcl %edx, %edi -; X86-NEXT: movl %edi, (%ecx,%esi) +; X86-NEXT: andl $-16, %esp +; X86-NEXT: subl $80, %esp +; X86-NEXT: movzbl 12(%ebp), %ecx +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $1, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl %ecx, %eax +; X86-NEXT: shrb $3, %al +; X86-NEXT: andb $12, %al +; X86-NEXT: negb %al +; X86-NEXT: movsbl %al, %eax +; X86-NEXT: movl 56(%esp,%eax), %esi +; X86-NEXT: movl 60(%esp,%eax), %edx +; X86-NEXT: shldl %cl, %esi, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 48(%esp,%eax), %edi +; X86-NEXT: movl 52(%esp,%eax), %ebx +; X86-NEXT: shldl %cl, %ebx, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edi, %ebx +; X86-NEXT: movl 8(%ebp), %eax +; X86-NEXT: shll %cl, %edi +; X86-NEXT: movl %eax, %ecx +; X86-NEXT: movl 8(%eax), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %esi, %eax +; X86-NEXT: movl (%ecx), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl %ecx, %esi +; X86-NEXT: movl %edx, %ecx +; X86-NEXT: andl %edi, %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl 12(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl 4(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ebx, %eax +; X86-NEXT: orl %edx, %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: xorl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: xorl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: xorl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Folded Reload +; X86-NEXT: xorl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: orl %ecx, %eax +; X86-NEXT: movl 8(%ebp), %eax +; X86-NEXT: movl %edx, 8(%eax) +; X86-NEXT: movl %esi, 12(%eax) +; X86-NEXT: movl %edi, (%eax) +; X86-NEXT: movl %ebx, 4(%eax) +; X86-NEXT: setne %al +; X86-NEXT: leal -12(%ebp), %esp ; X86-NEXT: popl %esi ; X86-NEXT: popl %edi +; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; -; X64-LABEL: complement_ne_i128: -; X64: # %bb.0: -; X64-NEXT: movl %esi, %ecx -; X64-NEXT: andl $96, %ecx -; X64-NEXT: shrl $3, %ecx -; X64-NEXT: movl (%rdi,%rcx), %edx -; X64-NEXT: btl %esi, %edx -; X64-NEXT: setb %al -; X64-NEXT: btcl %esi, %edx -; X64-NEXT: movl %edx, (%rdi,%rcx) -; X64-NEXT: retq +; SSE-LABEL: complement_ne_i128: +; SSE: # %bb.0: +; SSE-NEXT: movl %esi, %ecx +; SSE-NEXT: movl $1, %edx +; SSE-NEXT: xorl %esi, %esi +; SSE-NEXT: shldq %cl, %rdx, %rsi +; SSE-NEXT: shlq %cl, %rdx +; SSE-NEXT: xorl %eax, %eax +; SSE-NEXT: testb $64, %cl +; SSE-NEXT: cmovneq %rdx, %rsi +; SSE-NEXT: cmovneq %rax, %rdx +; SSE-NEXT: movq (%rdi), %rax +; SSE-NEXT: movq 8(%rdi), %rcx +; SSE-NEXT: movq %rcx, %r8 +; SSE-NEXT: andq %rsi, %r8 +; SSE-NEXT: movq %rax, %r9 +; SSE-NEXT: andq %rdx, %r9 +; SSE-NEXT: xorq %rcx, %rsi +; SSE-NEXT: xorq %rax, %rdx +; SSE-NEXT: orq %r8, %r9 +; SSE-NEXT: setne %al +; SSE-NEXT: movq %rdx, (%rdi) +; SSE-NEXT: movq %rsi, 8(%rdi) +; SSE-NEXT: retq +; +; AVX-LABEL: complement_ne_i128: +; AVX: # %bb.0: +; AVX-NEXT: movl %esi, %ecx +; AVX-NEXT: xorl %eax, %eax +; AVX-NEXT: movl $1, %edx +; AVX-NEXT: xorl %esi, %esi +; AVX-NEXT: shldq %cl, %rdx, %rsi +; AVX-NEXT: shlxq %rcx, %rdx, %rdx +; AVX-NEXT: testb $64, %cl +; AVX-NEXT: cmovneq %rdx, %rsi +; AVX-NEXT: cmovneq %rax, %rdx +; AVX-NEXT: movq (%rdi), %rax +; AVX-NEXT: movq 8(%rdi), %rcx +; AVX-NEXT: movq %rcx, %r8 +; AVX-NEXT: andq %rsi, %r8 +; AVX-NEXT: movq %rax, %r9 +; AVX-NEXT: andq %rdx, %r9 +; AVX-NEXT: xorq %rcx, %rsi +; AVX-NEXT: xorq %rax, %rdx +; AVX-NEXT: orq %r8, %r9 +; AVX-NEXT: setne %al +; AVX-NEXT: movq %rdx, (%rdi) +; AVX-NEXT: movq %rsi, 8(%rdi) +; AVX-NEXT: retq %rem = and i32 %position, 127 %ofs = zext nneg i32 %rem to i128 %bit = shl nuw i128 1, %ofs @@ -496,33 +755,124 @@ define i1 @complement_ne_i128(ptr %word, i32 %position) nounwind { define i1 @reset_eq_i128(ptr %word, i32 %position) nounwind { ; X86-LABEL: reset_eq_i128: ; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: pushl %ebx ; X86-NEXT: pushl %edi ; X86-NEXT: pushl %esi -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X86-NEXT: movl {{[0-9]+}}(%esp), %edx -; X86-NEXT: movl %edx, %esi -; X86-NEXT: andl $96, %esi -; X86-NEXT: shrl $3, %esi -; X86-NEXT: movl (%ecx,%esi), %edi -; X86-NEXT: btl %edx, %edi -; X86-NEXT: setae %al -; X86-NEXT: btrl %edx, %edi -; X86-NEXT: movl %edi, (%ecx,%esi) +; X86-NEXT: andl $-16, %esp +; X86-NEXT: subl $80, %esp +; X86-NEXT: movzbl 12(%ebp), %ecx +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $1, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl %ecx, %eax +; X86-NEXT: shrb $3, %al +; X86-NEXT: andb $12, %al +; X86-NEXT: negb %al +; X86-NEXT: movsbl %al, %eax +; X86-NEXT: movl 56(%esp,%eax), %edx +; X86-NEXT: movl 60(%esp,%eax), %esi +; X86-NEXT: shldl %cl, %edx, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 48(%esp,%eax), %esi +; X86-NEXT: movl 52(%esp,%eax), %edi +; X86-NEXT: shldl %cl, %edi, %edx +; X86-NEXT: shldl %cl, %esi, %edi +; X86-NEXT: movl 8(%ebp), %ebx +; X86-NEXT: shll %cl, %esi +; X86-NEXT: movl 8(%ebx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edx, %eax +; X86-NEXT: movl (%ebx), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %esi, %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 12(%ebx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl %edi, %ecx +; X86-NEXT: movl 4(%ebx), %ebx +; X86-NEXT: andl %ebx, %edi +; X86-NEXT: orl %eax, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: notl %eax +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: notl %ecx +; X86-NEXT: andl %ebx, %ecx +; X86-NEXT: notl %esi +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: movl 8(%ebp), %edi +; X86-NEXT: movl %edx, 8(%edi) +; X86-NEXT: movl %eax, 12(%edi) +; X86-NEXT: movl %esi, (%edi) +; X86-NEXT: movl %ecx, 4(%edi) +; X86-NEXT: sete %al +; X86-NEXT: leal -12(%ebp), %esp ; X86-NEXT: popl %esi ; X86-NEXT: popl %edi +; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; -; X64-LABEL: reset_eq_i128: -; X64: # %bb.0: -; X64-NEXT: movl %esi, %ecx -; X64-NEXT: andl $96, %ecx -; X64-NEXT: shrl $3, %ecx -; X64-NEXT: movl (%rdi,%rcx), %edx -; X64-NEXT: btl %esi, %edx -; X64-NEXT: setae %al -; X64-NEXT: btrl %esi, %edx -; X64-NEXT: movl %edx, (%rdi,%rcx) -; X64-NEXT: retq +; SSE-LABEL: reset_eq_i128: +; SSE: # %bb.0: +; SSE-NEXT: movl %esi, %ecx +; SSE-NEXT: movl $1, %edx +; SSE-NEXT: xorl %esi, %esi +; SSE-NEXT: shldq %cl, %rdx, %rsi +; SSE-NEXT: xorl %eax, %eax +; SSE-NEXT: shlq %cl, %rdx +; SSE-NEXT: testb $64, %cl +; SSE-NEXT: cmovneq %rdx, %rsi +; SSE-NEXT: cmovneq %rax, %rdx +; SSE-NEXT: movq (%rdi), %rax +; SSE-NEXT: movq 8(%rdi), %rcx +; SSE-NEXT: movq %rcx, %r8 +; SSE-NEXT: andq %rsi, %r8 +; SSE-NEXT: notq %rsi +; SSE-NEXT: movq %rax, %r9 +; SSE-NEXT: andq %rdx, %r9 +; SSE-NEXT: notq %rdx +; SSE-NEXT: andq %rcx, %rsi +; SSE-NEXT: andq %rax, %rdx +; SSE-NEXT: orq %r8, %r9 +; SSE-NEXT: sete %al +; SSE-NEXT: movq %rdx, (%rdi) +; SSE-NEXT: movq %rsi, 8(%rdi) +; SSE-NEXT: retq +; +; AVX-LABEL: reset_eq_i128: +; AVX: # %bb.0: +; AVX-NEXT: movl %esi, %ecx +; AVX-NEXT: xorl %eax, %eax +; AVX-NEXT: movl $1, %edx +; AVX-NEXT: xorl %esi, %esi +; AVX-NEXT: shldq %cl, %rdx, %rsi +; AVX-NEXT: shlxq %rcx, %rdx, %rdx +; AVX-NEXT: testb $64, %cl +; AVX-NEXT: cmovneq %rdx, %rsi +; AVX-NEXT: cmovneq %rax, %rdx +; AVX-NEXT: movq (%rdi), %rax +; AVX-NEXT: movq 8(%rdi), %rcx +; AVX-NEXT: andnq %rcx, %rsi, %r8 +; AVX-NEXT: andq %rsi, %rcx +; AVX-NEXT: andnq %rax, %rdx, %rsi +; AVX-NEXT: andq %rdx, %rax +; AVX-NEXT: orq %rcx, %rax +; AVX-NEXT: sete %al +; AVX-NEXT: movq %rsi, (%rdi) +; AVX-NEXT: movq %r8, 8(%rdi) +; AVX-NEXT: retq %rem = and i32 %position, 127 %ofs = zext nneg i32 %rem to i128 %bit = shl nuw i128 1, %ofs @@ -538,33 +888,124 @@ define i1 @reset_eq_i128(ptr %word, i32 %position) nounwind { define i1 @set_ne_i128(ptr %word, i32 %position) nounwind { ; X86-LABEL: set_ne_i128: ; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: pushl %ebx ; X86-NEXT: pushl %edi ; X86-NEXT: pushl %esi -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X86-NEXT: movl {{[0-9]+}}(%esp), %edx -; X86-NEXT: movl %edx, %esi -; X86-NEXT: andl $96, %esi -; X86-NEXT: shrl $3, %esi -; X86-NEXT: movl (%ecx,%esi), %edi -; X86-NEXT: btl %edx, %edi -; X86-NEXT: setb %al -; X86-NEXT: btsl %edx, %edi -; X86-NEXT: movl %edi, (%ecx,%esi) +; X86-NEXT: andl $-16, %esp +; X86-NEXT: subl $80, %esp +; X86-NEXT: movzbl 12(%ebp), %ecx +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $1, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl %ecx, %eax +; X86-NEXT: shrb $3, %al +; X86-NEXT: andb $12, %al +; X86-NEXT: negb %al +; X86-NEXT: movsbl %al, %eax +; X86-NEXT: movl 56(%esp,%eax), %esi +; X86-NEXT: movl 60(%esp,%eax), %edx +; X86-NEXT: shldl %cl, %esi, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 48(%esp,%eax), %edi +; X86-NEXT: movl 52(%esp,%eax), %ebx +; X86-NEXT: shldl %cl, %ebx, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edi, %ebx +; X86-NEXT: movl 8(%ebp), %eax +; X86-NEXT: shll %cl, %edi +; X86-NEXT: movl %eax, %ecx +; X86-NEXT: movl 8(%eax), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %esi, %eax +; X86-NEXT: movl (%ecx), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl %ecx, %esi +; X86-NEXT: movl %edx, %ecx +; X86-NEXT: andl %edi, %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl 12(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl 4(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ebx, %eax +; X86-NEXT: orl %edx, %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Folded Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: orl %ecx, %eax +; X86-NEXT: movl 8(%ebp), %eax +; X86-NEXT: movl %edx, 8(%eax) +; X86-NEXT: movl %esi, 12(%eax) +; X86-NEXT: movl %edi, (%eax) +; X86-NEXT: movl %ebx, 4(%eax) +; X86-NEXT: setne %al +; X86-NEXT: leal -12(%ebp), %esp ; X86-NEXT: popl %esi ; X86-NEXT: popl %edi +; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; -; X64-LABEL: set_ne_i128: -; X64: # %bb.0: -; X64-NEXT: movl %esi, %ecx -; X64-NEXT: andl $96, %ecx -; X64-NEXT: shrl $3, %ecx -; X64-NEXT: movl (%rdi,%rcx), %edx -; X64-NEXT: btl %esi, %edx -; X64-NEXT: setb %al -; X64-NEXT: btsl %esi, %edx -; X64-NEXT: movl %edx, (%rdi,%rcx) -; X64-NEXT: retq +; SSE-LABEL: set_ne_i128: +; SSE: # %bb.0: +; SSE-NEXT: movl %esi, %ecx +; SSE-NEXT: movl $1, %edx +; SSE-NEXT: xorl %esi, %esi +; SSE-NEXT: shldq %cl, %rdx, %rsi +; SSE-NEXT: shlq %cl, %rdx +; SSE-NEXT: xorl %eax, %eax +; SSE-NEXT: testb $64, %cl +; SSE-NEXT: cmovneq %rdx, %rsi +; SSE-NEXT: cmovneq %rax, %rdx +; SSE-NEXT: movq (%rdi), %rax +; SSE-NEXT: movq 8(%rdi), %rcx +; SSE-NEXT: movq %rcx, %r8 +; SSE-NEXT: andq %rsi, %r8 +; SSE-NEXT: movq %rax, %r9 +; SSE-NEXT: andq %rdx, %r9 +; SSE-NEXT: orq %rcx, %rsi +; SSE-NEXT: orq %rax, %rdx +; SSE-NEXT: orq %r8, %r9 +; SSE-NEXT: setne %al +; SSE-NEXT: movq %rdx, (%rdi) +; SSE-NEXT: movq %rsi, 8(%rdi) +; SSE-NEXT: retq +; +; AVX-LABEL: set_ne_i128: +; AVX: # %bb.0: +; AVX-NEXT: movl %esi, %ecx +; AVX-NEXT: xorl %eax, %eax +; AVX-NEXT: movl $1, %edx +; AVX-NEXT: xorl %esi, %esi +; AVX-NEXT: shldq %cl, %rdx, %rsi +; AVX-NEXT: shlxq %rcx, %rdx, %rdx +; AVX-NEXT: testb $64, %cl +; AVX-NEXT: cmovneq %rdx, %rsi +; AVX-NEXT: cmovneq %rax, %rdx +; AVX-NEXT: movq (%rdi), %rax +; AVX-NEXT: movq 8(%rdi), %rcx +; AVX-NEXT: movq %rcx, %r8 +; AVX-NEXT: andq %rsi, %r8 +; AVX-NEXT: movq %rax, %r9 +; AVX-NEXT: andq %rdx, %r9 +; AVX-NEXT: orq %rcx, %rsi +; AVX-NEXT: orq %rax, %rdx +; AVX-NEXT: orq %r8, %r9 +; AVX-NEXT: setne %al +; AVX-NEXT: movq %rdx, (%rdi) +; AVX-NEXT: movq %rsi, 8(%rdi) +; AVX-NEXT: retq %rem = and i32 %position, 127 %ofs = zext nneg i32 %rem to i128 %bit = shl nuw i128 1, %ofs @@ -579,55 +1020,218 @@ define i1 @set_ne_i128(ptr %word, i32 %position) nounwind { define i1 @init_eq_i128(ptr %word, i32 %position, i1 zeroext %value) nounwind { ; X86-LABEL: init_eq_i128: ; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp ; X86-NEXT: pushl %ebx ; X86-NEXT: pushl %edi ; X86-NEXT: pushl %esi -; X86-NEXT: movl {{[0-9]+}}(%esp), %edx -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X86-NEXT: movl %ecx, %esi -; X86-NEXT: andl $96, %esi -; X86-NEXT: shrl $3, %esi -; X86-NEXT: movl (%edx,%esi), %edi -; X86-NEXT: btl %ecx, %edi -; X86-NEXT: setae %al -; X86-NEXT: btrl %ecx, %edi -; X86-NEXT: movzbl {{[0-9]+}}(%esp), %ebx -; X86-NEXT: # kill: def $cl killed $cl killed $ecx -; X86-NEXT: shll %cl, %ebx +; X86-NEXT: andl $-16, %esp +; X86-NEXT: subl $128, %esp +; X86-NEXT: movzbl 12(%ebp), %ecx +; X86-NEXT: movzbl 16(%ebp), %eax +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $1, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl %ecx, %edx +; X86-NEXT: shrb $3, %dl +; X86-NEXT: andb $12, %dl +; X86-NEXT: negb %dl +; X86-NEXT: movsbl %dl, %esi +; X86-NEXT: movl 64(%esp,%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 68(%esp,%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 72(%esp,%esi), %ebx +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movzbl %al, %eax +; X86-NEXT: movl 76(%esp,%esi), %edi +; X86-NEXT: movl %eax, {{[0-9]+}}(%esp) +; X86-NEXT: movl %ebx, %eax +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: shldl %cl, %ebx, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: shll %cl, %edx +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl 8(%ebp), %esi +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%esi), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ebx, %eax +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl (%esi), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ebx, %edx +; X86-NEXT: orl %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %edi, %esi +; X86-NEXT: movl 8(%ebp), %ecx +; X86-NEXT: movl 12(%ecx), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ebx, %edi +; X86-NEXT: movl %eax, %ebx +; X86-NEXT: movl %eax, %edx +; X86-NEXT: movl 4(%ecx), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ecx, %ebx ; X86-NEXT: orl %edi, %ebx -; X86-NEXT: movl %ebx, (%edx,%esi) +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: notl %ecx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl 100(%esp,%ecx), %edi +; X86-NEXT: movl 104(%esp,%ecx), %ecx +; X86-NEXT: movl %ecx, %ebx +; X86-NEXT: movl %ecx, %eax +; X86-NEXT: movzbl 12(%ebp), %ecx +; X86-NEXT: shldl %cl, %edi, %ebx +; X86-NEXT: orl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: notl %esi +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: movl 108(%esp,%ebx), %ebx +; X86-NEXT: shldl %cl, %eax, %ebx +; X86-NEXT: orl %ebx, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: notl %eax +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: movl 96(%esp,%ebx), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shll %cl, %ebx +; X86-NEXT: orl %ebx, %eax +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: shldl %cl, %ebx, %edi +; X86-NEXT: orl %edi, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload +; X86-NEXT: movl 8(%ebp), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 8(%ecx) +; X86-NEXT: movl %esi, 12(%ecx) +; X86-NEXT: movl %eax, (%ecx) +; X86-NEXT: movl %edx, 4(%ecx) +; X86-NEXT: sete %al +; X86-NEXT: leal -12(%ebp), %esp ; X86-NEXT: popl %esi ; X86-NEXT: popl %edi ; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; ; SSE-LABEL: init_eq_i128: ; SSE: # %bb.0: ; SSE-NEXT: movl %esi, %ecx -; SSE-NEXT: andl $96, %esi -; SSE-NEXT: shrl $3, %esi -; SSE-NEXT: movl (%rdi,%rsi), %r8d -; SSE-NEXT: btl %ecx, %r8d -; SSE-NEXT: setae %al -; SSE-NEXT: shll %cl, %edx -; SSE-NEXT: btrl %ecx, %r8d -; SSE-NEXT: orl %r8d, %edx -; SSE-NEXT: movl %edx, (%rdi,%rsi) +; SSE-NEXT: movl $1, %esi +; SSE-NEXT: xorl %r8d, %r8d +; SSE-NEXT: shldq %cl, %rsi, %r8 +; SSE-NEXT: shlq %cl, %rsi +; SSE-NEXT: movl %edx, %eax +; SSE-NEXT: xorl %edx, %edx +; SSE-NEXT: shldq %cl, %rax, %rdx +; SSE-NEXT: shlq %cl, %rax +; SSE-NEXT: xorl %r9d, %r9d +; SSE-NEXT: testb $64, %cl +; SSE-NEXT: cmovneq %rsi, %r8 +; SSE-NEXT: cmovneq %r9, %rsi +; SSE-NEXT: cmovneq %rax, %rdx +; SSE-NEXT: cmovneq %r9, %rax +; SSE-NEXT: movq (%rdi), %rcx +; SSE-NEXT: movq 8(%rdi), %r9 +; SSE-NEXT: movq %r9, %r10 +; SSE-NEXT: andq %r8, %r10 +; SSE-NEXT: notq %r8 +; SSE-NEXT: movq %rcx, %r11 +; SSE-NEXT: andq %rsi, %r11 +; SSE-NEXT: notq %rsi +; SSE-NEXT: andq %r9, %r8 +; SSE-NEXT: orq %rdx, %r8 +; SSE-NEXT: andq %rcx, %rsi +; SSE-NEXT: orq %rax, %rsi +; SSE-NEXT: orq %r10, %r11 +; SSE-NEXT: sete %al +; SSE-NEXT: movq %rsi, (%rdi) +; SSE-NEXT: movq %r8, 8(%rdi) ; SSE-NEXT: retq ; -; AVX-LABEL: init_eq_i128: -; AVX: # %bb.0: -; AVX-NEXT: movl %esi, %ecx -; AVX-NEXT: andl $96, %ecx -; AVX-NEXT: shrl $3, %ecx -; AVX-NEXT: movl (%rdi,%rcx), %r8d -; AVX-NEXT: btl %esi, %r8d -; AVX-NEXT: setae %al -; AVX-NEXT: btrl %esi, %r8d -; AVX-NEXT: shlxl %esi, %edx, %edx -; AVX-NEXT: orl %r8d, %edx -; AVX-NEXT: movl %edx, (%rdi,%rcx) -; AVX-NEXT: retq +; AVX2-LABEL: init_eq_i128: +; AVX2: # %bb.0: +; AVX2-NEXT: movl %esi, %ecx +; AVX2-NEXT: movl $1, %esi +; AVX2-NEXT: xorl %eax, %eax +; AVX2-NEXT: shldq %cl, %rsi, %rax +; AVX2-NEXT: xorl %r8d, %r8d +; AVX2-NEXT: movl %edx, %edx +; AVX2-NEXT: xorl %r9d, %r9d +; AVX2-NEXT: shldq %cl, %rdx, %r9 +; AVX2-NEXT: shlxq %rcx, %rsi, %rsi +; AVX2-NEXT: testb $64, %cl +; AVX2-NEXT: cmovneq %rsi, %rax +; AVX2-NEXT: cmovneq %r8, %rsi +; AVX2-NEXT: shlxq %rcx, %rdx, %rcx +; AVX2-NEXT: cmovneq %rcx, %r9 +; AVX2-NEXT: cmovneq %r8, %rcx +; AVX2-NEXT: movq (%rdi), %rdx +; AVX2-NEXT: movq 8(%rdi), %r8 +; AVX2-NEXT: andnq %r8, %rax, %r10 +; AVX2-NEXT: andq %rax, %r8 +; AVX2-NEXT: andnq %rdx, %rsi, %r11 +; AVX2-NEXT: andq %rsi, %rdx +; AVX2-NEXT: orq %r9, %r10 +; AVX2-NEXT: orq %rcx, %r11 +; AVX2-NEXT: orq %r8, %rdx +; AVX2-NEXT: sete %al +; AVX2-NEXT: movq %r11, (%rdi) +; AVX2-NEXT: movq %r10, 8(%rdi) +; AVX2-NEXT: retq +; +; AVX512-LABEL: init_eq_i128: +; AVX512: # %bb.0: +; AVX512-NEXT: movl %esi, %ecx +; AVX512-NEXT: xorl %eax, %eax +; AVX512-NEXT: movl $1, %esi +; AVX512-NEXT: xorl %r8d, %r8d +; AVX512-NEXT: shldq %cl, %rsi, %r8 +; AVX512-NEXT: shlxq %rcx, %rsi, %rsi +; AVX512-NEXT: movl %edx, %edx +; AVX512-NEXT: xorl %r9d, %r9d +; AVX512-NEXT: shldq %cl, %rdx, %r9 +; AVX512-NEXT: testb $64, %cl +; AVX512-NEXT: cmovneq %rsi, %r8 +; AVX512-NEXT: cmovneq %rax, %rsi +; AVX512-NEXT: shlxq %rcx, %rdx, %rcx +; AVX512-NEXT: cmovneq %rcx, %r9 +; AVX512-NEXT: cmovneq %rax, %rcx +; AVX512-NEXT: movq (%rdi), %rax +; AVX512-NEXT: movq 8(%rdi), %rdx +; AVX512-NEXT: andnq %rdx, %r8, %r10 +; AVX512-NEXT: andq %r8, %rdx +; AVX512-NEXT: andnq %rax, %rsi, %r8 +; AVX512-NEXT: andq %rsi, %rax +; AVX512-NEXT: orq %r9, %r10 +; AVX512-NEXT: orq %rcx, %r8 +; AVX512-NEXT: orq %rdx, %rax +; AVX512-NEXT: sete %al +; AVX512-NEXT: movq %r8, (%rdi) +; AVX512-NEXT: movq %r10, 8(%rdi) +; AVX512-NEXT: retq %rem = and i32 %position, 127 %ofs = zext nneg i32 %rem to i128 %bit = shl nuw i128 1, %ofs @@ -648,25 +1252,344 @@ define i1 @init_eq_i128(ptr %word, i32 %position, i1 zeroext %value) nounwind { define i1 @test_ne_i512(ptr %word, i32 %position) nounwind { ; X86-LABEL: test_ne_i512: ; X86: # %bb.0: -; X86-NEXT: movl {{[0-9]+}}(%esp), %eax -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X86-NEXT: movl %ecx, %edx -; X86-NEXT: shrl $3, %edx -; X86-NEXT: andl $60, %edx -; X86-NEXT: movl (%eax,%edx), %eax -; X86-NEXT: btl %ecx, %eax -; X86-NEXT: setb %al +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: pushl %ebx +; X86-NEXT: pushl %edi +; X86-NEXT: pushl %esi +; X86-NEXT: andl $-16, %esp +; X86-NEXT: subl $224, %esp +; X86-NEXT: movl 12(%ebp), %ecx +; X86-NEXT: movl %ecx, %eax +; X86-NEXT: shrl $3, %eax +; X86-NEXT: andl $60, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: leal {{[0-9]+}}(%esp), %edx +; X86-NEXT: subl %eax, %edx +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $1, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl 24(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 28(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl $31, %ecx +; X86-NEXT: shldl %cl, %esi, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 56(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 60(%edx), %eax +; X86-NEXT: shldl %cl, %esi, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 12(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %esi, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 40(%edx), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 44(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edi, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 16(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 20(%edx), %edi +; X86-NEXT: movl %edi, %ebx +; X86-NEXT: shldl %cl, %eax, %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 32(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 36(%edx), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %esi, %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl 52(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl 4(%edx), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: shldl %cl, %edi, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: shldl %cl, %edi, %eax +; X86-NEXT: movl 8(%ebp), %ebx +; X86-NEXT: andl 40(%ebx), %eax +; X86-NEXT: andl 8(%ebx), %esi +; X86-NEXT: orl %eax, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 56(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: andl 24(%ebx), %edi +; X86-NEXT: orl %eax, %edi +; X86-NEXT: orl %esi, %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %ebx, %edi +; X86-NEXT: andl 44(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl 12(%ebx), %esi +; X86-NEXT: orl %eax, %esi +; X86-NEXT: movl %esi, %ebx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl 60(%edi), %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 28(%edi), %eax +; X86-NEXT: orl %esi, %eax +; X86-NEXT: orl %ebx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl (%edx), %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: shldl %cl, %edx, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: shldl %cl, %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: negl %edx +; X86-NEXT: movl 192(%esp,%edx), %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: shldl %cl, %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: shldl %cl, %ebx, %edx +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: shll %cl, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl 8(%ebp), %ebx +; X86-NEXT: andl 32(%ebx), %ecx +; X86-NEXT: andl (%ebx), %esi +; X86-NEXT: orl %ecx, %esi +; X86-NEXT: andl 16(%ebx), %edi +; X86-NEXT: andl 48(%ebx), %edx +; X86-NEXT: orl %edi, %edx +; X86-NEXT: orl %esi, %edx +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 36(%ebx), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl 4(%ebx), %esi +; X86-NEXT: orl %ecx, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 20(%ebx), %ecx +; X86-NEXT: andl 52(%ebx), %eax +; X86-NEXT: orl %ecx, %eax +; X86-NEXT: orl %esi, %eax +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: orl %edx, %eax +; X86-NEXT: setne %al +; X86-NEXT: leal -12(%ebp), %esp +; X86-NEXT: popl %esi +; X86-NEXT: popl %edi +; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; -; X64-LABEL: test_ne_i512: -; X64: # %bb.0: -; X64-NEXT: movl %esi, %eax -; X64-NEXT: shrl $3, %eax -; X64-NEXT: andl $60, %eax -; X64-NEXT: movl (%rdi,%rax), %eax -; X64-NEXT: btl %esi, %eax -; X64-NEXT: setb %al -; X64-NEXT: retq +; SSE-LABEL: test_ne_i512: +; SSE: # %bb.0: +; SSE-NEXT: pushq %r15 +; SSE-NEXT: pushq %r14 +; SSE-NEXT: pushq %rbx +; SSE-NEXT: xorps %xmm0, %xmm0 +; SSE-NEXT: movups %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movq $0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movq $1, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movl %esi, %ecx +; SSE-NEXT: andl $63, %ecx +; SSE-NEXT: shrl $3, %esi +; SSE-NEXT: andl $56, %esi +; SSE-NEXT: negl %esi +; SSE-NEXT: movslq %esi, %rbx +; SSE-NEXT: movq -48(%rsp,%rbx), %rdx +; SSE-NEXT: movq -40(%rsp,%rbx), %r14 +; SSE-NEXT: movq %r14, %rax +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq -16(%rsp,%rbx), %r11 +; SSE-NEXT: movq -8(%rsp,%rbx), %r10 +; SSE-NEXT: shldq %cl, %r11, %r10 +; SSE-NEXT: movq -32(%rsp,%rbx), %r9 +; SSE-NEXT: movq -24(%rsp,%rbx), %r15 +; SSE-NEXT: movq %r15, %r8 +; SSE-NEXT: shldq %cl, %r9, %r8 +; SSE-NEXT: movq -56(%rsp,%rbx), %rsi +; SSE-NEXT: shldq %cl, %rsi, %rdx +; SSE-NEXT: shldq %cl, %r15, %r11 +; SSE-NEXT: shldq %cl, %r14, %r9 +; SSE-NEXT: movq -64(%rsp,%rbx), %rbx +; SSE-NEXT: shldq %cl, %rbx, %rsi +; SSE-NEXT: # kill: def $cl killed $cl killed $ecx +; SSE-NEXT: shlq %cl, %rbx +; SSE-NEXT: andq 32(%rdi), %r9 +; SSE-NEXT: andq 48(%rdi), %r11 +; SSE-NEXT: andq 16(%rdi), %rdx +; SSE-NEXT: orq %r11, %rdx +; SSE-NEXT: andq 40(%rdi), %r8 +; SSE-NEXT: andq 56(%rdi), %r10 +; SSE-NEXT: andq 24(%rdi), %rax +; SSE-NEXT: orq %r10, %rax +; SSE-NEXT: andq (%rdi), %rbx +; SSE-NEXT: orq %r9, %rbx +; SSE-NEXT: orq %rdx, %rbx +; SSE-NEXT: andq 8(%rdi), %rsi +; SSE-NEXT: orq %r8, %rsi +; SSE-NEXT: orq %rax, %rsi +; SSE-NEXT: orq %rbx, %rsi +; SSE-NEXT: setne %al +; SSE-NEXT: popq %rbx +; SSE-NEXT: popq %r14 +; SSE-NEXT: popq %r15 +; SSE-NEXT: retq +; +; AVX2-LABEL: test_ne_i512: +; AVX2: # %bb.0: +; AVX2-NEXT: pushq %r15 +; AVX2-NEXT: pushq %r14 +; AVX2-NEXT: pushq %rbx +; AVX2-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovss {{.*#+}} xmm0 = [1,0,0,0] +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: movl %esi, %ecx +; AVX2-NEXT: andl $63, %ecx +; AVX2-NEXT: shrl $3, %esi +; AVX2-NEXT: andl $56, %esi +; AVX2-NEXT: negl %esi +; AVX2-NEXT: movslq %esi, %rsi +; AVX2-NEXT: movq -48(%rsp,%rsi), %rdx +; AVX2-NEXT: movq -40(%rsp,%rsi), %rbx +; AVX2-NEXT: movq %rbx, %rax +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq -16(%rsp,%rsi), %r11 +; AVX2-NEXT: movq -8(%rsp,%rsi), %r10 +; AVX2-NEXT: shldq %cl, %r11, %r10 +; AVX2-NEXT: movq -32(%rsp,%rsi), %r9 +; AVX2-NEXT: movq -24(%rsp,%rsi), %r14 +; AVX2-NEXT: movq %r14, %r8 +; AVX2-NEXT: shldq %cl, %r9, %r8 +; AVX2-NEXT: movq -64(%rsp,%rsi), %r15 +; AVX2-NEXT: movq -56(%rsp,%rsi), %rsi +; AVX2-NEXT: shldq %cl, %rsi, %rdx +; AVX2-NEXT: shldq %cl, %r14, %r11 +; AVX2-NEXT: shldq %cl, %rbx, %r9 +; AVX2-NEXT: shldq %cl, %r15, %rsi +; AVX2-NEXT: shlxq %rcx, %r15, %rcx +; AVX2-NEXT: andq 32(%rdi), %r9 +; AVX2-NEXT: andq 48(%rdi), %r11 +; AVX2-NEXT: andq 16(%rdi), %rdx +; AVX2-NEXT: andq 40(%rdi), %r8 +; AVX2-NEXT: andq 56(%rdi), %r10 +; AVX2-NEXT: andq 24(%rdi), %rax +; AVX2-NEXT: orq %r11, %rdx +; AVX2-NEXT: orq %r10, %rax +; AVX2-NEXT: andq (%rdi), %rcx +; AVX2-NEXT: orq %r9, %rcx +; AVX2-NEXT: orq %rdx, %rcx +; AVX2-NEXT: andq 8(%rdi), %rsi +; AVX2-NEXT: orq %r8, %rsi +; AVX2-NEXT: orq %rax, %rsi +; AVX2-NEXT: orq %rcx, %rsi +; AVX2-NEXT: setne %al +; AVX2-NEXT: popq %rbx +; AVX2-NEXT: popq %r14 +; AVX2-NEXT: popq %r15 +; AVX2-NEXT: vzeroupper +; AVX2-NEXT: retq +; +; AVX512-LABEL: test_ne_i512: +; AVX512: # %bb.0: +; AVX512-NEXT: pushq %r15 +; AVX512-NEXT: pushq %r14 +; AVX512-NEXT: pushq %rbx +; AVX512-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [1,0,0,0] +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: movl %esi, %ecx +; AVX512-NEXT: andl $63, %ecx +; AVX512-NEXT: shrl $3, %esi +; AVX512-NEXT: andl $56, %esi +; AVX512-NEXT: negl %esi +; AVX512-NEXT: movslq %esi, %rbx +; AVX512-NEXT: movq -48(%rsp,%rbx), %rdx +; AVX512-NEXT: movq -40(%rsp,%rbx), %r14 +; AVX512-NEXT: movq %r14, %rax +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq -16(%rsp,%rbx), %r11 +; AVX512-NEXT: movq -8(%rsp,%rbx), %r10 +; AVX512-NEXT: shldq %cl, %r11, %r10 +; AVX512-NEXT: movq -32(%rsp,%rbx), %r9 +; AVX512-NEXT: movq -24(%rsp,%rbx), %r15 +; AVX512-NEXT: movq %r15, %r8 +; AVX512-NEXT: shldq %cl, %r9, %r8 +; AVX512-NEXT: movq -56(%rsp,%rbx), %rsi +; AVX512-NEXT: shldq %cl, %rsi, %rdx +; AVX512-NEXT: shldq %cl, %r15, %r11 +; AVX512-NEXT: shldq %cl, %r14, %r9 +; AVX512-NEXT: movq -64(%rsp,%rbx), %rbx +; AVX512-NEXT: shldq %cl, %rbx, %rsi +; AVX512-NEXT: shlxq %rcx, %rbx, %rcx +; AVX512-NEXT: andq 32(%rdi), %r9 +; AVX512-NEXT: andq 48(%rdi), %r11 +; AVX512-NEXT: andq 16(%rdi), %rdx +; AVX512-NEXT: andq 40(%rdi), %r8 +; AVX512-NEXT: andq 56(%rdi), %r10 +; AVX512-NEXT: andq 24(%rdi), %rax +; AVX512-NEXT: orq %r11, %rdx +; AVX512-NEXT: orq %r10, %rax +; AVX512-NEXT: andq (%rdi), %rcx +; AVX512-NEXT: orq %r9, %rcx +; AVX512-NEXT: orq %rdx, %rcx +; AVX512-NEXT: andq 8(%rdi), %rsi +; AVX512-NEXT: orq %r8, %rsi +; AVX512-NEXT: orq %rax, %rsi +; AVX512-NEXT: orq %rcx, %rsi +; AVX512-NEXT: setne %al +; AVX512-NEXT: popq %rbx +; AVX512-NEXT: popq %r14 +; AVX512-NEXT: popq %r15 +; AVX512-NEXT: vzeroupper +; AVX512-NEXT: retq %rem = and i32 %position, 511 %ofs = zext nneg i32 %rem to i512 %bit = shl nuw i512 1, %ofs @@ -679,33 +1602,572 @@ define i1 @test_ne_i512(ptr %word, i32 %position) nounwind { define i1 @complement_ne_i512(ptr %word, i32 %position) nounwind { ; X86-LABEL: complement_ne_i512: ; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: pushl %ebx ; X86-NEXT: pushl %edi ; X86-NEXT: pushl %esi -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X86-NEXT: movl {{[0-9]+}}(%esp), %edx -; X86-NEXT: movl %edx, %esi -; X86-NEXT: shrl $3, %esi -; X86-NEXT: andl $60, %esi -; X86-NEXT: movl (%ecx,%esi), %edi -; X86-NEXT: btl %edx, %edi -; X86-NEXT: setb %al -; X86-NEXT: btcl %edx, %edi -; X86-NEXT: movl %edi, (%ecx,%esi) +; X86-NEXT: andl $-16, %esp +; X86-NEXT: subl $272, %esp # imm = 0x110 +; X86-NEXT: movl 12(%ebp), %ecx +; X86-NEXT: movl %ecx, %eax +; X86-NEXT: shrl $3, %eax +; X86-NEXT: andl $60, %eax +; X86-NEXT: movl %eax, (%esp) # 4-byte Spill +; X86-NEXT: leal {{[0-9]+}}(%esp), %edx +; X86-NEXT: subl %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $1, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl 24(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 28(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl $31, %ecx +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 56(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 60(%edx), %esi +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 12(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 40(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 44(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 16(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 20(%edx), %ebx +; X86-NEXT: movl %ebx, %esi +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 32(%edx), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 36(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edi, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl 52(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: shldl %cl, %esi, %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 4(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%ebp), %edx +; X86-NEXT: movl 40(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %esi, %eax +; X86-NEXT: movl 8(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ebx, %esi +; X86-NEXT: orl %eax, %esi +; X86-NEXT: movl %edx, %eax +; X86-NEXT: movl 56(%edx), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edi, %ebx +; X86-NEXT: movl 24(%edx), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: orl %ebx, %edx +; X86-NEXT: orl %esi, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 44(%eax), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: movl 12(%eax), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Folded Reload +; X86-NEXT: orl %esi, %ebx +; X86-NEXT: movl 60(%eax), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl %eax, %esi +; X86-NEXT: movl %edx, %eax +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl 28(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: orl %eax, %edx +; X86-NEXT: orl %ebx, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl (%eax), %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl (%esp), %eax # 4-byte Reload +; X86-NEXT: negl %eax +; X86-NEXT: movl 240(%esp,%eax), %esi +; X86-NEXT: shldl %cl, %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: shldl %cl, %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, (%esp) # 4-byte Spill +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: shll %cl, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%ebp), %esi +; X86-NEXT: movl 32(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edi, %eax +; X86-NEXT: movl (%esi), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edx, %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl 16(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ebx, %eax +; X86-NEXT: movl 48(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl (%esp), %edx # 4-byte Folded Reload +; X86-NEXT: orl %eax, %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl 36(%esi), %ebx +; X86-NEXT: movl %ebx, %eax +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl 4(%esi), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl %esi, %eax +; X86-NEXT: movl 20(%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl %esi, %edi +; X86-NEXT: movl 52(%eax), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: orl %edi, %eax +; X86-NEXT: orl %ecx, %eax +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: xorl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: xorl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: xorl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: xorl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: xorl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: xorl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: xorl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: xorl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: xorl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: xorl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: xorl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: xorl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: xorl %ecx, (%esp) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: xorl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: xorl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: xorl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload +; X86-NEXT: orl %edx, %eax +; X86-NEXT: movl 8(%ebp), %edx +; X86-NEXT: movl %ebx, 60(%edx) +; X86-NEXT: movl %edi, 56(%edx) +; X86-NEXT: movl %ecx, 52(%edx) +; X86-NEXT: movl %esi, 44(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 40(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 36(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 32(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 28(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 24(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 20(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 16(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 12(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 8(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 4(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, (%edx) +; X86-NEXT: movl (%esp), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 48(%edx) +; X86-NEXT: setne %al +; X86-NEXT: leal -12(%ebp), %esp ; X86-NEXT: popl %esi ; X86-NEXT: popl %edi +; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; -; X64-LABEL: complement_ne_i512: -; X64: # %bb.0: -; X64-NEXT: movl %esi, %ecx -; X64-NEXT: shrl $3, %ecx -; X64-NEXT: andl $60, %ecx -; X64-NEXT: movl (%rdi,%rcx), %edx -; X64-NEXT: btl %esi, %edx -; X64-NEXT: setb %al -; X64-NEXT: btcl %esi, %edx -; X64-NEXT: movl %edx, (%rdi,%rcx) -; X64-NEXT: retq +; SSE-LABEL: complement_ne_i512: +; SSE: # %bb.0: +; SSE-NEXT: pushq %rbp +; SSE-NEXT: pushq %r15 +; SSE-NEXT: pushq %r14 +; SSE-NEXT: pushq %r13 +; SSE-NEXT: pushq %r12 +; SSE-NEXT: pushq %rbx +; SSE-NEXT: subq $56, %rsp +; SSE-NEXT: xorps %xmm0, %xmm0 +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movq $0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movq $1, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movl %esi, %ecx +; SSE-NEXT: andl $63, %ecx +; SSE-NEXT: shrl $3, %esi +; SSE-NEXT: andl $56, %esi +; SSE-NEXT: negl %esi +; SSE-NEXT: movslq %esi, %rbx +; SSE-NEXT: movq (%rsp,%rbx), %rsi +; SSE-NEXT: movq 8(%rsp,%rbx), %r14 +; SSE-NEXT: movq %r14, %rax +; SSE-NEXT: shldq %cl, %rsi, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 32(%rsp,%rbx), %r8 +; SSE-NEXT: movq 40(%rsp,%rbx), %rbp +; SSE-NEXT: shldq %cl, %r8, %rbp +; SSE-NEXT: movq 16(%rsp,%rbx), %r9 +; SSE-NEXT: movq 24(%rsp,%rbx), %r15 +; SSE-NEXT: movq %r15, %r10 +; SSE-NEXT: shldq %cl, %r9, %r10 +; SSE-NEXT: movq -8(%rsp,%rbx), %r11 +; SSE-NEXT: shldq %cl, %r11, %rsi +; SSE-NEXT: shldq %cl, %r15, %r8 +; SSE-NEXT: shldq %cl, %r14, %r9 +; SSE-NEXT: movq -16(%rsp,%rbx), %rbx +; SSE-NEXT: shldq %cl, %rbx, %r11 +; SSE-NEXT: # kill: def $cl killed $cl killed $ecx +; SSE-NEXT: shlq %cl, %rbx +; SSE-NEXT: movq 24(%rdi), %r15 +; SSE-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 56(%rdi), %rcx +; SSE-NEXT: movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 16(%rdi), %r12 +; SSE-NEXT: movq %r12, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 48(%rdi), %r13 +; SSE-NEXT: movq %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %r8, %r13 +; SSE-NEXT: andq %rsi, %r12 +; SSE-NEXT: orq %r13, %r12 +; SSE-NEXT: movq %rcx, %r13 +; SSE-NEXT: andq %rbp, %r13 +; SSE-NEXT: andq %rax, %r15 +; SSE-NEXT: orq %r13, %r15 +; SSE-NEXT: movq 32(%rdi), %r14 +; SSE-NEXT: movq %r14, %rcx +; SSE-NEXT: andq %r9, %rcx +; SSE-NEXT: movq (%rdi), %r13 +; SSE-NEXT: movq %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %rbx, %r13 +; SSE-NEXT: orq %rcx, %r13 +; SSE-NEXT: orq %r12, %r13 +; SSE-NEXT: movq 40(%rdi), %rcx +; SSE-NEXT: movq %rcx, %r12 +; SSE-NEXT: andq %r10, %r12 +; SSE-NEXT: movq 8(%rdi), %rdx +; SSE-NEXT: movq %rdx, %rax +; SSE-NEXT: andq %r11, %rax +; SSE-NEXT: orq %r12, %rax +; SSE-NEXT: orq %r15, %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; SSE-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Folded Reload +; SSE-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %rsi # 8-byte Folded Reload +; SSE-NEXT: xorq %rcx, %r10 +; SSE-NEXT: xorq %r14, %r9 +; SSE-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Folded Reload +; SSE-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Folded Reload +; SSE-NEXT: xorq %rdx, %r11 +; SSE-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Folded Reload +; SSE-NEXT: orq %r13, %rax +; SSE-NEXT: movq %r8, 48(%rdi) +; SSE-NEXT: movq %rbp, 56(%rdi) +; SSE-NEXT: movq %r9, 32(%rdi) +; SSE-NEXT: movq %r10, 40(%rdi) +; SSE-NEXT: movq %rsi, 16(%rdi) +; SSE-NEXT: movq %r15, 24(%rdi) +; SSE-NEXT: movq %rbx, (%rdi) +; SSE-NEXT: movq %r11, 8(%rdi) +; SSE-NEXT: setne %al +; SSE-NEXT: addq $56, %rsp +; SSE-NEXT: popq %rbx +; SSE-NEXT: popq %r12 +; SSE-NEXT: popq %r13 +; SSE-NEXT: popq %r14 +; SSE-NEXT: popq %r15 +; SSE-NEXT: popq %rbp +; SSE-NEXT: retq +; +; AVX2-LABEL: complement_ne_i512: +; AVX2: # %bb.0: +; AVX2-NEXT: pushq %rbp +; AVX2-NEXT: pushq %r15 +; AVX2-NEXT: pushq %r14 +; AVX2-NEXT: pushq %r13 +; AVX2-NEXT: pushq %r12 +; AVX2-NEXT: pushq %rbx +; AVX2-NEXT: subq $72, %rsp +; AVX2-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovss {{.*#+}} xmm0 = [1,0,0,0] +; AVX2-NEXT: vmovups %ymm0, (%rsp) +; AVX2-NEXT: movl %esi, %ecx +; AVX2-NEXT: andl $63, %ecx +; AVX2-NEXT: shrl $3, %esi +; AVX2-NEXT: andl $56, %esi +; AVX2-NEXT: negl %esi +; AVX2-NEXT: movslq %esi, %rbx +; AVX2-NEXT: movq 16(%rsp,%rbx), %rsi +; AVX2-NEXT: movq 24(%rsp,%rbx), %rbp +; AVX2-NEXT: movq %rbp, %rax +; AVX2-NEXT: shldq %cl, %rsi, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 48(%rsp,%rbx), %r8 +; AVX2-NEXT: movq 56(%rsp,%rbx), %r13 +; AVX2-NEXT: shldq %cl, %r8, %r13 +; AVX2-NEXT: movq 32(%rsp,%rbx), %r9 +; AVX2-NEXT: movq 40(%rsp,%rbx), %r14 +; AVX2-NEXT: movq %r14, %r10 +; AVX2-NEXT: shldq %cl, %r9, %r10 +; AVX2-NEXT: movq 8(%rsp,%rbx), %r11 +; AVX2-NEXT: shldq %cl, %r11, %rsi +; AVX2-NEXT: shldq %cl, %r14, %r8 +; AVX2-NEXT: movq 16(%rdi), %r12 +; AVX2-NEXT: movq %r12, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 48(%rdi), %r14 +; AVX2-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: andq %r8, %r14 +; AVX2-NEXT: andq %rsi, %r12 +; AVX2-NEXT: orq %r14, %r12 +; AVX2-NEXT: movq 56(%rdi), %r15 +; AVX2-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: andq %r13, %r15 +; AVX2-NEXT: movq 24(%rdi), %r14 +; AVX2-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: andq %rax, %r14 +; AVX2-NEXT: orq %r15, %r14 +; AVX2-NEXT: shldq %cl, %rbp, %r9 +; AVX2-NEXT: movq (%rsp,%rbx), %rdx +; AVX2-NEXT: movq 32(%rdi), %r15 +; AVX2-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: andq %r9, %r15 +; AVX2-NEXT: shlxq %rcx, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq (%rdi), %rbx +; AVX2-NEXT: movq %rbx, %rbp +; AVX2-NEXT: andq %rax, %rbp +; AVX2-NEXT: orq %r15, %rbp +; AVX2-NEXT: orq %r12, %rbp +; AVX2-NEXT: # kill: def $cl killed $cl killed $rcx +; AVX2-NEXT: shldq %cl, %rdx, %r11 +; AVX2-NEXT: movq 40(%rdi), %rax +; AVX2-NEXT: movq %rax, %rcx +; AVX2-NEXT: andq %r10, %rcx +; AVX2-NEXT: movq 8(%rdi), %r15 +; AVX2-NEXT: movq %r15, %r12 +; AVX2-NEXT: andq %r11, %r12 +; AVX2-NEXT: orq %rcx, %r12 +; AVX2-NEXT: orq %r14, %r12 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload +; AVX2-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Folded Reload +; AVX2-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %rsi # 8-byte Folded Reload +; AVX2-NEXT: xorq %rax, %r10 +; AVX2-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Folded Reload +; AVX2-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Folded Reload +; AVX2-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Folded Reload +; AVX2-NEXT: xorq %r15, %r11 +; AVX2-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Folded Reload +; AVX2-NEXT: orq %rbp, %r12 +; AVX2-NEXT: movq %r8, 48(%rdi) +; AVX2-NEXT: movq %r13, 56(%rdi) +; AVX2-NEXT: movq %r9, 32(%rdi) +; AVX2-NEXT: movq %r10, 40(%rdi) +; AVX2-NEXT: movq %rsi, 16(%rdi) +; AVX2-NEXT: movq %rcx, 24(%rdi) +; AVX2-NEXT: movq %rbx, (%rdi) +; AVX2-NEXT: movq %r11, 8(%rdi) +; AVX2-NEXT: setne %al +; AVX2-NEXT: addq $72, %rsp +; AVX2-NEXT: popq %rbx +; AVX2-NEXT: popq %r12 +; AVX2-NEXT: popq %r13 +; AVX2-NEXT: popq %r14 +; AVX2-NEXT: popq %r15 +; AVX2-NEXT: popq %rbp +; AVX2-NEXT: vzeroupper +; AVX2-NEXT: retq +; +; AVX512-LABEL: complement_ne_i512: +; AVX512: # %bb.0: +; AVX512-NEXT: pushq %rbp +; AVX512-NEXT: pushq %r15 +; AVX512-NEXT: pushq %r14 +; AVX512-NEXT: pushq %r13 +; AVX512-NEXT: pushq %r12 +; AVX512-NEXT: pushq %rbx +; AVX512-NEXT: subq $72, %rsp +; AVX512-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [1,0,0,0] +; AVX512-NEXT: vmovups %ymm0, (%rsp) +; AVX512-NEXT: movl %esi, %ecx +; AVX512-NEXT: andl $63, %ecx +; AVX512-NEXT: shrl $3, %esi +; AVX512-NEXT: andl $56, %esi +; AVX512-NEXT: negl %esi +; AVX512-NEXT: movslq %esi, %rbx +; AVX512-NEXT: movq 16(%rsp,%rbx), %rsi +; AVX512-NEXT: movq 24(%rsp,%rbx), %rbp +; AVX512-NEXT: movq %rbp, %rax +; AVX512-NEXT: shldq %cl, %rsi, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 48(%rsp,%rbx), %r8 +; AVX512-NEXT: movq 56(%rsp,%rbx), %r13 +; AVX512-NEXT: shldq %cl, %r8, %r13 +; AVX512-NEXT: movq 32(%rsp,%rbx), %r9 +; AVX512-NEXT: movq 40(%rsp,%rbx), %r14 +; AVX512-NEXT: movq %r14, %r10 +; AVX512-NEXT: shldq %cl, %r9, %r10 +; AVX512-NEXT: movq 8(%rsp,%rbx), %r11 +; AVX512-NEXT: shldq %cl, %r11, %rsi +; AVX512-NEXT: shldq %cl, %r14, %r8 +; AVX512-NEXT: movq 16(%rdi), %r12 +; AVX512-NEXT: movq %r12, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 48(%rdi), %r14 +; AVX512-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: andq %r8, %r14 +; AVX512-NEXT: andq %rsi, %r12 +; AVX512-NEXT: orq %r14, %r12 +; AVX512-NEXT: movq 56(%rdi), %r15 +; AVX512-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: andq %r13, %r15 +; AVX512-NEXT: movq 24(%rdi), %r14 +; AVX512-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: andq %rax, %r14 +; AVX512-NEXT: orq %r15, %r14 +; AVX512-NEXT: shldq %cl, %rbp, %r9 +; AVX512-NEXT: movq (%rsp,%rbx), %rdx +; AVX512-NEXT: movq 32(%rdi), %r15 +; AVX512-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: andq %r9, %r15 +; AVX512-NEXT: shlxq %rcx, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq (%rdi), %rbx +; AVX512-NEXT: movq %rbx, %rbp +; AVX512-NEXT: andq %rax, %rbp +; AVX512-NEXT: orq %r15, %rbp +; AVX512-NEXT: orq %r12, %rbp +; AVX512-NEXT: # kill: def $cl killed $cl killed $rcx +; AVX512-NEXT: shldq %cl, %rdx, %r11 +; AVX512-NEXT: movq 40(%rdi), %rax +; AVX512-NEXT: movq %rax, %rcx +; AVX512-NEXT: andq %r10, %rcx +; AVX512-NEXT: movq 8(%rdi), %r15 +; AVX512-NEXT: movq %r15, %r12 +; AVX512-NEXT: andq %r11, %r12 +; AVX512-NEXT: orq %rcx, %r12 +; AVX512-NEXT: orq %r14, %r12 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload +; AVX512-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Folded Reload +; AVX512-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %rsi # 8-byte Folded Reload +; AVX512-NEXT: xorq %rax, %r10 +; AVX512-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Folded Reload +; AVX512-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Folded Reload +; AVX512-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Folded Reload +; AVX512-NEXT: xorq %r15, %r11 +; AVX512-NEXT: xorq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Folded Reload +; AVX512-NEXT: orq %rbp, %r12 +; AVX512-NEXT: movq %r8, 48(%rdi) +; AVX512-NEXT: movq %r13, 56(%rdi) +; AVX512-NEXT: movq %r9, 32(%rdi) +; AVX512-NEXT: movq %r10, 40(%rdi) +; AVX512-NEXT: movq %rsi, 16(%rdi) +; AVX512-NEXT: movq %rcx, 24(%rdi) +; AVX512-NEXT: movq %rbx, (%rdi) +; AVX512-NEXT: movq %r11, 8(%rdi) +; AVX512-NEXT: setne %al +; AVX512-NEXT: addq $72, %rsp +; AVX512-NEXT: popq %rbx +; AVX512-NEXT: popq %r12 +; AVX512-NEXT: popq %r13 +; AVX512-NEXT: popq %r14 +; AVX512-NEXT: popq %r15 +; AVX512-NEXT: popq %rbp +; AVX512-NEXT: vzeroupper +; AVX512-NEXT: retq %rem = and i32 %position, 511 %ofs = zext nneg i32 %rem to i512 %bit = shl nuw i512 1, %ofs @@ -720,33 +2182,606 @@ define i1 @complement_ne_i512(ptr %word, i32 %position) nounwind { define i1 @reset_eq_i512(ptr %word, i32 %position) nounwind { ; X86-LABEL: reset_eq_i512: ; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: pushl %ebx ; X86-NEXT: pushl %edi ; X86-NEXT: pushl %esi -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X86-NEXT: movl {{[0-9]+}}(%esp), %edx +; X86-NEXT: andl $-16, %esp +; X86-NEXT: subl $288, %esp # imm = 0x120 +; X86-NEXT: movl 12(%ebp), %ecx +; X86-NEXT: movl %ecx, %eax +; X86-NEXT: shrl $3, %eax +; X86-NEXT: andl $60, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: leal {{[0-9]+}}(%esp), %edi +; X86-NEXT: subl %eax, %edi +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $1, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl 4(%edi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%edi), %eax +; X86-NEXT: andl $31, %ecx +; X86-NEXT: movl %eax, %ebx +; X86-NEXT: shldl %cl, %edx, %ebx +; X86-NEXT: movl 12(%edi), %edx ; X86-NEXT: movl %edx, %esi -; X86-NEXT: shrl $3, %esi -; X86-NEXT: andl $60, %esi -; X86-NEXT: movl (%ecx,%esi), %edi -; X86-NEXT: btl %edx, %edi -; X86-NEXT: setae %al -; X86-NEXT: btrl %edx, %edi -; X86-NEXT: movl %edi, (%ecx,%esi) +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 16(%edi), %eax +; X86-NEXT: movl %eax, %esi +; X86-NEXT: shldl %cl, %edx, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 20(%edi), %edx +; X86-NEXT: movl %edx, %esi +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 24(%edi), %eax +; X86-NEXT: movl %eax, %esi +; X86-NEXT: shldl %cl, %edx, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 28(%edi), %edx +; X86-NEXT: movl %edx, %esi +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 32(%edi), %eax +; X86-NEXT: movl %eax, %esi +; X86-NEXT: shldl %cl, %edx, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 36(%edi), %esi +; X86-NEXT: movl %esi, %edx +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 40(%edi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %esi, %edx +; X86-NEXT: movl 8(%ebp), %esi +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 40(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %eax, %edx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %eax, %ebx +; X86-NEXT: orl %edx, %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 44(%edi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: movl 52(%edi), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 56(%edi), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: shldl %cl, %esi, %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%ebp), %esi +; X86-NEXT: movl 56(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %eax, %ebx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 24(%esi), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ecx, %eax +; X86-NEXT: orl %ebx, %eax +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl %esi, %ebx +; X86-NEXT: movl 44(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %eax, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 12(%esi), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ecx, %eax +; X86-NEXT: orl %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 60(%edi), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: shldl %cl, %esi, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 60(%ebx), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edx, %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 28(%ebx), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edx, %esi +; X86-NEXT: orl %eax, %esi +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl (%edi), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: negl %eax +; X86-NEXT: movl 256(%esp,%eax), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: shldl %cl, %edi, %eax +; X86-NEXT: movl %esi, %edi +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: shll %cl, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl %ebx, %esi +; X86-NEXT: movl 32(%ebx), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ecx, %edx +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl (%ebx), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ecx, %edi +; X86-NEXT: orl %edx, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 16(%esi), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ecx, %ebx +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 48(%esi), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ecx, %eax +; X86-NEXT: orl %ebx, %eax +; X86-NEXT: orl %edi, %eax +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: shldl %cl, %edx, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 36(%esi), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ebx, %edx +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 4(%esi), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ebx, %edi +; X86-NEXT: orl %edx, %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: shldl %cl, %edi, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 20(%esi), %edi +; X86-NEXT: andl %edi, %ecx +; X86-NEXT: movl %ecx, %esi +; X86-NEXT: movl %edx, %ecx +; X86-NEXT: movl 8(%ebp), %ebx +; X86-NEXT: movl 52(%ebx), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ebx, %edx +; X86-NEXT: orl %esi, %edx +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: notl %ebx +; X86-NEXT: andl %edi, %ebx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: notl %esi +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: notl %edi +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: notl %edi +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: notl %edi +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: notl %ecx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload +; X86-NEXT: orl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl 8(%ebp), %eax +; X86-NEXT: movl %edx, 60(%eax) +; X86-NEXT: movl %esi, 56(%eax) +; X86-NEXT: movl %ecx, 52(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl %ecx, 44(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl %ecx, 40(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl %ecx, 36(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl %ecx, 32(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl %ecx, 28(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl %ecx, 24(%eax) +; X86-NEXT: movl %ebx, 20(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl %ecx, 16(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl %ecx, 12(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl %ecx, 8(%eax) +; X86-NEXT: movl %edi, 4(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl %ecx, (%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl %ecx, 48(%eax) +; X86-NEXT: sete %al +; X86-NEXT: leal -12(%ebp), %esp ; X86-NEXT: popl %esi ; X86-NEXT: popl %edi +; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; -; X64-LABEL: reset_eq_i512: -; X64: # %bb.0: -; X64-NEXT: movl %esi, %ecx -; X64-NEXT: shrl $3, %ecx -; X64-NEXT: andl $60, %ecx -; X64-NEXT: movl (%rdi,%rcx), %edx -; X64-NEXT: btl %esi, %edx -; X64-NEXT: setae %al -; X64-NEXT: btrl %esi, %edx -; X64-NEXT: movl %edx, (%rdi,%rcx) -; X64-NEXT: retq +; SSE-LABEL: reset_eq_i512: +; SSE: # %bb.0: +; SSE-NEXT: pushq %rbp +; SSE-NEXT: pushq %r15 +; SSE-NEXT: pushq %r14 +; SSE-NEXT: pushq %r13 +; SSE-NEXT: pushq %r12 +; SSE-NEXT: pushq %rbx +; SSE-NEXT: subq $56, %rsp +; SSE-NEXT: xorps %xmm0, %xmm0 +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movq $0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movq $1, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movl %esi, %ecx +; SSE-NEXT: andl $63, %ecx +; SSE-NEXT: shrl $3, %esi +; SSE-NEXT: andl $56, %esi +; SSE-NEXT: negl %esi +; SSE-NEXT: movslq %esi, %rdx +; SSE-NEXT: movq (%rsp,%rdx), %r9 +; SSE-NEXT: movq 8(%rsp,%rdx), %r8 +; SSE-NEXT: movq %r8, %rsi +; SSE-NEXT: shldq %cl, %r9, %rsi +; SSE-NEXT: movq -8(%rsp,%rdx), %rax +; SSE-NEXT: shldq %cl, %rax, %r9 +; SSE-NEXT: movq 16(%rsp,%rdx), %r14 +; SSE-NEXT: movq 24(%rsp,%rdx), %r10 +; SSE-NEXT: movq %r10, %rbx +; SSE-NEXT: shldq %cl, %r14, %rbx +; SSE-NEXT: shldq %cl, %r8, %r14 +; SSE-NEXT: movq 32(%rsp,%rdx), %r13 +; SSE-NEXT: movq 40(%rsp,%rdx), %r12 +; SSE-NEXT: shldq %cl, %r13, %r12 +; SSE-NEXT: shldq %cl, %r10, %r13 +; SSE-NEXT: movq -16(%rsp,%rdx), %rdx +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: # kill: def $cl killed $cl killed $ecx +; SSE-NEXT: shlq %cl, %rdx +; SSE-NEXT: movq %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq %r12, %rbp +; SSE-NEXT: movq %r9, %r15 +; SSE-NEXT: movq %rsi, %r11 +; SSE-NEXT: movq 16(%rdi), %r8 +; SSE-NEXT: movq %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 48(%rdi), %rcx +; SSE-NEXT: movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %rcx, %r13 +; SSE-NEXT: andq %r8, %r9 +; SSE-NEXT: orq %r13, %r9 +; SSE-NEXT: movq 56(%rdi), %rcx +; SSE-NEXT: movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %rcx, %r12 +; SSE-NEXT: movq 24(%rdi), %r10 +; SSE-NEXT: andq %r10, %rsi +; SSE-NEXT: orq %r12, %rsi +; SSE-NEXT: movq %r14, %r13 +; SSE-NEXT: movq 32(%rdi), %rcx +; SSE-NEXT: movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %rcx, %r14 +; SSE-NEXT: movq %rdx, %r12 +; SSE-NEXT: movq (%rdi), %rcx +; SSE-NEXT: movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %rcx, %rdx +; SSE-NEXT: orq %r14, %rdx +; SSE-NEXT: orq %r9, %rdx +; SSE-NEXT: movq %rbx, %r14 +; SSE-NEXT: movq 40(%rdi), %rcx +; SSE-NEXT: andq %rcx, %rbx +; SSE-NEXT: movq %rax, %r9 +; SSE-NEXT: movq 8(%rdi), %r8 +; SSE-NEXT: andq %r8, %rax +; SSE-NEXT: orq %rbx, %rax +; SSE-NEXT: orq %rsi, %rax +; SSE-NEXT: notq %r11 +; SSE-NEXT: andq %r10, %r11 +; SSE-NEXT: notq %r15 +; SSE-NEXT: andq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Folded Reload +; SSE-NEXT: notq %r14 +; SSE-NEXT: andq %rcx, %r14 +; SSE-NEXT: notq %r13 +; SSE-NEXT: andq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Folded Reload +; SSE-NEXT: notq %rbp +; SSE-NEXT: andq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Folded Reload +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload +; SSE-NEXT: notq %rcx +; SSE-NEXT: andq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Folded Reload +; SSE-NEXT: notq %r9 +; SSE-NEXT: andq %r8, %r9 +; SSE-NEXT: notq %r12 +; SSE-NEXT: andq {{[-0-9]+}}(%r{{[sb]}}p), %r12 # 8-byte Folded Reload +; SSE-NEXT: orq %rdx, %rax +; SSE-NEXT: movq %rcx, 48(%rdi) +; SSE-NEXT: movq %rbp, 56(%rdi) +; SSE-NEXT: movq %r13, 32(%rdi) +; SSE-NEXT: movq %r14, 40(%rdi) +; SSE-NEXT: movq %r15, 16(%rdi) +; SSE-NEXT: movq %r11, 24(%rdi) +; SSE-NEXT: movq %r12, (%rdi) +; SSE-NEXT: movq %r9, 8(%rdi) +; SSE-NEXT: sete %al +; SSE-NEXT: addq $56, %rsp +; SSE-NEXT: popq %rbx +; SSE-NEXT: popq %r12 +; SSE-NEXT: popq %r13 +; SSE-NEXT: popq %r14 +; SSE-NEXT: popq %r15 +; SSE-NEXT: popq %rbp +; SSE-NEXT: retq +; +; AVX2-LABEL: reset_eq_i512: +; AVX2: # %bb.0: +; AVX2-NEXT: pushq %rbp +; AVX2-NEXT: pushq %r15 +; AVX2-NEXT: pushq %r14 +; AVX2-NEXT: pushq %r13 +; AVX2-NEXT: pushq %r12 +; AVX2-NEXT: pushq %rbx +; AVX2-NEXT: pushq %rax +; AVX2-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovss {{.*#+}} xmm0 = [1,0,0,0] +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: movl %esi, %ecx +; AVX2-NEXT: andl $63, %ecx +; AVX2-NEXT: shrl $3, %esi +; AVX2-NEXT: andl $56, %esi +; AVX2-NEXT: negl %esi +; AVX2-NEXT: movslq %esi, %rdx +; AVX2-NEXT: movq -48(%rsp,%rdx), %r8 +; AVX2-NEXT: movq -40(%rsp,%rdx), %rbx +; AVX2-NEXT: movq %rbx, %rax +; AVX2-NEXT: shldq %cl, %r8, %rax +; AVX2-NEXT: movq -16(%rsp,%rdx), %r10 +; AVX2-NEXT: movq -8(%rsp,%rdx), %rsi +; AVX2-NEXT: shldq %cl, %r10, %rsi +; AVX2-NEXT: movq -32(%rsp,%rdx), %r11 +; AVX2-NEXT: movq -24(%rsp,%rdx), %r14 +; AVX2-NEXT: movq %r14, %r9 +; AVX2-NEXT: shldq %cl, %r11, %r9 +; AVX2-NEXT: movq -64(%rsp,%rdx), %r15 +; AVX2-NEXT: movq -56(%rsp,%rdx), %rdx +; AVX2-NEXT: shldq %cl, %rdx, %r8 +; AVX2-NEXT: shldq %cl, %r14, %r10 +; AVX2-NEXT: shldq %cl, %rbx, %r11 +; AVX2-NEXT: shldq %cl, %r15, %rdx +; AVX2-NEXT: shlxq %rcx, %r15, %rcx +; AVX2-NEXT: movq 24(%rdi), %rbx +; AVX2-NEXT: movq 56(%rdi), %r14 +; AVX2-NEXT: movq 16(%rdi), %r15 +; AVX2-NEXT: movq 48(%rdi), %r13 +; AVX2-NEXT: movq 32(%rdi), %rbp +; AVX2-NEXT: andnq %rbp, %r11, %r12 +; AVX2-NEXT: andq %r11, %rbp +; AVX2-NEXT: andnq %r13, %r10, %r11 +; AVX2-NEXT: andq %r10, %r13 +; AVX2-NEXT: andnq %r15, %r8, %r10 +; AVX2-NEXT: andq %r8, %r15 +; AVX2-NEXT: movq 40(%rdi), %r8 +; AVX2-NEXT: orq %r13, %r15 +; AVX2-NEXT: andnq %r8, %r9, %r13 +; AVX2-NEXT: andq %r9, %r8 +; AVX2-NEXT: andnq %r14, %rsi, %r9 +; AVX2-NEXT: andq %rsi, %r14 +; AVX2-NEXT: andnq %rbx, %rax, %rsi +; AVX2-NEXT: andq %rax, %rbx +; AVX2-NEXT: movq (%rdi), %rax +; AVX2-NEXT: orq %r14, %rbx +; AVX2-NEXT: andnq %rax, %rcx, %r14 +; AVX2-NEXT: andq %rcx, %rax +; AVX2-NEXT: orq %rbp, %rax +; AVX2-NEXT: movq 8(%rdi), %rcx +; AVX2-NEXT: orq %r15, %rax +; AVX2-NEXT: andnq %rcx, %rdx, %r15 +; AVX2-NEXT: andq %rdx, %rcx +; AVX2-NEXT: orq %r8, %rcx +; AVX2-NEXT: orq %rbx, %rcx +; AVX2-NEXT: orq %rax, %rcx +; AVX2-NEXT: movq %r11, 48(%rdi) +; AVX2-NEXT: movq %r9, 56(%rdi) +; AVX2-NEXT: movq %r12, 32(%rdi) +; AVX2-NEXT: movq %r13, 40(%rdi) +; AVX2-NEXT: movq %r10, 16(%rdi) +; AVX2-NEXT: movq %rsi, 24(%rdi) +; AVX2-NEXT: movq %r14, (%rdi) +; AVX2-NEXT: movq %r15, 8(%rdi) +; AVX2-NEXT: sete %al +; AVX2-NEXT: addq $8, %rsp +; AVX2-NEXT: popq %rbx +; AVX2-NEXT: popq %r12 +; AVX2-NEXT: popq %r13 +; AVX2-NEXT: popq %r14 +; AVX2-NEXT: popq %r15 +; AVX2-NEXT: popq %rbp +; AVX2-NEXT: vzeroupper +; AVX2-NEXT: retq +; +; AVX512-LABEL: reset_eq_i512: +; AVX512: # %bb.0: +; AVX512-NEXT: pushq %rbp +; AVX512-NEXT: pushq %r15 +; AVX512-NEXT: pushq %r14 +; AVX512-NEXT: pushq %r13 +; AVX512-NEXT: pushq %r12 +; AVX512-NEXT: pushq %rbx +; AVX512-NEXT: pushq %rax +; AVX512-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [1,0,0,0] +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: movl %esi, %ecx +; AVX512-NEXT: andl $63, %ecx +; AVX512-NEXT: shrl $3, %esi +; AVX512-NEXT: andl $56, %esi +; AVX512-NEXT: negl %esi +; AVX512-NEXT: movslq %esi, %rbx +; AVX512-NEXT: movq -48(%rsp,%rbx), %r8 +; AVX512-NEXT: movq -40(%rsp,%rbx), %r14 +; AVX512-NEXT: movq %r14, %rax +; AVX512-NEXT: shldq %cl, %r8, %rax +; AVX512-NEXT: movq -16(%rsp,%rbx), %r10 +; AVX512-NEXT: movq -8(%rsp,%rbx), %rsi +; AVX512-NEXT: shldq %cl, %r10, %rsi +; AVX512-NEXT: movq -32(%rsp,%rbx), %r11 +; AVX512-NEXT: movq -24(%rsp,%rbx), %r15 +; AVX512-NEXT: movq %r15, %r9 +; AVX512-NEXT: shldq %cl, %r11, %r9 +; AVX512-NEXT: movq -56(%rsp,%rbx), %rdx +; AVX512-NEXT: shldq %cl, %rdx, %r8 +; AVX512-NEXT: shldq %cl, %r15, %r10 +; AVX512-NEXT: shldq %cl, %r14, %r11 +; AVX512-NEXT: movq -64(%rsp,%rbx), %rbx +; AVX512-NEXT: shldq %cl, %rbx, %rdx +; AVX512-NEXT: shlxq %rcx, %rbx, %rcx +; AVX512-NEXT: movq 24(%rdi), %rbx +; AVX512-NEXT: movq 56(%rdi), %r14 +; AVX512-NEXT: movq 16(%rdi), %r15 +; AVX512-NEXT: movq 48(%rdi), %r13 +; AVX512-NEXT: movq 32(%rdi), %rbp +; AVX512-NEXT: andnq %rbp, %r11, %r12 +; AVX512-NEXT: andq %r11, %rbp +; AVX512-NEXT: andnq %r13, %r10, %r11 +; AVX512-NEXT: andq %r10, %r13 +; AVX512-NEXT: andnq %r15, %r8, %r10 +; AVX512-NEXT: andq %r8, %r15 +; AVX512-NEXT: movq 40(%rdi), %r8 +; AVX512-NEXT: orq %r13, %r15 +; AVX512-NEXT: andnq %r8, %r9, %r13 +; AVX512-NEXT: andq %r9, %r8 +; AVX512-NEXT: andnq %r14, %rsi, %r9 +; AVX512-NEXT: andq %rsi, %r14 +; AVX512-NEXT: andnq %rbx, %rax, %rsi +; AVX512-NEXT: andq %rax, %rbx +; AVX512-NEXT: movq (%rdi), %rax +; AVX512-NEXT: orq %r14, %rbx +; AVX512-NEXT: andnq %rax, %rcx, %r14 +; AVX512-NEXT: andq %rcx, %rax +; AVX512-NEXT: orq %rbp, %rax +; AVX512-NEXT: movq 8(%rdi), %rcx +; AVX512-NEXT: orq %r15, %rax +; AVX512-NEXT: andnq %rcx, %rdx, %r15 +; AVX512-NEXT: andq %rdx, %rcx +; AVX512-NEXT: orq %r8, %rcx +; AVX512-NEXT: orq %rbx, %rcx +; AVX512-NEXT: orq %rax, %rcx +; AVX512-NEXT: movq %r11, 48(%rdi) +; AVX512-NEXT: movq %r9, 56(%rdi) +; AVX512-NEXT: movq %r12, 32(%rdi) +; AVX512-NEXT: movq %r13, 40(%rdi) +; AVX512-NEXT: movq %r10, 16(%rdi) +; AVX512-NEXT: movq %rsi, 24(%rdi) +; AVX512-NEXT: movq %r14, (%rdi) +; AVX512-NEXT: movq %r15, 8(%rdi) +; AVX512-NEXT: sete %al +; AVX512-NEXT: addq $8, %rsp +; AVX512-NEXT: popq %rbx +; AVX512-NEXT: popq %r12 +; AVX512-NEXT: popq %r13 +; AVX512-NEXT: popq %r14 +; AVX512-NEXT: popq %r15 +; AVX512-NEXT: popq %rbp +; AVX512-NEXT: vzeroupper +; AVX512-NEXT: retq %rem = and i32 %position, 511 %ofs = zext nneg i32 %rem to i512 %bit = shl nuw i512 1, %ofs @@ -762,33 +2797,572 @@ define i1 @reset_eq_i512(ptr %word, i32 %position) nounwind { define i1 @set_ne_i512(ptr %word, i32 %position) nounwind { ; X86-LABEL: set_ne_i512: ; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: pushl %ebx ; X86-NEXT: pushl %edi ; X86-NEXT: pushl %esi -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X86-NEXT: movl {{[0-9]+}}(%esp), %edx -; X86-NEXT: movl %edx, %esi -; X86-NEXT: shrl $3, %esi -; X86-NEXT: andl $60, %esi -; X86-NEXT: movl (%ecx,%esi), %edi -; X86-NEXT: btl %edx, %edi -; X86-NEXT: setb %al -; X86-NEXT: btsl %edx, %edi -; X86-NEXT: movl %edi, (%ecx,%esi) +; X86-NEXT: andl $-16, %esp +; X86-NEXT: subl $272, %esp # imm = 0x110 +; X86-NEXT: movl 12(%ebp), %ecx +; X86-NEXT: movl %ecx, %eax +; X86-NEXT: shrl $3, %eax +; X86-NEXT: andl $60, %eax +; X86-NEXT: movl %eax, (%esp) # 4-byte Spill +; X86-NEXT: leal {{[0-9]+}}(%esp), %edx +; X86-NEXT: subl %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $1, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl 24(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 28(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl $31, %ecx +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 56(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 60(%edx), %esi +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 12(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 40(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 44(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 16(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 20(%edx), %ebx +; X86-NEXT: movl %ebx, %esi +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 32(%edx), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 36(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edi, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl 52(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: shldl %cl, %esi, %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 4(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%ebp), %edx +; X86-NEXT: movl 40(%edx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %esi, %eax +; X86-NEXT: movl 8(%edx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ebx, %esi +; X86-NEXT: orl %eax, %esi +; X86-NEXT: movl %edx, %eax +; X86-NEXT: movl 56(%edx), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edi, %ebx +; X86-NEXT: movl 24(%edx), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: orl %ebx, %edx +; X86-NEXT: orl %esi, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 44(%eax), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: movl 12(%eax), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Folded Reload +; X86-NEXT: orl %esi, %ebx +; X86-NEXT: movl 60(%eax), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl %eax, %esi +; X86-NEXT: movl %edx, %eax +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl 28(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: orl %eax, %edx +; X86-NEXT: orl %ebx, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl (%eax), %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl (%esp), %eax # 4-byte Reload +; X86-NEXT: negl %eax +; X86-NEXT: movl 240(%esp,%eax), %esi +; X86-NEXT: shldl %cl, %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: shldl %cl, %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, (%esp) # 4-byte Spill +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: shll %cl, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%ebp), %esi +; X86-NEXT: movl 32(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edi, %eax +; X86-NEXT: movl (%esi), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edx, %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl 16(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ebx, %eax +; X86-NEXT: movl 48(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl (%esp), %edx # 4-byte Folded Reload +; X86-NEXT: orl %eax, %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl 36(%esi), %ebx +; X86-NEXT: movl %ebx, %eax +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl 4(%esi), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl %esi, %eax +; X86-NEXT: movl 20(%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl %esi, %edi +; X86-NEXT: movl 52(%eax), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: orl %edi, %eax +; X86-NEXT: orl %ecx, %eax +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: orl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: orl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: orl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: orl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: orl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: orl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: orl %ecx, (%esp) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: orl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: orl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload +; X86-NEXT: orl %edx, %eax +; X86-NEXT: movl 8(%ebp), %edx +; X86-NEXT: movl %ebx, 60(%edx) +; X86-NEXT: movl %edi, 56(%edx) +; X86-NEXT: movl %ecx, 52(%edx) +; X86-NEXT: movl %esi, 44(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 40(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 36(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 32(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 28(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 24(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 20(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 16(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 12(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 8(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 4(%edx) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, (%edx) +; X86-NEXT: movl (%esp), %eax # 4-byte Reload +; X86-NEXT: movl %eax, 48(%edx) +; X86-NEXT: setne %al +; X86-NEXT: leal -12(%ebp), %esp ; X86-NEXT: popl %esi ; X86-NEXT: popl %edi +; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; -; X64-LABEL: set_ne_i512: -; X64: # %bb.0: -; X64-NEXT: movl %esi, %ecx -; X64-NEXT: shrl $3, %ecx -; X64-NEXT: andl $60, %ecx -; X64-NEXT: movl (%rdi,%rcx), %edx -; X64-NEXT: btl %esi, %edx -; X64-NEXT: setb %al -; X64-NEXT: btsl %esi, %edx -; X64-NEXT: movl %edx, (%rdi,%rcx) -; X64-NEXT: retq +; SSE-LABEL: set_ne_i512: +; SSE: # %bb.0: +; SSE-NEXT: pushq %rbp +; SSE-NEXT: pushq %r15 +; SSE-NEXT: pushq %r14 +; SSE-NEXT: pushq %r13 +; SSE-NEXT: pushq %r12 +; SSE-NEXT: pushq %rbx +; SSE-NEXT: subq $56, %rsp +; SSE-NEXT: xorps %xmm0, %xmm0 +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movq $0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movq $1, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movl %esi, %ecx +; SSE-NEXT: andl $63, %ecx +; SSE-NEXT: shrl $3, %esi +; SSE-NEXT: andl $56, %esi +; SSE-NEXT: negl %esi +; SSE-NEXT: movslq %esi, %rbx +; SSE-NEXT: movq (%rsp,%rbx), %rsi +; SSE-NEXT: movq 8(%rsp,%rbx), %r14 +; SSE-NEXT: movq %r14, %rax +; SSE-NEXT: shldq %cl, %rsi, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 32(%rsp,%rbx), %r8 +; SSE-NEXT: movq 40(%rsp,%rbx), %rbp +; SSE-NEXT: shldq %cl, %r8, %rbp +; SSE-NEXT: movq 16(%rsp,%rbx), %r9 +; SSE-NEXT: movq 24(%rsp,%rbx), %r15 +; SSE-NEXT: movq %r15, %r10 +; SSE-NEXT: shldq %cl, %r9, %r10 +; SSE-NEXT: movq -8(%rsp,%rbx), %r11 +; SSE-NEXT: shldq %cl, %r11, %rsi +; SSE-NEXT: shldq %cl, %r15, %r8 +; SSE-NEXT: shldq %cl, %r14, %r9 +; SSE-NEXT: movq -16(%rsp,%rbx), %rbx +; SSE-NEXT: shldq %cl, %rbx, %r11 +; SSE-NEXT: # kill: def $cl killed $cl killed $ecx +; SSE-NEXT: shlq %cl, %rbx +; SSE-NEXT: movq 24(%rdi), %r15 +; SSE-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 56(%rdi), %rcx +; SSE-NEXT: movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 16(%rdi), %r12 +; SSE-NEXT: movq %r12, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 48(%rdi), %r13 +; SSE-NEXT: movq %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %r8, %r13 +; SSE-NEXT: andq %rsi, %r12 +; SSE-NEXT: orq %r13, %r12 +; SSE-NEXT: movq %rcx, %r13 +; SSE-NEXT: andq %rbp, %r13 +; SSE-NEXT: andq %rax, %r15 +; SSE-NEXT: orq %r13, %r15 +; SSE-NEXT: movq 32(%rdi), %r14 +; SSE-NEXT: movq %r14, %rcx +; SSE-NEXT: andq %r9, %rcx +; SSE-NEXT: movq (%rdi), %r13 +; SSE-NEXT: movq %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %rbx, %r13 +; SSE-NEXT: orq %rcx, %r13 +; SSE-NEXT: orq %r12, %r13 +; SSE-NEXT: movq 40(%rdi), %rcx +; SSE-NEXT: movq %rcx, %r12 +; SSE-NEXT: andq %r10, %r12 +; SSE-NEXT: movq 8(%rdi), %rdx +; SSE-NEXT: movq %rdx, %rax +; SSE-NEXT: andq %r11, %rax +; SSE-NEXT: orq %r12, %rax +; SSE-NEXT: orq %r15, %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Folded Reload +; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rsi # 8-byte Folded Reload +; SSE-NEXT: orq %rcx, %r10 +; SSE-NEXT: orq %r14, %r9 +; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Folded Reload +; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Folded Reload +; SSE-NEXT: orq %rdx, %r11 +; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Folded Reload +; SSE-NEXT: orq %r13, %rax +; SSE-NEXT: movq %r8, 48(%rdi) +; SSE-NEXT: movq %rbp, 56(%rdi) +; SSE-NEXT: movq %r9, 32(%rdi) +; SSE-NEXT: movq %r10, 40(%rdi) +; SSE-NEXT: movq %rsi, 16(%rdi) +; SSE-NEXT: movq %r15, 24(%rdi) +; SSE-NEXT: movq %rbx, (%rdi) +; SSE-NEXT: movq %r11, 8(%rdi) +; SSE-NEXT: setne %al +; SSE-NEXT: addq $56, %rsp +; SSE-NEXT: popq %rbx +; SSE-NEXT: popq %r12 +; SSE-NEXT: popq %r13 +; SSE-NEXT: popq %r14 +; SSE-NEXT: popq %r15 +; SSE-NEXT: popq %rbp +; SSE-NEXT: retq +; +; AVX2-LABEL: set_ne_i512: +; AVX2: # %bb.0: +; AVX2-NEXT: pushq %rbp +; AVX2-NEXT: pushq %r15 +; AVX2-NEXT: pushq %r14 +; AVX2-NEXT: pushq %r13 +; AVX2-NEXT: pushq %r12 +; AVX2-NEXT: pushq %rbx +; AVX2-NEXT: subq $72, %rsp +; AVX2-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovss {{.*#+}} xmm0 = [1,0,0,0] +; AVX2-NEXT: vmovups %ymm0, (%rsp) +; AVX2-NEXT: movl %esi, %ecx +; AVX2-NEXT: andl $63, %ecx +; AVX2-NEXT: shrl $3, %esi +; AVX2-NEXT: andl $56, %esi +; AVX2-NEXT: negl %esi +; AVX2-NEXT: movslq %esi, %rbx +; AVX2-NEXT: movq 16(%rsp,%rbx), %rsi +; AVX2-NEXT: movq 24(%rsp,%rbx), %rbp +; AVX2-NEXT: movq %rbp, %rax +; AVX2-NEXT: shldq %cl, %rsi, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 48(%rsp,%rbx), %r8 +; AVX2-NEXT: movq 56(%rsp,%rbx), %r13 +; AVX2-NEXT: shldq %cl, %r8, %r13 +; AVX2-NEXT: movq 32(%rsp,%rbx), %r9 +; AVX2-NEXT: movq 40(%rsp,%rbx), %r14 +; AVX2-NEXT: movq %r14, %r10 +; AVX2-NEXT: shldq %cl, %r9, %r10 +; AVX2-NEXT: movq 8(%rsp,%rbx), %r11 +; AVX2-NEXT: shldq %cl, %r11, %rsi +; AVX2-NEXT: shldq %cl, %r14, %r8 +; AVX2-NEXT: movq 16(%rdi), %r12 +; AVX2-NEXT: movq %r12, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 48(%rdi), %r14 +; AVX2-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: andq %r8, %r14 +; AVX2-NEXT: andq %rsi, %r12 +; AVX2-NEXT: orq %r14, %r12 +; AVX2-NEXT: movq 56(%rdi), %r15 +; AVX2-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: andq %r13, %r15 +; AVX2-NEXT: movq 24(%rdi), %r14 +; AVX2-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: andq %rax, %r14 +; AVX2-NEXT: orq %r15, %r14 +; AVX2-NEXT: shldq %cl, %rbp, %r9 +; AVX2-NEXT: movq (%rsp,%rbx), %rdx +; AVX2-NEXT: movq 32(%rdi), %r15 +; AVX2-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: andq %r9, %r15 +; AVX2-NEXT: shlxq %rcx, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq (%rdi), %rbx +; AVX2-NEXT: movq %rbx, %rbp +; AVX2-NEXT: andq %rax, %rbp +; AVX2-NEXT: orq %r15, %rbp +; AVX2-NEXT: orq %r12, %rbp +; AVX2-NEXT: # kill: def $cl killed $cl killed $rcx +; AVX2-NEXT: shldq %cl, %rdx, %r11 +; AVX2-NEXT: movq 40(%rdi), %rax +; AVX2-NEXT: movq %rax, %rcx +; AVX2-NEXT: andq %r10, %rcx +; AVX2-NEXT: movq 8(%rdi), %r15 +; AVX2-NEXT: movq %r15, %r12 +; AVX2-NEXT: andq %r11, %r12 +; AVX2-NEXT: orq %rcx, %r12 +; AVX2-NEXT: orq %r14, %r12 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload +; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Folded Reload +; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rsi # 8-byte Folded Reload +; AVX2-NEXT: orq %rax, %r10 +; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Folded Reload +; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Folded Reload +; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Folded Reload +; AVX2-NEXT: orq %r15, %r11 +; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Folded Reload +; AVX2-NEXT: orq %rbp, %r12 +; AVX2-NEXT: movq %r8, 48(%rdi) +; AVX2-NEXT: movq %r13, 56(%rdi) +; AVX2-NEXT: movq %r9, 32(%rdi) +; AVX2-NEXT: movq %r10, 40(%rdi) +; AVX2-NEXT: movq %rsi, 16(%rdi) +; AVX2-NEXT: movq %rcx, 24(%rdi) +; AVX2-NEXT: movq %rbx, (%rdi) +; AVX2-NEXT: movq %r11, 8(%rdi) +; AVX2-NEXT: setne %al +; AVX2-NEXT: addq $72, %rsp +; AVX2-NEXT: popq %rbx +; AVX2-NEXT: popq %r12 +; AVX2-NEXT: popq %r13 +; AVX2-NEXT: popq %r14 +; AVX2-NEXT: popq %r15 +; AVX2-NEXT: popq %rbp +; AVX2-NEXT: vzeroupper +; AVX2-NEXT: retq +; +; AVX512-LABEL: set_ne_i512: +; AVX512: # %bb.0: +; AVX512-NEXT: pushq %rbp +; AVX512-NEXT: pushq %r15 +; AVX512-NEXT: pushq %r14 +; AVX512-NEXT: pushq %r13 +; AVX512-NEXT: pushq %r12 +; AVX512-NEXT: pushq %rbx +; AVX512-NEXT: subq $72, %rsp +; AVX512-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [1,0,0,0] +; AVX512-NEXT: vmovups %ymm0, (%rsp) +; AVX512-NEXT: movl %esi, %ecx +; AVX512-NEXT: andl $63, %ecx +; AVX512-NEXT: shrl $3, %esi +; AVX512-NEXT: andl $56, %esi +; AVX512-NEXT: negl %esi +; AVX512-NEXT: movslq %esi, %rbx +; AVX512-NEXT: movq 16(%rsp,%rbx), %rsi +; AVX512-NEXT: movq 24(%rsp,%rbx), %rbp +; AVX512-NEXT: movq %rbp, %rax +; AVX512-NEXT: shldq %cl, %rsi, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 48(%rsp,%rbx), %r8 +; AVX512-NEXT: movq 56(%rsp,%rbx), %r13 +; AVX512-NEXT: shldq %cl, %r8, %r13 +; AVX512-NEXT: movq 32(%rsp,%rbx), %r9 +; AVX512-NEXT: movq 40(%rsp,%rbx), %r14 +; AVX512-NEXT: movq %r14, %r10 +; AVX512-NEXT: shldq %cl, %r9, %r10 +; AVX512-NEXT: movq 8(%rsp,%rbx), %r11 +; AVX512-NEXT: shldq %cl, %r11, %rsi +; AVX512-NEXT: shldq %cl, %r14, %r8 +; AVX512-NEXT: movq 16(%rdi), %r12 +; AVX512-NEXT: movq %r12, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 48(%rdi), %r14 +; AVX512-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: andq %r8, %r14 +; AVX512-NEXT: andq %rsi, %r12 +; AVX512-NEXT: orq %r14, %r12 +; AVX512-NEXT: movq 56(%rdi), %r15 +; AVX512-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: andq %r13, %r15 +; AVX512-NEXT: movq 24(%rdi), %r14 +; AVX512-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: andq %rax, %r14 +; AVX512-NEXT: orq %r15, %r14 +; AVX512-NEXT: shldq %cl, %rbp, %r9 +; AVX512-NEXT: movq (%rsp,%rbx), %rdx +; AVX512-NEXT: movq 32(%rdi), %r15 +; AVX512-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: andq %r9, %r15 +; AVX512-NEXT: shlxq %rcx, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq (%rdi), %rbx +; AVX512-NEXT: movq %rbx, %rbp +; AVX512-NEXT: andq %rax, %rbp +; AVX512-NEXT: orq %r15, %rbp +; AVX512-NEXT: orq %r12, %rbp +; AVX512-NEXT: # kill: def $cl killed $cl killed $rcx +; AVX512-NEXT: shldq %cl, %rdx, %r11 +; AVX512-NEXT: movq 40(%rdi), %rax +; AVX512-NEXT: movq %rax, %rcx +; AVX512-NEXT: andq %r10, %rcx +; AVX512-NEXT: movq 8(%rdi), %r15 +; AVX512-NEXT: movq %r15, %r12 +; AVX512-NEXT: andq %r11, %r12 +; AVX512-NEXT: orq %rcx, %r12 +; AVX512-NEXT: orq %r14, %r12 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload +; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Folded Reload +; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rsi # 8-byte Folded Reload +; AVX512-NEXT: orq %rax, %r10 +; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Folded Reload +; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Folded Reload +; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Folded Reload +; AVX512-NEXT: orq %r15, %r11 +; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Folded Reload +; AVX512-NEXT: orq %rbp, %r12 +; AVX512-NEXT: movq %r8, 48(%rdi) +; AVX512-NEXT: movq %r13, 56(%rdi) +; AVX512-NEXT: movq %r9, 32(%rdi) +; AVX512-NEXT: movq %r10, 40(%rdi) +; AVX512-NEXT: movq %rsi, 16(%rdi) +; AVX512-NEXT: movq %rcx, 24(%rdi) +; AVX512-NEXT: movq %rbx, (%rdi) +; AVX512-NEXT: movq %r11, 8(%rdi) +; AVX512-NEXT: setne %al +; AVX512-NEXT: addq $72, %rsp +; AVX512-NEXT: popq %rbx +; AVX512-NEXT: popq %r12 +; AVX512-NEXT: popq %r13 +; AVX512-NEXT: popq %r14 +; AVX512-NEXT: popq %r15 +; AVX512-NEXT: popq %rbp +; AVX512-NEXT: vzeroupper +; AVX512-NEXT: retq %rem = and i32 %position, 511 %ofs = zext nneg i32 %rem to i512 %bit = shl nuw i512 1, %ofs @@ -803,55 +3377,883 @@ define i1 @set_ne_i512(ptr %word, i32 %position) nounwind { define i1 @init_eq_i512(ptr %word, i32 %position, i1 zeroext %value) nounwind { ; X86-LABEL: init_eq_i512: ; X86: # %bb.0: +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp ; X86-NEXT: pushl %ebx ; X86-NEXT: pushl %edi ; X86-NEXT: pushl %esi -; X86-NEXT: movl {{[0-9]+}}(%esp), %edx -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X86-NEXT: movl %ecx, %esi -; X86-NEXT: shrl $3, %esi -; X86-NEXT: andl $60, %esi -; X86-NEXT: movl (%edx,%esi), %edi -; X86-NEXT: btl %ecx, %edi -; X86-NEXT: setae %al -; X86-NEXT: btrl %ecx, %edi -; X86-NEXT: movzbl {{[0-9]+}}(%esp), %ebx +; X86-NEXT: andl $-16, %esp +; X86-NEXT: subl $432, %esp # imm = 0x1B0 +; X86-NEXT: movl 12(%ebp), %ecx +; X86-NEXT: movl %ecx, %edx +; X86-NEXT: shrl $3, %edx +; X86-NEXT: andl $60, %edx +; X86-NEXT: leal {{[0-9]+}}(%esp), %esi +; X86-NEXT: subl %edx, %esi +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $1, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl 56(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 60(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 52(%esi), %eax +; X86-NEXT: movl 48(%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 40(%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 44(%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 36(%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 32(%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 28(%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 24(%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 20(%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 16(%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 12(%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl (%esi), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 4(%esi), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movzbl 16(%ebp), %ebx +; X86-NEXT: movzbl %bl, %esi +; X86-NEXT: movl %esi, {{[0-9]+}}(%esp) +; X86-NEXT: leal {{[0-9]+}}(%esp), %esi +; X86-NEXT: subl %edx, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl $31, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: shldl %cl, %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: shldl %cl, %ebx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: shldl %cl, %esi, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: shldl %cl, %edx, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: shldl %cl, %esi, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill ; X86-NEXT: # kill: def $cl killed $cl killed $ecx -; X86-NEXT: shll %cl, %ebx -; X86-NEXT: orl %edi, %ebx -; X86-NEXT: movl %ebx, (%edx,%esi) +; X86-NEXT: movl %ebx, %edx +; X86-NEXT: shldl %cl, %edi, %edx +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl 8(%ebp), %ebx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 48(%ebx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %esi, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 16(%ebx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %eax, %esi +; X86-NEXT: orl %edx, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 40(%ebx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %eax, %edx +; X86-NEXT: movl %edx, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%ebx), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edx, %eax +; X86-NEXT: orl %esi, %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 56(%ebx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %esi, %edx +; X86-NEXT: movl %edx, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 24(%ebx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %esi, %edx +; X86-NEXT: orl %edi, %edx +; X86-NEXT: orl %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 52(%ebx), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edx, %eax +; X86-NEXT: movl %eax, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 20(%ebx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %esi, %eax +; X86-NEXT: orl %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 44(%ebx), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edx, %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 12(%ebx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %esi, %edi +; X86-NEXT: orl %eax, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 60(%ebx), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edx, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 28(%ebx), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %eax, %edx +; X86-NEXT: orl %esi, %edx +; X86-NEXT: orl %edi, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: shll %cl, %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 32(%ebx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %esi, %ecx +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl (%ebx), %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edi, %eax +; X86-NEXT: orl %ecx, %eax +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 36(%ebx), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ecx, %eax +; X86-NEXT: movl %eax, %ecx +; X86-NEXT: movl %edx, %eax +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 4(%ebx), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edx, %eax +; X86-NEXT: orl %ecx, %eax +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl 56(%edi), %ebx +; X86-NEXT: movl 60(%edi), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: shldl %cl, %ebx, %eax +; X86-NEXT: orl %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl 52(%edi), %eax +; X86-NEXT: shldl %cl, %eax, %ebx +; X86-NEXT: orl %ebx, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl 48(%edi), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %esi, %eax +; X86-NEXT: orl %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: notl %eax +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl %eax, %edx +; X86-NEXT: movl 40(%edi), %ebx +; X86-NEXT: movl 44(%edi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %ebx, %eax +; X86-NEXT: orl %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl 36(%edi), %eax +; X86-NEXT: shldl %cl, %eax, %ebx +; X86-NEXT: orl %ebx, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl 32(%edi), %ebx +; X86-NEXT: shldl %cl, %ebx, %eax +; X86-NEXT: orl %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl 28(%edi), %eax +; X86-NEXT: shldl %cl, %eax, %ebx +; X86-NEXT: orl %ebx, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl 24(%edi), %ebx +; X86-NEXT: shldl %cl, %ebx, %eax +; X86-NEXT: orl %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl 20(%edi), %eax +; X86-NEXT: shldl %cl, %eax, %ebx +; X86-NEXT: orl %ebx, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl 16(%edi), %ebx +; X86-NEXT: shldl %cl, %ebx, %eax +; X86-NEXT: orl %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl 12(%edi), %eax +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: shldl %cl, %eax, %ebx +; X86-NEXT: orl %ebx, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: notl %esi +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: movl 8(%edi), %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: orl %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: notl %eax +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl 4(%edi), %ebx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: shldl %cl, %ebx, %edx +; X86-NEXT: orl %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: shldl %cl, %esi, %eax +; X86-NEXT: orl %eax, %edx +; X86-NEXT: movl %edx, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: notl %edx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload +; X86-NEXT: movl (%edi), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %ebx +; X86-NEXT: orl %ebx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: notl %edi +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: shll %cl, %eax +; X86-NEXT: orl %eax, %edi +; X86-NEXT: movl %edi, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Folded Reload +; X86-NEXT: movl 8(%ebp), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 60(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 56(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 52(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 44(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 40(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 36(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 32(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 28(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 24(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 20(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 16(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 12(%eax) +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl %edi, 8(%eax) +; X86-NEXT: movl %edx, 4(%eax) +; X86-NEXT: movl %ecx, (%eax) +; X86-NEXT: movl %esi, 48(%eax) +; X86-NEXT: sete %al +; X86-NEXT: leal -12(%ebp), %esp ; X86-NEXT: popl %esi ; X86-NEXT: popl %edi ; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; ; SSE-LABEL: init_eq_i512: ; SSE: # %bb.0: +; SSE-NEXT: pushq %rbp +; SSE-NEXT: pushq %r15 +; SSE-NEXT: pushq %r14 +; SSE-NEXT: pushq %r13 +; SSE-NEXT: pushq %r12 +; SSE-NEXT: pushq %rbx +; SSE-NEXT: subq $216, %rsp +; SSE-NEXT: xorps %xmm0, %xmm0 +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movq $0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movq $1, {{[0-9]+}}(%rsp) ; SSE-NEXT: movl %esi, %ecx +; SSE-NEXT: andl $63, %ecx ; SSE-NEXT: shrl $3, %esi -; SSE-NEXT: andl $60, %esi -; SSE-NEXT: movl (%rdi,%rsi), %r8d -; SSE-NEXT: btl %ecx, %r8d -; SSE-NEXT: setae %al -; SSE-NEXT: shll %cl, %edx -; SSE-NEXT: btrl %ecx, %r8d -; SSE-NEXT: orl %r8d, %edx -; SSE-NEXT: movl %edx, (%rdi,%rsi) +; SSE-NEXT: andl $56, %esi +; SSE-NEXT: negl %esi +; SSE-NEXT: movslq %esi, %r10 +; SSE-NEXT: movq 184(%rsp,%r10), %r11 +; SSE-NEXT: movq 192(%rsp,%r10), %rsi +; SSE-NEXT: movq %rsi, %r13 +; SSE-NEXT: shldq %cl, %r11, %r13 +; SSE-NEXT: movq 200(%rsp,%r10), %r15 +; SSE-NEXT: shldq %cl, %rsi, %r15 +; SSE-NEXT: movq 168(%rsp,%r10), %rbx +; SSE-NEXT: movq 176(%rsp,%r10), %rsi +; SSE-NEXT: movq %rsi, %r14 +; SSE-NEXT: shldq %cl, %rbx, %r14 +; SSE-NEXT: shldq %cl, %rsi, %r11 +; SSE-NEXT: movq 152(%rsp,%r10), %rax +; SSE-NEXT: movq 160(%rsp,%r10), %r8 +; SSE-NEXT: movq %r8, %r12 +; SSE-NEXT: shldq %cl, %rax, %r12 +; SSE-NEXT: shldq %cl, %r8, %rbx +; SSE-NEXT: movq 144(%rsp,%r10), %r9 +; SSE-NEXT: movq %r9, %r8 +; SSE-NEXT: shlq %cl, %r8 +; SSE-NEXT: shldq %cl, %r9, %rax +; SSE-NEXT: movq %r12, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movl %edx, %edx +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, (%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp) +; SSE-NEXT: movq %rdx, {{[0-9]+}}(%rsp) +; SSE-NEXT: movq $0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movq 16(%rdi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 48(%rdi), %rsi +; SSE-NEXT: movq %rsi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %rsi, %r13 +; SSE-NEXT: andq %rdx, %r12 +; SSE-NEXT: orq %r13, %r12 +; SSE-NEXT: movq %r15, %rsi +; SSE-NEXT: movq 56(%rdi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %rdx, %r15 +; SSE-NEXT: movq %rbx, %r13 +; SSE-NEXT: movq 24(%rdi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %rdx, %rbx +; SSE-NEXT: orq %r15, %rbx +; SSE-NEXT: movq %r14, %rbp +; SSE-NEXT: movq 32(%rdi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %rdx, %r14 +; SSE-NEXT: movq %r8, %r15 +; SSE-NEXT: movq (%rdi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %rdx, %r8 +; SSE-NEXT: orq %r14, %r8 +; SSE-NEXT: orq %r12, %r8 +; SSE-NEXT: movq %r11, %r12 +; SSE-NEXT: movq 40(%rdi), %r9 +; SSE-NEXT: andq %r9, %r11 +; SSE-NEXT: movq %rax, %r14 +; SSE-NEXT: movq 8(%rdi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq %rdx, %rax +; SSE-NEXT: orq %r11, %rax +; SSE-NEXT: orq %rbx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: notq %rax +; SSE-NEXT: andq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Folded Reload +; SSE-NEXT: movq %rax, %rdx +; SSE-NEXT: movq 56(%rsp,%r10), %r11 +; SSE-NEXT: movq 64(%rsp,%r10), %rax +; SSE-NEXT: movq %rax, %rbx +; SSE-NEXT: shldq %cl, %r11, %rbx +; SSE-NEXT: orq %rbx, %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: notq %rsi +; SSE-NEXT: movq 72(%rsp,%r10), %rbx +; SSE-NEXT: shldq %cl, %rax, %rbx +; SSE-NEXT: andq {{[-0-9]+}}(%r{{[sb]}}p), %rsi # 8-byte Folded Reload +; SSE-NEXT: orq %rbx, %rsi +; SSE-NEXT: notq %rbp +; SSE-NEXT: andq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Folded Reload +; SSE-NEXT: movq 40(%rsp,%r10), %rax +; SSE-NEXT: movq 48(%rsp,%r10), %rdx +; SSE-NEXT: movq %rdx, %rbx +; SSE-NEXT: shldq %cl, %rax, %rbx +; SSE-NEXT: orq %rbx, %rbp +; SSE-NEXT: notq %r12 +; SSE-NEXT: andq %r9, %r12 +; SSE-NEXT: shldq %cl, %rdx, %r11 +; SSE-NEXT: movq 24(%rsp,%r10), %r9 +; SSE-NEXT: movq 32(%rsp,%r10), %rdx +; SSE-NEXT: movq %rdx, %rbx +; SSE-NEXT: shldq %cl, %r9, %rbx +; SSE-NEXT: orq %r11, %r12 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Reload +; SSE-NEXT: notq %r11 +; SSE-NEXT: andq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Folded Reload +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: orq %rbx, %r11 +; SSE-NEXT: notq %r13 +; SSE-NEXT: andq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Folded Reload +; SSE-NEXT: orq %rax, %r13 +; SSE-NEXT: notq %r15 +; SSE-NEXT: andq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Folded Reload +; SSE-NEXT: movq 16(%rsp,%r10), %rax +; SSE-NEXT: movq %rax, %rdx +; SSE-NEXT: shlq %cl, %rdx +; SSE-NEXT: orq %rdx, %r15 +; SSE-NEXT: notq %r14 +; SSE-NEXT: andq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Folded Reload +; SSE-NEXT: # kill: def $cl killed $cl killed $ecx +; SSE-NEXT: shldq %cl, %rax, %r9 +; SSE-NEXT: orq %r9, %r14 +; SSE-NEXT: orq %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: movq %rax, 48(%rdi) +; SSE-NEXT: movq %rsi, 56(%rdi) +; SSE-NEXT: movq %rbp, 32(%rdi) +; SSE-NEXT: movq %r12, 40(%rdi) +; SSE-NEXT: movq %r11, 16(%rdi) +; SSE-NEXT: movq %r13, 24(%rdi) +; SSE-NEXT: movq %r15, (%rdi) +; SSE-NEXT: movq %r14, 8(%rdi) +; SSE-NEXT: sete %al +; SSE-NEXT: addq $216, %rsp +; SSE-NEXT: popq %rbx +; SSE-NEXT: popq %r12 +; SSE-NEXT: popq %r13 +; SSE-NEXT: popq %r14 +; SSE-NEXT: popq %r15 +; SSE-NEXT: popq %rbp ; SSE-NEXT: retq ; -; AVX-LABEL: init_eq_i512: -; AVX: # %bb.0: -; AVX-NEXT: movl %esi, %ecx -; AVX-NEXT: shrl $3, %ecx -; AVX-NEXT: andl $60, %ecx -; AVX-NEXT: movl (%rdi,%rcx), %r8d -; AVX-NEXT: btl %esi, %r8d -; AVX-NEXT: setae %al -; AVX-NEXT: btrl %esi, %r8d -; AVX-NEXT: shlxl %esi, %edx, %edx -; AVX-NEXT: orl %r8d, %edx -; AVX-NEXT: movl %edx, (%rdi,%rcx) -; AVX-NEXT: retq +; AVX2-LABEL: init_eq_i512: +; AVX2: # %bb.0: +; AVX2-NEXT: pushq %rbp +; AVX2-NEXT: pushq %r15 +; AVX2-NEXT: pushq %r14 +; AVX2-NEXT: pushq %r13 +; AVX2-NEXT: pushq %r12 +; AVX2-NEXT: pushq %rbx +; AVX2-NEXT: subq $200, %rsp +; AVX2-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovss {{.*#+}} xmm1 = [1,0,0,0] +; AVX2-NEXT: vmovups %ymm1, {{[0-9]+}}(%rsp) +; AVX2-NEXT: movl %esi, %r8d +; AVX2-NEXT: andl $63, %r8d +; AVX2-NEXT: shrl $3, %esi +; AVX2-NEXT: andl $56, %esi +; AVX2-NEXT: negl %esi +; AVX2-NEXT: movslq %esi, %rsi +; AVX2-NEXT: movq 144(%rsp,%rsi), %r11 +; AVX2-NEXT: movq 152(%rsp,%rsi), %r12 +; AVX2-NEXT: movq %r12, %r10 +; AVX2-NEXT: movl %r8d, %ecx +; AVX2-NEXT: shldq %cl, %r11, %r10 +; AVX2-NEXT: movq 176(%rsp,%rsi), %r14 +; AVX2-NEXT: movq 184(%rsp,%rsi), %r9 +; AVX2-NEXT: shldq %cl, %r14, %r9 +; AVX2-NEXT: movq 160(%rsp,%rsi), %r15 +; AVX2-NEXT: movq 168(%rsp,%rsi), %r13 +; AVX2-NEXT: movq %r13, %rbx +; AVX2-NEXT: shldq %cl, %r15, %rbx +; AVX2-NEXT: movq 128(%rsp,%rsi), %rbp +; AVX2-NEXT: movq %rbp, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 136(%rsp,%rsi), %rax +; AVX2-NEXT: shldq %cl, %rax, %r11 +; AVX2-NEXT: shldq %cl, %r13, %r14 +; AVX2-NEXT: shldq %cl, %r12, %r15 +; AVX2-NEXT: shldq %cl, %rbp, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movl %edx, %edx +; AVX2-NEXT: vxorps %xmm1, %xmm1, %xmm1 +; AVX2-NEXT: vmovups %xmm1, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX2-NEXT: movq %rdx, (%rsp) +; AVX2-NEXT: movq $0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: movq 16(%rdi), %r12 +; AVX2-NEXT: movq 48(%rdi), %rbp +; AVX2-NEXT: movq 32(%rdi), %r13 +; AVX2-NEXT: andnq %r13, %r15, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: andq %r15, %r13 +; AVX2-NEXT: andnq %rbp, %r14, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: andq %r14, %rbp +; AVX2-NEXT: andnq %r12, %r11, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: andq %r11, %r12 +; AVX2-NEXT: movq 40(%rdi), %rax +; AVX2-NEXT: orq %rbp, %r12 +; AVX2-NEXT: andnq %rax, %rbx, %rcx +; AVX2-NEXT: movq %rcx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq %rax, %rbp +; AVX2-NEXT: andq %rbx, %rbp +; AVX2-NEXT: movq 56(%rdi), %rcx +; AVX2-NEXT: andnq %rcx, %r9, %rbx +; AVX2-NEXT: andq %r9, %rcx +; AVX2-NEXT: movq 24(%rdi), %rax +; AVX2-NEXT: andnq %rax, %r10, %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: andq %r10, %rax +; AVX2-NEXT: orq %rcx, %rax +; AVX2-NEXT: shlxq %r8, {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Folded Reload +; AVX2-NEXT: movq (%rdi), %r10 +; AVX2-NEXT: andnq %r10, %rcx, %r15 +; AVX2-NEXT: andq %rcx, %r10 +; AVX2-NEXT: movq 40(%rsp,%rsi), %rdx +; AVX2-NEXT: movq 48(%rsp,%rsi), %r11 +; AVX2-NEXT: movq %r11, %r9 +; AVX2-NEXT: movl %r8d, %ecx +; AVX2-NEXT: shldq %cl, %rdx, %r9 +; AVX2-NEXT: orq %r13, %r10 +; AVX2-NEXT: orq %r12, %r10 +; AVX2-NEXT: movq 8(%rdi), %r13 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload +; AVX2-NEXT: andnq %r13, %rcx, %r12 +; AVX2-NEXT: andq %rcx, %r13 +; AVX2-NEXT: orq %rbp, %r13 +; AVX2-NEXT: orq %rax, %r13 +; AVX2-NEXT: movq 56(%rsp,%rsi), %rax +; AVX2-NEXT: movl %r8d, %ecx +; AVX2-NEXT: shldq %cl, %r11, %rax +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; AVX2-NEXT: orq %r9, %r14 +; AVX2-NEXT: orq %rax, %rbx +; AVX2-NEXT: movq %rbx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 24(%rsp,%rsi), %rax +; AVX2-NEXT: movq 32(%rsp,%rsi), %r9 +; AVX2-NEXT: movq %r9, %r11 +; AVX2-NEXT: shldq %cl, %rax, %r11 +; AVX2-NEXT: shldq %cl, %r9, %rdx +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Reload +; AVX2-NEXT: orq %r11, %rbp +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload +; AVX2-NEXT: orq %rdx, %rbx +; AVX2-NEXT: movq 8(%rsp,%rsi), %rdx +; AVX2-NEXT: movq 16(%rsp,%rsi), %r9 +; AVX2-NEXT: movq %r9, %r11 +; AVX2-NEXT: shldq %cl, %rdx, %r11 +; AVX2-NEXT: shldq %cl, %r9, %rax +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; AVX2-NEXT: orq %r11, %r9 +; AVX2-NEXT: movq (%rsp,%rsi), %rsi +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Reload +; AVX2-NEXT: orq %rax, %r11 +; AVX2-NEXT: shlxq %r8, %rsi, %rax +; AVX2-NEXT: shldq %cl, %rsi, %rdx +; AVX2-NEXT: orq %rax, %r15 +; AVX2-NEXT: orq %rdx, %r12 +; AVX2-NEXT: orq %r10, %r13 +; AVX2-NEXT: movq %r14, 48(%rdi) +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: movq %rax, 56(%rdi) +; AVX2-NEXT: movq %rbp, 32(%rdi) +; AVX2-NEXT: movq %rbx, 40(%rdi) +; AVX2-NEXT: movq %r9, 16(%rdi) +; AVX2-NEXT: movq %r11, 24(%rdi) +; AVX2-NEXT: movq %r15, (%rdi) +; AVX2-NEXT: movq %r12, 8(%rdi) +; AVX2-NEXT: sete %al +; AVX2-NEXT: addq $200, %rsp +; AVX2-NEXT: popq %rbx +; AVX2-NEXT: popq %r12 +; AVX2-NEXT: popq %r13 +; AVX2-NEXT: popq %r14 +; AVX2-NEXT: popq %r15 +; AVX2-NEXT: popq %rbp +; AVX2-NEXT: vzeroupper +; AVX2-NEXT: retq +; +; AVX512-LABEL: init_eq_i512: +; AVX512: # %bb.0: +; AVX512-NEXT: pushq %rbp +; AVX512-NEXT: pushq %r15 +; AVX512-NEXT: pushq %r14 +; AVX512-NEXT: pushq %r13 +; AVX512-NEXT: pushq %r12 +; AVX512-NEXT: pushq %rbx +; AVX512-NEXT: subq $184, %rsp +; AVX512-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovaps {{.*#+}} xmm1 = [1,0,0,0] +; AVX512-NEXT: vmovups %ymm1, {{[0-9]+}}(%rsp) +; AVX512-NEXT: movl %esi, %ecx +; AVX512-NEXT: andl $63, %ecx +; AVX512-NEXT: shrl $3, %esi +; AVX512-NEXT: andl $56, %esi +; AVX512-NEXT: negl %esi +; AVX512-NEXT: movslq %esi, %rsi +; AVX512-NEXT: movq 128(%rsp,%rsi), %r10 +; AVX512-NEXT: movq 136(%rsp,%rsi), %r12 +; AVX512-NEXT: movq %r12, %rax +; AVX512-NEXT: shldq %cl, %r10, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 160(%rsp,%rsi), %r14 +; AVX512-NEXT: movq 168(%rsp,%rsi), %rax +; AVX512-NEXT: shldq %cl, %r14, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 144(%rsp,%rsi), %r15 +; AVX512-NEXT: movq 152(%rsp,%rsi), %r11 +; AVX512-NEXT: movq %r11, %rbx +; AVX512-NEXT: shldq %cl, %r15, %rbx +; AVX512-NEXT: movq 120(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rax, %r10 +; AVX512-NEXT: shldq %cl, %r11, %r14 +; AVX512-NEXT: movq %rdi, %r9 +; AVX512-NEXT: movq 112(%rsp,%rsi), %r11 +; AVX512-NEXT: shldq %cl, %r12, %r15 +; AVX512-NEXT: movl %edx, %edx +; AVX512-NEXT: vxorps %xmm1, %xmm1, %xmm1 +; AVX512-NEXT: vmovups %xmm1, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: movq %rdx, -{{[0-9]+}}(%rsp) +; AVX512-NEXT: movq $0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: movq 16(%rdi), %r12 +; AVX512-NEXT: movq 48(%rdi), %r13 +; AVX512-NEXT: movq 32(%rdi), %rbp +; AVX512-NEXT: andnq %rbp, %r15, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: andq %r15, %rbp +; AVX512-NEXT: andnq %r13, %r14, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: andq %r14, %r13 +; AVX512-NEXT: andnq %r12, %r10, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: andq %r10, %r12 +; AVX512-NEXT: movq 40(%rdi), %r8 +; AVX512-NEXT: orq %r13, %r12 +; AVX512-NEXT: andnq %r8, %rbx, %rdi +; AVX512-NEXT: andq %rbx, %r8 +; AVX512-NEXT: movq 56(%r9), %r13 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; AVX512-NEXT: andnq %r13, %rdx, %r10 +; AVX512-NEXT: andq %rdx, %r13 +; AVX512-NEXT: movq 24(%r9), %rax +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; AVX512-NEXT: andnq %rax, %rdx, %r15 +; AVX512-NEXT: andq %rdx, %rax +; AVX512-NEXT: orq %r13, %rax +; AVX512-NEXT: shlxq %rcx, %r11, %r13 +; AVX512-NEXT: movq (%r9), %rdx +; AVX512-NEXT: andnq %rdx, %r13, %r14 +; AVX512-NEXT: andq %r13, %rdx +; AVX512-NEXT: orq %rbp, %rdx +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r11, %rbp +; AVX512-NEXT: orq %r12, %rdx +; AVX512-NEXT: movq 8(%r9), %r13 +; AVX512-NEXT: andnq %r13, %rbp, %rbx +; AVX512-NEXT: andq %rbp, %r13 +; AVX512-NEXT: orq %r8, %r13 +; AVX512-NEXT: movq 24(%rsp,%rsi), %r8 +; AVX512-NEXT: orq %rax, %r13 +; AVX512-NEXT: movq 32(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, %r12 +; AVX512-NEXT: shldq %cl, %r8, %r12 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Reload +; AVX512-NEXT: orq %r12, %r11 +; AVX512-NEXT: movq 40(%rsp,%rsi), %r12 +; AVX512-NEXT: shldq %cl, %rax, %r12 +; AVX512-NEXT: orq %r12, %r10 +; AVX512-NEXT: movq %r10, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 8(%rsp,%rsi), %rax +; AVX512-NEXT: movq 16(%rsp,%rsi), %r12 +; AVX512-NEXT: movq %r12, %rbp +; AVX512-NEXT: shldq %cl, %rax, %rbp +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX512-NEXT: orq %rbp, %r10 +; AVX512-NEXT: shldq %cl, %r12, %r8 +; AVX512-NEXT: orq %r8, %rdi +; AVX512-NEXT: movq %rdi, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq -8(%rsp,%rsi), %r8 +; AVX512-NEXT: movq (%rsp,%rsi), %r12 +; AVX512-NEXT: movq %r12, %rbp +; AVX512-NEXT: shldq %cl, %r8, %rbp +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdi # 8-byte Reload +; AVX512-NEXT: orq %rbp, %rdi +; AVX512-NEXT: movq -16(%rsp,%rsi), %rsi +; AVX512-NEXT: shldq %cl, %r12, %rax +; AVX512-NEXT: orq %rax, %r15 +; AVX512-NEXT: shlxq %rcx, %rsi, %rax +; AVX512-NEXT: # kill: def $cl killed $cl killed $rcx +; AVX512-NEXT: shldq %cl, %rsi, %r8 +; AVX512-NEXT: orq %rax, %r14 +; AVX512-NEXT: orq %r8, %rbx +; AVX512-NEXT: orq %rdx, %r13 +; AVX512-NEXT: movq %r11, 48(%r9) +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: movq %rax, 56(%r9) +; AVX512-NEXT: movq %r10, 32(%r9) +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: movq %rax, 40(%r9) +; AVX512-NEXT: movq %rdi, 16(%r9) +; AVX512-NEXT: movq %r15, 24(%r9) +; AVX512-NEXT: movq %r14, (%r9) +; AVX512-NEXT: movq %rbx, 8(%r9) +; AVX512-NEXT: sete %al +; AVX512-NEXT: addq $184, %rsp +; AVX512-NEXT: popq %rbx +; AVX512-NEXT: popq %r12 +; AVX512-NEXT: popq %r13 +; AVX512-NEXT: popq %r14 +; AVX512-NEXT: popq %r15 +; AVX512-NEXT: popq %rbp +; AVX512-NEXT: vzeroupper +; AVX512-NEXT: retq %rem = and i32 %position, 511 %ofs = zext nneg i32 %rem to i512 %bit = shl nuw i512 1, %ofs @@ -872,25 +4274,2749 @@ define i1 @init_eq_i512(ptr %word, i32 %position, i1 zeroext %value) nounwind { define i1 @test_ne_i4096(ptr %word, i32 %position) nounwind { ; X86-LABEL: test_ne_i4096: ; X86: # %bb.0: -; X86-NEXT: movl {{[0-9]+}}(%esp), %eax -; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx -; X86-NEXT: movl %ecx, %edx -; X86-NEXT: andl $4064, %edx # imm = 0xFE0 -; X86-NEXT: shrl $3, %edx -; X86-NEXT: movl (%eax,%edx), %eax -; X86-NEXT: btl %ecx, %eax -; X86-NEXT: setb %al +; X86-NEXT: pushl %ebp +; X86-NEXT: movl %esp, %ebp +; X86-NEXT: pushl %ebx +; X86-NEXT: pushl %edi +; X86-NEXT: pushl %esi +; X86-NEXT: andl $-16, %esp +; X86-NEXT: subl $1792, %esp # imm = 0x700 +; X86-NEXT: movl 12(%ebp), %ebx +; X86-NEXT: movl %ebx, %ecx +; X86-NEXT: shrl $3, %ecx +; X86-NEXT: andl $508, %ecx # imm = 0x1FC +; X86-NEXT: leal {{[0-9]+}}(%esp), %esi +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: subl %ecx, %esi +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $1, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl $0, {{[0-9]+}}(%esp) +; X86-NEXT: movl 248(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 252(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl $31, %ebx +; X86-NEXT: movl %ebx, %ecx +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 504(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 508(%esi), %edx +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 120(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 124(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 376(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 380(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 184(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 188(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 440(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 444(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 56(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 60(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 312(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 316(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 216(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 220(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 472(%esi), %edi +; X86-NEXT: movl 476(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edi, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 88(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 92(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 344(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 348(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 152(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 156(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 408(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 412(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 24(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 28(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 280(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 284(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 232(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 236(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 488(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 492(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 104(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 108(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 360(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 364(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 168(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 172(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 424(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 428(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 40(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 44(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 296(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 300(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 200(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 204(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 456(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 460(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 72(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 76(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 328(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 332(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 136(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 140(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 392(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 396(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 12(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 264(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 268(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 240(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 244(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 496(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 500(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 112(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 116(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 368(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 372(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 176(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 180(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 432(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 436(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 48(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 52(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 304(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 308(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 208(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 212(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 464(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 468(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 80(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 84(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 336(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 340(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 144(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 148(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 400(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 404(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 16(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 20(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 272(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 276(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 224(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 228(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 480(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 484(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 96(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 100(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 352(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 356(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 160(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 164(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 416(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 420(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 32(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 36(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 288(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 292(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 192(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 196(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 448(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 452(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 64(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 68(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 320(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 324(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 128(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 132(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl %edi, %edx +; X86-NEXT: movl 256(%esi), %edi +; X86-NEXT: movl 260(%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edi, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: shldl %cl, %edi, %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl 388(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl 4(%esi), %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shrdl $1, %eax, %edi +; X86-NEXT: shrl %eax +; X86-NEXT: movl %ebx, %edx +; X86-NEXT: movl %eax, %ebx +; X86-NEXT: notb %cl +; X86-NEXT: shrdl %cl, %eax, %edi +; X86-NEXT: shrl %cl, %ebx +; X86-NEXT: movb $32, %cl +; X86-NEXT: testb %cl, %cl +; X86-NEXT: movl (%esi), %eax +; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 8(%ebp), %eax +; X86-NEXT: jne .LBB20_2 +; X86-NEXT: # %bb.1: +; X86-NEXT: movl %edi, %ebx +; X86-NEXT: .LBB20_2: +; X86-NEXT: movl %edx, %ecx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: shll %cl, %edx +; X86-NEXT: orl %ebx, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 320(%eax), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl 64(%eax), %esi +; X86-NEXT: orl %ecx, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 448(%eax), %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 192(%eax), %ecx +; X86-NEXT: orl %edx, %ecx +; X86-NEXT: orl %esi, %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 288(%eax), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 32(%eax), %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 416(%eax), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: andl 160(%eax), %edi +; X86-NEXT: orl %ecx, %edi +; X86-NEXT: orl %edx, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 352(%eax), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl 96(%eax), %esi +; X86-NEXT: orl %ecx, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 480(%eax), %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 224(%eax), %ecx +; X86-NEXT: orl %edx, %ecx +; X86-NEXT: orl %esi, %ecx +; X86-NEXT: orl %edi, %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 272(%eax), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 16(%eax), %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 400(%eax), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl 144(%eax), %esi +; X86-NEXT: orl %ecx, %esi +; X86-NEXT: orl %edx, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 336(%eax), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 80(%eax), %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 464(%eax), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: andl 208(%eax), %edi +; X86-NEXT: orl %ecx, %edi +; X86-NEXT: orl %edx, %edi +; X86-NEXT: orl %esi, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 304(%eax), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 48(%eax), %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 432(%eax), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl 176(%eax), %esi +; X86-NEXT: orl %ecx, %esi +; X86-NEXT: orl %edx, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 368(%eax), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 112(%eax), %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 496(%eax), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: andl 240(%eax), %ebx +; X86-NEXT: orl %ecx, %ebx +; X86-NEXT: orl %edx, %ebx +; X86-NEXT: orl %esi, %ebx +; X86-NEXT: orl %edi, %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 264(%eax), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 8(%eax), %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl %eax, %ebx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 392(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 136(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: orl %edx, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 328(%ebx), %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 72(%ebx), %eax +; X86-NEXT: orl %edx, %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: andl 456(%ebx), %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl 200(%ebx), %esi +; X86-NEXT: orl %edi, %esi +; X86-NEXT: orl %eax, %esi +; X86-NEXT: orl %ecx, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: andl 296(%ebx), %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 40(%ebx), %eax +; X86-NEXT: orl %edi, %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: andl 424(%ebx), %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 168(%ebx), %edx +; X86-NEXT: orl %edi, %edx +; X86-NEXT: orl %eax, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 360(%ebx), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 104(%ebx), %eax +; X86-NEXT: orl %ecx, %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 488(%ebx), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: andl 232(%ebx), %edi +; X86-NEXT: orl %ecx, %edi +; X86-NEXT: orl %eax, %edi +; X86-NEXT: orl %edx, %edi +; X86-NEXT: orl %esi, %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 280(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 24(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 408(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 152(%ebx), %edx +; X86-NEXT: orl %eax, %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 344(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 88(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 472(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl 216(%ebx), %esi +; X86-NEXT: orl %eax, %esi +; X86-NEXT: orl %ecx, %esi +; X86-NEXT: orl %edx, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 312(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 56(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 440(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 184(%ebx), %edx +; X86-NEXT: orl %eax, %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 376(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 120(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 504(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: andl 248(%ebx), %edi +; X86-NEXT: orl %eax, %edi +; X86-NEXT: orl %ecx, %edi +; X86-NEXT: orl %edx, %edi +; X86-NEXT: orl %esi, %edi +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 324(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 68(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 452(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 196(%ebx), %edx +; X86-NEXT: orl %eax, %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 292(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 36(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 420(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 164(%ebx), %edx +; X86-NEXT: orl %eax, %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 356(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 100(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 484(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl 228(%ebx), %esi +; X86-NEXT: orl %eax, %esi +; X86-NEXT: orl %ecx, %esi +; X86-NEXT: orl %edx, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 276(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 20(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 404(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 148(%ebx), %edx +; X86-NEXT: orl %eax, %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 340(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 84(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 468(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl 212(%ebx), %esi +; X86-NEXT: orl %eax, %esi +; X86-NEXT: orl %ecx, %esi +; X86-NEXT: orl %edx, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 308(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 52(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 436(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 180(%ebx), %edx +; X86-NEXT: orl %eax, %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 372(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 116(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 500(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: andl 244(%ebx), %edi +; X86-NEXT: orl %eax, %edi +; X86-NEXT: orl %ecx, %edi +; X86-NEXT: orl %edx, %edi +; X86-NEXT: orl %esi, %edi +; X86-NEXT: movl %edi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 268(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 12(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 396(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 140(%ebx), %edx +; X86-NEXT: orl %eax, %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 332(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 76(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 460(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: andl 204(%ebx), %edi +; X86-NEXT: orl %eax, %edi +; X86-NEXT: orl %ecx, %edi +; X86-NEXT: orl %edx, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 300(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 44(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 428(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 172(%ebx), %edx +; X86-NEXT: orl %eax, %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 364(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 108(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 492(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: andl 236(%ebx), %esi +; X86-NEXT: orl %eax, %esi +; X86-NEXT: orl %ecx, %esi +; X86-NEXT: orl %edx, %esi +; X86-NEXT: orl %edi, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 284(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 28(%ebx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 412(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: andl 156(%ebx), %edi +; X86-NEXT: orl %eax, %edi +; X86-NEXT: orl %ecx, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 348(%ebx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 92(%ebx), %edx +; X86-NEXT: orl %eax, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 476(%ebx), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 220(%ebx), %eax +; X86-NEXT: orl %ecx, %eax +; X86-NEXT: orl %edx, %eax +; X86-NEXT: orl %edi, %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 316(%ebx), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 60(%ebx), %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 444(%ebx), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Reload +; X86-NEXT: andl 188(%ebx), %edi +; X86-NEXT: orl %ecx, %edi +; X86-NEXT: orl %edx, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 380(%ebx), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: andl 124(%ebx), %edx +; X86-NEXT: orl %ecx, %edx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 508(%ebx), %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload +; X86-NEXT: movl 8(%ebp), %esi +; X86-NEXT: andl 252(%esi), %ebx +; X86-NEXT: orl %ecx, %ebx +; X86-NEXT: orl %edx, %ebx +; X86-NEXT: orl %edi, %ebx +; X86-NEXT: orl %eax, %ebx +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: negl %ecx +; X86-NEXT: movl 1648(%esp,%ecx), %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Reload +; X86-NEXT: shldl %cl, %edi, %esi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: shldl %cl, %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Folded Spill +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload +; X86-NEXT: shldl %cl, %edx, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: movl 8(%ebp), %edx +; X86-NEXT: andl 128(%edx), %ecx +; X86-NEXT: andl 384(%edx), %edi +; X86-NEXT: orl %ecx, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: # kill: def $cl killed $cl killed $ecx +; X86-NEXT: shll %cl, %eax +; X86-NEXT: andl (%edx), %eax +; X86-NEXT: orl %eax, %edi +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 256(%edx), %eax +; X86-NEXT: orl %eax, %edi +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 260(%edx), %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: andl 4(%edx), %ecx +; X86-NEXT: orl %eax, %ecx +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload +; X86-NEXT: andl 132(%edx), %eax +; X86-NEXT: andl 388(%edx), %esi +; X86-NEXT: orl %eax, %esi +; X86-NEXT: orl %ecx, %esi +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: orl {{[-0-9]+}}(%e{{[sb]}}p), %esi # 4-byte Folded Reload +; X86-NEXT: orl %ebx, %esi +; X86-NEXT: orl %edi, %esi +; X86-NEXT: setne %al +; X86-NEXT: leal -12(%ebp), %esp +; X86-NEXT: popl %esi +; X86-NEXT: popl %edi +; X86-NEXT: popl %ebx +; X86-NEXT: popl %ebp ; X86-NEXT: retl ; -; X64-LABEL: test_ne_i4096: -; X64: # %bb.0: -; X64-NEXT: movl %esi, %eax -; X64-NEXT: andl $4064, %eax # imm = 0xFE0 -; X64-NEXT: shrl $3, %eax -; X64-NEXT: movl (%rdi,%rax), %eax -; X64-NEXT: btl %esi, %eax -; X64-NEXT: setb %al -; X64-NEXT: retq +; SSE-LABEL: test_ne_i4096: +; SSE: # %bb.0: +; SSE-NEXT: pushq %rbp +; SSE-NEXT: pushq %r15 +; SSE-NEXT: pushq %r14 +; SSE-NEXT: pushq %r13 +; SSE-NEXT: pushq %r12 +; SSE-NEXT: pushq %rbx +; SSE-NEXT: subq $1576, %rsp # imm = 0x628 +; SSE-NEXT: movl %esi, %ecx +; SSE-NEXT: movl %esi, %eax +; SSE-NEXT: andl $4032, %eax # imm = 0xFC0 +; SSE-NEXT: xorps %xmm0, %xmm0 +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movups %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movaps %xmm0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movq $0, {{[0-9]+}}(%rsp) +; SSE-NEXT: movq $1, {{[0-9]+}}(%rsp) +; SSE-NEXT: andl $63, %ecx +; SSE-NEXT: shrl $3, %eax +; SSE-NEXT: negl %eax +; SSE-NEXT: movslq %eax, %rsi +; SSE-NEXT: movq 1296(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1304(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1552(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1560(%rsp,%rsi), %rax +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1168(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1176(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1424(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1432(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1232(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1240(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1488(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1496(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1104(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1112(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1360(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, (%rsp) # 8-byte Spill +; SSE-NEXT: movq 1368(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1264(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1272(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1520(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1528(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1136(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1144(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1392(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1400(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1200(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1208(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1456(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1464(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1072(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1080(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1328(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1336(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1280(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1288(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1536(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1544(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1152(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1160(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1408(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1416(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1216(%rsp,%rsi), %r11 +; SSE-NEXT: movq 1224(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %r11, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1472(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1480(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1088(%rsp,%rsi), %r9 +; SSE-NEXT: movq 1096(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %r9, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1344(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1352(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1248(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1256(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rax, %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1504(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1512(%rsp,%rsi), %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rdx, %rax +; SSE-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1120(%rsp,%rsi), %rax +; SSE-NEXT: movq 1128(%rsp,%rsi), %r8 +; SSE-NEXT: movq %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: shldq %cl, %rax, %r8 +; SSE-NEXT: movq %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1376(%rsp,%rsi), %r13 +; SSE-NEXT: movq 1384(%rsp,%rsi), %rbx +; SSE-NEXT: movq %rbx, %r8 +; SSE-NEXT: shldq %cl, %r13, %r8 +; SSE-NEXT: movq %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1184(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1192(%rsp,%rsi), %r15 +; SSE-NEXT: movq %r15, %r14 +; SSE-NEXT: shldq %cl, %rdx, %r14 +; SSE-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1440(%rsp,%rsi), %r10 +; SSE-NEXT: movq 1448(%rsp,%rsi), %rdx +; SSE-NEXT: movq %rdx, %r14 +; SSE-NEXT: shldq %cl, %r10, %r14 +; SSE-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1312(%rsp,%rsi), %r14 +; SSE-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq 1320(%rsp,%rsi), %rbp +; SSE-NEXT: movq %rbp, %r12 +; SSE-NEXT: shldq %cl, %r14, %r12 +; SSE-NEXT: movq %r12, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r8, (%rsp) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: shldq %cl, %rbx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: shldq %cl, %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq 1064(%rsp,%rsi), %rbx +; SSE-NEXT: shldq %cl, %rbx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; SSE-NEXT: shldq %cl, %rbp, %r14 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; SSE-NEXT: shldq %cl, %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; SSE-NEXT: shldq %cl, %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; SSE-NEXT: shldq %cl, %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; SSE-NEXT: shldq %cl, %rdx, %r11 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r15, %rdx +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r15, %r9 +; SSE-NEXT: movq %r9, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r15, %r8 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Reload +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r15, %rbp +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r15, %r9 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r15, %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r15, %r13 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r12 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r12, %r15 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r12 # 8-byte Reload +; SSE-NEXT: shldq %cl, %r12, %r10 +; SSE-NEXT: andq 384(%rdi), %r10 +; SSE-NEXT: andq 128(%rdi), %r15 +; SSE-NEXT: andq 320(%rdi), %r13 +; SSE-NEXT: andq 64(%rdi), %rax +; SSE-NEXT: orq %r10, %r15 +; SSE-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: orq %r13, %rax +; SSE-NEXT: andq 448(%rdi), %r9 +; SSE-NEXT: andq 192(%rdi), %rbp +; SSE-NEXT: orq %r9, %rbp +; SSE-NEXT: orq %rax, %rbp +; SSE-NEXT: movq %rbp, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: andq 288(%rdi), %r8 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; SSE-NEXT: andq 32(%rdi), %r9 +; SSE-NEXT: andq 416(%rdi), %rdx +; SSE-NEXT: andq 160(%rdi), %r11 +; SSE-NEXT: orq %r8, %r9 +; SSE-NEXT: orq %rdx, %r11 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; SSE-NEXT: andq 352(%rdi), %rdx +; SSE-NEXT: orq %r9, %r11 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 96(%rdi), %rax +; SSE-NEXT: orq %rdx, %rax +; SSE-NEXT: movq %rax, %rdx +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 480(%rdi), %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: andq 224(%rdi), %r8 +; SSE-NEXT: orq %rax, %r8 +; SSE-NEXT: orq %rdx, %r8 +; SSE-NEXT: andq 272(%rdi), %r14 +; SSE-NEXT: orq %r11, %r8 +; SSE-NEXT: movq %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 16(%rdi), %rax +; SSE-NEXT: orq %r14, %rax +; SSE-NEXT: movq %rax, %r8 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; SSE-NEXT: andq 400(%rdi), %rdx +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 144(%rdi), %rax +; SSE-NEXT: orq %rdx, %rax +; SSE-NEXT: orq %r8, %rax +; SSE-NEXT: movq %rax, %r8 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; SSE-NEXT: andq 336(%rdi), %r9 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 80(%rdi), %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; SSE-NEXT: andq 464(%rdi), %rdx +; SSE-NEXT: orq %r9, %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Reload +; SSE-NEXT: andq 208(%rdi), %r11 +; SSE-NEXT: orq %rdx, %r11 +; SSE-NEXT: orq %rax, %r11 +; SSE-NEXT: orq %r8, %r11 +; SSE-NEXT: movq (%rsp), %rdx # 8-byte Reload +; SSE-NEXT: andq 304(%rdi), %rdx +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 48(%rdi), %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; SSE-NEXT: andq 432(%rdi), %r9 +; SSE-NEXT: orq %rdx, %rax +; SSE-NEXT: movq %rax, %r10 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: andq 176(%rdi), %r8 +; SSE-NEXT: orq %r9, %r8 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; SSE-NEXT: andq 368(%rdi), %r9 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 112(%rdi), %rax +; SSE-NEXT: orq %r10, %r8 +; SSE-NEXT: movq %r8, %r10 +; SSE-NEXT: orq %r9, %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: andq 496(%rdi), %r8 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Reload +; SSE-NEXT: andq 240(%rdi), %rbp +; SSE-NEXT: orq %r8, %rbp +; SSE-NEXT: orq %rax, %rbp +; SSE-NEXT: orq %r10, %rbp +; SSE-NEXT: orq %r11, %rbp +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 392(%rdi), %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r12 # 8-byte Reload +; SSE-NEXT: andq 136(%rdi), %r12 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; SSE-NEXT: andq 328(%rdi), %rdx +; SSE-NEXT: orq %rax, %r12 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 72(%rdi), %rax +; SSE-NEXT: orq %rdx, %rax +; SSE-NEXT: movq %rax, %rdx +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 456(%rdi), %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; SSE-NEXT: andq 200(%rdi), %r13 +; SSE-NEXT: orq %rax, %r13 +; SSE-NEXT: orq %rdx, %r13 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; SSE-NEXT: andq 296(%rdi), %rdx +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 40(%rdi), %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: andq 424(%rdi), %r8 +; SSE-NEXT: orq %rdx, %rax +; SSE-NEXT: movq %rax, %r9 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; SSE-NEXT: andq 168(%rdi), %rdx +; SSE-NEXT: orq %r8, %rdx +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: andq 360(%rdi), %r8 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 104(%rdi), %rax +; SSE-NEXT: orq %r9, %rdx +; SSE-NEXT: orq %r8, %rax +; SSE-NEXT: movq %rax, %r8 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 488(%rdi), %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; SSE-NEXT: andq 232(%rdi), %r15 +; SSE-NEXT: orq %rax, %r15 +; SSE-NEXT: orq %r8, %r15 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: andq 280(%rdi), %r8 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 24(%rdi), %rax +; SSE-NEXT: orq %rdx, %r15 +; SSE-NEXT: orq %r8, %rax +; SSE-NEXT: movq %rax, %r10 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: andq 408(%rdi), %r8 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 152(%rdi), %rax +; SSE-NEXT: orq %r8, %rax +; SSE-NEXT: orq %r10, %rax +; SSE-NEXT: movq %rax, %r10 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Reload +; SSE-NEXT: andq 344(%rdi), %r11 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: andq 88(%rdi), %r8 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 472(%rdi), %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; SSE-NEXT: andq 216(%rdi), %r14 +; SSE-NEXT: orq %r11, %r8 +; SSE-NEXT: orq %rax, %r14 +; SSE-NEXT: orq %r8, %r14 +; SSE-NEXT: orq %r10, %r14 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Reload +; SSE-NEXT: andq 312(%rdi), %r11 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; SSE-NEXT: andq 56(%rdi), %r10 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: andq 440(%rdi), %r8 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; SSE-NEXT: andq 184(%rdi), %r9 +; SSE-NEXT: orq %r11, %r10 +; SSE-NEXT: orq %r8, %r9 +; SSE-NEXT: orq %r10, %r9 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rdx # 8-byte Reload +; SSE-NEXT: shldq %cl, %rax, %rdx +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; SSE-NEXT: andq 376(%rdi), %r10 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; SSE-NEXT: andq 120(%rdi), %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Reload +; SSE-NEXT: andq 504(%rdi), %r11 +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; SSE-NEXT: andq 248(%rdi), %r8 +; SSE-NEXT: orq %r10, %rax +; SSE-NEXT: movq %rax, %r10 +; SSE-NEXT: orq %r11, %r8 +; SSE-NEXT: movq 1056(%rsp,%rsi), %rax +; SSE-NEXT: shldq %cl, %rax, %rbx +; SSE-NEXT: # kill: def $cl killed $cl killed $ecx +; SSE-NEXT: shlq %cl, %rax +; SSE-NEXT: orq %r10, %r8 +; SSE-NEXT: orq %r9, %r8 +; SSE-NEXT: andq 256(%rdi), %rdx +; SSE-NEXT: orq %r14, %r8 +; SSE-NEXT: andq (%rdi), %rax +; SSE-NEXT: orq %rdx, %rax +; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Folded Reload +; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Folded Reload +; SSE-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Folded Reload +; SSE-NEXT: orq %rbp, %rax +; SSE-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload +; SSE-NEXT: andq 264(%rdi), %rcx +; SSE-NEXT: andq 8(%rdi), %rbx +; SSE-NEXT: orq %rcx, %rbx +; SSE-NEXT: orq %r12, %rbx +; SSE-NEXT: orq %r13, %rbx +; SSE-NEXT: orq %r15, %rbx +; SSE-NEXT: orq %r8, %rbx +; SSE-NEXT: orq %rax, %rbx +; SSE-NEXT: setne %al +; SSE-NEXT: addq $1576, %rsp # imm = 0x628 +; SSE-NEXT: popq %rbx +; SSE-NEXT: popq %r12 +; SSE-NEXT: popq %r13 +; SSE-NEXT: popq %r14 +; SSE-NEXT: popq %r15 +; SSE-NEXT: popq %rbp +; SSE-NEXT: retq +; +; AVX2-LABEL: test_ne_i4096: +; AVX2: # %bb.0: +; AVX2-NEXT: pushq %rbp +; AVX2-NEXT: pushq %r15 +; AVX2-NEXT: pushq %r14 +; AVX2-NEXT: pushq %r13 +; AVX2-NEXT: pushq %r12 +; AVX2-NEXT: pushq %rbx +; AVX2-NEXT: subq $1560, %rsp # imm = 0x618 +; AVX2-NEXT: movl %esi, %ecx +; AVX2-NEXT: movl %esi, %eax +; AVX2-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: andl $4032, %eax # imm = 0xFC0 +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: vmovss {{.*#+}} xmm0 = [1,0,0,0] +; AVX2-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX2-NEXT: andl $63, %ecx +; AVX2-NEXT: shrl $3, %eax +; AVX2-NEXT: negl %eax +; AVX2-NEXT: movslq %eax, %rsi +; AVX2-NEXT: movq 1280(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1288(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1536(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1544(%rsp,%rsi), %rax +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1152(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1160(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1408(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1416(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1216(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, (%rsp) # 8-byte Spill +; AVX2-NEXT: movq 1224(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1472(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1480(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1088(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1096(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1344(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1352(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1248(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1256(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1504(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1512(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1120(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1128(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1376(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1384(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1184(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1192(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1440(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1448(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1056(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1064(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1312(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1320(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1264(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1272(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1520(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1528(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1136(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1144(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1392(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1400(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1200(%rsp,%rsi), %r11 +; AVX2-NEXT: movq 1208(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %r11, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1456(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1464(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1072(%rsp,%rsi), %r12 +; AVX2-NEXT: movq 1080(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %r12, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1328(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1336(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rdx, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1232(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1240(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rax, %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1488(%rsp,%rsi), %rbp +; AVX2-NEXT: movq 1496(%rsp,%rsi), %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rbp, %rax +; AVX2-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1104(%rsp,%rsi), %rax +; AVX2-NEXT: movq 1112(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: shldq %cl, %rax, %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1360(%rsp,%rsi), %r10 +; AVX2-NEXT: movq 1368(%rsp,%rsi), %r8 +; AVX2-NEXT: movq %r8, %rdx +; AVX2-NEXT: shldq %cl, %r10, %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1168(%rsp,%rsi), %r9 +; AVX2-NEXT: movq %r9, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1176(%rsp,%rsi), %rbx +; AVX2-NEXT: movq %rbx, %rdx +; AVX2-NEXT: shldq %cl, %r9, %rdx +; AVX2-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1424(%rsp,%rsi), %r9 +; AVX2-NEXT: movq 1432(%rsp,%rsi), %rdx +; AVX2-NEXT: movq %rdx, %r14 +; AVX2-NEXT: shldq %cl, %r9, %r14 +; AVX2-NEXT: movq %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1296(%rsp,%rsi), %r15 +; AVX2-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq 1304(%rsp,%rsi), %r14 +; AVX2-NEXT: movq %r14, %r13 +; AVX2-NEXT: shldq %cl, %r15, %r13 +; AVX2-NEXT: movq %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r15, (%rsp) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: shldq %cl, %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: shldq %cl, %rbx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: shldq %cl, %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq 1048(%rsp,%rsi), %rdx +; AVX2-NEXT: shldq %cl, %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: shldq %cl, %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload +; AVX2-NEXT: shldq %cl, %rbx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload +; AVX2-NEXT: shldq %cl, %rbx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload +; AVX2-NEXT: shldq %cl, %rbx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r14, %rbx +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r14, %r11 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r14, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r14, %r12 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r14, %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r14, %r13 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r14, %rbp +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r14, %rax +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r14, %r10 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r15, %r14 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r15, %r9 +; AVX2-NEXT: andq 384(%rdi), %r9 +; AVX2-NEXT: andq 128(%rdi), %r14 +; AVX2-NEXT: andq 320(%rdi), %r10 +; AVX2-NEXT: orq %r9, %r14 +; AVX2-NEXT: movq %r14, %r15 +; AVX2-NEXT: andq 64(%rdi), %rax +; AVX2-NEXT: orq %r10, %rax +; AVX2-NEXT: andq 448(%rdi), %rbp +; AVX2-NEXT: andq 192(%rdi), %r13 +; AVX2-NEXT: orq %rbp, %r13 +; AVX2-NEXT: orq %rax, %r13 +; AVX2-NEXT: movq %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX2-NEXT: andq 288(%rdi), %r8 +; AVX2-NEXT: andq 32(%rdi), %r12 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 416(%rdi), %rax +; AVX2-NEXT: orq %r8, %r12 +; AVX2-NEXT: andq 160(%rdi), %r11 +; AVX2-NEXT: orq %rax, %r11 +; AVX2-NEXT: andq 352(%rdi), %rbx +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 96(%rdi), %rax +; AVX2-NEXT: orq %r12, %r11 +; AVX2-NEXT: orq %rbx, %rax +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX2-NEXT: andq 480(%rdi), %r10 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX2-NEXT: andq 224(%rdi), %r13 +; AVX2-NEXT: orq %r10, %r13 +; AVX2-NEXT: orq %rax, %r13 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 272(%rdi), %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 16(%rdi), %rax +; AVX2-NEXT: orq %r11, %r13 +; AVX2-NEXT: orq %r8, %rax +; AVX2-NEXT: movq %rax, %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; AVX2-NEXT: andq 400(%rdi), %r9 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 144(%rdi), %rax +; AVX2-NEXT: orq %r9, %rax +; AVX2-NEXT: orq %r8, %rax +; AVX2-NEXT: movq %rax, %r9 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX2-NEXT: andq 336(%rdi), %r10 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 80(%rdi), %rax +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 464(%rdi), %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Reload +; AVX2-NEXT: andq 208(%rdi), %r11 +; AVX2-NEXT: orq %r10, %rax +; AVX2-NEXT: orq %r8, %r11 +; AVX2-NEXT: orq %rax, %r11 +; AVX2-NEXT: orq %r9, %r11 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; AVX2-NEXT: andq 304(%rdi), %r9 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 48(%rdi), %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX2-NEXT: andq 432(%rdi), %r10 +; AVX2-NEXT: movq (%rsp), %rax # 8-byte Reload +; AVX2-NEXT: andq 176(%rdi), %rax +; AVX2-NEXT: orq %r9, %r8 +; AVX2-NEXT: movq %r8, %r9 +; AVX2-NEXT: orq %r10, %rax +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 368(%rdi), %r8 +; AVX2-NEXT: orq %r9, %rax +; AVX2-NEXT: movq %rax, %r10 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 112(%rdi), %rax +; AVX2-NEXT: orq %r8, %rax +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 496(%rdi), %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; AVX2-NEXT: andq 240(%rdi), %r9 +; AVX2-NEXT: orq %r8, %r9 +; AVX2-NEXT: orq %rax, %r9 +; AVX2-NEXT: orq %r10, %r9 +; AVX2-NEXT: orq %r11, %r9 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX2-NEXT: andq 392(%rdi), %r10 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Reload +; AVX2-NEXT: andq 136(%rdi), %rbp +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 328(%rdi), %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 72(%rdi), %rax +; AVX2-NEXT: orq %r10, %rbp +; AVX2-NEXT: orq %r8, %rax +; AVX2-NEXT: movq %rax, %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 456(%rdi), %rax +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r12 # 8-byte Reload +; AVX2-NEXT: andq 200(%rdi), %r12 +; AVX2-NEXT: orq %rax, %r12 +; AVX2-NEXT: orq %r8, %r12 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX2-NEXT: andq 296(%rdi), %r10 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 40(%rdi), %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Reload +; AVX2-NEXT: andq 424(%rdi), %r11 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 168(%rdi), %rax +; AVX2-NEXT: orq %r10, %r8 +; AVX2-NEXT: movq %r8, %r10 +; AVX2-NEXT: orq %r11, %rax +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 360(%rdi), %r8 +; AVX2-NEXT: orq %r10, %rax +; AVX2-NEXT: movq %rax, %r10 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 104(%rdi), %rax +; AVX2-NEXT: orq %r8, %rax +; AVX2-NEXT: movq %rax, %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 488(%rdi), %rax +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; AVX2-NEXT: andq 232(%rdi), %r14 +; AVX2-NEXT: orq %rax, %r14 +; AVX2-NEXT: orq %r8, %r14 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 280(%rdi), %r8 +; AVX2-NEXT: orq %r10, %r14 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 24(%rdi), %rax +; AVX2-NEXT: orq %r8, %rax +; AVX2-NEXT: movq %rax, %r10 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 408(%rdi), %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 152(%rdi), %rax +; AVX2-NEXT: orq %r8, %rax +; AVX2-NEXT: orq %r10, %rax +; AVX2-NEXT: movq %rax, %r10 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Reload +; AVX2-NEXT: andq 344(%rdi), %r11 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 88(%rdi), %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 472(%rdi), %rax +; AVX2-NEXT: orq %r11, %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload +; AVX2-NEXT: andq 216(%rdi), %rbx +; AVX2-NEXT: orq %rax, %rbx +; AVX2-NEXT: orq %r8, %rbx +; AVX2-NEXT: orq %r10, %rbx +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 312(%rdi), %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 56(%rdi), %rax +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX2-NEXT: andq 440(%rdi), %r10 +; AVX2-NEXT: orq %r8, %rax +; AVX2-NEXT: movq %rax, %r11 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 184(%rdi), %r8 +; AVX2-NEXT: orq %r10, %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX2-NEXT: andq 376(%rdi), %r10 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 120(%rdi), %rax +; AVX2-NEXT: orq %r11, %r8 +; AVX2-NEXT: movq %r8, %r11 +; AVX2-NEXT: orq %r10, %rax +; AVX2-NEXT: movq %rax, %r10 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: andq 504(%rdi), %r8 +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX2-NEXT: andq 248(%rdi), %rax +; AVX2-NEXT: orq %r8, %rax +; AVX2-NEXT: orq %r10, %rax +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX2-NEXT: shldq %cl, %r8, %r10 +; AVX2-NEXT: orq %r11, %rax +; AVX2-NEXT: movq 1040(%rsp,%rsi), %rsi +; AVX2-NEXT: orq %rbx, %rax +; AVX2-NEXT: movq %rax, %r8 +; AVX2-NEXT: shlxq %rcx, %rsi, %rax +; AVX2-NEXT: andq 256(%rdi), %r10 +; AVX2-NEXT: andq (%rdi), %rax +; AVX2-NEXT: orq %r10, %rax +; AVX2-NEXT: orq %r15, %rax +; AVX2-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Folded Reload +; AVX2-NEXT: orq %r13, %rax +; AVX2-NEXT: # kill: def $cl killed $cl killed $rcx +; AVX2-NEXT: shldq %cl, %rsi, %rdx +; AVX2-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rcx # 8-byte Reload +; AVX2-NEXT: andq 264(%rdi), %rcx +; AVX2-NEXT: andq 8(%rdi), %rdx +; AVX2-NEXT: orq %r9, %rax +; AVX2-NEXT: orq %rcx, %rdx +; AVX2-NEXT: orq %rbp, %rdx +; AVX2-NEXT: orq %r12, %rdx +; AVX2-NEXT: orq %r14, %rdx +; AVX2-NEXT: orq %r8, %rdx +; AVX2-NEXT: orq %rax, %rdx +; AVX2-NEXT: setne %al +; AVX2-NEXT: addq $1560, %rsp # imm = 0x618 +; AVX2-NEXT: popq %rbx +; AVX2-NEXT: popq %r12 +; AVX2-NEXT: popq %r13 +; AVX2-NEXT: popq %r14 +; AVX2-NEXT: popq %r15 +; AVX2-NEXT: popq %rbp +; AVX2-NEXT: vzeroupper +; AVX2-NEXT: retq +; +; AVX512-LABEL: test_ne_i4096: +; AVX512: # %bb.0: +; AVX512-NEXT: pushq %rbp +; AVX512-NEXT: pushq %r15 +; AVX512-NEXT: pushq %r14 +; AVX512-NEXT: pushq %r13 +; AVX512-NEXT: pushq %r12 +; AVX512-NEXT: pushq %rbx +; AVX512-NEXT: subq $1560, %rsp # imm = 0x618 +; AVX512-NEXT: movl %esi, %ecx +; AVX512-NEXT: movl %esi, %eax +; AVX512-NEXT: andl $4032, %eax # imm = 0xFC0 +; AVX512-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [1,0,0,0] +; AVX512-NEXT: vmovups %ymm0, {{[0-9]+}}(%rsp) +; AVX512-NEXT: andl $63, %ecx +; AVX512-NEXT: shrl $3, %eax +; AVX512-NEXT: negl %eax +; AVX512-NEXT: movslq %eax, %rsi +; AVX512-NEXT: movq 1280(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1288(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1536(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1544(%rsp,%rsi), %rax +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1152(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1160(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1408(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1416(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1216(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, (%rsp) # 8-byte Spill +; AVX512-NEXT: movq 1224(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1472(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1480(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1088(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1096(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1344(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1352(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1248(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1256(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1504(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1512(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1120(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1128(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1376(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1384(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1184(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1192(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1440(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1448(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1056(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1064(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1312(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1320(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1264(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1272(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1520(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1528(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1136(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1144(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1392(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1400(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1200(%rsp,%rsi), %r10 +; AVX512-NEXT: movq 1208(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %r10, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1456(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1464(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1072(%rsp,%rsi), %r14 +; AVX512-NEXT: movq 1080(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %r14, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1328(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1336(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rdx, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1232(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1240(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rax, %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1488(%rsp,%rsi), %r12 +; AVX512-NEXT: movq 1496(%rsp,%rsi), %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %r12, %rax +; AVX512-NEXT: movq %rax, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1104(%rsp,%rsi), %rax +; AVX512-NEXT: movq 1112(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: shldq %cl, %rax, %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1360(%rsp,%rsi), %r11 +; AVX512-NEXT: movq 1368(%rsp,%rsi), %rbx +; AVX512-NEXT: movq %rbx, %rdx +; AVX512-NEXT: shldq %cl, %r11, %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1168(%rsp,%rsi), %r9 +; AVX512-NEXT: movq %r9, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1176(%rsp,%rsi), %r8 +; AVX512-NEXT: movq %r8, %rdx +; AVX512-NEXT: shldq %cl, %r9, %rdx +; AVX512-NEXT: movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1424(%rsp,%rsi), %r9 +; AVX512-NEXT: movq 1432(%rsp,%rsi), %rdx +; AVX512-NEXT: movq %rdx, %r15 +; AVX512-NEXT: shldq %cl, %r9, %r15 +; AVX512-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1296(%rsp,%rsi), %rbp +; AVX512-NEXT: movq %rbp, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq 1304(%rsp,%rsi), %r15 +; AVX512-NEXT: movq %r15, %r13 +; AVX512-NEXT: shldq %cl, %rbp, %r13 +; AVX512-NEXT: movq %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r13, (%rsp) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r13, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: shldq %cl, %rbx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: shldq %cl, %r8, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: shldq %cl, %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq 1048(%rsp,%rsi), %rdx +; AVX512-NEXT: shldq %cl, %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload +; AVX512-NEXT: shldq %cl, %rbx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload +; AVX512-NEXT: shldq %cl, %rbx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload +; AVX512-NEXT: shldq %cl, %rbx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r15, %rbx +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r15, %r10 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Folded Spill +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r15, %r14 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r15, %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r13 # 8-byte Reload +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r15, %r13 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r15, %r12 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r15, %rax +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %r15, %r11 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Reload +; AVX512-NEXT: shldq %cl, %rbp, %r15 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Reload +; AVX512-NEXT: shldq %cl, %rbp, %r9 +; AVX512-NEXT: andq 384(%rdi), %r9 +; AVX512-NEXT: andq 128(%rdi), %r15 +; AVX512-NEXT: orq %r9, %r15 +; AVX512-NEXT: movq %r15, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill +; AVX512-NEXT: andq 320(%rdi), %r11 +; AVX512-NEXT: andq 64(%rdi), %rax +; AVX512-NEXT: orq %r11, %rax +; AVX512-NEXT: andq 448(%rdi), %r12 +; AVX512-NEXT: andq 192(%rdi), %r13 +; AVX512-NEXT: orq %r12, %r13 +; AVX512-NEXT: orq %rax, %r13 +; AVX512-NEXT: andq 288(%rdi), %r8 +; AVX512-NEXT: andq 32(%rdi), %r14 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 416(%rdi), %rax +; AVX512-NEXT: orq %r8, %r14 +; AVX512-NEXT: andq 160(%rdi), %r10 +; AVX512-NEXT: orq %rax, %r10 +; AVX512-NEXT: andq 352(%rdi), %rbx +; AVX512-NEXT: orq %r14, %r10 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 96(%rdi), %rax +; AVX512-NEXT: orq %rbx, %rax +; AVX512-NEXT: movq %rax, %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 480(%rdi), %rax +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r15 # 8-byte Reload +; AVX512-NEXT: andq 224(%rdi), %r15 +; AVX512-NEXT: orq %rax, %r15 +; AVX512-NEXT: orq %r8, %r15 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: andq 272(%rdi), %r8 +; AVX512-NEXT: orq %r10, %r15 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 16(%rdi), %rax +; AVX512-NEXT: orq %r8, %rax +; AVX512-NEXT: movq %rax, %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; AVX512-NEXT: andq 400(%rdi), %r9 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 144(%rdi), %rax +; AVX512-NEXT: orq %r9, %rax +; AVX512-NEXT: orq %r8, %rax +; AVX512-NEXT: movq %rax, %r9 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX512-NEXT: andq 336(%rdi), %r10 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 80(%rdi), %rax +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: andq 464(%rdi), %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Reload +; AVX512-NEXT: andq 208(%rdi), %r11 +; AVX512-NEXT: orq %r10, %rax +; AVX512-NEXT: orq %r8, %r11 +; AVX512-NEXT: orq %rax, %r11 +; AVX512-NEXT: orq %r9, %r11 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX512-NEXT: andq 304(%rdi), %r10 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 48(%rdi), %rax +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; AVX512-NEXT: andq 432(%rdi), %r9 +; AVX512-NEXT: movq (%rsp), %r8 # 8-byte Reload +; AVX512-NEXT: andq 176(%rdi), %r8 +; AVX512-NEXT: orq %r10, %rax +; AVX512-NEXT: movq %rax, %r10 +; AVX512-NEXT: orq %r9, %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; AVX512-NEXT: andq 368(%rdi), %r9 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 112(%rdi), %rax +; AVX512-NEXT: orq %r10, %r8 +; AVX512-NEXT: movq %r8, %r10 +; AVX512-NEXT: orq %r9, %rax +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: andq 496(%rdi), %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r9 # 8-byte Reload +; AVX512-NEXT: andq 240(%rdi), %r9 +; AVX512-NEXT: orq %r8, %r9 +; AVX512-NEXT: orq %rax, %r9 +; AVX512-NEXT: orq %r10, %r9 +; AVX512-NEXT: orq %r11, %r9 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX512-NEXT: andq 392(%rdi), %r10 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbp # 8-byte Reload +; AVX512-NEXT: andq 136(%rdi), %rbp +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: andq 328(%rdi), %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 72(%rdi), %rax +; AVX512-NEXT: orq %r10, %rbp +; AVX512-NEXT: orq %r8, %rax +; AVX512-NEXT: movq %rax, %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 456(%rdi), %rax +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r12 # 8-byte Reload +; AVX512-NEXT: andq 200(%rdi), %r12 +; AVX512-NEXT: orq %rax, %r12 +; AVX512-NEXT: orq %r8, %r12 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: andq 296(%rdi), %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 40(%rdi), %rax +; AVX512-NEXT: orq %r8, %rax +; AVX512-NEXT: movq %rax, %r10 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: andq 424(%rdi), %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 168(%rdi), %rax +; AVX512-NEXT: orq %r8, %rax +; AVX512-NEXT: orq %r10, %rax +; AVX512-NEXT: movq %rax, %r10 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: andq 360(%rdi), %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 104(%rdi), %rax +; AVX512-NEXT: orq %r8, %rax +; AVX512-NEXT: movq %rax, %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 488(%rdi), %rax +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r14 # 8-byte Reload +; AVX512-NEXT: andq 232(%rdi), %r14 +; AVX512-NEXT: orq %rax, %r14 +; AVX512-NEXT: orq %r8, %r14 +; AVX512-NEXT: orq %r10, %r14 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: andq 280(%rdi), %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 24(%rdi), %rax +; AVX512-NEXT: orq %r8, %rax +; AVX512-NEXT: movq %rax, %r10 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: andq 408(%rdi), %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 152(%rdi), %rax +; AVX512-NEXT: orq %r8, %rax +; AVX512-NEXT: orq %r10, %rax +; AVX512-NEXT: movq %rax, %r10 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r11 # 8-byte Reload +; AVX512-NEXT: andq 344(%rdi), %r11 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: andq 88(%rdi), %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 472(%rdi), %rax +; AVX512-NEXT: orq %r11, %r8 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rbx # 8-byte Reload +; AVX512-NEXT: andq 216(%rdi), %rbx +; AVX512-NEXT: orq %rax, %rbx +; AVX512-NEXT: orq %r8, %rbx +; AVX512-NEXT: orq %r10, %rbx +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX512-NEXT: andq 312(%rdi), %r10 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 56(%rdi), %rax +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: andq 440(%rdi), %r8 +; AVX512-NEXT: orq %r10, %rax +; AVX512-NEXT: movq %rax, %r10 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 184(%rdi), %rax +; AVX512-NEXT: orq %r8, %rax +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: andq 376(%rdi), %r8 +; AVX512-NEXT: orq %r10, %rax +; AVX512-NEXT: movq %rax, %r11 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 120(%rdi), %rax +; AVX512-NEXT: orq %r8, %rax +; AVX512-NEXT: movq %rax, %r10 +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 504(%rdi), %rax +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r8 # 8-byte Reload +; AVX512-NEXT: andq 248(%rdi), %r8 +; AVX512-NEXT: orq %rax, %r8 +; AVX512-NEXT: orq %r10, %r8 +; AVX512-NEXT: orq %r11, %r8 +; AVX512-NEXT: movq 1040(%rsp,%rsi), %rax +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rsi # 8-byte Reload +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %r10 # 8-byte Reload +; AVX512-NEXT: shldq %cl, %rsi, %r10 +; AVX512-NEXT: orq %rbx, %r8 +; AVX512-NEXT: shlxq %rcx, %rax, %rsi +; AVX512-NEXT: andq 256(%rdi), %r10 +; AVX512-NEXT: andq (%rdi), %rsi +; AVX512-NEXT: orq %r10, %rsi +; AVX512-NEXT: orq {{[-0-9]+}}(%r{{[sb]}}p), %rsi # 8-byte Folded Reload +; AVX512-NEXT: orq %r13, %rsi +; AVX512-NEXT: orq %r15, %rsi +; AVX512-NEXT: # kill: def $cl killed $cl killed $rcx +; AVX512-NEXT: shldq %cl, %rax, %rdx +; AVX512-NEXT: orq %r9, %rsi +; AVX512-NEXT: movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload +; AVX512-NEXT: andq 264(%rdi), %rax +; AVX512-NEXT: andq 8(%rdi), %rdx +; AVX512-NEXT: orq %rax, %rdx +; AVX512-NEXT: orq %rbp, %rdx +; AVX512-NEXT: orq %r12, %rdx +; AVX512-NEXT: orq %r14, %rdx +; AVX512-NEXT: orq %r8, %rdx +; AVX512-NEXT: orq %rsi, %rdx +; AVX512-NEXT: setne %al +; AVX512-NEXT: addq $1560, %rsp # imm = 0x618 +; AVX512-NEXT: popq %rbx +; AVX512-NEXT: popq %r12 +; AVX512-NEXT: popq %r13 +; AVX512-NEXT: popq %r14 +; AVX512-NEXT: popq %r15 +; AVX512-NEXT: popq %rbp +; AVX512-NEXT: vzeroupper +; AVX512-NEXT: retq %rem = and i32 %position, 4095 %ofs = zext nneg i32 %rem to i4096 %bit = shl nuw i4096 1, %ofs @@ -1035,8 +7161,8 @@ define i32 @reset_multiload_i128(ptr %word, i32 %position, ptr %p) nounwind { ; X86-NEXT: pushl %edi ; X86-NEXT: pushl %esi ; X86-NEXT: andl $-16, %esp -; X86-NEXT: subl $64, %esp -; X86-NEXT: movl 12(%ebp), %ecx +; X86-NEXT: subl $80, %esp +; X86-NEXT: movzbl 12(%ebp), %ecx ; X86-NEXT: movl $0, {{[0-9]+}}(%esp) ; X86-NEXT: movl $0, {{[0-9]+}}(%esp) ; X86-NEXT: movl $0, {{[0-9]+}}(%esp) @@ -1049,41 +7175,52 @@ define i32 @reset_multiload_i128(ptr %word, i32 %position, ptr %p) nounwind { ; X86-NEXT: shrb $3, %al ; X86-NEXT: andb $12, %al ; X86-NEXT: negb %al -; X86-NEXT: movsbl %al, %edi -; X86-NEXT: movl 36(%esp,%edi), %edx -; X86-NEXT: movl 40(%esp,%edi), %ebx -; X86-NEXT: movl %ebx, %esi -; X86-NEXT: shldl %cl, %edx, %esi -; X86-NEXT: movl 32(%esp,%edi), %eax -; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill -; X86-NEXT: movl 44(%esp,%edi), %edi -; X86-NEXT: shldl %cl, %ebx, %edi -; X86-NEXT: movl %eax, %ebx -; X86-NEXT: # kill: def $cl killed $cl killed $ecx -; X86-NEXT: shll %cl, %ebx -; X86-NEXT: notl %ebx -; X86-NEXT: movl 16(%ebp), %eax -; X86-NEXT: movl (%eax), %eax -; X86-NEXT: movl %eax, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill -; X86-NEXT: movl 12(%ebp), %eax -; X86-NEXT: andl $96, %eax -; X86-NEXT: shrl $3, %eax -; X86-NEXT: movl 8(%ebp), %ecx -; X86-NEXT: movl (%ecx,%eax), %eax -; X86-NEXT: andl %ebx, (%ecx) -; X86-NEXT: movl 12(%ebp), %ecx +; X86-NEXT: movsbl %al, %eax +; X86-NEXT: movl 56(%esp,%eax), %esi +; X86-NEXT: movl 60(%esp,%eax), %edx +; X86-NEXT: shldl %cl, %esi, %edx +; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: movl 48(%esp,%eax), %edi +; X86-NEXT: movl 52(%esp,%eax), %eax +; X86-NEXT: shldl %cl, %eax, %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: shldl %cl, %edi, %eax +; X86-NEXT: movl 8(%ebp), %ebx +; X86-NEXT: shll %cl, %edi +; X86-NEXT: movl 8(%ebx), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %esi, %ecx +; X86-NEXT: movl (%ebx), %esi +; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %edi, %esi +; X86-NEXT: orl %ecx, %esi +; X86-NEXT: movl 12(%ebx), %ecx +; X86-NEXT: movl %ecx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload +; X86-NEXT: movl %eax, %edx +; X86-NEXT: movl 4(%ebx), %ebx +; X86-NEXT: movl %ebx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill +; X86-NEXT: andl %ebx, %eax +; X86-NEXT: orl %ecx, %eax +; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Reload +; X86-NEXT: notl %ecx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %ecx # 4-byte Folded Reload ; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Reload -; X86-NEXT: shldl %cl, %ebx, %edx +; X86-NEXT: notl %ebx +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %ebx # 4-byte Folded Reload ; X86-NEXT: notl %edx -; X86-NEXT: movl 8(%ebp), %ebx -; X86-NEXT: andl %edx, 4(%ebx) -; X86-NEXT: notl %esi -; X86-NEXT: andl %esi, 8(%ebx) +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Folded Reload ; X86-NEXT: notl %edi -; X86-NEXT: andl %edi, 12(%ebx) -; X86-NEXT: btl %ecx, %eax -; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload -; X86-NEXT: jae .LBB22_2 +; X86-NEXT: andl {{[-0-9]+}}(%e{{[sb]}}p), %edi # 4-byte Folded Reload +; X86-NEXT: orl %esi, %eax +; X86-NEXT: movl 16(%ebp), %eax +; X86-NEXT: movl (%eax), %eax +; X86-NEXT: movl 8(%ebp), %esi +; X86-NEXT: movl %ebx, 8(%esi) +; X86-NEXT: movl %ecx, 12(%esi) +; X86-NEXT: movl %edi, (%esi) +; X86-NEXT: movl %edx, 4(%esi) +; X86-NEXT: je .LBB22_2 ; X86-NEXT: # %bb.1: ; X86-NEXT: xorl %eax, %eax ; X86-NEXT: .LBB22_2: @@ -1105,75 +7242,52 @@ define i32 @reset_multiload_i128(ptr %word, i32 %position, ptr %p) nounwind { ; SSE-NEXT: testb $64, %cl ; SSE-NEXT: cmovneq %rsi, %r8 ; SSE-NEXT: cmovneq %rax, %rsi +; SSE-NEXT: movq (%rdi), %rcx +; SSE-NEXT: movq 8(%rdi), %r9 +; SSE-NEXT: movq %r9, %r10 +; SSE-NEXT: andq %r8, %r10 ; SSE-NEXT: notq %r8 +; SSE-NEXT: movq %rcx, %r11 +; SSE-NEXT: andq %rsi, %r11 ; SSE-NEXT: notq %rsi -; SSE-NEXT: movl %ecx, %r9d -; SSE-NEXT: andl $96, %r9d -; SSE-NEXT: shrl $3, %r9d -; SSE-NEXT: movl (%rdi,%r9), %r9d -; SSE-NEXT: btl %ecx, %r9d -; SSE-NEXT: jb .LBB22_2 +; SSE-NEXT: andq %r9, %r8 +; SSE-NEXT: andq %rcx, %rsi +; SSE-NEXT: orq %r10, %r11 +; SSE-NEXT: jne .LBB22_2 ; SSE-NEXT: # %bb.1: ; SSE-NEXT: movl (%rdx), %eax ; SSE-NEXT: .LBB22_2: -; SSE-NEXT: andq %r8, 8(%rdi) -; SSE-NEXT: andq %rsi, (%rdi) +; SSE-NEXT: movq %rsi, (%rdi) +; SSE-NEXT: movq %r8, 8(%rdi) ; SSE-NEXT: # kill: def $eax killed $eax killed $rax ; SSE-NEXT: retq ; -; AVX2-LABEL: reset_multiload_i128: -; AVX2: # %bb.0: -; AVX2-NEXT: movl %esi, %ecx -; AVX2-NEXT: xorl %eax, %eax -; AVX2-NEXT: movl $1, %r8d -; AVX2-NEXT: xorl %esi, %esi -; AVX2-NEXT: shldq %cl, %r8, %rsi -; AVX2-NEXT: shlxq %rcx, %r8, %r8 -; AVX2-NEXT: testb $64, %cl -; AVX2-NEXT: cmovneq %r8, %rsi -; AVX2-NEXT: cmovneq %rax, %r8 -; AVX2-NEXT: notq %rsi -; AVX2-NEXT: notq %r8 -; AVX2-NEXT: movl %ecx, %r9d -; AVX2-NEXT: andl $96, %r9d -; AVX2-NEXT: shrl $3, %r9d -; AVX2-NEXT: movl (%rdi,%r9), %r9d -; AVX2-NEXT: btl %ecx, %r9d -; AVX2-NEXT: jb .LBB22_2 -; AVX2-NEXT: # %bb.1: -; AVX2-NEXT: movl (%rdx), %eax -; AVX2-NEXT: .LBB22_2: -; AVX2-NEXT: andq %rsi, 8(%rdi) -; AVX2-NEXT: andq %r8, (%rdi) -; AVX2-NEXT: # kill: def $eax killed $eax killed $rax -; AVX2-NEXT: retq -; -; AVX512-LABEL: reset_multiload_i128: -; AVX512: # %bb.0: -; AVX512-NEXT: movl %esi, %ecx -; AVX512-NEXT: movl $1, %r8d -; AVX512-NEXT: xorl %esi, %esi -; AVX512-NEXT: shldq %cl, %r8, %rsi -; AVX512-NEXT: xorl %eax, %eax -; AVX512-NEXT: shlxq %rcx, %r8, %r8 -; AVX512-NEXT: testb $64, %cl -; AVX512-NEXT: cmovneq %r8, %rsi -; AVX512-NEXT: cmovneq %rax, %r8 -; AVX512-NEXT: notq %rsi -; AVX512-NEXT: notq %r8 -; AVX512-NEXT: movl %ecx, %r9d -; AVX512-NEXT: andl $96, %r9d -; AVX512-NEXT: shrl $3, %r9d -; AVX512-NEXT: movl (%rdi,%r9), %r9d -; AVX512-NEXT: btl %ecx, %r9d -; AVX512-NEXT: jb .LBB22_2 -; AVX512-NEXT: # %bb.1: -; AVX512-NEXT: movl (%rdx), %eax -; AVX512-NEXT: .LBB22_2: -; AVX512-NEXT: andq %rsi, 8(%rdi) -; AVX512-NEXT: andq %r8, (%rdi) -; AVX512-NEXT: # kill: def $eax killed $eax killed $rax -; AVX512-NEXT: retq +; AVX-LABEL: reset_multiload_i128: +; AVX: # %bb.0: +; AVX-NEXT: movl %esi, %ecx +; AVX-NEXT: movl $1, %esi +; AVX-NEXT: xorl %r8d, %r8d +; AVX-NEXT: shldq %cl, %rsi, %r8 +; AVX-NEXT: xorl %eax, %eax +; AVX-NEXT: shlxq %rcx, %rsi, %r9 +; AVX-NEXT: testb $64, %cl +; AVX-NEXT: cmovneq %r9, %r8 +; AVX-NEXT: cmovneq %rax, %r9 +; AVX-NEXT: movq (%rdi), %r10 +; AVX-NEXT: movq 8(%rdi), %r11 +; AVX-NEXT: andnq %r11, %r8, %rcx +; AVX-NEXT: andq %r8, %r11 +; AVX-NEXT: andnq %r10, %r9, %rsi +; AVX-NEXT: andq %r9, %r10 +; AVX-NEXT: orq %r11, %r10 +; AVX-NEXT: jne .LBB22_2 +; AVX-NEXT: # %bb.1: +; AVX-NEXT: movl (%rdx), %eax +; AVX-NEXT: .LBB22_2: +; AVX-NEXT: movq %rsi, (%rdi) +; AVX-NEXT: movq %rcx, 8(%rdi) +; AVX-NEXT: # kill: def $eax killed $eax killed $rax +; AVX-NEXT: retq %rem = and i32 %position, 127 %ofs = zext nneg i32 %rem to i128 %bit = shl nuw i128 1, %ofs diff --git a/llvm/test/DebugInfo/PDB/Native/pdb-native-index-overflow.test b/llvm/test/DebugInfo/PDB/Native/pdb-native-index-overflow.test new file mode 100755 index 0000000..aa3f6dc --- /dev/null +++ b/llvm/test/DebugInfo/PDB/Native/pdb-native-index-overflow.test @@ -0,0 +1,13 @@ +; Test that the native PDB reader isn't crashed by index value bigger than +; number of types in TPI or IPI stream +; RUN: llvm-pdbutil dump %p/../Inputs/empty.pdb --type-index=20000000\ +; RUN: | FileCheck -check-prefixes=TYPES,NOT_FOUND %s +; RUN: llvm-pdbutil dump %p/../Inputs/empty.pdb --id-index=20000000\ +; RUN: | FileCheck -check-prefixes=IDS,NOT_FOUND %s + +TYPES: Types (TPI Stream) +IDS: Types (IPI Stream) +NOT_FOUND:============================================================ +NOT_FOUND: Showing 1 records. +NOT_FOUND: Type 0x1312D00 doesn't exist in TPI stream + diff --git a/llvm/test/Transforms/InstCombine/vec_extract_var_elt-inseltpoison.ll b/llvm/test/Transforms/InstCombine/vec_extract_var_elt-inseltpoison.ll deleted file mode 100644 index 9fcac80..0000000 --- a/llvm/test/Transforms/InstCombine/vec_extract_var_elt-inseltpoison.ll +++ /dev/null @@ -1,26 +0,0 @@ -; RUN: opt < %s -passes=instcombine -S | FileCheck %s - -define void @test (float %b, ptr %p) { -; CHECK: extractelement -; CHECK: fptosi - %1 = load <8 x float> , ptr %p - %2 = bitcast <8 x float> %1 to <8 x i32> - %3 = bitcast <8 x i32> %2 to <8 x float> - %a = fptosi <8 x float> %3 to <8 x i32> - %4 = fptosi float %b to i32 - %5 = add i32 %4, -2 - %6 = extractelement <8 x i32> %a, i32 %5 - %7 = insertelement <8 x i32> poison, i32 %6, i32 7 - %8 = sitofp <8 x i32> %7 to <8 x float> - store <8 x float> %8, ptr %p - ret void -} - -; PR18600 -define i32 @test2(i32 %i) { - %e = extractelement <4 x i32> bitcast (<2 x i64> <i64 1, i64 2> to <4 x i32>), i32 %i - ret i32 %e - -; CHECK-LABEL: @test2 -; CHECK: extractelement -} diff --git a/llvm/test/Transforms/InstCombine/vec_extract_var_elt.ll b/llvm/test/Transforms/InstCombine/vec_extract_var_elt.ll index 32bf4da..205b4b8 100644 --- a/llvm/test/Transforms/InstCombine/vec_extract_var_elt.ll +++ b/llvm/test/Transforms/InstCombine/vec_extract_var_elt.ll @@ -1,26 +1,81 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6 ; RUN: opt < %s -passes=instcombine -S | FileCheck %s -define void @test (float %b, ptr %p) { -; CHECK: extractelement -; CHECK: fptosi - %1 = load <8 x float> , ptr %p +define void @test_poison(float %b, ptr %p) { +; CHECK-LABEL: define void @test_poison( +; CHECK-SAME: float [[B:%.*]], ptr [[P:%.*]]) { +; CHECK-NEXT: [[TMP1:%.*]] = load <8 x float>, ptr [[P]], align 32 +; CHECK-NEXT: [[TMP2:%.*]] = fptosi float [[B]] to i32 +; CHECK-NEXT: [[TMP3:%.*]] = add i32 [[TMP2]], -2 +; CHECK-NEXT: [[TMP4:%.*]] = extractelement <8 x float> [[TMP1]], i32 [[TMP3]] +; CHECK-NEXT: [[TMP5:%.*]] = fptosi float [[TMP4]] to i32 +; CHECK-NEXT: [[TMP6:%.*]] = insertelement <8 x i32> poison, i32 [[TMP5]], i64 7 +; CHECK-NEXT: [[TMP7:%.*]] = sitofp <8 x i32> [[TMP6]] to <8 x float> +; CHECK-NEXT: store <8 x float> [[TMP7]], ptr [[P]], align 32 +; CHECK-NEXT: ret void +; + %1 = load <8 x float>, ptr %p %2 = bitcast <8 x float> %1 to <8 x i32> %3 = bitcast <8 x i32> %2 to <8 x float> %a = fptosi <8 x float> %3 to <8 x i32> %4 = fptosi float %b to i32 %5 = add i32 %4, -2 %6 = extractelement <8 x i32> %a, i32 %5 - %7 = insertelement <8 x i32> undef, i32 %6, i32 7 + %7 = insertelement <8 x i32> poison, i32 %6, i32 7 %8 = sitofp <8 x i32> %7 to <8 x float> store <8 x float> %8, ptr %p - ret void + ret void } ; PR18600 -define i32 @test2(i32 %i) { +define i32 @test_bitcast(i32 %i) { +; CHECK-LABEL: define i32 @test_bitcast( +; CHECK-SAME: i32 [[I:%.*]]) { +; CHECK-NEXT: [[E:%.*]] = extractelement <4 x i32> <i32 1, i32 0, i32 2, i32 0>, i32 [[I]] +; CHECK-NEXT: ret i32 [[E]] +; %e = extractelement <4 x i32> bitcast (<2 x i64> <i64 1, i64 2> to <4 x i32>), i32 %i ret i32 %e +} + +declare void @use(i32) -; CHECK-LABEL: @test2 -; CHECK: extractelement +define void @test_loop(<4 x float> %in) { +; CHECK-LABEL: define void @test_loop( +; CHECK-SAME: <4 x float> [[IN:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*]]: +; CHECK-NEXT: [[R:%.*]] = call <4 x float> @llvm.x86.sse41.round.ps(<4 x float> [[IN]], i32 9) +; CHECK-NEXT: br label %[[LOOP:.*]] +; CHECK: [[LOOP]]: +; CHECK-NEXT: [[I:%.*]] = phi i32 [ 0, %[[ENTRY]] ], [ [[NEXT:%.*]], %[[LATCH:.*]] ] +; CHECK-NEXT: [[COND:%.*]] = icmp samesign ult i32 [[I]], 4 +; CHECK-NEXT: br i1 [[COND]], label %[[BODY:.*]], label %[[DONE:.*]] +; CHECK: [[BODY]]: +; CHECK-NEXT: [[TMP0:%.*]] = extractelement <4 x float> [[R]], i32 [[I]] +; CHECK-NEXT: [[ELEM:%.*]] = fptosi float [[TMP0]] to i32 +; CHECK-NEXT: call void @use(i32 [[ELEM]]) +; CHECK-NEXT: br label %[[LATCH]] +; CHECK: [[LATCH]]: +; CHECK-NEXT: [[NEXT]] = add nuw nsw i32 [[I]], 1 +; CHECK-NEXT: br label %[[LOOP]] +; CHECK: [[DONE]]: +; CHECK-NEXT: ret void +; +entry: + %r = call <4 x float> @llvm.x86.sse41.round.ps(<4 x float> %in, i32 9) + %vi = fptosi <4 x float> %r to <4 x i32> + br label %loop +loop: + %i = phi i32 [ 0, %entry ], [ %next, %latch ] + %cond = icmp ult i32 %i, 4 + br i1 %cond, label %body, label %done +body: + %elem = extractelement <4 x i32> %vi, i32 %i + call void @use(i32 %elem) + br label %latch +latch: + %next = add i32 %i, 1 + br label %loop +done: + ret void } diff --git a/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/check_empty.ll b/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/check_empty.ll new file mode 100644 index 0000000..bfd216d --- /dev/null +++ b/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/check_empty.ll @@ -0,0 +1,29 @@ +; RUN: opt < %s -S | FileCheck %s + +; Test whether UTC checks empty lines instead of skipping them. +define i32 @test(i32 %x) { +entry: + br label %block1 + +block1: + %cmp = icmp eq i32 %x, 0 + br i1 %cmp, label %block2, label %exit1 + +block2: + br i1 %cmp, label %block3, label %exit2 + +block3: + br i1 %cmp, label %exit3, label %exit4 + +exit1: + ret i32 0 + +exit2: + ret i32 %x + +exit3: + ret i32 %x + +exit4: + ret i32 %x +} diff --git a/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/check_empty.ll.expected b/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/check_empty.ll.expected new file mode 100644 index 0000000..c5f822d --- /dev/null +++ b/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/check_empty.ll.expected @@ -0,0 +1,57 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 7 +; RUN: opt < %s -S | FileCheck %s + +; Test whether UTC checks empty lines instead of skipping them. +define i32 @test(i32 %x) { +; CHECK-LABEL: define i32 @test( +; CHECK-SAME: i32 [[X:%.*]]) { +; CHECK-NEXT: [[ENTRY:.*:]] +; CHECK-NEXT: br label %[[BLOCK1:.*]] +; CHECK-EMPTY: +; CHECK-NEXT: [[BLOCK1]]: +; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[X]], 0 +; CHECK-NEXT: br i1 [[CMP]], label %[[BLOCK2:.*]], label %[[EXIT1:.*]] +; CHECK-EMPTY: +; CHECK-NEXT: [[BLOCK2]]: +; CHECK-NEXT: br i1 [[CMP]], label %[[BLOCK3:.*]], label %[[EXIT2:.*]] +; CHECK-EMPTY: +; CHECK-NEXT: [[BLOCK3]]: +; CHECK-NEXT: br i1 [[CMP]], label %[[EXIT3:.*]], label %[[EXIT4:.*]] +; CHECK-EMPTY: +; CHECK-NEXT: [[EXIT1]]: +; CHECK-NEXT: ret i32 0 +; CHECK-EMPTY: +; CHECK-NEXT: [[EXIT2]]: +; CHECK-NEXT: ret i32 [[X]] +; CHECK-EMPTY: +; CHECK-NEXT: [[EXIT3]]: +; CHECK-NEXT: ret i32 [[X]] +; CHECK-EMPTY: +; CHECK-NEXT: [[EXIT4]]: +; CHECK-NEXT: ret i32 [[X]] +; +entry: + br label %block1 + +block1: + %cmp = icmp eq i32 %x, 0 + br i1 %cmp, label %block2, label %exit1 + +block2: + br i1 %cmp, label %block3, label %exit2 + +block3: + br i1 %cmp, label %exit3, label %exit4 + +exit1: + ret i32 0 + +exit2: + ret i32 %x + +exit3: + ret i32 %x + +exit4: + ret i32 %x +} diff --git a/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/switch_case.ll.expected b/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/switch_case.ll.expected index b1977e7..8cab0bb 100644 --- a/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/switch_case.ll.expected +++ b/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/switch_case.ll.expected @@ -12,13 +12,17 @@ define i8 @testi8(i8 %x) { ; CHECK-NEXT: i8 2, label %[[CASE3:.*]] ; CHECK-NEXT: i8 3, label %[[CASE3]] ; CHECK-NEXT: ] -; CHECK: [[DEFAULT]]: +; CHECK-EMPTY: +; CHECK-NEXT: [[DEFAULT]]: ; CHECK-NEXT: ret i8 0 -; CHECK: [[CASE1]]: +; CHECK-EMPTY: +; CHECK-NEXT: [[CASE1]]: ; CHECK-NEXT: ret i8 1 -; CHECK: [[CASE2]]: +; CHECK-EMPTY: +; CHECK-NEXT: [[CASE2]]: ; CHECK-NEXT: ret i8 2 -; CHECK: [[CASE3]]: +; CHECK-EMPTY: +; CHECK-NEXT: [[CASE3]]: ; CHECK-NEXT: ret i8 3 ; switch i8 %x, label %default [ @@ -46,13 +50,17 @@ define i32 @testi32(i32 %x) { ; CHECK-NEXT: i32 2, label %[[CASE3:.*]] ; CHECK-NEXT: i32 3, label %[[CASE3]] ; CHECK-NEXT: ] -; CHECK: [[DEFAULT]]: +; CHECK-EMPTY: +; CHECK-NEXT: [[DEFAULT]]: ; CHECK-NEXT: ret i32 0 -; CHECK: [[CASE1]]: +; CHECK-EMPTY: +; CHECK-NEXT: [[CASE1]]: ; CHECK-NEXT: ret i32 1 -; CHECK: [[CASE2]]: +; CHECK-EMPTY: +; CHECK-NEXT: [[CASE2]]: ; CHECK-NEXT: ret i32 2 -; CHECK: [[CASE3]]: +; CHECK-EMPTY: +; CHECK-NEXT: [[CASE3]]: ; CHECK-NEXT: ret i32 3 ; switch i32 %x, label %default [ @@ -80,13 +88,17 @@ define i128 @testi128(i128 %x) { ; CHECK-NEXT: i128 2, label %[[CASE3:.*]] ; CHECK-NEXT: i128 3, label %[[CASE3]] ; CHECK-NEXT: ] -; CHECK: [[DEFAULT]]: +; CHECK-EMPTY: +; CHECK-NEXT: [[DEFAULT]]: ; CHECK-NEXT: ret i128 0 -; CHECK: [[CASE1]]: +; CHECK-EMPTY: +; CHECK-NEXT: [[CASE1]]: ; CHECK-NEXT: ret i128 1 -; CHECK: [[CASE2]]: +; CHECK-EMPTY: +; CHECK-NEXT: [[CASE2]]: ; CHECK-NEXT: ret i128 2 -; CHECK: [[CASE3]]: +; CHECK-EMPTY: +; CHECK-NEXT: [[CASE3]]: ; CHECK-NEXT: ret i128 3 ; switch i128 %x, label %default [ diff --git a/llvm/test/tools/UpdateTestChecks/update_test_checks/check_empty.test b/llvm/test/tools/UpdateTestChecks/update_test_checks/check_empty.test new file mode 100644 index 0000000..670bda2 --- /dev/null +++ b/llvm/test/tools/UpdateTestChecks/update_test_checks/check_empty.test @@ -0,0 +1,3 @@ +## test whether the UTC generates CHECK-EMPTY for blank lines +# RUN: cp -f %S/Inputs/check_empty.ll %t.ll && %update_test_checks %t.ll --version 7 +# RUN: diff -u %t.ll %S/Inputs/check_empty.ll.expected diff --git a/llvm/unittests/ExecutionEngine/Orc/CMakeLists.txt b/llvm/unittests/ExecutionEngine/Orc/CMakeLists.txt index b06aa25..7b563d7 100644 --- a/llvm/unittests/ExecutionEngine/Orc/CMakeLists.txt +++ b/llvm/unittests/ExecutionEngine/Orc/CMakeLists.txt @@ -26,6 +26,7 @@ add_llvm_unittest(OrcJITTests IndirectionUtilsTest.cpp JITTargetMachineBuilderTest.cpp LazyCallThroughAndReexportsTest.cpp + LibraryResolverTest.cpp LookupAndRecordAddrsTest.cpp MachOPlatformTest.cpp MapperJITLinkMemoryManagerTest.cpp diff --git a/llvm/unittests/ExecutionEngine/Orc/Inputs/A/A_linux.yaml b/llvm/unittests/ExecutionEngine/Orc/Inputs/A/A_linux.yaml new file mode 100644 index 0000000..afd1d9e --- /dev/null +++ b/llvm/unittests/ExecutionEngine/Orc/Inputs/A/A_linux.yaml @@ -0,0 +1,460 @@ +--- !ELF +FileHeader: + Class: ELFCLASS64 + Data: ELFDATA2LSB + Type: ET_DYN + Machine: EM_X86_64 +ProgramHeaders: + - Type: PT_LOAD + Flags: [ PF_R ] + FirstSec: .note.gnu.property + LastSec: .rela.plt + Align: 0x1000 + Offset: 0x0 + - Type: PT_LOAD + Flags: [ PF_X, PF_R ] + FirstSec: .init + LastSec: .fini + VAddr: 0x1000 + Align: 0x1000 + Offset: 0x1000 + - Type: PT_LOAD + Flags: [ PF_R ] + FirstSec: .rodata + LastSec: .eh_frame + VAddr: 0x2000 + Align: 0x1000 + Offset: 0x2000 + - Type: PT_LOAD + Flags: [ PF_W, PF_R ] + FirstSec: .init_array + LastSec: .bss + VAddr: 0x3E10 + Align: 0x1000 + Offset: 0x2E10 + - Type: PT_DYNAMIC + Flags: [ PF_W, PF_R ] + FirstSec: .dynamic + LastSec: .dynamic + VAddr: 0x3E20 + Align: 0x8 + Offset: 0x2E20 + - Type: PT_NOTE + Flags: [ PF_R ] + FirstSec: .note.gnu.property + LastSec: .note.gnu.property + VAddr: 0x2A8 + Align: 0x8 + Offset: 0x2A8 + - Type: PT_NOTE + Flags: [ PF_R ] + FirstSec: .note.gnu.build-id + LastSec: .note.gnu.build-id + VAddr: 0x2C8 + Align: 0x4 + Offset: 0x2C8 + - Type: PT_GNU_PROPERTY + Flags: [ PF_R ] + FirstSec: .note.gnu.property + LastSec: .note.gnu.property + VAddr: 0x2A8 + Align: 0x8 + Offset: 0x2A8 + - Type: PT_GNU_EH_FRAME + Flags: [ PF_R ] + FirstSec: .eh_frame_hdr + LastSec: .eh_frame_hdr + VAddr: 0x2010 + Align: 0x4 + Offset: 0x2010 + - Type: PT_GNU_STACK + Flags: [ PF_W, PF_R ] + Align: 0x10 + Offset: 0x0 + - Type: PT_GNU_RELRO + Flags: [ PF_R ] + FirstSec: .init_array + LastSec: .got + VAddr: 0x3E10 + Offset: 0x2E10 +Sections: + - Name: .note.gnu.property + Type: SHT_NOTE + Flags: [ SHF_ALLOC ] + Address: 0x2A8 + AddressAlign: 0x8 + Notes: + - Name: GNU + Desc: 020000C0040000000300000000000000 + Type: NT_GNU_PROPERTY_TYPE_0 + - Name: .note.gnu.build-id + Type: SHT_NOTE + Flags: [ SHF_ALLOC ] + Address: 0x2C8 + AddressAlign: 0x4 + Notes: + - Name: GNU + Desc: 73604396C95840D5C380A0950F085A778F94EE7C + Type: NT_PRPSINFO + - Name: .gnu.hash + Type: SHT_GNU_HASH + Flags: [ SHF_ALLOC ] + Address: 0x2F0 + Link: .dynsym + AddressAlign: 0x8 + Header: + SymNdx: 0x6 + Shift2: 0x6 + BloomFilter: [ 0x400000080000 ] + HashBuckets: [ 0x0, 0x6 ] + HashValues: [ 0x7C9DCB93 ] + - Name: .dynsym + Type: SHT_DYNSYM + Flags: [ SHF_ALLOC ] + Address: 0x318 + Link: .dynstr + AddressAlign: 0x8 + - Name: .dynstr + Type: SHT_STRTAB + Flags: [ SHF_ALLOC ] + Address: 0x3C0 + AddressAlign: 0x1 + - Name: .gnu.version + Type: SHT_GNU_versym + Flags: [ SHF_ALLOC ] + Address: 0x436 + Link: .dynsym + AddressAlign: 0x2 + Entries: [ 0, 1, 2, 1, 1, 2, 1 ] + - Name: .gnu.version_r + Type: SHT_GNU_verneed + Flags: [ SHF_ALLOC ] + Address: 0x448 + Link: .dynstr + AddressAlign: 0x8 + Dependencies: + - Version: 1 + File: libc.so.6 + Entries: + - Name: GLIBC_2.2.5 + Hash: 157882997 + Flags: 0 + Other: 2 + - Name: .rela.dyn + Type: SHT_RELA + Flags: [ SHF_ALLOC ] + Address: 0x468 + Link: .dynsym + AddressAlign: 0x8 + Relocations: + - Offset: 0x3E10 + Type: R_X86_64_RELATIVE + Addend: 4368 + - Offset: 0x3E18 + Type: R_X86_64_RELATIVE + Addend: 4304 + - Offset: 0x4020 + Type: R_X86_64_RELATIVE + Addend: 16416 + - Offset: 0x3FE0 + Symbol: _ITM_deregisterTMCloneTable + Type: R_X86_64_GLOB_DAT + - Offset: 0x3FE8 + Symbol: __gmon_start__ + Type: R_X86_64_GLOB_DAT + - Offset: 0x3FF0 + Symbol: _ITM_registerTMCloneTable + Type: R_X86_64_GLOB_DAT + - Offset: 0x3FF8 + Symbol: __cxa_finalize + Type: R_X86_64_GLOB_DAT + - Name: .rela.plt + Type: SHT_RELA + Flags: [ SHF_ALLOC, SHF_INFO_LINK ] + Address: 0x510 + Link: .dynsym + AddressAlign: 0x8 + Info: .got.plt + Relocations: + - Offset: 0x4018 + Symbol: puts + Type: R_X86_64_JUMP_SLOT + - Name: .init + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1000 + AddressAlign: 0x4 + Offset: 0x1000 + Content: F30F1EFA4883EC08488B05D92F00004885C07402FFD04883C408C3 + - Name: .plt + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1020 + AddressAlign: 0x10 + EntSize: 0x10 + Content: FF35E22F0000F2FF25E32F00000F1F00F30F1EFA6800000000F2E9E1FFFFFF90 + - Name: .plt.got + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1040 + AddressAlign: 0x10 + EntSize: 0x10 + Content: F30F1EFAF2FF25AD2F00000F1F440000 + - Name: .plt.sec + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1050 + AddressAlign: 0x10 + EntSize: 0x10 + Content: F30F1EFAF2FF25BD2F00000F1F440000 + - Name: .text + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1060 + AddressAlign: 0x10 + Content: 488D3DC12F0000488D05BA2F00004839F87415488B05662F00004885C07409FFE00F1F8000000000C30F1F8000000000488D3D912F0000488D358A2F00004829FE4889F048C1EE3F48C1F8034801C648D1FE7414488B05352F00004885C07408FFE0660F1F440000C30F1F8000000000F30F1EFA803D4D2F000000752B5548833D122F0000004889E5740C488B3D2E2F0000E849FFFFFFE864FFFFFFC605252F0000015DC30F1F00C30F1F8000000000F30F1EFAE977FFFFFFF30F1EFA554889E5488D05D80E00004889C7E820FFFFFF905DC3 + - Name: .fini + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1134 + AddressAlign: 0x4 + Content: F30F1EFA4883EC084883C408C3 + - Name: .rodata + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC ] + Address: 0x2000 + AddressAlign: 0x1 + Offset: 0x2000 + Content: 48656C6C6F2066726F6D204100 + - Name: .eh_frame_hdr + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC ] + Address: 0x2010 + AddressAlign: 0x4 + Content: 011B033B2C0000000400000010F0FFFF4800000030F0FFFF7000000040F0FFFF8800000009F1FFFFA0000000 + - Name: .eh_frame + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC ] + Address: 0x2040 + AddressAlign: 0x8 + Content: 1400000000000000017A5200017810011B0C070890010000240000001C000000C0EFFFFF20000000000E10460E184A0F0B770880003F1A3A2A332422000000001400000044000000B8EFFFFF100000000000000000000000140000005C000000B0EFFFFF1000000000000000000000001C0000007400000061F0FFFF1A00000000450E108602430D06510C070800000000000000 + - Name: .init_array + Type: SHT_INIT_ARRAY + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3E10 + AddressAlign: 0x8 + EntSize: 0x8 + Offset: 0x2E10 + Content: '1011000000000000' + - Name: .fini_array + Type: SHT_FINI_ARRAY + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3E18 + AddressAlign: 0x8 + EntSize: 0x8 + Content: D010000000000000 + - Name: .dynamic + Type: SHT_DYNAMIC + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3E20 + Link: .dynstr + AddressAlign: 0x8 + Entries: + - Tag: DT_NEEDED + Value: 0x5F + - Tag: DT_INIT + Value: 0x1000 + - Tag: DT_FINI + Value: 0x1134 + - Tag: DT_INIT_ARRAY + Value: 0x3E10 + - Tag: DT_INIT_ARRAYSZ + Value: 0x8 + - Tag: DT_FINI_ARRAY + Value: 0x3E18 + - Tag: DT_FINI_ARRAYSZ + Value: 0x8 + - Tag: DT_GNU_HASH + Value: 0x2F0 + - Tag: DT_STRTAB + Value: 0x3C0 + - Tag: DT_SYMTAB + Value: 0x318 + - Tag: DT_STRSZ + Value: 0x75 + - Tag: DT_SYMENT + Value: 0x18 + - Tag: DT_PLTGOT + Value: 0x4000 + - Tag: DT_PLTRELSZ + Value: 0x18 + - Tag: DT_PLTREL + Value: 0x7 + - Tag: DT_JMPREL + Value: 0x510 + - Tag: DT_RELA + Value: 0x468 + - Tag: DT_RELASZ + Value: 0xA8 + - Tag: DT_RELAENT + Value: 0x18 + - Tag: DT_VERNEED + Value: 0x448 + - Tag: DT_VERNEEDNUM + Value: 0x1 + - Tag: DT_VERSYM + Value: 0x436 + - Tag: DT_RELACOUNT + Value: 0x3 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Name: .got + Type: SHT_PROGBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3FE0 + AddressAlign: 0x8 + EntSize: 0x8 + Content: '0000000000000000000000000000000000000000000000000000000000000000' + - Name: .got.plt + Type: SHT_PROGBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x4000 + AddressAlign: 0x8 + EntSize: 0x8 + Content: '203E000000000000000000000000000000000000000000003010000000000000' + - Name: .data + Type: SHT_PROGBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x4020 + AddressAlign: 0x8 + Content: '2040000000000000' + - Name: .bss + Type: SHT_NOBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x4028 + AddressAlign: 0x1 + Size: 0x8 + - Name: .comment + Type: SHT_PROGBITS + Flags: [ SHF_MERGE, SHF_STRINGS ] + AddressAlign: 0x1 + EntSize: 0x1 + Content: 4743433A20285562756E74752031312E342E302D317562756E7475317E32322E30342E32292031312E342E3000 +Symbols: + - Name: crtstuff.c + Type: STT_FILE + Index: SHN_ABS + - Name: deregister_tm_clones + Type: STT_FUNC + Section: .text + Value: 0x1060 + - Name: register_tm_clones + Type: STT_FUNC + Section: .text + Value: 0x1090 + - Name: __do_global_dtors_aux + Type: STT_FUNC + Section: .text + Value: 0x10D0 + - Name: completed.0 + Type: STT_OBJECT + Section: .bss + Value: 0x4028 + Size: 0x1 + - Name: __do_global_dtors_aux_fini_array_entry + Type: STT_OBJECT + Section: .fini_array + Value: 0x3E18 + - Name: frame_dummy + Type: STT_FUNC + Section: .text + Value: 0x1110 + - Name: __frame_dummy_init_array_entry + Type: STT_OBJECT + Section: .init_array + Value: 0x3E10 + - Name: libA.c + Type: STT_FILE + Index: SHN_ABS + - Name: 'crtstuff.c (1)' + Type: STT_FILE + Index: SHN_ABS + - Name: __FRAME_END__ + Type: STT_OBJECT + Section: .eh_frame + Value: 0x20D0 + - Type: STT_FILE + Index: SHN_ABS + - Name: _fini + Type: STT_FUNC + Section: .fini + Value: 0x1134 + - Name: __dso_handle + Type: STT_OBJECT + Section: .data + Value: 0x4020 + - Name: _DYNAMIC + Type: STT_OBJECT + Section: .dynamic + Value: 0x3E20 + - Name: __GNU_EH_FRAME_HDR + Section: .eh_frame_hdr + Value: 0x2010 + - Name: __TMC_END__ + Type: STT_OBJECT + Section: .data + Value: 0x4028 + - Name: _GLOBAL_OFFSET_TABLE_ + Type: STT_OBJECT + Section: .got.plt + Value: 0x4000 + - Name: _init + Type: STT_FUNC + Section: .init + Value: 0x1000 + - Name: _ITM_deregisterTMCloneTable + Binding: STB_WEAK + - Name: 'puts@GLIBC_2.2.5' + Type: STT_FUNC + Binding: STB_GLOBAL + - Name: sayA + Type: STT_FUNC + Section: .text + Binding: STB_GLOBAL + Value: 0x1119 + Size: 0x1A + - Name: __gmon_start__ + Binding: STB_WEAK + - Name: _ITM_registerTMCloneTable + Binding: STB_WEAK + - Name: '__cxa_finalize@GLIBC_2.2.5' + Type: STT_FUNC + Binding: STB_WEAK +DynamicSymbols: + - Name: _ITM_deregisterTMCloneTable + Binding: STB_WEAK + - Name: puts + Type: STT_FUNC + Binding: STB_GLOBAL + - Name: __gmon_start__ + Binding: STB_WEAK + - Name: _ITM_registerTMCloneTable + Binding: STB_WEAK + - Name: __cxa_finalize + Type: STT_FUNC + Binding: STB_WEAK + - Name: sayA + Type: STT_FUNC + Section: .text + Binding: STB_GLOBAL + Value: 0x1119 + Size: 0x1A +... diff --git a/llvm/unittests/ExecutionEngine/Orc/Inputs/A/A_macho.yaml b/llvm/unittests/ExecutionEngine/Orc/Inputs/A/A_macho.yaml new file mode 100644 index 0000000..2e851a90 --- /dev/null +++ b/llvm/unittests/ExecutionEngine/Orc/Inputs/A/A_macho.yaml @@ -0,0 +1,723 @@ +--- !fat-mach-o +FatHeader: + magic: 0xCAFEBABE + nfat_arch: 3 +FatArchs: + - cputype: 0x1000007 + cpusubtype: 0x3 + offset: 0x1000 + size: 8376 + align: 12 + - cputype: 0x100000C + cpusubtype: 0x0 + offset: 0x4000 + size: 33376 + align: 14 + - cputype: 0x100000C + cpusubtype: 0x80000002 + offset: 0x10000 + size: 33376 + align: 14 +Slices: + - !mach-o + FileHeader: + magic: 0xFEEDFACF + cputype: 0x1000007 + cpusubtype: 0x3 + filetype: 0x6 + ncmds: 14 + sizeofcmds: 960 + flags: 0x100085 + reserved: 0x0 + LoadCommands: + - cmd: LC_SEGMENT_64 + cmdsize: 392 + segname: __TEXT + vmaddr: 0 + vmsize: 4096 + fileoff: 0 + filesize: 4096 + maxprot: 5 + initprot: 5 + nsects: 4 + flags: 0 + Sections: + - sectname: __text + segname: __TEXT + addr: 0xF80 + size: 20 + offset: 0xF80 + align: 4 + reloff: 0x0 + nreloc: 0 + flags: 0x80000400 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 554889E5488D3D0F000000B000E8020000005DC3 + - sectname: __stubs + segname: __TEXT + addr: 0xF94 + size: 6 + offset: 0xF94 + align: 1 + reloff: 0x0 + nreloc: 0 + flags: 0x80000408 + reserved1: 0x0 + reserved2: 0x6 + reserved3: 0x0 + content: FF2566000000 + - sectname: __cstring + segname: __TEXT + addr: 0xF9A + size: 14 + offset: 0xF9A + align: 0 + reloff: 0x0 + nreloc: 0 + flags: 0x2 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 48656C6C6F2066726F6D20410A00 + - sectname: __unwind_info + segname: __TEXT + addr: 0xFA8 + size: 88 + offset: 0xFA8 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x0 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 010000001C000000000000001C000000000000001C00000002000000800F00004000000040000000940F00000000000040000000000000000000000000000000030000000C00010010000100000000000000000100000000 + - cmd: LC_SEGMENT_64 + cmdsize: 152 + segname: __DATA_CONST + vmaddr: 4096 + vmsize: 4096 + fileoff: 4096 + filesize: 4096 + maxprot: 3 + initprot: 3 + nsects: 1 + flags: 16 + Sections: + - sectname: __got + segname: __DATA_CONST + addr: 0x1000 + size: 8 + offset: 0x1000 + align: 3 + reloff: 0x0 + nreloc: 0 + flags: 0x6 + reserved1: 0x1 + reserved2: 0x0 + reserved3: 0x0 + content: '0000000000000080' + - cmd: LC_SEGMENT_64 + cmdsize: 72 + segname: __LINKEDIT + vmaddr: 8192 + vmsize: 4096 + fileoff: 8192 + filesize: 184 + maxprot: 1 + initprot: 1 + nsects: 0 + flags: 0 + - cmd: LC_ID_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 1 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libA.dylib' + ZeroPadBytes: 7 + - cmd: LC_DYLD_CHAINED_FIXUPS + cmdsize: 16 + dataoff: 8192 + datasize: 96 + - cmd: LC_DYLD_EXPORTS_TRIE + cmdsize: 16 + dataoff: 8288 + datasize: 24 + - cmd: LC_SYMTAB + cmdsize: 24 + symoff: 8320 + nsyms: 2 + stroff: 8360 + strsize: 16 + - cmd: LC_DYSYMTAB + cmdsize: 80 + ilocalsym: 0 + nlocalsym: 0 + iextdefsym: 0 + nextdefsym: 1 + iundefsym: 1 + nundefsym: 1 + tocoff: 0 + ntoc: 0 + modtaboff: 0 + nmodtab: 0 + extrefsymoff: 0 + nextrefsyms: 0 + indirectsymoff: 8352 + nindirectsyms: 2 + extreloff: 0 + nextrel: 0 + locreloff: 0 + nlocrel: 0 + - cmd: LC_UUID + cmdsize: 24 + uuid: ADFFA141-C3EE-37CD-B1E7-906D69F81BCB + - cmd: LC_BUILD_VERSION + cmdsize: 32 + platform: 1 + minos: 983040 + sdk: 983552 + ntools: 1 + Tools: + - tool: 3 + version: 73074435 + - cmd: LC_SOURCE_VERSION + cmdsize: 16 + version: 0 + - cmd: LC_LOAD_DYLIB + cmdsize: 56 + dylib: + name: 24 + timestamp: 2 + current_version: 88539136 + compatibility_version: 65536 + Content: '/usr/lib/libSystem.B.dylib' + ZeroPadBytes: 6 + - cmd: LC_FUNCTION_STARTS + cmdsize: 16 + dataoff: 8312 + datasize: 8 + - cmd: LC_DATA_IN_CODE + cmdsize: 16 + dataoff: 8320 + datasize: 0 + LinkEditData: + ExportTrie: + TerminalSize: 0 + NodeOffset: 0 + Name: '' + Flags: 0x0 + Address: 0x0 + Other: 0x0 + ImportName: '' + Children: + - TerminalSize: 3 + NodeOffset: 13 + Name: _sayA + Flags: 0x0 + Address: 0xF80 + Other: 0x0 + ImportName: '' + NameList: + - n_strx: 2 + n_type: 0xF + n_sect: 1 + n_desc: 0 + n_value: 3968 + - n_strx: 8 + n_type: 0x1 + n_sect: 0 + n_desc: 256 + n_value: 0 + StringTable: + - ' ' + - _sayA + - _printf + IndirectSymbols: [ 0x1, 0x1 ] + FunctionStarts: [ 0xF80 ] + ChainedFixups: [ 0x0, 0x0, 0x0, 0x0, 0x20, 0x0, 0x0, 0x0, 0x48, + 0x0, 0x0, 0x0, 0x50, 0x0, 0x0, 0x0, 0x1, 0x0, + 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x3, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x18, 0x0, 0x0, 0x0, 0x0, 0x10, 0x6, 0x0, + 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x1, 0x2, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x5F, 0x70, 0x72, + 0x69, 0x6E, 0x74, 0x66, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0 ] + - !mach-o + FileHeader: + magic: 0xFEEDFACF + cputype: 0x100000C + cpusubtype: 0x0 + filetype: 0x6 + ncmds: 15 + sizeofcmds: 976 + flags: 0x100085 + reserved: 0x0 + LoadCommands: + - cmd: LC_SEGMENT_64 + cmdsize: 392 + segname: __TEXT + vmaddr: 0 + vmsize: 16384 + fileoff: 0 + filesize: 16384 + maxprot: 5 + initprot: 5 + nsects: 4 + flags: 0 + Sections: + - sectname: __text + segname: __TEXT + addr: 0x3F70 + size: 28 + offset: 0x3F70 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000400 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: FD7BBFA9FD0300910000009000603E9103000094FD7BC1A8C0035FD6 + - sectname: __stubs + segname: __TEXT + addr: 0x3F8C + size: 12 + offset: 0x3F8C + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000408 + reserved1: 0x0 + reserved2: 0xC + reserved3: 0x0 + content: 100000B0100240F900021FD6 + - sectname: __cstring + segname: __TEXT + addr: 0x3F98 + size: 14 + offset: 0x3F98 + align: 0 + reloff: 0x0 + nreloc: 0 + flags: 0x2 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 48656C6C6F2066726F6D20410A00 + - sectname: __unwind_info + segname: __TEXT + addr: 0x3FA8 + size: 88 + offset: 0x3FA8 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x0 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 010000001C000000000000001C000000000000001C00000002000000703F000040000000400000008C3F00000000000040000000000000000000000000000000030000000C00010010000100000000000000000400000000 + - cmd: LC_SEGMENT_64 + cmdsize: 152 + segname: __DATA_CONST + vmaddr: 16384 + vmsize: 16384 + fileoff: 16384 + filesize: 16384 + maxprot: 3 + initprot: 3 + nsects: 1 + flags: 16 + Sections: + - sectname: __got + segname: __DATA_CONST + addr: 0x4000 + size: 8 + offset: 0x4000 + align: 3 + reloff: 0x0 + nreloc: 0 + flags: 0x6 + reserved1: 0x1 + reserved2: 0x0 + reserved3: 0x0 + content: '0000000000000080' + - cmd: LC_SEGMENT_64 + cmdsize: 72 + segname: __LINKEDIT + vmaddr: 32768 + vmsize: 16384 + fileoff: 32768 + filesize: 608 + maxprot: 1 + initprot: 1 + nsects: 0 + flags: 0 + - cmd: LC_ID_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 1 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libA.dylib' + ZeroPadBytes: 7 + - cmd: LC_DYLD_CHAINED_FIXUPS + cmdsize: 16 + dataoff: 32768 + datasize: 96 + - cmd: LC_DYLD_EXPORTS_TRIE + cmdsize: 16 + dataoff: 32864 + datasize: 24 + - cmd: LC_SYMTAB + cmdsize: 24 + symoff: 32896 + nsyms: 2 + stroff: 32936 + strsize: 16 + - cmd: LC_DYSYMTAB + cmdsize: 80 + ilocalsym: 0 + nlocalsym: 0 + iextdefsym: 0 + nextdefsym: 1 + iundefsym: 1 + nundefsym: 1 + tocoff: 0 + ntoc: 0 + modtaboff: 0 + nmodtab: 0 + extrefsymoff: 0 + nextrefsyms: 0 + indirectsymoff: 32928 + nindirectsyms: 2 + extreloff: 0 + nextrel: 0 + locreloff: 0 + nlocrel: 0 + - cmd: LC_UUID + cmdsize: 24 + uuid: C45227E0-C6C0-3137-969B-36AABF9D5487 + - cmd: LC_BUILD_VERSION + cmdsize: 32 + platform: 1 + minos: 983040 + sdk: 983552 + ntools: 1 + Tools: + - tool: 3 + version: 73074435 + - cmd: LC_SOURCE_VERSION + cmdsize: 16 + version: 0 + - cmd: LC_LOAD_DYLIB + cmdsize: 56 + dylib: + name: 24 + timestamp: 2 + current_version: 88539136 + compatibility_version: 65536 + Content: '/usr/lib/libSystem.B.dylib' + ZeroPadBytes: 6 + - cmd: LC_FUNCTION_STARTS + cmdsize: 16 + dataoff: 32888 + datasize: 8 + - cmd: LC_DATA_IN_CODE + cmdsize: 16 + dataoff: 32896 + datasize: 0 + - cmd: LC_CODE_SIGNATURE + cmdsize: 16 + dataoff: 32960 + datasize: 416 + LinkEditData: + ExportTrie: + TerminalSize: 0 + NodeOffset: 0 + Name: '' + Flags: 0x0 + Address: 0x0 + Other: 0x0 + ImportName: '' + Children: + - TerminalSize: 3 + NodeOffset: 13 + Name: _sayA + Flags: 0x0 + Address: 0x3F70 + Other: 0x0 + ImportName: '' + NameList: + - n_strx: 2 + n_type: 0xF + n_sect: 1 + n_desc: 0 + n_value: 16240 + - n_strx: 8 + n_type: 0x1 + n_sect: 0 + n_desc: 256 + n_value: 0 + StringTable: + - ' ' + - _sayA + - _printf + IndirectSymbols: [ 0x1, 0x1 ] + FunctionStarts: [ 0x3F70 ] + ChainedFixups: [ 0x0, 0x0, 0x0, 0x0, 0x20, 0x0, 0x0, 0x0, 0x48, + 0x0, 0x0, 0x0, 0x50, 0x0, 0x0, 0x0, 0x1, 0x0, + 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x3, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x18, 0x0, 0x0, 0x0, 0x0, 0x40, 0x6, 0x0, + 0x0, 0x40, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x1, 0x2, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x5F, 0x70, 0x72, + 0x69, 0x6E, 0x74, 0x66, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0 ] + - !mach-o + FileHeader: + magic: 0xFEEDFACF + cputype: 0x100000C + cpusubtype: 0x80000002 + filetype: 0x6 + ncmds: 15 + sizeofcmds: 976 + flags: 0x100085 + reserved: 0x0 + LoadCommands: + - cmd: LC_SEGMENT_64 + cmdsize: 392 + segname: __TEXT + vmaddr: 0 + vmsize: 16384 + fileoff: 0 + filesize: 16384 + maxprot: 5 + initprot: 5 + nsects: 4 + flags: 0 + Sections: + - sectname: __text + segname: __TEXT + addr: 0x3F68 + size: 32 + offset: 0x3F68 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000400 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 7F2303D5FD7BBFA9FD0300910000009000603E9103000094FD7BC1A8FF0F5FD6 + - sectname: __auth_stubs + segname: __TEXT + addr: 0x3F88 + size: 16 + offset: 0x3F88 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000408 + reserved1: 0x0 + reserved2: 0x10 + reserved3: 0x0 + content: 110000B031020091300240F9110A1FD7 + - sectname: __cstring + segname: __TEXT + addr: 0x3F98 + size: 14 + offset: 0x3F98 + align: 0 + reloff: 0x0 + nreloc: 0 + flags: 0x2 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 48656C6C6F2066726F6D20410A00 + - sectname: __unwind_info + segname: __TEXT + addr: 0x3FA8 + size: 88 + offset: 0x3FA8 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x0 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 010000001C000000000000001C000000000000001C00000002000000683F00004000000040000000883F00000000000040000000000000000000000000000000030000000C00010010000100000000000000000400000000 + - cmd: LC_SEGMENT_64 + cmdsize: 152 + segname: __DATA_CONST + vmaddr: 16384 + vmsize: 16384 + fileoff: 16384 + filesize: 16384 + maxprot: 3 + initprot: 3 + nsects: 1 + flags: 16 + Sections: + - sectname: __auth_got + segname: __DATA_CONST + addr: 0x4000 + size: 8 + offset: 0x4000 + align: 3 + reloff: 0x0 + nreloc: 0 + flags: 0x6 + reserved1: 0x1 + reserved2: 0x0 + reserved3: 0x0 + content: 00000000000001C0 + - cmd: LC_SEGMENT_64 + cmdsize: 72 + segname: __LINKEDIT + vmaddr: 32768 + vmsize: 16384 + fileoff: 32768 + filesize: 608 + maxprot: 1 + initprot: 1 + nsects: 0 + flags: 0 + - cmd: LC_ID_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 1 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libA.dylib' + ZeroPadBytes: 7 + - cmd: LC_DYLD_CHAINED_FIXUPS + cmdsize: 16 + dataoff: 32768 + datasize: 96 + - cmd: LC_DYLD_EXPORTS_TRIE + cmdsize: 16 + dataoff: 32864 + datasize: 24 + - cmd: LC_SYMTAB + cmdsize: 24 + symoff: 32896 + nsyms: 2 + stroff: 32936 + strsize: 16 + - cmd: LC_DYSYMTAB + cmdsize: 80 + ilocalsym: 0 + nlocalsym: 0 + iextdefsym: 0 + nextdefsym: 1 + iundefsym: 1 + nundefsym: 1 + tocoff: 0 + ntoc: 0 + modtaboff: 0 + nmodtab: 0 + extrefsymoff: 0 + nextrefsyms: 0 + indirectsymoff: 32928 + nindirectsyms: 2 + extreloff: 0 + nextrel: 0 + locreloff: 0 + nlocrel: 0 + - cmd: LC_UUID + cmdsize: 24 + uuid: C9DC00C2-E721-365C-9C2D-E9FDB7C838BB + - cmd: LC_BUILD_VERSION + cmdsize: 32 + platform: 1 + minos: 983040 + sdk: 983552 + ntools: 1 + Tools: + - tool: 3 + version: 73074435 + - cmd: LC_SOURCE_VERSION + cmdsize: 16 + version: 0 + - cmd: LC_LOAD_DYLIB + cmdsize: 56 + dylib: + name: 24 + timestamp: 2 + current_version: 88539136 + compatibility_version: 65536 + Content: '/usr/lib/libSystem.B.dylib' + ZeroPadBytes: 6 + - cmd: LC_FUNCTION_STARTS + cmdsize: 16 + dataoff: 32888 + datasize: 8 + - cmd: LC_DATA_IN_CODE + cmdsize: 16 + dataoff: 32896 + datasize: 0 + - cmd: LC_CODE_SIGNATURE + cmdsize: 16 + dataoff: 32960 + datasize: 416 + LinkEditData: + ExportTrie: + TerminalSize: 0 + NodeOffset: 0 + Name: '' + Flags: 0x0 + Address: 0x0 + Other: 0x0 + ImportName: '' + Children: + - TerminalSize: 3 + NodeOffset: 13 + Name: _sayA + Flags: 0x0 + Address: 0x3F68 + Other: 0x0 + ImportName: '' + NameList: + - n_strx: 2 + n_type: 0xF + n_sect: 1 + n_desc: 0 + n_value: 16232 + - n_strx: 8 + n_type: 0x1 + n_sect: 0 + n_desc: 256 + n_value: 0 + StringTable: + - ' ' + - _sayA + - _printf + IndirectSymbols: [ 0x1, 0x1 ] + FunctionStarts: [ 0x3F68 ] + ChainedFixups: [ 0x0, 0x0, 0x0, 0x0, 0x20, 0x0, 0x0, 0x0, 0x48, + 0x0, 0x0, 0x0, 0x50, 0x0, 0x0, 0x0, 0x1, 0x0, + 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x3, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x18, 0x0, 0x0, 0x0, 0x0, 0x40, 0xC, 0x0, + 0x0, 0x40, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x1, 0x2, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x5F, 0x70, 0x72, + 0x69, 0x6E, 0x74, 0x66, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0 ] +... diff --git a/llvm/unittests/ExecutionEngine/Orc/Inputs/B/B_linux.yaml b/llvm/unittests/ExecutionEngine/Orc/Inputs/B/B_linux.yaml new file mode 100644 index 0000000..fe4393e --- /dev/null +++ b/llvm/unittests/ExecutionEngine/Orc/Inputs/B/B_linux.yaml @@ -0,0 +1,460 @@ +--- !ELF +FileHeader: + Class: ELFCLASS64 + Data: ELFDATA2LSB + Type: ET_DYN + Machine: EM_X86_64 +ProgramHeaders: + - Type: PT_LOAD + Flags: [ PF_R ] + FirstSec: .note.gnu.property + LastSec: .rela.plt + Align: 0x1000 + Offset: 0x0 + - Type: PT_LOAD + Flags: [ PF_X, PF_R ] + FirstSec: .init + LastSec: .fini + VAddr: 0x1000 + Align: 0x1000 + Offset: 0x1000 + - Type: PT_LOAD + Flags: [ PF_R ] + FirstSec: .rodata + LastSec: .eh_frame + VAddr: 0x2000 + Align: 0x1000 + Offset: 0x2000 + - Type: PT_LOAD + Flags: [ PF_W, PF_R ] + FirstSec: .init_array + LastSec: .bss + VAddr: 0x3E10 + Align: 0x1000 + Offset: 0x2E10 + - Type: PT_DYNAMIC + Flags: [ PF_W, PF_R ] + FirstSec: .dynamic + LastSec: .dynamic + VAddr: 0x3E20 + Align: 0x8 + Offset: 0x2E20 + - Type: PT_NOTE + Flags: [ PF_R ] + FirstSec: .note.gnu.property + LastSec: .note.gnu.property + VAddr: 0x2A8 + Align: 0x8 + Offset: 0x2A8 + - Type: PT_NOTE + Flags: [ PF_R ] + FirstSec: .note.gnu.build-id + LastSec: .note.gnu.build-id + VAddr: 0x2C8 + Align: 0x4 + Offset: 0x2C8 + - Type: PT_GNU_PROPERTY + Flags: [ PF_R ] + FirstSec: .note.gnu.property + LastSec: .note.gnu.property + VAddr: 0x2A8 + Align: 0x8 + Offset: 0x2A8 + - Type: PT_GNU_EH_FRAME + Flags: [ PF_R ] + FirstSec: .eh_frame_hdr + LastSec: .eh_frame_hdr + VAddr: 0x2010 + Align: 0x4 + Offset: 0x2010 + - Type: PT_GNU_STACK + Flags: [ PF_W, PF_R ] + Align: 0x10 + Offset: 0x0 + - Type: PT_GNU_RELRO + Flags: [ PF_R ] + FirstSec: .init_array + LastSec: .got + VAddr: 0x3E10 + Offset: 0x2E10 +Sections: + - Name: .note.gnu.property + Type: SHT_NOTE + Flags: [ SHF_ALLOC ] + Address: 0x2A8 + AddressAlign: 0x8 + Notes: + - Name: GNU + Desc: 020000C0040000000300000000000000 + Type: NT_GNU_PROPERTY_TYPE_0 + - Name: .note.gnu.build-id + Type: SHT_NOTE + Flags: [ SHF_ALLOC ] + Address: 0x2C8 + AddressAlign: 0x4 + Notes: + - Name: GNU + Desc: 6337F7C1BF21A1DE17630C55602EB4CAC50435BB + Type: NT_PRPSINFO + - Name: .gnu.hash + Type: SHT_GNU_HASH + Flags: [ SHF_ALLOC ] + Address: 0x2F0 + Link: .dynsym + AddressAlign: 0x8 + Header: + SymNdx: 0x6 + Shift2: 0x6 + BloomFilter: [ 0x400000100000 ] + HashBuckets: [ 0x6, 0x0 ] + HashValues: [ 0x7C9DCB95 ] + - Name: .dynsym + Type: SHT_DYNSYM + Flags: [ SHF_ALLOC ] + Address: 0x318 + Link: .dynstr + AddressAlign: 0x8 + - Name: .dynstr + Type: SHT_STRTAB + Flags: [ SHF_ALLOC ] + Address: 0x3C0 + AddressAlign: 0x1 + - Name: .gnu.version + Type: SHT_GNU_versym + Flags: [ SHF_ALLOC ] + Address: 0x436 + Link: .dynsym + AddressAlign: 0x2 + Entries: [ 0, 1, 2, 1, 1, 2, 1 ] + - Name: .gnu.version_r + Type: SHT_GNU_verneed + Flags: [ SHF_ALLOC ] + Address: 0x448 + Link: .dynstr + AddressAlign: 0x8 + Dependencies: + - Version: 1 + File: libc.so.6 + Entries: + - Name: GLIBC_2.2.5 + Hash: 157882997 + Flags: 0 + Other: 2 + - Name: .rela.dyn + Type: SHT_RELA + Flags: [ SHF_ALLOC ] + Address: 0x468 + Link: .dynsym + AddressAlign: 0x8 + Relocations: + - Offset: 0x3E10 + Type: R_X86_64_RELATIVE + Addend: 4368 + - Offset: 0x3E18 + Type: R_X86_64_RELATIVE + Addend: 4304 + - Offset: 0x4020 + Type: R_X86_64_RELATIVE + Addend: 16416 + - Offset: 0x3FE0 + Symbol: _ITM_deregisterTMCloneTable + Type: R_X86_64_GLOB_DAT + - Offset: 0x3FE8 + Symbol: __gmon_start__ + Type: R_X86_64_GLOB_DAT + - Offset: 0x3FF0 + Symbol: _ITM_registerTMCloneTable + Type: R_X86_64_GLOB_DAT + - Offset: 0x3FF8 + Symbol: __cxa_finalize + Type: R_X86_64_GLOB_DAT + - Name: .rela.plt + Type: SHT_RELA + Flags: [ SHF_ALLOC, SHF_INFO_LINK ] + Address: 0x510 + Link: .dynsym + AddressAlign: 0x8 + Info: .got.plt + Relocations: + - Offset: 0x4018 + Symbol: puts + Type: R_X86_64_JUMP_SLOT + - Name: .init + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1000 + AddressAlign: 0x4 + Offset: 0x1000 + Content: F30F1EFA4883EC08488B05D92F00004885C07402FFD04883C408C3 + - Name: .plt + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1020 + AddressAlign: 0x10 + EntSize: 0x10 + Content: FF35E22F0000F2FF25E32F00000F1F00F30F1EFA6800000000F2E9E1FFFFFF90 + - Name: .plt.got + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1040 + AddressAlign: 0x10 + EntSize: 0x10 + Content: F30F1EFAF2FF25AD2F00000F1F440000 + - Name: .plt.sec + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1050 + AddressAlign: 0x10 + EntSize: 0x10 + Content: F30F1EFAF2FF25BD2F00000F1F440000 + - Name: .text + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1060 + AddressAlign: 0x10 + Content: 488D3DC12F0000488D05BA2F00004839F87415488B05662F00004885C07409FFE00F1F8000000000C30F1F8000000000488D3D912F0000488D358A2F00004829FE4889F048C1EE3F48C1F8034801C648D1FE7414488B05352F00004885C07408FFE0660F1F440000C30F1F8000000000F30F1EFA803D4D2F000000752B5548833D122F0000004889E5740C488B3D2E2F0000E849FFFFFFE864FFFFFFC605252F0000015DC30F1F00C30F1F8000000000F30F1EFAE977FFFFFFF30F1EFA554889E5488D05D80E00004889C7E820FFFFFF905DC3 + - Name: .fini + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1134 + AddressAlign: 0x4 + Content: F30F1EFA4883EC084883C408C3 + - Name: .rodata + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC ] + Address: 0x2000 + AddressAlign: 0x1 + Offset: 0x2000 + Content: 48656C6C6F2066726F6D204200 + - Name: .eh_frame_hdr + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC ] + Address: 0x2010 + AddressAlign: 0x4 + Content: 011B033B2C0000000400000010F0FFFF4800000030F0FFFF7000000040F0FFFF8800000009F1FFFFA0000000 + - Name: .eh_frame + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC ] + Address: 0x2040 + AddressAlign: 0x8 + Content: 1400000000000000017A5200017810011B0C070890010000240000001C000000C0EFFFFF20000000000E10460E184A0F0B770880003F1A3A2A332422000000001400000044000000B8EFFFFF100000000000000000000000140000005C000000B0EFFFFF1000000000000000000000001C0000007400000061F0FFFF1A00000000450E108602430D06510C070800000000000000 + - Name: .init_array + Type: SHT_INIT_ARRAY + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3E10 + AddressAlign: 0x8 + EntSize: 0x8 + Offset: 0x2E10 + Content: '1011000000000000' + - Name: .fini_array + Type: SHT_FINI_ARRAY + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3E18 + AddressAlign: 0x8 + EntSize: 0x8 + Content: D010000000000000 + - Name: .dynamic + Type: SHT_DYNAMIC + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3E20 + Link: .dynstr + AddressAlign: 0x8 + Entries: + - Tag: DT_NEEDED + Value: 0x5F + - Tag: DT_INIT + Value: 0x1000 + - Tag: DT_FINI + Value: 0x1134 + - Tag: DT_INIT_ARRAY + Value: 0x3E10 + - Tag: DT_INIT_ARRAYSZ + Value: 0x8 + - Tag: DT_FINI_ARRAY + Value: 0x3E18 + - Tag: DT_FINI_ARRAYSZ + Value: 0x8 + - Tag: DT_GNU_HASH + Value: 0x2F0 + - Tag: DT_STRTAB + Value: 0x3C0 + - Tag: DT_SYMTAB + Value: 0x318 + - Tag: DT_STRSZ + Value: 0x75 + - Tag: DT_SYMENT + Value: 0x18 + - Tag: DT_PLTGOT + Value: 0x4000 + - Tag: DT_PLTRELSZ + Value: 0x18 + - Tag: DT_PLTREL + Value: 0x7 + - Tag: DT_JMPREL + Value: 0x510 + - Tag: DT_RELA + Value: 0x468 + - Tag: DT_RELASZ + Value: 0xA8 + - Tag: DT_RELAENT + Value: 0x18 + - Tag: DT_VERNEED + Value: 0x448 + - Tag: DT_VERNEEDNUM + Value: 0x1 + - Tag: DT_VERSYM + Value: 0x436 + - Tag: DT_RELACOUNT + Value: 0x3 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Name: .got + Type: SHT_PROGBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3FE0 + AddressAlign: 0x8 + EntSize: 0x8 + Content: '0000000000000000000000000000000000000000000000000000000000000000' + - Name: .got.plt + Type: SHT_PROGBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x4000 + AddressAlign: 0x8 + EntSize: 0x8 + Content: '203E000000000000000000000000000000000000000000003010000000000000' + - Name: .data + Type: SHT_PROGBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x4020 + AddressAlign: 0x8 + Content: '2040000000000000' + - Name: .bss + Type: SHT_NOBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x4028 + AddressAlign: 0x1 + Size: 0x8 + - Name: .comment + Type: SHT_PROGBITS + Flags: [ SHF_MERGE, SHF_STRINGS ] + AddressAlign: 0x1 + EntSize: 0x1 + Content: 4743433A20285562756E74752031312E342E302D317562756E7475317E32322E30342E32292031312E342E3000 +Symbols: + - Name: crtstuff.c + Type: STT_FILE + Index: SHN_ABS + - Name: deregister_tm_clones + Type: STT_FUNC + Section: .text + Value: 0x1060 + - Name: register_tm_clones + Type: STT_FUNC + Section: .text + Value: 0x1090 + - Name: __do_global_dtors_aux + Type: STT_FUNC + Section: .text + Value: 0x10D0 + - Name: completed.0 + Type: STT_OBJECT + Section: .bss + Value: 0x4028 + Size: 0x1 + - Name: __do_global_dtors_aux_fini_array_entry + Type: STT_OBJECT + Section: .fini_array + Value: 0x3E18 + - Name: frame_dummy + Type: STT_FUNC + Section: .text + Value: 0x1110 + - Name: __frame_dummy_init_array_entry + Type: STT_OBJECT + Section: .init_array + Value: 0x3E10 + - Name: libB.c + Type: STT_FILE + Index: SHN_ABS + - Name: 'crtstuff.c (1)' + Type: STT_FILE + Index: SHN_ABS + - Name: __FRAME_END__ + Type: STT_OBJECT + Section: .eh_frame + Value: 0x20D0 + - Type: STT_FILE + Index: SHN_ABS + - Name: _fini + Type: STT_FUNC + Section: .fini + Value: 0x1134 + - Name: __dso_handle + Type: STT_OBJECT + Section: .data + Value: 0x4020 + - Name: _DYNAMIC + Type: STT_OBJECT + Section: .dynamic + Value: 0x3E20 + - Name: __GNU_EH_FRAME_HDR + Section: .eh_frame_hdr + Value: 0x2010 + - Name: __TMC_END__ + Type: STT_OBJECT + Section: .data + Value: 0x4028 + - Name: _GLOBAL_OFFSET_TABLE_ + Type: STT_OBJECT + Section: .got.plt + Value: 0x4000 + - Name: _init + Type: STT_FUNC + Section: .init + Value: 0x1000 + - Name: _ITM_deregisterTMCloneTable + Binding: STB_WEAK + - Name: 'puts@GLIBC_2.2.5' + Type: STT_FUNC + Binding: STB_GLOBAL + - Name: __gmon_start__ + Binding: STB_WEAK + - Name: sayB + Type: STT_FUNC + Section: .text + Binding: STB_GLOBAL + Value: 0x1119 + Size: 0x1A + - Name: _ITM_registerTMCloneTable + Binding: STB_WEAK + - Name: '__cxa_finalize@GLIBC_2.2.5' + Type: STT_FUNC + Binding: STB_WEAK +DynamicSymbols: + - Name: _ITM_deregisterTMCloneTable + Binding: STB_WEAK + - Name: puts + Type: STT_FUNC + Binding: STB_GLOBAL + - Name: __gmon_start__ + Binding: STB_WEAK + - Name: _ITM_registerTMCloneTable + Binding: STB_WEAK + - Name: __cxa_finalize + Type: STT_FUNC + Binding: STB_WEAK + - Name: sayB + Type: STT_FUNC + Section: .text + Binding: STB_GLOBAL + Value: 0x1119 + Size: 0x1A +... diff --git a/llvm/unittests/ExecutionEngine/Orc/Inputs/B/B_macho.yaml b/llvm/unittests/ExecutionEngine/Orc/Inputs/B/B_macho.yaml new file mode 100644 index 0000000..3d57c4f --- /dev/null +++ b/llvm/unittests/ExecutionEngine/Orc/Inputs/B/B_macho.yaml @@ -0,0 +1,723 @@ +--- !fat-mach-o +FatHeader: + magic: 0xCAFEBABE + nfat_arch: 3 +FatArchs: + - cputype: 0x1000007 + cpusubtype: 0x3 + offset: 0x1000 + size: 8376 + align: 12 + - cputype: 0x100000C + cpusubtype: 0x0 + offset: 0x4000 + size: 33376 + align: 14 + - cputype: 0x100000C + cpusubtype: 0x80000002 + offset: 0x10000 + size: 33376 + align: 14 +Slices: + - !mach-o + FileHeader: + magic: 0xFEEDFACF + cputype: 0x1000007 + cpusubtype: 0x3 + filetype: 0x6 + ncmds: 14 + sizeofcmds: 960 + flags: 0x100085 + reserved: 0x0 + LoadCommands: + - cmd: LC_SEGMENT_64 + cmdsize: 392 + segname: __TEXT + vmaddr: 0 + vmsize: 4096 + fileoff: 0 + filesize: 4096 + maxprot: 5 + initprot: 5 + nsects: 4 + flags: 0 + Sections: + - sectname: __text + segname: __TEXT + addr: 0xF80 + size: 20 + offset: 0xF80 + align: 4 + reloff: 0x0 + nreloc: 0 + flags: 0x80000400 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 554889E5488D3D0F000000B000E8020000005DC3 + - sectname: __stubs + segname: __TEXT + addr: 0xF94 + size: 6 + offset: 0xF94 + align: 1 + reloff: 0x0 + nreloc: 0 + flags: 0x80000408 + reserved1: 0x0 + reserved2: 0x6 + reserved3: 0x0 + content: FF2566000000 + - sectname: __cstring + segname: __TEXT + addr: 0xF9A + size: 14 + offset: 0xF9A + align: 0 + reloff: 0x0 + nreloc: 0 + flags: 0x2 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 48656C6C6F2066726F6D20420A00 + - sectname: __unwind_info + segname: __TEXT + addr: 0xFA8 + size: 88 + offset: 0xFA8 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x0 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 010000001C000000000000001C000000000000001C00000002000000800F00004000000040000000940F00000000000040000000000000000000000000000000030000000C00010010000100000000000000000100000000 + - cmd: LC_SEGMENT_64 + cmdsize: 152 + segname: __DATA_CONST + vmaddr: 4096 + vmsize: 4096 + fileoff: 4096 + filesize: 4096 + maxprot: 3 + initprot: 3 + nsects: 1 + flags: 16 + Sections: + - sectname: __got + segname: __DATA_CONST + addr: 0x1000 + size: 8 + offset: 0x1000 + align: 3 + reloff: 0x0 + nreloc: 0 + flags: 0x6 + reserved1: 0x1 + reserved2: 0x0 + reserved3: 0x0 + content: '0000000000000080' + - cmd: LC_SEGMENT_64 + cmdsize: 72 + segname: __LINKEDIT + vmaddr: 8192 + vmsize: 4096 + fileoff: 8192 + filesize: 184 + maxprot: 1 + initprot: 1 + nsects: 0 + flags: 0 + - cmd: LC_ID_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 1 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libB.dylib' + ZeroPadBytes: 7 + - cmd: LC_DYLD_CHAINED_FIXUPS + cmdsize: 16 + dataoff: 8192 + datasize: 96 + - cmd: LC_DYLD_EXPORTS_TRIE + cmdsize: 16 + dataoff: 8288 + datasize: 24 + - cmd: LC_SYMTAB + cmdsize: 24 + symoff: 8320 + nsyms: 2 + stroff: 8360 + strsize: 16 + - cmd: LC_DYSYMTAB + cmdsize: 80 + ilocalsym: 0 + nlocalsym: 0 + iextdefsym: 0 + nextdefsym: 1 + iundefsym: 1 + nundefsym: 1 + tocoff: 0 + ntoc: 0 + modtaboff: 0 + nmodtab: 0 + extrefsymoff: 0 + nextrefsyms: 0 + indirectsymoff: 8352 + nindirectsyms: 2 + extreloff: 0 + nextrel: 0 + locreloff: 0 + nlocrel: 0 + - cmd: LC_UUID + cmdsize: 24 + uuid: 88B60B3C-13D3-3D7E-AEED-5F3E991FDF08 + - cmd: LC_BUILD_VERSION + cmdsize: 32 + platform: 1 + minos: 983040 + sdk: 983552 + ntools: 1 + Tools: + - tool: 3 + version: 73074435 + - cmd: LC_SOURCE_VERSION + cmdsize: 16 + version: 0 + - cmd: LC_LOAD_DYLIB + cmdsize: 56 + dylib: + name: 24 + timestamp: 2 + current_version: 88539136 + compatibility_version: 65536 + Content: '/usr/lib/libSystem.B.dylib' + ZeroPadBytes: 6 + - cmd: LC_FUNCTION_STARTS + cmdsize: 16 + dataoff: 8312 + datasize: 8 + - cmd: LC_DATA_IN_CODE + cmdsize: 16 + dataoff: 8320 + datasize: 0 + LinkEditData: + ExportTrie: + TerminalSize: 0 + NodeOffset: 0 + Name: '' + Flags: 0x0 + Address: 0x0 + Other: 0x0 + ImportName: '' + Children: + - TerminalSize: 3 + NodeOffset: 13 + Name: _sayB + Flags: 0x0 + Address: 0xF80 + Other: 0x0 + ImportName: '' + NameList: + - n_strx: 2 + n_type: 0xF + n_sect: 1 + n_desc: 0 + n_value: 3968 + - n_strx: 8 + n_type: 0x1 + n_sect: 0 + n_desc: 256 + n_value: 0 + StringTable: + - ' ' + - _sayB + - _printf + IndirectSymbols: [ 0x1, 0x1 ] + FunctionStarts: [ 0xF80 ] + ChainedFixups: [ 0x0, 0x0, 0x0, 0x0, 0x20, 0x0, 0x0, 0x0, 0x48, + 0x0, 0x0, 0x0, 0x50, 0x0, 0x0, 0x0, 0x1, 0x0, + 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x3, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x18, 0x0, 0x0, 0x0, 0x0, 0x10, 0x6, 0x0, + 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x1, 0x2, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x5F, 0x70, 0x72, + 0x69, 0x6E, 0x74, 0x66, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0 ] + - !mach-o + FileHeader: + magic: 0xFEEDFACF + cputype: 0x100000C + cpusubtype: 0x0 + filetype: 0x6 + ncmds: 15 + sizeofcmds: 976 + flags: 0x100085 + reserved: 0x0 + LoadCommands: + - cmd: LC_SEGMENT_64 + cmdsize: 392 + segname: __TEXT + vmaddr: 0 + vmsize: 16384 + fileoff: 0 + filesize: 16384 + maxprot: 5 + initprot: 5 + nsects: 4 + flags: 0 + Sections: + - sectname: __text + segname: __TEXT + addr: 0x3F70 + size: 28 + offset: 0x3F70 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000400 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: FD7BBFA9FD0300910000009000603E9103000094FD7BC1A8C0035FD6 + - sectname: __stubs + segname: __TEXT + addr: 0x3F8C + size: 12 + offset: 0x3F8C + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000408 + reserved1: 0x0 + reserved2: 0xC + reserved3: 0x0 + content: 100000B0100240F900021FD6 + - sectname: __cstring + segname: __TEXT + addr: 0x3F98 + size: 14 + offset: 0x3F98 + align: 0 + reloff: 0x0 + nreloc: 0 + flags: 0x2 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 48656C6C6F2066726F6D20420A00 + - sectname: __unwind_info + segname: __TEXT + addr: 0x3FA8 + size: 88 + offset: 0x3FA8 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x0 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 010000001C000000000000001C000000000000001C00000002000000703F000040000000400000008C3F00000000000040000000000000000000000000000000030000000C00010010000100000000000000000400000000 + - cmd: LC_SEGMENT_64 + cmdsize: 152 + segname: __DATA_CONST + vmaddr: 16384 + vmsize: 16384 + fileoff: 16384 + filesize: 16384 + maxprot: 3 + initprot: 3 + nsects: 1 + flags: 16 + Sections: + - sectname: __got + segname: __DATA_CONST + addr: 0x4000 + size: 8 + offset: 0x4000 + align: 3 + reloff: 0x0 + nreloc: 0 + flags: 0x6 + reserved1: 0x1 + reserved2: 0x0 + reserved3: 0x0 + content: '0000000000000080' + - cmd: LC_SEGMENT_64 + cmdsize: 72 + segname: __LINKEDIT + vmaddr: 32768 + vmsize: 16384 + fileoff: 32768 + filesize: 608 + maxprot: 1 + initprot: 1 + nsects: 0 + flags: 0 + - cmd: LC_ID_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 1 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libB.dylib' + ZeroPadBytes: 7 + - cmd: LC_DYLD_CHAINED_FIXUPS + cmdsize: 16 + dataoff: 32768 + datasize: 96 + - cmd: LC_DYLD_EXPORTS_TRIE + cmdsize: 16 + dataoff: 32864 + datasize: 24 + - cmd: LC_SYMTAB + cmdsize: 24 + symoff: 32896 + nsyms: 2 + stroff: 32936 + strsize: 16 + - cmd: LC_DYSYMTAB + cmdsize: 80 + ilocalsym: 0 + nlocalsym: 0 + iextdefsym: 0 + nextdefsym: 1 + iundefsym: 1 + nundefsym: 1 + tocoff: 0 + ntoc: 0 + modtaboff: 0 + nmodtab: 0 + extrefsymoff: 0 + nextrefsyms: 0 + indirectsymoff: 32928 + nindirectsyms: 2 + extreloff: 0 + nextrel: 0 + locreloff: 0 + nlocrel: 0 + - cmd: LC_UUID + cmdsize: 24 + uuid: 90C3787A-22E1-35AE-9284-97A4842F88AF + - cmd: LC_BUILD_VERSION + cmdsize: 32 + platform: 1 + minos: 983040 + sdk: 983552 + ntools: 1 + Tools: + - tool: 3 + version: 73074435 + - cmd: LC_SOURCE_VERSION + cmdsize: 16 + version: 0 + - cmd: LC_LOAD_DYLIB + cmdsize: 56 + dylib: + name: 24 + timestamp: 2 + current_version: 88539136 + compatibility_version: 65536 + Content: '/usr/lib/libSystem.B.dylib' + ZeroPadBytes: 6 + - cmd: LC_FUNCTION_STARTS + cmdsize: 16 + dataoff: 32888 + datasize: 8 + - cmd: LC_DATA_IN_CODE + cmdsize: 16 + dataoff: 32896 + datasize: 0 + - cmd: LC_CODE_SIGNATURE + cmdsize: 16 + dataoff: 32960 + datasize: 416 + LinkEditData: + ExportTrie: + TerminalSize: 0 + NodeOffset: 0 + Name: '' + Flags: 0x0 + Address: 0x0 + Other: 0x0 + ImportName: '' + Children: + - TerminalSize: 3 + NodeOffset: 13 + Name: _sayB + Flags: 0x0 + Address: 0x3F70 + Other: 0x0 + ImportName: '' + NameList: + - n_strx: 2 + n_type: 0xF + n_sect: 1 + n_desc: 0 + n_value: 16240 + - n_strx: 8 + n_type: 0x1 + n_sect: 0 + n_desc: 256 + n_value: 0 + StringTable: + - ' ' + - _sayB + - _printf + IndirectSymbols: [ 0x1, 0x1 ] + FunctionStarts: [ 0x3F70 ] + ChainedFixups: [ 0x0, 0x0, 0x0, 0x0, 0x20, 0x0, 0x0, 0x0, 0x48, + 0x0, 0x0, 0x0, 0x50, 0x0, 0x0, 0x0, 0x1, 0x0, + 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x3, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x18, 0x0, 0x0, 0x0, 0x0, 0x40, 0x6, 0x0, + 0x0, 0x40, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x1, 0x2, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x5F, 0x70, 0x72, + 0x69, 0x6E, 0x74, 0x66, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0 ] + - !mach-o + FileHeader: + magic: 0xFEEDFACF + cputype: 0x100000C + cpusubtype: 0x80000002 + filetype: 0x6 + ncmds: 15 + sizeofcmds: 976 + flags: 0x100085 + reserved: 0x0 + LoadCommands: + - cmd: LC_SEGMENT_64 + cmdsize: 392 + segname: __TEXT + vmaddr: 0 + vmsize: 16384 + fileoff: 0 + filesize: 16384 + maxprot: 5 + initprot: 5 + nsects: 4 + flags: 0 + Sections: + - sectname: __text + segname: __TEXT + addr: 0x3F68 + size: 32 + offset: 0x3F68 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000400 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 7F2303D5FD7BBFA9FD0300910000009000603E9103000094FD7BC1A8FF0F5FD6 + - sectname: __auth_stubs + segname: __TEXT + addr: 0x3F88 + size: 16 + offset: 0x3F88 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000408 + reserved1: 0x0 + reserved2: 0x10 + reserved3: 0x0 + content: 110000B031020091300240F9110A1FD7 + - sectname: __cstring + segname: __TEXT + addr: 0x3F98 + size: 14 + offset: 0x3F98 + align: 0 + reloff: 0x0 + nreloc: 0 + flags: 0x2 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 48656C6C6F2066726F6D20420A00 + - sectname: __unwind_info + segname: __TEXT + addr: 0x3FA8 + size: 88 + offset: 0x3FA8 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x0 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 010000001C000000000000001C000000000000001C00000002000000683F00004000000040000000883F00000000000040000000000000000000000000000000030000000C00010010000100000000000000000400000000 + - cmd: LC_SEGMENT_64 + cmdsize: 152 + segname: __DATA_CONST + vmaddr: 16384 + vmsize: 16384 + fileoff: 16384 + filesize: 16384 + maxprot: 3 + initprot: 3 + nsects: 1 + flags: 16 + Sections: + - sectname: __auth_got + segname: __DATA_CONST + addr: 0x4000 + size: 8 + offset: 0x4000 + align: 3 + reloff: 0x0 + nreloc: 0 + flags: 0x6 + reserved1: 0x1 + reserved2: 0x0 + reserved3: 0x0 + content: 00000000000001C0 + - cmd: LC_SEGMENT_64 + cmdsize: 72 + segname: __LINKEDIT + vmaddr: 32768 + vmsize: 16384 + fileoff: 32768 + filesize: 608 + maxprot: 1 + initprot: 1 + nsects: 0 + flags: 0 + - cmd: LC_ID_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 1 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libB.dylib' + ZeroPadBytes: 7 + - cmd: LC_DYLD_CHAINED_FIXUPS + cmdsize: 16 + dataoff: 32768 + datasize: 96 + - cmd: LC_DYLD_EXPORTS_TRIE + cmdsize: 16 + dataoff: 32864 + datasize: 24 + - cmd: LC_SYMTAB + cmdsize: 24 + symoff: 32896 + nsyms: 2 + stroff: 32936 + strsize: 16 + - cmd: LC_DYSYMTAB + cmdsize: 80 + ilocalsym: 0 + nlocalsym: 0 + iextdefsym: 0 + nextdefsym: 1 + iundefsym: 1 + nundefsym: 1 + tocoff: 0 + ntoc: 0 + modtaboff: 0 + nmodtab: 0 + extrefsymoff: 0 + nextrefsyms: 0 + indirectsymoff: 32928 + nindirectsyms: 2 + extreloff: 0 + nextrel: 0 + locreloff: 0 + nlocrel: 0 + - cmd: LC_UUID + cmdsize: 24 + uuid: 76B41B3A-00EC-388B-A432-478A96772CC4 + - cmd: LC_BUILD_VERSION + cmdsize: 32 + platform: 1 + minos: 983040 + sdk: 983552 + ntools: 1 + Tools: + - tool: 3 + version: 73074435 + - cmd: LC_SOURCE_VERSION + cmdsize: 16 + version: 0 + - cmd: LC_LOAD_DYLIB + cmdsize: 56 + dylib: + name: 24 + timestamp: 2 + current_version: 88539136 + compatibility_version: 65536 + Content: '/usr/lib/libSystem.B.dylib' + ZeroPadBytes: 6 + - cmd: LC_FUNCTION_STARTS + cmdsize: 16 + dataoff: 32888 + datasize: 8 + - cmd: LC_DATA_IN_CODE + cmdsize: 16 + dataoff: 32896 + datasize: 0 + - cmd: LC_CODE_SIGNATURE + cmdsize: 16 + dataoff: 32960 + datasize: 416 + LinkEditData: + ExportTrie: + TerminalSize: 0 + NodeOffset: 0 + Name: '' + Flags: 0x0 + Address: 0x0 + Other: 0x0 + ImportName: '' + Children: + - TerminalSize: 3 + NodeOffset: 13 + Name: _sayB + Flags: 0x0 + Address: 0x3F68 + Other: 0x0 + ImportName: '' + NameList: + - n_strx: 2 + n_type: 0xF + n_sect: 1 + n_desc: 0 + n_value: 16232 + - n_strx: 8 + n_type: 0x1 + n_sect: 0 + n_desc: 256 + n_value: 0 + StringTable: + - ' ' + - _sayB + - _printf + IndirectSymbols: [ 0x1, 0x1 ] + FunctionStarts: [ 0x3F68 ] + ChainedFixups: [ 0x0, 0x0, 0x0, 0x0, 0x20, 0x0, 0x0, 0x0, 0x48, + 0x0, 0x0, 0x0, 0x50, 0x0, 0x0, 0x0, 0x1, 0x0, + 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x3, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x18, 0x0, 0x0, 0x0, 0x0, 0x40, 0xC, 0x0, + 0x0, 0x40, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x1, 0x2, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x5F, 0x70, 0x72, + 0x69, 0x6E, 0x74, 0x66, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0 ] +... diff --git a/llvm/unittests/ExecutionEngine/Orc/Inputs/C/C_linux.yaml b/llvm/unittests/ExecutionEngine/Orc/Inputs/C/C_linux.yaml new file mode 100644 index 0000000..3fabf9a --- /dev/null +++ b/llvm/unittests/ExecutionEngine/Orc/Inputs/C/C_linux.yaml @@ -0,0 +1,450 @@ +--- !ELF +FileHeader: + Class: ELFCLASS64 + Data: ELFDATA2LSB + Type: ET_DYN + Machine: EM_X86_64 +ProgramHeaders: + - Type: PT_LOAD + Flags: [ PF_R ] + FirstSec: .note.gnu.property + LastSec: .rela.plt + Align: 0x1000 + Offset: 0x0 + - Type: PT_LOAD + Flags: [ PF_X, PF_R ] + FirstSec: .init + LastSec: .fini + VAddr: 0x1000 + Align: 0x1000 + Offset: 0x1000 + - Type: PT_LOAD + Flags: [ PF_R ] + FirstSec: .eh_frame_hdr + LastSec: .eh_frame + VAddr: 0x2000 + Align: 0x1000 + Offset: 0x2000 + - Type: PT_LOAD + Flags: [ PF_W, PF_R ] + FirstSec: .init_array + LastSec: .bss + VAddr: 0x3E10 + Align: 0x1000 + Offset: 0x2E10 + - Type: PT_DYNAMIC + Flags: [ PF_W, PF_R ] + FirstSec: .dynamic + LastSec: .dynamic + VAddr: 0x3E20 + Align: 0x8 + Offset: 0x2E20 + - Type: PT_NOTE + Flags: [ PF_R ] + FirstSec: .note.gnu.property + LastSec: .note.gnu.property + VAddr: 0x2A8 + Align: 0x8 + Offset: 0x2A8 + - Type: PT_NOTE + Flags: [ PF_R ] + FirstSec: .note.gnu.build-id + LastSec: .note.gnu.build-id + VAddr: 0x2C8 + Align: 0x4 + Offset: 0x2C8 + - Type: PT_GNU_PROPERTY + Flags: [ PF_R ] + FirstSec: .note.gnu.property + LastSec: .note.gnu.property + VAddr: 0x2A8 + Align: 0x8 + Offset: 0x2A8 + - Type: PT_GNU_EH_FRAME + Flags: [ PF_R ] + FirstSec: .eh_frame_hdr + LastSec: .eh_frame_hdr + VAddr: 0x2000 + Align: 0x4 + Offset: 0x2000 + - Type: PT_GNU_STACK + Flags: [ PF_W, PF_R ] + Align: 0x10 + Offset: 0x0 + - Type: PT_GNU_RELRO + Flags: [ PF_R ] + FirstSec: .init_array + LastSec: .got + VAddr: 0x3E10 + Offset: 0x2E10 +Sections: + - Name: .note.gnu.property + Type: SHT_NOTE + Flags: [ SHF_ALLOC ] + Address: 0x2A8 + AddressAlign: 0x8 + Notes: + - Name: GNU + Desc: 020000C0040000000300000000000000 + Type: NT_GNU_PROPERTY_TYPE_0 + - Name: .note.gnu.build-id + Type: SHT_NOTE + Flags: [ SHF_ALLOC ] + Address: 0x2C8 + AddressAlign: 0x4 + Notes: + - Name: GNU + Desc: 0318D63E46BF31CEFF90D5C7F0475D9F78676EC8 + Type: NT_PRPSINFO + - Name: .gnu.hash + Type: SHT_GNU_HASH + Flags: [ SHF_ALLOC ] + Address: 0x2F0 + Link: .dynsym + AddressAlign: 0x8 + Header: + SymNdx: 0x8 + Shift2: 0x6 + BloomFilter: [ 0x400000200000 ] + HashBuckets: [ 0x0, 0x8 ] + HashValues: [ 0x7C9DCB95 ] + - Name: .dynsym + Type: SHT_DYNSYM + Flags: [ SHF_ALLOC ] + Address: 0x318 + Link: .dynstr + AddressAlign: 0x8 + - Name: .dynstr + Type: SHT_STRTAB + Flags: [ SHF_ALLOC ] + Address: 0x3F0 + AddressAlign: 0x1 + Content: "6C6962412E736F006C6962422E736F006C69625A2E736F00244F524947494E2F2E2E2F413A244F524947494E2F2E2E2F423A244F524947494E2F2E2E2F5A" + - Name: .rela.dyn + Type: SHT_RELA + Flags: [ SHF_ALLOC ] + Address: 0x498 + Link: .dynsym + AddressAlign: 0x8 + Relocations: + - Offset: 0x3E10 + Type: R_X86_64_RELATIVE + Addend: 4432 + - Offset: 0x3E18 + Type: R_X86_64_RELATIVE + Addend: 4368 + - Offset: 0x4030 + Type: R_X86_64_RELATIVE + Addend: 16432 + - Offset: 0x3FE0 + Symbol: __cxa_finalize + Type: R_X86_64_GLOB_DAT + - Offset: 0x3FE8 + Symbol: _ITM_registerTMCloneTable + Type: R_X86_64_GLOB_DAT + - Offset: 0x3FF0 + Symbol: _ITM_deregisterTMCloneTable + Type: R_X86_64_GLOB_DAT + - Offset: 0x3FF8 + Symbol: __gmon_start__ + Type: R_X86_64_GLOB_DAT + - Name: .rela.plt + Type: SHT_RELA + Flags: [ SHF_ALLOC, SHF_INFO_LINK ] + Address: 0x540 + Link: .dynsym + AddressAlign: 0x8 + Info: .got.plt + Relocations: + - Offset: 0x4018 + Symbol: sayA + Type: R_X86_64_JUMP_SLOT + - Offset: 0x4020 + Symbol: sayB + Type: R_X86_64_JUMP_SLOT + - Offset: 0x4028 + Symbol: sayZ + Type: R_X86_64_JUMP_SLOT + - Name: .init + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1000 + AddressAlign: 0x4 + Offset: 0x1000 + Content: F30F1EFA4883EC08488B05E92F00004885C07402FFD04883C408C3 + - Name: .plt + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1020 + AddressAlign: 0x10 + EntSize: 0x10 + Content: FF35E22F0000F2FF25E32F00000F1F00F30F1EFA6800000000F2E9E1FFFFFF90F30F1EFA6801000000F2E9D1FFFFFF90F30F1EFA6802000000F2E9C1FFFFFF90 + - Name: .plt.got + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1060 + AddressAlign: 0x10 + EntSize: 0x10 + Content: F30F1EFAF2FF25752F00000F1F440000 + - Name: .plt.sec + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1070 + AddressAlign: 0x10 + EntSize: 0x10 + Content: F30F1EFAF2FF259D2F00000F1F440000F30F1EFAF2FF25952F00000F1F440000F30F1EFAF2FF258D2F00000F1F440000 + - Name: .text + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x10A0 + AddressAlign: 0x10 + Content: 488D3D912F0000488D058A2F00004839F87415488B05362F00004885C07409FFE00F1F8000000000C30F1F8000000000488D3D612F0000488D355A2F00004829FE4889F048C1EE3F48C1F8034801C648D1FE7414488B05ED2E00004885C07408FFE0660F1F440000C30F1F8000000000F30F1EFA803D1D2F000000752B5548833DBA2E0000004889E5740C488B3DFE2E0000E829FFFFFFE864FFFFFFC605F52E0000015DC30F1F00C30F1F8000000000F30F1EFAE977FFFFFFF30F1EFA554889E5B800000000E805FFFFFFB800000000E80BFFFFFFB800000000E811FFFFFF905DC3 + - Name: .fini + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1184 + AddressAlign: 0x4 + Content: F30F1EFA4883EC084883C408C3 + - Name: .eh_frame_hdr + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC ] + Address: 0x2000 + AddressAlign: 0x4 + Offset: 0x2000 + Content: 011B033B2C0000000400000020F0FFFF4800000060F0FFFF7000000070F0FFFF8800000059F1FFFFA0000000 + - Name: .eh_frame + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC ] + Address: 0x2030 + AddressAlign: 0x8 + Content: 1400000000000000017A5200017810011B0C070890010000240000001C000000D0EFFFFF40000000000E10460E184A0F0B770880003F1A3A2A332422000000001400000044000000E8EFFFFF100000000000000000000000140000005C000000E0EFFFFF3000000000000000000000001C00000074000000B1F0FFFF2900000000450E108602430D06600C070800000000000000 + - Name: .init_array + Type: SHT_INIT_ARRAY + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3E10 + AddressAlign: 0x8 + EntSize: 0x8 + Offset: 0x2E10 + Content: '5011000000000000' + - Name: .fini_array + Type: SHT_FINI_ARRAY + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3E18 + AddressAlign: 0x8 + EntSize: 0x8 + Content: '1011000000000000' + - Name: .dynamic + Type: SHT_DYNAMIC + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3E20 + Link: .dynstr + AddressAlign: 0x8 + Entries: + - Tag: DT_NEEDED + Value: 0x0 + - Tag: DT_NEEDED + Value: 0x8 + - Tag: DT_NEEDED + Value: 0x10 + - Tag: DT_RUNPATH + Value: 0x18 + - Tag: DT_INIT + Value: 0x1000 + - Tag: DT_FINI + Value: 0x1184 + - Tag: DT_INIT_ARRAY + Value: 0x3E10 + - Tag: DT_INIT_ARRAYSZ + Value: 0x8 + - Tag: DT_FINI_ARRAY + Value: 0x3E18 + - Tag: DT_FINI_ARRAYSZ + Value: 0x8 + - Tag: DT_GNU_HASH + Value: 0x2F0 + - Tag: DT_STRTAB + Value: 0x3F0 + - Tag: DT_SYMTAB + Value: 0x318 + - Tag: DT_STRSZ + Value: 0xA8 + - Tag: DT_SYMENT + Value: 0x18 + - Tag: DT_PLTGOT + Value: 0x4000 + - Tag: DT_PLTRELSZ + Value: 0x48 + - Tag: DT_PLTREL + Value: 0x7 + - Tag: DT_JMPREL + Value: 0x540 + - Tag: DT_RELA + Value: 0x498 + - Tag: DT_RELASZ + Value: 0xA8 + - Tag: DT_RELAENT + Value: 0x18 + - Tag: DT_RELACOUNT + Value: 0x3 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Name: .got + Type: SHT_PROGBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3FE0 + AddressAlign: 0x8 + EntSize: 0x8 + Content: '0000000000000000000000000000000000000000000000000000000000000000' + - Name: .got.plt + Type: SHT_PROGBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x4000 + AddressAlign: 0x8 + EntSize: 0x8 + Content: '203E00000000000000000000000000000000000000000000301000000000000040100000000000005010000000000000' + - Name: .data + Type: SHT_PROGBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x4030 + AddressAlign: 0x8 + Content: '3040000000000000' + - Name: .bss + Type: SHT_NOBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x4038 + AddressAlign: 0x1 + Size: 0x8 + - Name: .comment + Type: SHT_PROGBITS + Flags: [ SHF_MERGE, SHF_STRINGS ] + AddressAlign: 0x1 + EntSize: 0x1 + Content: 4743433A20285562756E74752031312E342E302D317562756E7475317E32322E30342E32292031312E342E3000 +Symbols: + - Name: crtstuff.c + Type: STT_FILE + Index: SHN_ABS + - Name: deregister_tm_clones + Type: STT_FUNC + Section: .text + Value: 0x10A0 + - Name: register_tm_clones + Type: STT_FUNC + Section: .text + Value: 0x10D0 + - Name: __do_global_dtors_aux + Type: STT_FUNC + Section: .text + Value: 0x1110 + - Name: completed.0 + Type: STT_OBJECT + Section: .bss + Value: 0x4038 + Size: 0x1 + - Name: __do_global_dtors_aux_fini_array_entry + Type: STT_OBJECT + Section: .fini_array + Value: 0x3E18 + - Name: frame_dummy + Type: STT_FUNC + Section: .text + Value: 0x1150 + - Name: __frame_dummy_init_array_entry + Type: STT_OBJECT + Section: .init_array + Value: 0x3E10 + - Name: libC.c + Type: STT_FILE + Index: SHN_ABS + - Name: 'crtstuff.c (1)' + Type: STT_FILE + Index: SHN_ABS + - Name: __FRAME_END__ + Type: STT_OBJECT + Section: .eh_frame + Value: 0x20C0 + - Type: STT_FILE + Index: SHN_ABS + - Name: _DYNAMIC + Type: STT_OBJECT + Section: .dynamic + Value: 0x3E20 + - Name: __TMC_END__ + Type: STT_OBJECT + Section: .data + Value: 0x4038 + - Name: __dso_handle + Type: STT_OBJECT + Section: .data + Value: 0x4030 + - Name: _init + Type: STT_FUNC + Section: .init + Value: 0x1000 + - Name: __GNU_EH_FRAME_HDR + Section: .eh_frame_hdr + Value: 0x2000 + - Name: _fini + Type: STT_FUNC + Section: .fini + Value: 0x1184 + - Name: _GLOBAL_OFFSET_TABLE_ + Type: STT_OBJECT + Section: .got.plt + Value: 0x4000 + - Name: __cxa_finalize + Binding: STB_WEAK + - Name: sayC + Type: STT_FUNC + Section: .text + Binding: STB_GLOBAL + Value: 0x1159 + Size: 0x29 + - Name: _ITM_registerTMCloneTable + Binding: STB_WEAK + - Name: _ITM_deregisterTMCloneTable + Binding: STB_WEAK + - Name: sayA + Type: STT_FUNC + Binding: STB_GLOBAL + - Name: sayB + Type: STT_FUNC + Binding: STB_GLOBAL + - Name: sayZ + Type: STT_FUNC + Binding: STB_GLOBAL + - Name: __gmon_start__ + Binding: STB_WEAK +DynamicSymbols: + - Name: __cxa_finalize + Binding: STB_WEAK + - Name: _ITM_registerTMCloneTable + Binding: STB_WEAK + - Name: _ITM_deregisterTMCloneTable + Binding: STB_WEAK + - Name: sayA + Type: STT_FUNC + Binding: STB_GLOBAL + - Name: sayB + Type: STT_FUNC + Binding: STB_GLOBAL + - Name: sayZ + Type: STT_FUNC + Binding: STB_GLOBAL + - Name: __gmon_start__ + Binding: STB_WEAK + - Name: sayC + Type: STT_FUNC + Section: .text + Binding: STB_GLOBAL + Value: 0x1159 + Size: 0x29 +... diff --git a/llvm/unittests/ExecutionEngine/Orc/Inputs/C/C_macho.yaml b/llvm/unittests/ExecutionEngine/Orc/Inputs/C/C_macho.yaml new file mode 100644 index 0000000..ba33483 --- /dev/null +++ b/llvm/unittests/ExecutionEngine/Orc/Inputs/C/C_macho.yaml @@ -0,0 +1,870 @@ +--- !fat-mach-o +FatHeader: + magic: 0xCAFEBABE + nfat_arch: 3 +FatArchs: + - cputype: 0x1000007 + cpusubtype: 0x3 + offset: 0x1000 + size: 8456 + align: 12 + - cputype: 0x100000C + cpusubtype: 0x0 + offset: 0x4000 + size: 33456 + align: 14 + - cputype: 0x100000C + cpusubtype: 0x80000002 + offset: 0x10000 + size: 33456 + align: 14 +Slices: + - !mach-o + FileHeader: + magic: 0xFEEDFACF + cputype: 0x1000007 + cpusubtype: 0x3 + filetype: 0x6 + ncmds: 20 + sizeofcmds: 1120 + flags: 0x100085 + reserved: 0x0 + LoadCommands: + - cmd: LC_SEGMENT_64 + cmdsize: 312 + segname: __TEXT + vmaddr: 0 + vmsize: 4096 + fileoff: 0 + filesize: 4096 + maxprot: 5 + initprot: 5 + nsects: 3 + flags: 0 + Sections: + - sectname: __text + segname: __TEXT + addr: 0xF70 + size: 27 + offset: 0xF70 + align: 4 + reloff: 0x0 + nreloc: 0 + flags: 0x80000400 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 554889E5B000E811000000B000E810000000B000E80F0000005DC3 + - sectname: __stubs + segname: __TEXT + addr: 0xF8C + size: 18 + offset: 0xF8C + align: 1 + reloff: 0x0 + nreloc: 0 + flags: 0x80000408 + reserved1: 0x0 + reserved2: 0x6 + reserved3: 0x0 + content: FF256E000000FF2570000000FF2572000000 + - sectname: __unwind_info + segname: __TEXT + addr: 0xFA0 + size: 88 + offset: 0xFA0 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x0 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 010000001C000000000000001C000000000000001C00000002000000700F000040000000400000008B0F00000000000040000000000000000000000000000000030000000C00010010000100000000000000000100000000 + - cmd: LC_SEGMENT_64 + cmdsize: 152 + segname: __DATA_CONST + vmaddr: 4096 + vmsize: 4096 + fileoff: 4096 + filesize: 4096 + maxprot: 3 + initprot: 3 + nsects: 1 + flags: 16 + Sections: + - sectname: __got + segname: __DATA_CONST + addr: 0x1000 + size: 24 + offset: 0x1000 + align: 3 + reloff: 0x0 + nreloc: 0 + flags: 0x6 + reserved1: 0x3 + reserved2: 0x0 + reserved3: 0x0 + content: '000000000000108001000000000010800200000000000080' + - cmd: LC_SEGMENT_64 + cmdsize: 72 + segname: __LINKEDIT + vmaddr: 8192 + vmsize: 4096 + fileoff: 8192 + filesize: 264 + maxprot: 1 + initprot: 1 + nsects: 0 + flags: 0 + - cmd: LC_ID_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 1 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libC.dylib' + ZeroPadBytes: 7 + - cmd: LC_DYLD_CHAINED_FIXUPS + cmdsize: 16 + dataoff: 8192 + datasize: 112 + - cmd: LC_DYLD_EXPORTS_TRIE + cmdsize: 16 + dataoff: 8304 + datasize: 24 + - cmd: LC_SYMTAB + cmdsize: 24 + symoff: 8336 + nsyms: 4 + stroff: 8424 + strsize: 32 + - cmd: LC_DYSYMTAB + cmdsize: 80 + ilocalsym: 0 + nlocalsym: 0 + iextdefsym: 0 + nextdefsym: 1 + iundefsym: 1 + nundefsym: 3 + tocoff: 0 + ntoc: 0 + modtaboff: 0 + nmodtab: 0 + extrefsymoff: 0 + nextrefsyms: 0 + indirectsymoff: 8400 + nindirectsyms: 6 + extreloff: 0 + nextrel: 0 + locreloff: 0 + nlocrel: 0 + - cmd: LC_UUID + cmdsize: 24 + uuid: 2AA1F9E9-F250-366F-B382-51A91DE06BED + - cmd: LC_BUILD_VERSION + cmdsize: 32 + platform: 1 + minos: 983040 + sdk: 983552 + ntools: 1 + Tools: + - tool: 3 + version: 73074435 + - cmd: LC_SOURCE_VERSION + cmdsize: 16 + version: 0 + - cmd: LC_LOAD_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 2 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libA.dylib' + ZeroPadBytes: 7 + - cmd: LC_LOAD_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 2 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libB.dylib' + ZeroPadBytes: 7 + - cmd: LC_LOAD_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 2 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libZ.dylib' + ZeroPadBytes: 7 + - cmd: LC_LOAD_DYLIB + cmdsize: 56 + dylib: + name: 24 + timestamp: 2 + current_version: 88539136 + compatibility_version: 65536 + Content: '/usr/lib/libSystem.B.dylib' + ZeroPadBytes: 6 + - cmd: LC_RPATH + cmdsize: 32 + path: 12 + Content: '@loader_path/../A' + ZeroPadBytes: 3 + - cmd: LC_RPATH + cmdsize: 32 + path: 12 + Content: '@loader_path/../B' + ZeroPadBytes: 3 + - cmd: LC_RPATH + cmdsize: 32 + path: 12 + Content: '@loader_path/../Z' + ZeroPadBytes: 3 + - cmd: LC_FUNCTION_STARTS + cmdsize: 16 + dataoff: 8328 + datasize: 8 + - cmd: LC_DATA_IN_CODE + cmdsize: 16 + dataoff: 8336 + datasize: 0 + LinkEditData: + ExportTrie: + TerminalSize: 0 + NodeOffset: 0 + Name: '' + Flags: 0x0 + Address: 0x0 + Other: 0x0 + ImportName: '' + Children: + - TerminalSize: 3 + NodeOffset: 13 + Name: _sayC + Flags: 0x0 + Address: 0xF70 + Other: 0x0 + ImportName: '' + NameList: + - n_strx: 2 + n_type: 0xF + n_sect: 1 + n_desc: 0 + n_value: 3952 + - n_strx: 8 + n_type: 0x1 + n_sect: 0 + n_desc: 256 + n_value: 0 + - n_strx: 14 + n_type: 0x1 + n_sect: 0 + n_desc: 512 + n_value: 0 + - n_strx: 20 + n_type: 0x1 + n_sect: 0 + n_desc: 768 + n_value: 0 + StringTable: + - ' ' + - _sayC + - _sayA + - _sayB + - _sayZ + - '' + - '' + - '' + - '' + - '' + - '' + IndirectSymbols: [ 0x1, 0x2, 0x3, 0x1, 0x2, 0x3 ] + FunctionStarts: [ 0xF70 ] + ChainedFixups: [ 0x0, 0x0, 0x0, 0x0, 0x20, 0x0, 0x0, 0x0, 0x48, + 0x0, 0x0, 0x0, 0x58, 0x0, 0x0, 0x0, 0x3, 0x0, + 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x3, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x18, 0x0, 0x0, 0x0, 0x0, 0x10, 0x6, 0x0, + 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x1, 0x2, 0x0, + 0x0, 0x2, 0xE, 0x0, 0x0, 0x3, 0x1A, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x0, 0x5F, 0x73, 0x61, 0x79, + 0x41, 0x0, 0x5F, 0x73, 0x61, 0x79, 0x42, 0x0, + 0x5F, 0x73, 0x61, 0x79, 0x5A, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0 ] + - !mach-o + FileHeader: + magic: 0xFEEDFACF + cputype: 0x100000C + cpusubtype: 0x0 + filetype: 0x6 + ncmds: 21 + sizeofcmds: 1136 + flags: 0x100085 + reserved: 0x0 + LoadCommands: + - cmd: LC_SEGMENT_64 + cmdsize: 312 + segname: __TEXT + vmaddr: 0 + vmsize: 16384 + fileoff: 0 + filesize: 16384 + maxprot: 5 + initprot: 5 + nsects: 3 + flags: 0 + Sections: + - sectname: __text + segname: __TEXT + addr: 0x3F68 + size: 28 + offset: 0x3F68 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000400 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: FD7BBFA9FD030091050000940700009409000094FD7BC1A8C0035FD6 + - sectname: __stubs + segname: __TEXT + addr: 0x3F84 + size: 36 + offset: 0x3F84 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000408 + reserved1: 0x0 + reserved2: 0xC + reserved3: 0x0 + content: 100000B0100240F900021FD6100000B0100640F900021FD6100000B0100A40F900021FD6 + - sectname: __unwind_info + segname: __TEXT + addr: 0x3FA8 + size: 88 + offset: 0x3FA8 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x0 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 010000001C000000000000001C000000000000001C00000002000000683F00004000000040000000843F00000000000040000000000000000000000000000000030000000C00010010000100000000000000000400000000 + - cmd: LC_SEGMENT_64 + cmdsize: 152 + segname: __DATA_CONST + vmaddr: 16384 + vmsize: 16384 + fileoff: 16384 + filesize: 16384 + maxprot: 3 + initprot: 3 + nsects: 1 + flags: 16 + Sections: + - sectname: __got + segname: __DATA_CONST + addr: 0x4000 + size: 24 + offset: 0x4000 + align: 3 + reloff: 0x0 + nreloc: 0 + flags: 0x6 + reserved1: 0x3 + reserved2: 0x0 + reserved3: 0x0 + content: '000000000000108001000000000010800200000000000080' + - cmd: LC_SEGMENT_64 + cmdsize: 72 + segname: __LINKEDIT + vmaddr: 32768 + vmsize: 16384 + fileoff: 32768 + filesize: 688 + maxprot: 1 + initprot: 1 + nsects: 0 + flags: 0 + - cmd: LC_ID_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 1 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libC.dylib' + ZeroPadBytes: 7 + - cmd: LC_DYLD_CHAINED_FIXUPS + cmdsize: 16 + dataoff: 32768 + datasize: 112 + - cmd: LC_DYLD_EXPORTS_TRIE + cmdsize: 16 + dataoff: 32880 + datasize: 24 + - cmd: LC_SYMTAB + cmdsize: 24 + symoff: 32912 + nsyms: 4 + stroff: 33000 + strsize: 32 + - cmd: LC_DYSYMTAB + cmdsize: 80 + ilocalsym: 0 + nlocalsym: 0 + iextdefsym: 0 + nextdefsym: 1 + iundefsym: 1 + nundefsym: 3 + tocoff: 0 + ntoc: 0 + modtaboff: 0 + nmodtab: 0 + extrefsymoff: 0 + nextrefsyms: 0 + indirectsymoff: 32976 + nindirectsyms: 6 + extreloff: 0 + nextrel: 0 + locreloff: 0 + nlocrel: 0 + - cmd: LC_UUID + cmdsize: 24 + uuid: 02B69690-925D-35EE-A8AB-6D99813D2A16 + - cmd: LC_BUILD_VERSION + cmdsize: 32 + platform: 1 + minos: 983040 + sdk: 983552 + ntools: 1 + Tools: + - tool: 3 + version: 73074435 + - cmd: LC_SOURCE_VERSION + cmdsize: 16 + version: 0 + - cmd: LC_LOAD_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 2 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libA.dylib' + ZeroPadBytes: 7 + - cmd: LC_LOAD_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 2 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libB.dylib' + ZeroPadBytes: 7 + - cmd: LC_LOAD_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 2 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libZ.dylib' + ZeroPadBytes: 7 + - cmd: LC_LOAD_DYLIB + cmdsize: 56 + dylib: + name: 24 + timestamp: 2 + current_version: 88539136 + compatibility_version: 65536 + Content: '/usr/lib/libSystem.B.dylib' + ZeroPadBytes: 6 + - cmd: LC_RPATH + cmdsize: 32 + path: 12 + Content: '@loader_path/../A' + ZeroPadBytes: 3 + - cmd: LC_RPATH + cmdsize: 32 + path: 12 + Content: '@loader_path/../B' + ZeroPadBytes: 3 + - cmd: LC_RPATH + cmdsize: 32 + path: 12 + Content: '@loader_path/../Z' + ZeroPadBytes: 3 + - cmd: LC_FUNCTION_STARTS + cmdsize: 16 + dataoff: 32904 + datasize: 8 + - cmd: LC_DATA_IN_CODE + cmdsize: 16 + dataoff: 32912 + datasize: 0 + - cmd: LC_CODE_SIGNATURE + cmdsize: 16 + dataoff: 33040 + datasize: 416 + LinkEditData: + ExportTrie: + TerminalSize: 0 + NodeOffset: 0 + Name: '' + Flags: 0x0 + Address: 0x0 + Other: 0x0 + ImportName: '' + Children: + - TerminalSize: 3 + NodeOffset: 13 + Name: _sayC + Flags: 0x0 + Address: 0x3F68 + Other: 0x0 + ImportName: '' + NameList: + - n_strx: 2 + n_type: 0xF + n_sect: 1 + n_desc: 0 + n_value: 16232 + - n_strx: 8 + n_type: 0x1 + n_sect: 0 + n_desc: 256 + n_value: 0 + - n_strx: 14 + n_type: 0x1 + n_sect: 0 + n_desc: 512 + n_value: 0 + - n_strx: 20 + n_type: 0x1 + n_sect: 0 + n_desc: 768 + n_value: 0 + StringTable: + - ' ' + - _sayC + - _sayA + - _sayB + - _sayZ + - '' + - '' + - '' + - '' + - '' + - '' + IndirectSymbols: [ 0x1, 0x2, 0x3, 0x1, 0x2, 0x3 ] + FunctionStarts: [ 0x3F68 ] + ChainedFixups: [ 0x0, 0x0, 0x0, 0x0, 0x20, 0x0, 0x0, 0x0, 0x48, + 0x0, 0x0, 0x0, 0x58, 0x0, 0x0, 0x0, 0x3, 0x0, + 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x3, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x18, 0x0, 0x0, 0x0, 0x0, 0x40, 0x6, 0x0, + 0x0, 0x40, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x1, 0x2, 0x0, + 0x0, 0x2, 0xE, 0x0, 0x0, 0x3, 0x1A, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x0, 0x5F, 0x73, 0x61, 0x79, + 0x41, 0x0, 0x5F, 0x73, 0x61, 0x79, 0x42, 0x0, + 0x5F, 0x73, 0x61, 0x79, 0x5A, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0 ] + - !mach-o + FileHeader: + magic: 0xFEEDFACF + cputype: 0x100000C + cpusubtype: 0x80000002 + filetype: 0x6 + ncmds: 21 + sizeofcmds: 1136 + flags: 0x100085 + reserved: 0x0 + LoadCommands: + - cmd: LC_SEGMENT_64 + cmdsize: 312 + segname: __TEXT + vmaddr: 0 + vmsize: 16384 + fileoff: 0 + filesize: 16384 + maxprot: 5 + initprot: 5 + nsects: 3 + flags: 0 + Sections: + - sectname: __text + segname: __TEXT + addr: 0x3F58 + size: 32 + offset: 0x3F58 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000400 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 7F2303D5FD7BBFA9FD03009105000094080000940B000094FD7BC1A8FF0F5FD6 + - sectname: __auth_stubs + segname: __TEXT + addr: 0x3F78 + size: 48 + offset: 0x3F78 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000408 + reserved1: 0x0 + reserved2: 0x10 + reserved3: 0x0 + content: 110000B031020091300240F9110A1FD7110000B031220091300240F9110A1FD7110000B031420091300240F9110A1FD7 + - sectname: __unwind_info + segname: __TEXT + addr: 0x3FA8 + size: 88 + offset: 0x3FA8 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x0 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 010000001C000000000000001C000000000000001C00000002000000583F00004000000040000000783F00000000000040000000000000000000000000000000030000000C00010010000100000000000000000400000000 + - cmd: LC_SEGMENT_64 + cmdsize: 152 + segname: __DATA_CONST + vmaddr: 16384 + vmsize: 16384 + fileoff: 16384 + filesize: 16384 + maxprot: 3 + initprot: 3 + nsects: 1 + flags: 16 + Sections: + - sectname: __auth_got + segname: __DATA_CONST + addr: 0x4000 + size: 24 + offset: 0x4000 + align: 3 + reloff: 0x0 + nreloc: 0 + flags: 0x6 + reserved1: 0x3 + reserved2: 0x0 + reserved3: 0x0 + content: 00000000000009C001000000000009C002000000000001C0 + - cmd: LC_SEGMENT_64 + cmdsize: 72 + segname: __LINKEDIT + vmaddr: 32768 + vmsize: 16384 + fileoff: 32768 + filesize: 688 + maxprot: 1 + initprot: 1 + nsects: 0 + flags: 0 + - cmd: LC_ID_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 1 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libC.dylib' + ZeroPadBytes: 7 + - cmd: LC_DYLD_CHAINED_FIXUPS + cmdsize: 16 + dataoff: 32768 + datasize: 112 + - cmd: LC_DYLD_EXPORTS_TRIE + cmdsize: 16 + dataoff: 32880 + datasize: 24 + - cmd: LC_SYMTAB + cmdsize: 24 + symoff: 32912 + nsyms: 4 + stroff: 33000 + strsize: 32 + - cmd: LC_DYSYMTAB + cmdsize: 80 + ilocalsym: 0 + nlocalsym: 0 + iextdefsym: 0 + nextdefsym: 1 + iundefsym: 1 + nundefsym: 3 + tocoff: 0 + ntoc: 0 + modtaboff: 0 + nmodtab: 0 + extrefsymoff: 0 + nextrefsyms: 0 + indirectsymoff: 32976 + nindirectsyms: 6 + extreloff: 0 + nextrel: 0 + locreloff: 0 + nlocrel: 0 + - cmd: LC_UUID + cmdsize: 24 + uuid: F54076AA-8888-3DED-8BDF-BC7FB3E6FE8A + - cmd: LC_BUILD_VERSION + cmdsize: 32 + platform: 1 + minos: 983040 + sdk: 983552 + ntools: 1 + Tools: + - tool: 3 + version: 73074435 + - cmd: LC_SOURCE_VERSION + cmdsize: 16 + version: 0 + - cmd: LC_LOAD_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 2 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libA.dylib' + ZeroPadBytes: 7 + - cmd: LC_LOAD_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 2 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libB.dylib' + ZeroPadBytes: 7 + - cmd: LC_LOAD_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 2 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libZ.dylib' + ZeroPadBytes: 7 + - cmd: LC_LOAD_DYLIB + cmdsize: 56 + dylib: + name: 24 + timestamp: 2 + current_version: 88539136 + compatibility_version: 65536 + Content: '/usr/lib/libSystem.B.dylib' + ZeroPadBytes: 6 + - cmd: LC_RPATH + cmdsize: 32 + path: 12 + Content: '@loader_path/../A' + ZeroPadBytes: 3 + - cmd: LC_RPATH + cmdsize: 32 + path: 12 + Content: '@loader_path/../B' + ZeroPadBytes: 3 + - cmd: LC_RPATH + cmdsize: 32 + path: 12 + Content: '@loader_path/../Z' + ZeroPadBytes: 3 + - cmd: LC_FUNCTION_STARTS + cmdsize: 16 + dataoff: 32904 + datasize: 8 + - cmd: LC_DATA_IN_CODE + cmdsize: 16 + dataoff: 32912 + datasize: 0 + - cmd: LC_CODE_SIGNATURE + cmdsize: 16 + dataoff: 33040 + datasize: 416 + LinkEditData: + ExportTrie: + TerminalSize: 0 + NodeOffset: 0 + Name: '' + Flags: 0x0 + Address: 0x0 + Other: 0x0 + ImportName: '' + Children: + - TerminalSize: 3 + NodeOffset: 13 + Name: _sayC + Flags: 0x0 + Address: 0x3F58 + Other: 0x0 + ImportName: '' + NameList: + - n_strx: 2 + n_type: 0xF + n_sect: 1 + n_desc: 0 + n_value: 16216 + - n_strx: 8 + n_type: 0x1 + n_sect: 0 + n_desc: 256 + n_value: 0 + - n_strx: 14 + n_type: 0x1 + n_sect: 0 + n_desc: 512 + n_value: 0 + - n_strx: 20 + n_type: 0x1 + n_sect: 0 + n_desc: 768 + n_value: 0 + StringTable: + - ' ' + - _sayC + - _sayA + - _sayB + - _sayZ + - '' + - '' + - '' + - '' + - '' + - '' + IndirectSymbols: [ 0x1, 0x2, 0x3, 0x1, 0x2, 0x3 ] + FunctionStarts: [ 0x3F58 ] + ChainedFixups: [ 0x0, 0x0, 0x0, 0x0, 0x20, 0x0, 0x0, 0x0, 0x48, + 0x0, 0x0, 0x0, 0x58, 0x0, 0x0, 0x0, 0x3, 0x0, + 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x3, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x18, 0x0, 0x0, 0x0, 0x0, 0x40, 0xC, 0x0, + 0x0, 0x40, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x1, 0x2, 0x0, + 0x0, 0x2, 0xE, 0x0, 0x0, 0x3, 0x1A, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x0, 0x5F, 0x73, 0x61, 0x79, + 0x41, 0x0, 0x5F, 0x73, 0x61, 0x79, 0x42, 0x0, + 0x5F, 0x73, 0x61, 0x79, 0x5A, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0 ] +... diff --git a/llvm/unittests/ExecutionEngine/Orc/Inputs/Z/Z_linux.yaml b/llvm/unittests/ExecutionEngine/Orc/Inputs/Z/Z_linux.yaml new file mode 100644 index 0000000..5561f29 --- /dev/null +++ b/llvm/unittests/ExecutionEngine/Orc/Inputs/Z/Z_linux.yaml @@ -0,0 +1,460 @@ +--- !ELF +FileHeader: + Class: ELFCLASS64 + Data: ELFDATA2LSB + Type: ET_DYN + Machine: EM_X86_64 +ProgramHeaders: + - Type: PT_LOAD + Flags: [ PF_R ] + FirstSec: .note.gnu.property + LastSec: .rela.plt + Align: 0x1000 + Offset: 0x0 + - Type: PT_LOAD + Flags: [ PF_X, PF_R ] + FirstSec: .init + LastSec: .fini + VAddr: 0x1000 + Align: 0x1000 + Offset: 0x1000 + - Type: PT_LOAD + Flags: [ PF_R ] + FirstSec: .rodata + LastSec: .eh_frame + VAddr: 0x2000 + Align: 0x1000 + Offset: 0x2000 + - Type: PT_LOAD + Flags: [ PF_W, PF_R ] + FirstSec: .init_array + LastSec: .bss + VAddr: 0x3E10 + Align: 0x1000 + Offset: 0x2E10 + - Type: PT_DYNAMIC + Flags: [ PF_W, PF_R ] + FirstSec: .dynamic + LastSec: .dynamic + VAddr: 0x3E20 + Align: 0x8 + Offset: 0x2E20 + - Type: PT_NOTE + Flags: [ PF_R ] + FirstSec: .note.gnu.property + LastSec: .note.gnu.property + VAddr: 0x2A8 + Align: 0x8 + Offset: 0x2A8 + - Type: PT_NOTE + Flags: [ PF_R ] + FirstSec: .note.gnu.build-id + LastSec: .note.gnu.build-id + VAddr: 0x2C8 + Align: 0x4 + Offset: 0x2C8 + - Type: PT_GNU_PROPERTY + Flags: [ PF_R ] + FirstSec: .note.gnu.property + LastSec: .note.gnu.property + VAddr: 0x2A8 + Align: 0x8 + Offset: 0x2A8 + - Type: PT_GNU_EH_FRAME + Flags: [ PF_R ] + FirstSec: .eh_frame_hdr + LastSec: .eh_frame_hdr + VAddr: 0x2010 + Align: 0x4 + Offset: 0x2010 + - Type: PT_GNU_STACK + Flags: [ PF_W, PF_R ] + Align: 0x10 + Offset: 0x0 + - Type: PT_GNU_RELRO + Flags: [ PF_R ] + FirstSec: .init_array + LastSec: .got + VAddr: 0x3E10 + Offset: 0x2E10 +Sections: + - Name: .note.gnu.property + Type: SHT_NOTE + Flags: [ SHF_ALLOC ] + Address: 0x2A8 + AddressAlign: 0x8 + Notes: + - Name: GNU + Desc: 020000C0040000000300000000000000 + Type: NT_GNU_PROPERTY_TYPE_0 + - Name: .note.gnu.build-id + Type: SHT_NOTE + Flags: [ SHF_ALLOC ] + Address: 0x2C8 + AddressAlign: 0x4 + Notes: + - Name: GNU + Desc: 640A4A3AC0DF6BA3DAC3B51CCD727245117E0B30 + Type: NT_PRPSINFO + - Name: .gnu.hash + Type: SHT_GNU_HASH + Flags: [ SHF_ALLOC ] + Address: 0x2F0 + Link: .dynsym + AddressAlign: 0x8 + Header: + SymNdx: 0x6 + Shift2: 0x6 + BloomFilter: [ 0x500000000000 ] + HashBuckets: [ 0x6, 0x0 ] + HashValues: [ 0x7C9DCBAD ] + - Name: .dynsym + Type: SHT_DYNSYM + Flags: [ SHF_ALLOC ] + Address: 0x318 + Link: .dynstr + AddressAlign: 0x8 + - Name: .dynstr + Type: SHT_STRTAB + Flags: [ SHF_ALLOC ] + Address: 0x3C0 + AddressAlign: 0x1 + - Name: .gnu.version + Type: SHT_GNU_versym + Flags: [ SHF_ALLOC ] + Address: 0x436 + Link: .dynsym + AddressAlign: 0x2 + Entries: [ 0, 1, 2, 1, 1, 2, 1 ] + - Name: .gnu.version_r + Type: SHT_GNU_verneed + Flags: [ SHF_ALLOC ] + Address: 0x448 + Link: .dynstr + AddressAlign: 0x8 + Dependencies: + - Version: 1 + File: libc.so.6 + Entries: + - Name: GLIBC_2.2.5 + Hash: 157882997 + Flags: 0 + Other: 2 + - Name: .rela.dyn + Type: SHT_RELA + Flags: [ SHF_ALLOC ] + Address: 0x468 + Link: .dynsym + AddressAlign: 0x8 + Relocations: + - Offset: 0x3E10 + Type: R_X86_64_RELATIVE + Addend: 4368 + - Offset: 0x3E18 + Type: R_X86_64_RELATIVE + Addend: 4304 + - Offset: 0x4020 + Type: R_X86_64_RELATIVE + Addend: 16416 + - Offset: 0x3FE0 + Symbol: _ITM_deregisterTMCloneTable + Type: R_X86_64_GLOB_DAT + - Offset: 0x3FE8 + Symbol: __gmon_start__ + Type: R_X86_64_GLOB_DAT + - Offset: 0x3FF0 + Symbol: _ITM_registerTMCloneTable + Type: R_X86_64_GLOB_DAT + - Offset: 0x3FF8 + Symbol: __cxa_finalize + Type: R_X86_64_GLOB_DAT + - Name: .rela.plt + Type: SHT_RELA + Flags: [ SHF_ALLOC, SHF_INFO_LINK ] + Address: 0x510 + Link: .dynsym + AddressAlign: 0x8 + Info: .got.plt + Relocations: + - Offset: 0x4018 + Symbol: puts + Type: R_X86_64_JUMP_SLOT + - Name: .init + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1000 + AddressAlign: 0x4 + Offset: 0x1000 + Content: F30F1EFA4883EC08488B05D92F00004885C07402FFD04883C408C3 + - Name: .plt + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1020 + AddressAlign: 0x10 + EntSize: 0x10 + Content: FF35E22F0000F2FF25E32F00000F1F00F30F1EFA6800000000F2E9E1FFFFFF90 + - Name: .plt.got + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1040 + AddressAlign: 0x10 + EntSize: 0x10 + Content: F30F1EFAF2FF25AD2F00000F1F440000 + - Name: .plt.sec + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1050 + AddressAlign: 0x10 + EntSize: 0x10 + Content: F30F1EFAF2FF25BD2F00000F1F440000 + - Name: .text + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1060 + AddressAlign: 0x10 + Content: 488D3DC12F0000488D05BA2F00004839F87415488B05662F00004885C07409FFE00F1F8000000000C30F1F8000000000488D3D912F0000488D358A2F00004829FE4889F048C1EE3F48C1F8034801C648D1FE7414488B05352F00004885C07408FFE0660F1F440000C30F1F8000000000F30F1EFA803D4D2F000000752B5548833D122F0000004889E5740C488B3D2E2F0000E849FFFFFFE864FFFFFFC605252F0000015DC30F1F00C30F1F8000000000F30F1EFAE977FFFFFFF30F1EFA554889E5488D05D80E00004889C7E820FFFFFF905DC3 + - Name: .fini + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC, SHF_EXECINSTR ] + Address: 0x1134 + AddressAlign: 0x4 + Content: F30F1EFA4883EC084883C408C3 + - Name: .rodata + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC ] + Address: 0x2000 + AddressAlign: 0x1 + Offset: 0x2000 + Content: 48656C6C6F2066726F6D205A00 + - Name: .eh_frame_hdr + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC ] + Address: 0x2010 + AddressAlign: 0x4 + Content: 011B033B2C0000000400000010F0FFFF4800000030F0FFFF7000000040F0FFFF8800000009F1FFFFA0000000 + - Name: .eh_frame + Type: SHT_PROGBITS + Flags: [ SHF_ALLOC ] + Address: 0x2040 + AddressAlign: 0x8 + Content: 1400000000000000017A5200017810011B0C070890010000240000001C000000C0EFFFFF20000000000E10460E184A0F0B770880003F1A3A2A332422000000001400000044000000B8EFFFFF100000000000000000000000140000005C000000B0EFFFFF1000000000000000000000001C0000007400000061F0FFFF1A00000000450E108602430D06510C070800000000000000 + - Name: .init_array + Type: SHT_INIT_ARRAY + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3E10 + AddressAlign: 0x8 + EntSize: 0x8 + Offset: 0x2E10 + Content: '1011000000000000' + - Name: .fini_array + Type: SHT_FINI_ARRAY + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3E18 + AddressAlign: 0x8 + EntSize: 0x8 + Content: D010000000000000 + - Name: .dynamic + Type: SHT_DYNAMIC + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3E20 + Link: .dynstr + AddressAlign: 0x8 + Entries: + - Tag: DT_NEEDED + Value: 0x5F + - Tag: DT_INIT + Value: 0x1000 + - Tag: DT_FINI + Value: 0x1134 + - Tag: DT_INIT_ARRAY + Value: 0x3E10 + - Tag: DT_INIT_ARRAYSZ + Value: 0x8 + - Tag: DT_FINI_ARRAY + Value: 0x3E18 + - Tag: DT_FINI_ARRAYSZ + Value: 0x8 + - Tag: DT_GNU_HASH + Value: 0x2F0 + - Tag: DT_STRTAB + Value: 0x3C0 + - Tag: DT_SYMTAB + Value: 0x318 + - Tag: DT_STRSZ + Value: 0x75 + - Tag: DT_SYMENT + Value: 0x18 + - Tag: DT_PLTGOT + Value: 0x4000 + - Tag: DT_PLTRELSZ + Value: 0x18 + - Tag: DT_PLTREL + Value: 0x7 + - Tag: DT_JMPREL + Value: 0x510 + - Tag: DT_RELA + Value: 0x468 + - Tag: DT_RELASZ + Value: 0xA8 + - Tag: DT_RELAENT + Value: 0x18 + - Tag: DT_VERNEED + Value: 0x448 + - Tag: DT_VERNEEDNUM + Value: 0x1 + - Tag: DT_VERSYM + Value: 0x436 + - Tag: DT_RELACOUNT + Value: 0x3 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Tag: DT_NULL + Value: 0x0 + - Name: .got + Type: SHT_PROGBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x3FE0 + AddressAlign: 0x8 + EntSize: 0x8 + Content: '0000000000000000000000000000000000000000000000000000000000000000' + - Name: .got.plt + Type: SHT_PROGBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x4000 + AddressAlign: 0x8 + EntSize: 0x8 + Content: '203E000000000000000000000000000000000000000000003010000000000000' + - Name: .data + Type: SHT_PROGBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x4020 + AddressAlign: 0x8 + Content: '2040000000000000' + - Name: .bss + Type: SHT_NOBITS + Flags: [ SHF_WRITE, SHF_ALLOC ] + Address: 0x4028 + AddressAlign: 0x1 + Size: 0x8 + - Name: .comment + Type: SHT_PROGBITS + Flags: [ SHF_MERGE, SHF_STRINGS ] + AddressAlign: 0x1 + EntSize: 0x1 + Content: 4743433A20285562756E74752031312E342E302D317562756E7475317E32322E30342E32292031312E342E3000 +Symbols: + - Name: crtstuff.c + Type: STT_FILE + Index: SHN_ABS + - Name: deregister_tm_clones + Type: STT_FUNC + Section: .text + Value: 0x1060 + - Name: register_tm_clones + Type: STT_FUNC + Section: .text + Value: 0x1090 + - Name: __do_global_dtors_aux + Type: STT_FUNC + Section: .text + Value: 0x10D0 + - Name: completed.0 + Type: STT_OBJECT + Section: .bss + Value: 0x4028 + Size: 0x1 + - Name: __do_global_dtors_aux_fini_array_entry + Type: STT_OBJECT + Section: .fini_array + Value: 0x3E18 + - Name: frame_dummy + Type: STT_FUNC + Section: .text + Value: 0x1110 + - Name: __frame_dummy_init_array_entry + Type: STT_OBJECT + Section: .init_array + Value: 0x3E10 + - Name: libZ.c + Type: STT_FILE + Index: SHN_ABS + - Name: 'crtstuff.c (1)' + Type: STT_FILE + Index: SHN_ABS + - Name: __FRAME_END__ + Type: STT_OBJECT + Section: .eh_frame + Value: 0x20D0 + - Type: STT_FILE + Index: SHN_ABS + - Name: _fini + Type: STT_FUNC + Section: .fini + Value: 0x1134 + - Name: __dso_handle + Type: STT_OBJECT + Section: .data + Value: 0x4020 + - Name: _DYNAMIC + Type: STT_OBJECT + Section: .dynamic + Value: 0x3E20 + - Name: __GNU_EH_FRAME_HDR + Section: .eh_frame_hdr + Value: 0x2010 + - Name: __TMC_END__ + Type: STT_OBJECT + Section: .data + Value: 0x4028 + - Name: _GLOBAL_OFFSET_TABLE_ + Type: STT_OBJECT + Section: .got.plt + Value: 0x4000 + - Name: _init + Type: STT_FUNC + Section: .init + Value: 0x1000 + - Name: _ITM_deregisterTMCloneTable + Binding: STB_WEAK + - Name: 'puts@GLIBC_2.2.5' + Type: STT_FUNC + Binding: STB_GLOBAL + - Name: __gmon_start__ + Binding: STB_WEAK + - Name: sayZ + Type: STT_FUNC + Section: .text + Binding: STB_GLOBAL + Value: 0x1119 + Size: 0x1A + - Name: _ITM_registerTMCloneTable + Binding: STB_WEAK + - Name: '__cxa_finalize@GLIBC_2.2.5' + Type: STT_FUNC + Binding: STB_WEAK +DynamicSymbols: + - Name: _ITM_deregisterTMCloneTable + Binding: STB_WEAK + - Name: puts + Type: STT_FUNC + Binding: STB_GLOBAL + - Name: __gmon_start__ + Binding: STB_WEAK + - Name: _ITM_registerTMCloneTable + Binding: STB_WEAK + - Name: __cxa_finalize + Type: STT_FUNC + Binding: STB_WEAK + - Name: sayZ + Type: STT_FUNC + Section: .text + Binding: STB_GLOBAL + Value: 0x1119 + Size: 0x1A +... diff --git a/llvm/unittests/ExecutionEngine/Orc/Inputs/Z/Z_macho.yaml b/llvm/unittests/ExecutionEngine/Orc/Inputs/Z/Z_macho.yaml new file mode 100644 index 0000000..c0c1826 --- /dev/null +++ b/llvm/unittests/ExecutionEngine/Orc/Inputs/Z/Z_macho.yaml @@ -0,0 +1,723 @@ +--- !fat-mach-o +FatHeader: + magic: 0xCAFEBABE + nfat_arch: 3 +FatArchs: + - cputype: 0x1000007 + cpusubtype: 0x3 + offset: 0x1000 + size: 8376 + align: 12 + - cputype: 0x100000C + cpusubtype: 0x0 + offset: 0x4000 + size: 33376 + align: 14 + - cputype: 0x100000C + cpusubtype: 0x80000002 + offset: 0x10000 + size: 33376 + align: 14 +Slices: + - !mach-o + FileHeader: + magic: 0xFEEDFACF + cputype: 0x1000007 + cpusubtype: 0x3 + filetype: 0x6 + ncmds: 14 + sizeofcmds: 960 + flags: 0x100085 + reserved: 0x0 + LoadCommands: + - cmd: LC_SEGMENT_64 + cmdsize: 392 + segname: __TEXT + vmaddr: 0 + vmsize: 4096 + fileoff: 0 + filesize: 4096 + maxprot: 5 + initprot: 5 + nsects: 4 + flags: 0 + Sections: + - sectname: __text + segname: __TEXT + addr: 0xF80 + size: 20 + offset: 0xF80 + align: 4 + reloff: 0x0 + nreloc: 0 + flags: 0x80000400 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 554889E5488D3D0F000000B000E8020000005DC3 + - sectname: __stubs + segname: __TEXT + addr: 0xF94 + size: 6 + offset: 0xF94 + align: 1 + reloff: 0x0 + nreloc: 0 + flags: 0x80000408 + reserved1: 0x0 + reserved2: 0x6 + reserved3: 0x0 + content: FF2566000000 + - sectname: __cstring + segname: __TEXT + addr: 0xF9A + size: 14 + offset: 0xF9A + align: 0 + reloff: 0x0 + nreloc: 0 + flags: 0x2 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 48656C6C6F2066726F6D205A0A00 + - sectname: __unwind_info + segname: __TEXT + addr: 0xFA8 + size: 88 + offset: 0xFA8 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x0 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 010000001C000000000000001C000000000000001C00000002000000800F00004000000040000000940F00000000000040000000000000000000000000000000030000000C00010010000100000000000000000100000000 + - cmd: LC_SEGMENT_64 + cmdsize: 152 + segname: __DATA_CONST + vmaddr: 4096 + vmsize: 4096 + fileoff: 4096 + filesize: 4096 + maxprot: 3 + initprot: 3 + nsects: 1 + flags: 16 + Sections: + - sectname: __got + segname: __DATA_CONST + addr: 0x1000 + size: 8 + offset: 0x1000 + align: 3 + reloff: 0x0 + nreloc: 0 + flags: 0x6 + reserved1: 0x1 + reserved2: 0x0 + reserved3: 0x0 + content: '0000000000000080' + - cmd: LC_SEGMENT_64 + cmdsize: 72 + segname: __LINKEDIT + vmaddr: 8192 + vmsize: 4096 + fileoff: 8192 + filesize: 184 + maxprot: 1 + initprot: 1 + nsects: 0 + flags: 0 + - cmd: LC_ID_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 1 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libZ.dylib' + ZeroPadBytes: 7 + - cmd: LC_DYLD_CHAINED_FIXUPS + cmdsize: 16 + dataoff: 8192 + datasize: 96 + - cmd: LC_DYLD_EXPORTS_TRIE + cmdsize: 16 + dataoff: 8288 + datasize: 24 + - cmd: LC_SYMTAB + cmdsize: 24 + symoff: 8320 + nsyms: 2 + stroff: 8360 + strsize: 16 + - cmd: LC_DYSYMTAB + cmdsize: 80 + ilocalsym: 0 + nlocalsym: 0 + iextdefsym: 0 + nextdefsym: 1 + iundefsym: 1 + nundefsym: 1 + tocoff: 0 + ntoc: 0 + modtaboff: 0 + nmodtab: 0 + extrefsymoff: 0 + nextrefsyms: 0 + indirectsymoff: 8352 + nindirectsyms: 2 + extreloff: 0 + nextrel: 0 + locreloff: 0 + nlocrel: 0 + - cmd: LC_UUID + cmdsize: 24 + uuid: 399E203C-FF9A-3B80-872C-85F3A759A78B + - cmd: LC_BUILD_VERSION + cmdsize: 32 + platform: 1 + minos: 983040 + sdk: 983552 + ntools: 1 + Tools: + - tool: 3 + version: 73074435 + - cmd: LC_SOURCE_VERSION + cmdsize: 16 + version: 0 + - cmd: LC_LOAD_DYLIB + cmdsize: 56 + dylib: + name: 24 + timestamp: 2 + current_version: 88539136 + compatibility_version: 65536 + Content: '/usr/lib/libSystem.B.dylib' + ZeroPadBytes: 6 + - cmd: LC_FUNCTION_STARTS + cmdsize: 16 + dataoff: 8312 + datasize: 8 + - cmd: LC_DATA_IN_CODE + cmdsize: 16 + dataoff: 8320 + datasize: 0 + LinkEditData: + ExportTrie: + TerminalSize: 0 + NodeOffset: 0 + Name: '' + Flags: 0x0 + Address: 0x0 + Other: 0x0 + ImportName: '' + Children: + - TerminalSize: 3 + NodeOffset: 13 + Name: _sayZ + Flags: 0x0 + Address: 0xF80 + Other: 0x0 + ImportName: '' + NameList: + - n_strx: 2 + n_type: 0xF + n_sect: 1 + n_desc: 0 + n_value: 3968 + - n_strx: 8 + n_type: 0x1 + n_sect: 0 + n_desc: 256 + n_value: 0 + StringTable: + - ' ' + - _sayZ + - _printf + IndirectSymbols: [ 0x1, 0x1 ] + FunctionStarts: [ 0xF80 ] + ChainedFixups: [ 0x0, 0x0, 0x0, 0x0, 0x20, 0x0, 0x0, 0x0, 0x48, + 0x0, 0x0, 0x0, 0x50, 0x0, 0x0, 0x0, 0x1, 0x0, + 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x3, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x18, 0x0, 0x0, 0x0, 0x0, 0x10, 0x6, 0x0, + 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x1, 0x2, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x5F, 0x70, 0x72, + 0x69, 0x6E, 0x74, 0x66, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0 ] + - !mach-o + FileHeader: + magic: 0xFEEDFACF + cputype: 0x100000C + cpusubtype: 0x0 + filetype: 0x6 + ncmds: 15 + sizeofcmds: 976 + flags: 0x100085 + reserved: 0x0 + LoadCommands: + - cmd: LC_SEGMENT_64 + cmdsize: 392 + segname: __TEXT + vmaddr: 0 + vmsize: 16384 + fileoff: 0 + filesize: 16384 + maxprot: 5 + initprot: 5 + nsects: 4 + flags: 0 + Sections: + - sectname: __text + segname: __TEXT + addr: 0x3F70 + size: 28 + offset: 0x3F70 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000400 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: FD7BBFA9FD0300910000009000603E9103000094FD7BC1A8C0035FD6 + - sectname: __stubs + segname: __TEXT + addr: 0x3F8C + size: 12 + offset: 0x3F8C + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000408 + reserved1: 0x0 + reserved2: 0xC + reserved3: 0x0 + content: 100000B0100240F900021FD6 + - sectname: __cstring + segname: __TEXT + addr: 0x3F98 + size: 14 + offset: 0x3F98 + align: 0 + reloff: 0x0 + nreloc: 0 + flags: 0x2 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 48656C6C6F2066726F6D205A0A00 + - sectname: __unwind_info + segname: __TEXT + addr: 0x3FA8 + size: 88 + offset: 0x3FA8 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x0 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 010000001C000000000000001C000000000000001C00000002000000703F000040000000400000008C3F00000000000040000000000000000000000000000000030000000C00010010000100000000000000000400000000 + - cmd: LC_SEGMENT_64 + cmdsize: 152 + segname: __DATA_CONST + vmaddr: 16384 + vmsize: 16384 + fileoff: 16384 + filesize: 16384 + maxprot: 3 + initprot: 3 + nsects: 1 + flags: 16 + Sections: + - sectname: __got + segname: __DATA_CONST + addr: 0x4000 + size: 8 + offset: 0x4000 + align: 3 + reloff: 0x0 + nreloc: 0 + flags: 0x6 + reserved1: 0x1 + reserved2: 0x0 + reserved3: 0x0 + content: '0000000000000080' + - cmd: LC_SEGMENT_64 + cmdsize: 72 + segname: __LINKEDIT + vmaddr: 32768 + vmsize: 16384 + fileoff: 32768 + filesize: 608 + maxprot: 1 + initprot: 1 + nsects: 0 + flags: 0 + - cmd: LC_ID_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 1 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libZ.dylib' + ZeroPadBytes: 7 + - cmd: LC_DYLD_CHAINED_FIXUPS + cmdsize: 16 + dataoff: 32768 + datasize: 96 + - cmd: LC_DYLD_EXPORTS_TRIE + cmdsize: 16 + dataoff: 32864 + datasize: 24 + - cmd: LC_SYMTAB + cmdsize: 24 + symoff: 32896 + nsyms: 2 + stroff: 32936 + strsize: 16 + - cmd: LC_DYSYMTAB + cmdsize: 80 + ilocalsym: 0 + nlocalsym: 0 + iextdefsym: 0 + nextdefsym: 1 + iundefsym: 1 + nundefsym: 1 + tocoff: 0 + ntoc: 0 + modtaboff: 0 + nmodtab: 0 + extrefsymoff: 0 + nextrefsyms: 0 + indirectsymoff: 32928 + nindirectsyms: 2 + extreloff: 0 + nextrel: 0 + locreloff: 0 + nlocrel: 0 + - cmd: LC_UUID + cmdsize: 24 + uuid: 6E8E78AF-EDB2-3830-BE1E-013390302CC5 + - cmd: LC_BUILD_VERSION + cmdsize: 32 + platform: 1 + minos: 983040 + sdk: 983552 + ntools: 1 + Tools: + - tool: 3 + version: 73074435 + - cmd: LC_SOURCE_VERSION + cmdsize: 16 + version: 0 + - cmd: LC_LOAD_DYLIB + cmdsize: 56 + dylib: + name: 24 + timestamp: 2 + current_version: 88539136 + compatibility_version: 65536 + Content: '/usr/lib/libSystem.B.dylib' + ZeroPadBytes: 6 + - cmd: LC_FUNCTION_STARTS + cmdsize: 16 + dataoff: 32888 + datasize: 8 + - cmd: LC_DATA_IN_CODE + cmdsize: 16 + dataoff: 32896 + datasize: 0 + - cmd: LC_CODE_SIGNATURE + cmdsize: 16 + dataoff: 32960 + datasize: 416 + LinkEditData: + ExportTrie: + TerminalSize: 0 + NodeOffset: 0 + Name: '' + Flags: 0x0 + Address: 0x0 + Other: 0x0 + ImportName: '' + Children: + - TerminalSize: 3 + NodeOffset: 13 + Name: _sayZ + Flags: 0x0 + Address: 0x3F70 + Other: 0x0 + ImportName: '' + NameList: + - n_strx: 2 + n_type: 0xF + n_sect: 1 + n_desc: 0 + n_value: 16240 + - n_strx: 8 + n_type: 0x1 + n_sect: 0 + n_desc: 256 + n_value: 0 + StringTable: + - ' ' + - _sayZ + - _printf + IndirectSymbols: [ 0x1, 0x1 ] + FunctionStarts: [ 0x3F70 ] + ChainedFixups: [ 0x0, 0x0, 0x0, 0x0, 0x20, 0x0, 0x0, 0x0, 0x48, + 0x0, 0x0, 0x0, 0x50, 0x0, 0x0, 0x0, 0x1, 0x0, + 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x3, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x18, 0x0, 0x0, 0x0, 0x0, 0x40, 0x6, 0x0, + 0x0, 0x40, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x1, 0x2, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x5F, 0x70, 0x72, + 0x69, 0x6E, 0x74, 0x66, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0 ] + - !mach-o + FileHeader: + magic: 0xFEEDFACF + cputype: 0x100000C + cpusubtype: 0x80000002 + filetype: 0x6 + ncmds: 15 + sizeofcmds: 976 + flags: 0x100085 + reserved: 0x0 + LoadCommands: + - cmd: LC_SEGMENT_64 + cmdsize: 392 + segname: __TEXT + vmaddr: 0 + vmsize: 16384 + fileoff: 0 + filesize: 16384 + maxprot: 5 + initprot: 5 + nsects: 4 + flags: 0 + Sections: + - sectname: __text + segname: __TEXT + addr: 0x3F68 + size: 32 + offset: 0x3F68 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000400 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 7F2303D5FD7BBFA9FD0300910000009000603E9103000094FD7BC1A8FF0F5FD6 + - sectname: __auth_stubs + segname: __TEXT + addr: 0x3F88 + size: 16 + offset: 0x3F88 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x80000408 + reserved1: 0x0 + reserved2: 0x10 + reserved3: 0x0 + content: 110000B031020091300240F9110A1FD7 + - sectname: __cstring + segname: __TEXT + addr: 0x3F98 + size: 14 + offset: 0x3F98 + align: 0 + reloff: 0x0 + nreloc: 0 + flags: 0x2 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 48656C6C6F2066726F6D205A0A00 + - sectname: __unwind_info + segname: __TEXT + addr: 0x3FA8 + size: 88 + offset: 0x3FA8 + align: 2 + reloff: 0x0 + nreloc: 0 + flags: 0x0 + reserved1: 0x0 + reserved2: 0x0 + reserved3: 0x0 + content: 010000001C000000000000001C000000000000001C00000002000000683F00004000000040000000883F00000000000040000000000000000000000000000000030000000C00010010000100000000000000000400000000 + - cmd: LC_SEGMENT_64 + cmdsize: 152 + segname: __DATA_CONST + vmaddr: 16384 + vmsize: 16384 + fileoff: 16384 + filesize: 16384 + maxprot: 3 + initprot: 3 + nsects: 1 + flags: 16 + Sections: + - sectname: __auth_got + segname: __DATA_CONST + addr: 0x4000 + size: 8 + offset: 0x4000 + align: 3 + reloff: 0x0 + nreloc: 0 + flags: 0x6 + reserved1: 0x1 + reserved2: 0x0 + reserved3: 0x0 + content: 00000000000001C0 + - cmd: LC_SEGMENT_64 + cmdsize: 72 + segname: __LINKEDIT + vmaddr: 32768 + vmsize: 16384 + fileoff: 32768 + filesize: 608 + maxprot: 1 + initprot: 1 + nsects: 0 + flags: 0 + - cmd: LC_ID_DYLIB + cmdsize: 48 + dylib: + name: 24 + timestamp: 1 + current_version: 0 + compatibility_version: 0 + Content: '@rpath/libZ.dylib' + ZeroPadBytes: 7 + - cmd: LC_DYLD_CHAINED_FIXUPS + cmdsize: 16 + dataoff: 32768 + datasize: 96 + - cmd: LC_DYLD_EXPORTS_TRIE + cmdsize: 16 + dataoff: 32864 + datasize: 24 + - cmd: LC_SYMTAB + cmdsize: 24 + symoff: 32896 + nsyms: 2 + stroff: 32936 + strsize: 16 + - cmd: LC_DYSYMTAB + cmdsize: 80 + ilocalsym: 0 + nlocalsym: 0 + iextdefsym: 0 + nextdefsym: 1 + iundefsym: 1 + nundefsym: 1 + tocoff: 0 + ntoc: 0 + modtaboff: 0 + nmodtab: 0 + extrefsymoff: 0 + nextrefsyms: 0 + indirectsymoff: 32928 + nindirectsyms: 2 + extreloff: 0 + nextrel: 0 + locreloff: 0 + nlocrel: 0 + - cmd: LC_UUID + cmdsize: 24 + uuid: E74F368D-238F-31FA-BF40-FA2964FED986 + - cmd: LC_BUILD_VERSION + cmdsize: 32 + platform: 1 + minos: 983040 + sdk: 983552 + ntools: 1 + Tools: + - tool: 3 + version: 73074435 + - cmd: LC_SOURCE_VERSION + cmdsize: 16 + version: 0 + - cmd: LC_LOAD_DYLIB + cmdsize: 56 + dylib: + name: 24 + timestamp: 2 + current_version: 88539136 + compatibility_version: 65536 + Content: '/usr/lib/libSystem.B.dylib' + ZeroPadBytes: 6 + - cmd: LC_FUNCTION_STARTS + cmdsize: 16 + dataoff: 32888 + datasize: 8 + - cmd: LC_DATA_IN_CODE + cmdsize: 16 + dataoff: 32896 + datasize: 0 + - cmd: LC_CODE_SIGNATURE + cmdsize: 16 + dataoff: 32960 + datasize: 416 + LinkEditData: + ExportTrie: + TerminalSize: 0 + NodeOffset: 0 + Name: '' + Flags: 0x0 + Address: 0x0 + Other: 0x0 + ImportName: '' + Children: + - TerminalSize: 3 + NodeOffset: 13 + Name: _sayZ + Flags: 0x0 + Address: 0x3F68 + Other: 0x0 + ImportName: '' + NameList: + - n_strx: 2 + n_type: 0xF + n_sect: 1 + n_desc: 0 + n_value: 16232 + - n_strx: 8 + n_type: 0x1 + n_sect: 0 + n_desc: 256 + n_value: 0 + StringTable: + - ' ' + - _sayZ + - _printf + IndirectSymbols: [ 0x1, 0x1 ] + FunctionStarts: [ 0x3F68 ] + ChainedFixups: [ 0x0, 0x0, 0x0, 0x0, 0x20, 0x0, 0x0, 0x0, 0x48, + 0x0, 0x0, 0x0, 0x50, 0x0, 0x0, 0x0, 0x1, 0x0, + 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x3, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x10, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x18, 0x0, 0x0, 0x0, 0x0, 0x40, 0xC, 0x0, + 0x0, 0x40, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x1, 0x2, 0x0, + 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x5F, 0x70, 0x72, + 0x69, 0x6E, 0x74, 0x66, 0x0, 0x0, 0x0, 0x0, 0x0, + 0x0, 0x0, 0x0 ] +... diff --git a/llvm/unittests/ExecutionEngine/Orc/LibraryResolverTest.cpp b/llvm/unittests/ExecutionEngine/Orc/LibraryResolverTest.cpp new file mode 100644 index 0000000..2a396da --- /dev/null +++ b/llvm/unittests/ExecutionEngine/Orc/LibraryResolverTest.cpp @@ -0,0 +1,764 @@ +//===- LibraryResolverTest.cpp - Unit tests for LibraryResolver -===// +// +// 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 +// +//===----------------------------------------------------------------------===// + +#include "llvm/ExecutionEngine/Orc/TargetProcess/LibraryResolver.h" +#include "llvm/ExecutionEngine/Orc/JITTargetMachineBuilder.h" +#include "llvm/ExecutionEngine/Orc/TargetProcess/LibraryScanner.h" +#include "llvm/ObjectYAML/MachOYAML.h" +#include "llvm/ObjectYAML/yaml2obj.h" +#include "llvm/Support/FileSystem.h" +#include "llvm/Support/MemoryBuffer.h" +#include "llvm/Support/Path.h" +#include "llvm/Support/YAMLParser.h" +#include "llvm/Support/YAMLTraits.h" +#include "llvm/Support/raw_ostream.h" + +#include "llvm/Testing/Support/SupportHelpers.h" + +#include "gtest/gtest.h" + +#include <algorithm> +#include <optional> +#include <string> +#include <vector> + +using namespace llvm; +using namespace llvm::orc; + +// Disabled due to test setup issue — YAML to shared library creation seems +// invalid on some build bots. (PR #165360) Not related to code logic. +#if 0 +// TODO: Add COFF (Windows) support for these tests. +// this facility also works correctly on Windows (COFF), +// so we should eventually enable and run these tests for that platform as well. +namespace { + +#if defined(__APPLE__) +constexpr const char *ext = ".dylib"; +#elif defined(_WIN32) +constexpr const char *ext = ".dll"; +#else +constexpr const char *ext = ".so"; +#endif + +bool EnvReady = false; + +Triple getTargetTriple() { + auto JTMB = JITTargetMachineBuilder::detectHost(); + if (!JTMB) { + consumeError(JTMB.takeError()); + return Triple(); + } + return JTMB->getTargetTriple(); +} + +static bool CheckHostSupport() { + auto Triple = getTargetTriple(); + // TODO: Extend support to COFF (Windows) once test setup and YAML conversion + // are verified. + if (!Triple.isOSBinFormatMachO() && + !(Triple.isOSBinFormatELF() && Triple.getArch() == Triple::x86_64)) + return false; + + return true; +} + +std::string getYamlFilePlatformExt() { + auto Triple = getTargetTriple(); + if (Triple.isOSBinFormatMachO()) + return "_macho"; + else if (Triple.isOSBinFormatELF()) + return "_linux"; + + return ""; +} + +unsigned getYamlDocNum() { + // auto Triple = getTargetTriple(); + // if (Triple.isOSBinFormatELF()) + // return 1; + + return 1; +} + +class LibraryTestEnvironment : public ::testing::Environment { + std::vector<std::string> CreatedDylibsDir; + std::vector<std::string> CreatedDylibs; + SmallVector<char, 128> DirPath; + +public: + void SetUp() override { + if (!CheckHostSupport()) { + EnvReady = false; + return; + } + + StringRef ThisFile = __FILE__; + SmallVector<char, 128> InputDirPath(ThisFile.begin(), ThisFile.end()); + sys::path::remove_filename(InputDirPath); + sys::path::append(InputDirPath, "Inputs"); + if (!sys::fs::exists(InputDirPath)) + return; + + SmallString<128> UniqueDir; + sys::path::append(UniqueDir, InputDirPath); + std::error_code EC = sys::fs::createUniqueDirectory(UniqueDir, DirPath); + + if (EC) + return; + + // given yamlPath + DylibPath, validate + convert + auto processYamlToDylib = [&](const SmallVector<char, 128> &YamlPath, + const SmallVector<char, 128> &DylibPath, + unsigned DocNum) -> bool { + if (!sys::fs::exists(YamlPath)) { + errs() << "YAML file missing: " + << StringRef(YamlPath.data(), YamlPath.size()) << "\n"; + EnvReady = false; + return false; + } + + auto BufOrErr = MemoryBuffer::getFile(YamlPath); + if (!BufOrErr) { + errs() << "Failed to read " + << StringRef(YamlPath.data(), YamlPath.size()) << ": " + << BufOrErr.getError().message() << "\n"; + EnvReady = false; + return false; + } + + yaml::Input yin(BufOrErr->get()->getBuffer()); + std::error_code EC; + raw_fd_ostream outFile(StringRef(DylibPath.data(), DylibPath.size()), EC, + sys::fs::OF_None); + + if (EC) { + errs() << "Failed to open " + << StringRef(DylibPath.data(), DylibPath.size()) + << " for writing: " << EC.message() << "\n"; + EnvReady = false; + return false; + } + + if (!yaml::convertYAML( + yin, outFile, + [](const Twine &M) { + // Handle or ignore errors here + errs() << "Yaml Error :" << M << "\n"; + }, + DocNum)) { + errs() << "Failed to convert " + << StringRef(YamlPath.data(), YamlPath.size()) << " to " + << StringRef(DylibPath.data(), DylibPath.size()) << "\n"; + EnvReady = false; + return false; + } + + CreatedDylibsDir.push_back(std::string(sys::path::parent_path( + StringRef(DylibPath.data(), DylibPath.size())))); + CreatedDylibs.push_back(std::string(DylibPath.begin(), DylibPath.end())); + return true; + }; + + std::vector<const char *> LibDirs = {"Z", "A", "B", "C"}; + + unsigned DocNum = getYamlDocNum(); + std::string YamlPltExt = getYamlFilePlatformExt(); + for (const auto &LibdirName : LibDirs) { + // YAML path + SmallVector<char, 128> YamlPath(InputDirPath.begin(), InputDirPath.end()); + SmallVector<char, 128> YamlFileName; + YamlFileName.append(LibdirName, LibdirName + strlen(LibdirName)); + YamlFileName.append(YamlPltExt.begin(), YamlPltExt.end()); + sys::path::append(YamlPath, LibdirName, YamlFileName); + sys::path::replace_extension(YamlPath, ".yaml"); + + // dylib path + SmallVector<char, 128> DylibPath(DirPath.begin(), DirPath.end()); + SmallVector<char, 128> DylibFileName; + StringRef prefix("lib"); + DylibFileName.append(prefix.begin(), prefix.end()); + DylibFileName.append(LibdirName, LibdirName + strlen(LibdirName)); + + sys::path::append(DylibPath, LibdirName); + if (!sys::fs::exists(DylibPath)) { + auto EC = sys::fs::create_directory(DylibPath); + if (EC) + return; + } + sys::path::append(DylibPath, DylibFileName); + sys::path::replace_extension(DylibPath, ext); + if (!processYamlToDylib(YamlPath, DylibPath, DocNum)) + return; + } + + EnvReady = true; + } + + void TearDown() override { sys::fs::remove_directories(DirPath); } + + std::string getBaseDir() const { + return std::string(DirPath.begin(), DirPath.end()); + } + + std::vector<std::string> getDylibPaths() const { return CreatedDylibs; } +}; + +static LibraryTestEnvironment *GlobalEnv = + static_cast<LibraryTestEnvironment *>( + ::testing::AddGlobalTestEnvironment(new LibraryTestEnvironment())); + +inline std::string libPath(const std::string &BaseDir, + const std::string &name) { +#if defined(__APPLE__) + return BaseDir + "/" + name + ".dylib"; +#elif defined(_WIN32) + return BaseDir + "/" + name + ".dll"; +#else + return BaseDir + "/" + name + ".so"; +#endif +} + +inline std::string withext(const std::string &lib) { + SmallString<128> P(lib); + sys::path::replace_extension(P, ext); + return P.str().str(); +} + +inline std::string platformSymbolName(const std::string &name) { +#if defined(__APPLE__) + return "_" + name; // macOS prepends underscore +#else + return name; +#endif +} + +struct TestLibrary { + std::string path; + std::vector<std::string> Syms; +}; + +class LibraryResolverIT : public ::testing::Test { +protected: + std::string BaseDir; + std::unordered_map<std::string, TestLibrary> libs; + + void addLib(const std::string &name) { + SmallString<512> path; + std::error_code EC = + sys::fs::real_path(libPath(BaseDir, name + "/lib" + name), path); + if (EC || path.empty() || !sys::fs::exists(path)) + GTEST_SKIP(); + libs[name] = {path.str().str(), {platformSymbolName("say" + name)}}; + } + + void SetUp() override { + if (!EnvReady || GlobalEnv == nullptr) + GTEST_SKIP() << "Skipping test: environment setup failed."; + + { + SmallString<512> path; + std::error_code EC = sys::fs::real_path(GlobalEnv->getBaseDir(), path); + if (path.empty() || EC) + GTEST_SKIP() << "Base directory resolution failed: " << EC.message(); + BaseDir = path.str().str(); + } + + for (const auto &P : GlobalEnv->getDylibPaths()) { + if (!sys::fs::exists(P)) + GTEST_SKIP() << "Missing dylib path: " << P; + } + + const std::vector<std::string> libNames = {"A", "B", "C", "Z"}; + for (const auto &name : libNames) + addLib(name); + + if (!EnvReady) + GTEST_SKIP() << "Skipping test: environment setup failed."; + } + + const std::vector<std::string> &sym(const std::string &key) { + return libs[key].Syms; + } + const std::string &lib(const std::string &key) { return libs[key].path; } + const std::string libdir(const std::string &key) { + SmallString<512> P(libs[key].path); + sys::path::remove_filename(P); + return P.str().str(); + } + const std::string libname(const std::string &key) { + return sys::path::filename(libs[key].path).str(); + } +}; + +// Helper: allow either "sayA" or "_sayA" depending on how your +// SymbolEnumerator reports. +static bool matchesEitherUnderscore(const std::string &got, + const std::string &bare) { + return got == bare || got == ("_" + bare); +} + +// Helper: normalize path ending check (we only care that it resolved to the +// right dylib) +static bool endsWith(const std::string &s, const std::string &suffix) { + if (s.size() < suffix.size()) + return false; + return std::equal(suffix.rbegin(), suffix.rend(), s.rbegin()); +} + +TEST_F(LibraryResolverIT, EnumerateSymbols_ExportsOnly_DefaultFlags) { + const std::string libC = lib("C"); + SymbolEnumeratorOptions Opts = SymbolEnumeratorOptions::defaultOptions(); + + std::vector<std::string> seen; + auto onEach = [&](llvm::StringRef sym) -> EnumerateResult { + seen.emplace_back(sym.str()); + return EnumerateResult::Continue; + }; + + ASSERT_TRUE(SymbolEnumerator::enumerateSymbols(libC, onEach, Opts)); + + // sayC is exported, others are undefined → only sayC expected + EXPECT_TRUE(any_of(seen, [&](const std::string &s) { + return matchesEitherUnderscore(s, "sayC"); + })); + EXPECT_FALSE(any_of(seen, [&](const std::string &s) { + return matchesEitherUnderscore(s, "sayA"); + })); + EXPECT_FALSE(any_of(seen, [&](const std::string &s) { + return matchesEitherUnderscore(s, "sayB"); + })); +} + +TEST_F(LibraryResolverIT, EnumerateSymbols_IncludesUndefineds) { + const std::string libC = lib("C"); + + SymbolEnumeratorOptions Opts; + Opts.FilterFlags = + SymbolEnumeratorOptions::IgnoreWeak | + SymbolEnumeratorOptions::IgnoreIndirect; // no IgnoreUndefined + + std::vector<std::string> seen; + auto onEach = [&](llvm::StringRef sym) -> EnumerateResult { + seen.emplace_back(sym.str()); + return EnumerateResult::Continue; + }; + + ASSERT_TRUE(SymbolEnumerator::enumerateSymbols(libC, onEach, Opts)); + + // Now we should see both sayC (export) and the undefined refs sayA, sayB, + // sayZ + EXPECT_TRUE(any_of(seen, [&](const std::string &s) { + return matchesEitherUnderscore(s, "sayC"); + })); + EXPECT_TRUE(any_of(seen, [&](const std::string &s) { + return matchesEitherUnderscore(s, "sayA"); + })); + EXPECT_TRUE(any_of(seen, [&](const std::string &s) { + return matchesEitherUnderscore(s, "sayB"); + })); +} + +// Full resolution via LibraryResolutionDriver/LibraryResolver --- +TEST_F(LibraryResolverIT, DriverResolvesSymbolsToCorrectLibraries) { + // Create the resolver from real base paths (our fixtures dir) + auto Stup = LibraryResolver::Setup::create({BaseDir}); + + // Full system behavior: no mocks + auto Driver = LibraryResolutionDriver::create(Stup); + ASSERT_NE(Driver, nullptr); + + // Tell the Driver about the scan path kinds (User/System) as your + // production code expects. + Driver->addScanPath(libdir("A"), PathType::User); + Driver->addScanPath(libdir("B"), PathType::User); + Driver->addScanPath(libdir("Z"), PathType::User); + + // Symbols to resolve (bare names; class handles underscore differences + // internally) + std::vector<std::string> Syms = {platformSymbolName("sayA"), + platformSymbolName("sayB"), + platformSymbolName("sayZ")}; + + bool CallbackRan = false; + Driver->resolveSymbols(Syms, [&](SymbolQuery &Q) { + CallbackRan = true; + + // sayA should resolve to A.dylib + { + auto lib = Q.getResolvedLib(platformSymbolName("sayA")); + ASSERT_TRUE(lib.has_value()) << "sayA should be resolved"; + EXPECT_TRUE(endsWith(lib->str(), libname("A"))) + << "sayA resolved to: " << lib->str(); + } + + // sayB should resolve to B.dylib + { + auto lib = Q.getResolvedLib(platformSymbolName("sayB")); + ASSERT_TRUE(lib.has_value()) << "sayB should be resolved"; + EXPECT_TRUE(endsWith(lib->str(), libname("B"))) + << "sayB resolved to: " << lib->str(); + } + + // sayZ should resolve to B.dylib + { + auto lib = Q.getResolvedLib(platformSymbolName("sayZ")); + ASSERT_TRUE(lib.has_value()) << "sayZ should be resolved"; + EXPECT_TRUE(endsWith(lib->str(), libname("Z"))) + << "sayZ resolved to: " << lib->str(); + } + + EXPECT_TRUE(Q.allResolved()); + }); + + EXPECT_TRUE(CallbackRan); +} + +// stress SymbolQuery with the real resolve flow +// And resolve libC dependency libA, libB, libZ --- +TEST_F(LibraryResolverIT, ResolveManySymbols) { + auto Stup = LibraryResolver::Setup::create({BaseDir}); + auto Driver = LibraryResolutionDriver::create(Stup); + ASSERT_NE(Driver, nullptr); + Driver->addScanPath(libdir("C"), PathType::User); + + // Many duplicates to provoke concurrent updates inside SymbolQuery + std::vector<std::string> Syms = { + platformSymbolName("sayA"), platformSymbolName("sayB"), + platformSymbolName("sayA"), platformSymbolName("sayB"), + platformSymbolName("sayZ"), platformSymbolName("sayZ"), + platformSymbolName("sayZ"), platformSymbolName("sayZ"), + platformSymbolName("sayA"), platformSymbolName("sayB"), + platformSymbolName("sayA"), platformSymbolName("sayB")}; + + bool CallbackRan = false; + Driver->resolveSymbols(Syms, [&](SymbolQuery &Q) { + CallbackRan = true; + EXPECT_TRUE(Q.isResolved(platformSymbolName("sayA"))); + EXPECT_TRUE(Q.isResolved(platformSymbolName("sayB"))); + EXPECT_TRUE(Q.isResolved(platformSymbolName("sayZ"))); + + auto A = Q.getResolvedLib(platformSymbolName("sayA")); + auto B = Q.getResolvedLib(platformSymbolName("sayB")); + auto Z = Q.getResolvedLib(platformSymbolName("sayZ")); + ASSERT_TRUE(A.has_value()); + ASSERT_TRUE(B.has_value()); + ASSERT_TRUE(Z.has_value()); + EXPECT_TRUE(endsWith(A->str(), libname("A"))); + EXPECT_TRUE(endsWith(B->str(), libname("B"))); + EXPECT_TRUE(endsWith(Z->str(), libname("Z"))); + EXPECT_TRUE(Q.allResolved()); + }); + + EXPECT_TRUE(CallbackRan); +} + +TEST_F(LibraryResolverIT, ScanAndResolveDependencyGraph) { + auto LibPathCache = std::make_shared<LibraryPathCache>(); + auto PResolver = std::make_shared<PathResolver>(LibPathCache); + LibraryScanHelper ScanH({}, LibPathCache, PResolver); + + ScanH.addBasePath(libdir("C"), PathType::User); + + LibraryManager LibMgr; + LibraryScanner Scanner(ScanH, LibMgr); + + Scanner.scanNext(PathType::User, 0); + + size_t numLibs = 0; + LibMgr.forEachLibrary([&](const LibraryInfo &L) { + numLibs++; + return true; + }); + + EXPECT_GT(numLibs, 0u) << "Expected at least one library scanned"; + + // Validate that each scanned library path is resolvable + std::error_code EC; + LibMgr.forEachLibrary([&](const LibraryInfo &L) { + auto R = PResolver->resolve(L.getFullPath(), EC); + EXPECT_TRUE(R.has_value()); + EXPECT_FALSE(EC); + return true; + }); +} + +TEST_F(LibraryResolverIT, ScanEmptyPath) { + auto LibPathCache = std::make_shared<LibraryPathCache>(); + auto PResolver = std::make_shared<PathResolver>(LibPathCache); + LibraryScanHelper ScanH({}, LibPathCache, PResolver); + + ScanH.addBasePath("/tmp/empty", PathType::User); + + LibraryManager LibMgr; + LibraryScanner Scanner(ScanH, LibMgr); + + Scanner.scanNext(PathType::User, 0); + + size_t count = 0; + LibMgr.forEachLibrary([&](const LibraryInfo &) { + count++; + return true; + }); + EXPECT_EQ(count, 0u); +} + +TEST_F(LibraryResolverIT, PathResolverResolvesKnownPaths) { + auto LibPathCache = std::make_shared<LibraryPathCache>(); + auto PResolver = std::make_shared<PathResolver>(LibPathCache); + + std::error_code EC; + auto Missing = PResolver->resolve("temp/foo/bar", EC); + EXPECT_FALSE(Missing.has_value()) << "Unexpectedly resolved a bogus path"; + EXPECT_TRUE(EC) << "Expected error resolving path"; + + auto DirPath = PResolver->resolve(BaseDir, EC); + ASSERT_TRUE(DirPath.has_value()); + EXPECT_FALSE(EC) << "Expected no error resolving path"; + EXPECT_EQ(*DirPath, BaseDir); + + auto DylibPath = PResolver->resolve(lib("C"), EC); + ASSERT_TRUE(DylibPath.has_value()); + EXPECT_FALSE(EC) << "Expected no error resolving path"; + EXPECT_EQ(*DylibPath, lib("C")); +} + +TEST_F(LibraryResolverIT, PathResolverNormalizesDotAndDotDot) { + auto LibPathCache = std::make_shared<LibraryPathCache>(); + auto PResolver = std::make_shared<PathResolver>(LibPathCache); + + std::error_code EC; + + // e.g. BaseDir + "/./C/../C/C.dylib" → BaseDir + "/C.dylib" + std::string Messy = BaseDir + "/C/./../C/./libC" + ext; + auto Resolved = PResolver->resolve(Messy, EC); + ASSERT_TRUE(Resolved.has_value()); + EXPECT_FALSE(EC); + EXPECT_EQ(*Resolved, lib("C")) << "Expected realpath to collapse . and .."; +} + +#if !defined(_WIN32) +TEST_F(LibraryResolverIT, PathResolverFollowsSymlinks) { + auto LibPathCache = std::make_shared<LibraryPathCache>(); + auto PResolver = std::make_shared<PathResolver>(LibPathCache); + + std::error_code EC; + + // Create a symlink temp -> BaseDir (only if filesystem allows it) + std::string linkName = BaseDir + withext("/link_to_C"); + std::string target = lib("C"); + if (::symlink(target.c_str(), linkName.c_str()) != 0) + GTEST_SKIP() << "Failed to create symlink: " << strerror(errno); + + auto resolved = PResolver->resolve(linkName, EC); + ASSERT_TRUE(resolved.has_value()); + EXPECT_FALSE(EC); + EXPECT_EQ(*resolved, target); + + (void)::unlink(linkName.c_str()); // cleanup +} + +TEST_F(LibraryResolverIT, PathResolverCachesResults) { + auto LibPathCache = std::make_shared<LibraryPathCache>(); + auto PResolver = std::make_shared<PathResolver>(LibPathCache); + + SmallString<128> TmpDylib; + std::error_code EC; + EC = sys::fs::createUniqueFile(withext("A-copy"), TmpDylib); + if (EC) + GTEST_SKIP() << "Failed to create temp dylib" << EC.message(); + + EC = sys::fs::copy_file(lib("A"), TmpDylib); + if (EC) + GTEST_SKIP() << "Failed to copy libA: " << EC.message(); + EC.clear(); + + // First resolve -> should populate LibPathCache + auto first = PResolver->resolve(TmpDylib, EC); + ASSERT_TRUE(first.has_value()); + + // Forcefully remove the file from disk + (void)::unlink(TmpDylib.c_str()); + + // Second resolve -> should still succeed from LibPathCache + auto second = PResolver->resolve(TmpDylib, EC); + EXPECT_TRUE(second.has_value()); + EXPECT_EQ(*second, *first); +} +#endif + +TEST_F(LibraryResolverIT, LoaderPathSubstitutionAndResolve) { + auto LibPathCache = std::make_shared<LibraryPathCache>(); + auto PResolver = std::make_shared<PathResolver>(LibPathCache); + + DylibSubstitutor substitutor; + substitutor.configure(libdir("C")); +#if defined(__APPLE__) + // Substitute @loader_path with BaseDir + std::string substituted = + substitutor.substitute(withext("@loader_path/libC")); +#elif defined(__linux__) + // Substitute $origin with BaseDir + std::string substituted = substitutor.substitute(withext("$ORIGIN/libC")); +#endif + ASSERT_FALSE(substituted.empty()); + EXPECT_EQ(substituted, lib("C")); + + // Now try resolving the substituted path + std::error_code EC; + auto resolved = PResolver->resolve(substituted, EC); + ASSERT_TRUE(resolved.has_value()) << "Expected to resolve substituted dylib"; + EXPECT_EQ(*resolved, lib("C")); + EXPECT_FALSE(EC) << "Expected no error resolving substituted dylib"; +} + +TEST_F(LibraryResolverIT, ResolveFromUsrOrSystemPaths) { + auto LibPathCache = std::make_shared<LibraryPathCache>(); + auto PResolver = std::make_shared<PathResolver>(LibPathCache); + + DylibPathValidator validator(*PResolver); + + std::vector<std::string> Paths = {"/foo/bar/", "temp/foo", libdir("C"), + libdir("A"), libdir("B"), libdir("Z")}; + + SmallVector<StringRef> P(Paths.begin(), Paths.end()); + + DylibResolver Resolver(validator); + Resolver.configure("", {{P, SearchPathType::UsrOrSys}}); + + // Check "C" + auto ValOptC = Resolver.resolve("libC", true); + EXPECT_TRUE(ValOptC.has_value()); + EXPECT_EQ(*ValOptC, lib("C")); + + auto ValOptCdylib = Resolver.resolve(withext("libC")); + EXPECT_TRUE(ValOptCdylib.has_value()); + EXPECT_EQ(*ValOptCdylib, lib("C")); + + // Check "A" + auto ValOptA = Resolver.resolve("libA", true); + EXPECT_TRUE(ValOptA.has_value()); + EXPECT_EQ(*ValOptA, lib("A")); + + auto ValOptAdylib = Resolver.resolve(withext("libA")); + EXPECT_TRUE(ValOptAdylib.has_value()); + EXPECT_EQ(*ValOptAdylib, lib("A")); + + // Check "B" + auto ValOptB = Resolver.resolve("libB", true); + EXPECT_TRUE(ValOptB.has_value()); + EXPECT_EQ(*ValOptB, lib("B")); + + auto ValOptBdylib = Resolver.resolve(withext("libB")); + EXPECT_TRUE(ValOptBdylib.has_value()); + EXPECT_EQ(*ValOptBdylib, lib("B")); + + // Check "Z" + auto ValOptZ = Resolver.resolve("libZ", true); + EXPECT_TRUE(ValOptZ.has_value()); + EXPECT_EQ(*ValOptZ, lib("Z")); + + auto ValOptZdylib = Resolver.resolve(withext("libZ")); + EXPECT_TRUE(ValOptZdylib.has_value()); + EXPECT_EQ(*ValOptZdylib, lib("Z")); +} + +#if defined(__APPLE__) +TEST_F(LibraryResolverIT, ResolveViaLoaderPathAndRPathSubstitution) { + auto LibPathCache = std::make_shared<LibraryPathCache>(); + auto PResolver = std::make_shared<PathResolver>(LibPathCache); + + DylibPathValidator validator(*PResolver); + + std::vector<std::string> Paths = {"@loader_path/../A", "@loader_path/../B", + "@loader_path/../C", "@loader_path/../Z"}; + + SmallVector<StringRef> P(Paths.begin(), Paths.end()); + + DylibResolver Resolver(validator); + + // Use only RPath config + Resolver.configure(lib("C"), {{P, SearchPathType::RPath}}); + + // --- Check A --- + auto ValOptA = Resolver.resolve("@rpath/libA", true); + EXPECT_TRUE(ValOptA.has_value()); + EXPECT_EQ(*ValOptA, lib("A")); + + auto ValOptAdylib = Resolver.resolve(withext("@rpath/libA")); + EXPECT_TRUE(ValOptAdylib.has_value()); + EXPECT_EQ(*ValOptAdylib, lib("A")); + + // --- Check B --- + auto ValOptB = Resolver.resolve("@rpath/libB", true); + EXPECT_TRUE(ValOptB.has_value()); + EXPECT_EQ(*ValOptB, lib("B")); + + auto ValOptBdylib = Resolver.resolve(withext("@rpath/libB")); + EXPECT_TRUE(ValOptBdylib.has_value()); + EXPECT_EQ(*ValOptBdylib, lib("B")); + + // --- Check Z --- + auto ValOptZ = Resolver.resolve("@rpath/libZ", true); + EXPECT_TRUE(ValOptZ.has_value()); + EXPECT_EQ(*ValOptZ, lib("Z")); + + auto ValOptZdylib = Resolver.resolve(withext("@rpath/libZ")); + EXPECT_TRUE(ValOptZdylib.has_value()); + EXPECT_EQ(*ValOptZdylib, lib("Z")); +} +#endif + +#if defined(__linux__) +TEST_F(LibraryResolverIT, ResolveViaOriginAndRPathSubstitution) { + auto LibPathCache = std::make_shared<LibraryPathCache>(); + auto PResolver = std::make_shared<PathResolver>(LibPathCache); + + DylibPathValidator validator(*PResolver); + + // On Linux, $ORIGIN works like @loader_path + std::vector<std::string> Paths = {"$ORIGIN/../A", "$ORIGIN/../B", + "$ORIGIN/../C", "$ORIGIN/../Z"}; + + SmallVector<StringRef> P(Paths.begin(), Paths.end()); + + DylibResolver Resolver(validator); + + // Use only RPath config + Resolver.configure(lib("C"), {{P, SearchPathType::RunPath}}); + + // --- Check A --- + auto ValOptA = Resolver.resolve("libA", true); + EXPECT_TRUE(ValOptA.has_value()); + EXPECT_EQ(*ValOptA, lib("A")); + + auto valOptASO = Resolver.resolve(withext("libA")); + EXPECT_TRUE(valOptASO.has_value()); + EXPECT_EQ(*valOptASO, lib("A")); + + // --- Check B --- + auto ValOptB = Resolver.resolve("libB", true); + EXPECT_TRUE(ValOptB.has_value()); + EXPECT_EQ(*ValOptB, lib("B")); + + auto valOptBSO = Resolver.resolve(withext("libB")); + EXPECT_TRUE(valOptBSO.has_value()); + EXPECT_EQ(*valOptBSO, lib("B")); + + // --- Check Z --- + auto ValOptZ = Resolver.resolve("libZ", true); + EXPECT_TRUE(ValOptZ.has_value()); + EXPECT_EQ(*ValOptZ, lib("Z")); + + auto valOptZSO = Resolver.resolve(withext("libZ")); + EXPECT_TRUE(valOptZSO.has_value()); + EXPECT_EQ(*valOptZSO, lib("Z")); +} +#endif +} // namespace +#endif // defined(__APPLE__) diff --git a/llvm/unittests/Transforms/Vectorize/VPlanVerifierTest.cpp b/llvm/unittests/Transforms/Vectorize/VPlanVerifierTest.cpp index 50ad4d5..4680282 100644 --- a/llvm/unittests/Transforms/Vectorize/VPlanVerifierTest.cpp +++ b/llvm/unittests/Transforms/Vectorize/VPlanVerifierTest.cpp @@ -21,7 +21,7 @@ using VPVerifierTest = VPlanTestBase; namespace { TEST_F(VPVerifierTest, VPInstructionUseBeforeDefSameBB) { VPlan &Plan = getPlan(); - VPValue *Zero = Plan.getOrAddLiveIn(ConstantInt::get(Type::getInt32Ty(C), 0)); + VPValue *Zero = Plan.getConstantInt(32, 0); VPInstruction *DefI = new VPInstruction(Instruction::Add, {Zero}); VPInstruction *UseI = new VPInstruction(Instruction::Sub, {DefI}); auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {}); @@ -56,7 +56,7 @@ TEST_F(VPVerifierTest, VPInstructionUseBeforeDefSameBB) { TEST_F(VPVerifierTest, VPInstructionUseBeforeDefDifferentBB) { VPlan &Plan = getPlan(); - VPValue *Zero = Plan.getOrAddLiveIn(ConstantInt::get(Type::getInt32Ty(C), 0)); + VPValue *Zero = Plan.getConstantInt(32, 0); VPInstruction *DefI = new VPInstruction(Instruction::Add, {Zero}); VPInstruction *UseI = new VPInstruction(Instruction::Sub, {DefI}); auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {}); @@ -184,7 +184,7 @@ TEST_F(VPVerifierTest, VPPhiIncomingValueDoesntDominateIncomingBlock) { TEST_F(VPVerifierTest, DuplicateSuccessorsOutsideRegion) { VPlan &Plan = getPlan(); - VPValue *Zero = Plan.getOrAddLiveIn(ConstantInt::get(Type::getInt32Ty(C), 0)); + VPValue *Zero = Plan.getConstantInt(32, 0); VPInstruction *I1 = new VPInstruction(Instruction::Add, {Zero}); auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {}); VPInstruction *BranchOnCond = @@ -218,7 +218,7 @@ TEST_F(VPVerifierTest, DuplicateSuccessorsOutsideRegion) { TEST_F(VPVerifierTest, DuplicateSuccessorsInsideRegion) { VPlan &Plan = getPlan(); - VPValue *Zero = Plan.getOrAddLiveIn(ConstantInt::get(Type::getInt32Ty(C), 0)); + VPValue *Zero = Plan.getConstantInt(32, 0); VPInstruction *I1 = new VPInstruction(Instruction::Add, {Zero}); auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {}); VPInstruction *BranchOnCond = @@ -259,7 +259,7 @@ TEST_F(VPVerifierTest, BlockOutsideRegionWithParent) { VPBasicBlock *VPBB1 = Plan.getEntry(); VPBasicBlock *VPBB2 = Plan.createVPBasicBlock(""); - VPValue *Zero = Plan.getOrAddLiveIn(ConstantInt::get(Type::getInt32Ty(C), 0)); + VPValue *Zero = Plan.getConstantInt(32, 0); auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {}); VPBB2->appendRecipe(CanIV); @@ -288,7 +288,7 @@ TEST_F(VPVerifierTest, BlockOutsideRegionWithParent) { TEST_F(VPVerifierTest, NonHeaderPHIInHeader) { VPlan &Plan = getPlan(); - VPValue *Zero = Plan.getOrAddLiveIn(ConstantInt::get(Type::getInt32Ty(C), 0)); + VPValue *Zero = Plan.getConstantInt(32, 0); auto *CanIV = new VPCanonicalIVPHIRecipe(Zero, {}); auto *BranchOnCond = new VPInstruction(VPInstruction::BranchOnCond, {CanIV}); @@ -351,8 +351,7 @@ TEST_F(VPIRVerifierTest, testVerifyIRPhi) { BasicBlock *LoopHeader = F->getEntryBlock().getSingleSuccessor(); auto Plan = buildVPlan(LoopHeader); - Plan->getExitBlocks()[0]->front().addOperand( - Plan->getOrAddLiveIn(ConstantInt::get(Type::getInt32Ty(*Ctx), 0))); + Plan->getExitBlocks()[0]->front().addOperand(Plan->getConstantInt(32, 0)); #if GTEST_HAS_STREAM_REDIRECTION ::testing::internal::CaptureStderr(); diff --git a/llvm/utils/UpdateTestChecks/common.py b/llvm/utils/UpdateTestChecks/common.py index 119303c..2dad16a 100644 --- a/llvm/utils/UpdateTestChecks/common.py +++ b/llvm/utils/UpdateTestChecks/common.py @@ -29,7 +29,7 @@ Version changelog: 'none' and 'all'. 'smart' is the default. 5: Basic block labels are matched by FileCheck expressions 6: The semantics of TBAA checks has been incorporated in the check lines. -7: Indent switch-cases correctly. +7: Indent switch-cases correctly; CHECK-EMPTY instead of skipping blank lines. """ DEFAULT_VERSION = 6 @@ -2280,6 +2280,14 @@ def add_checks( # For IR output, change all defs to FileCheck variables, so we're immune # to variable naming fashions. else: + if ginfo.get_version() >= 7: + # Record the indices of blank lines in the function body preemptively. + blank_line_indices = { + i for i, line in enumerate(func_body) if line.strip() == "" + } + else: + blank_line_indices = set() + func_body = generalize_check_lines( func_body, ginfo, @@ -2305,9 +2313,18 @@ def add_checks( is_blank_line = False - for func_line in func_body: + for idx, func_line in enumerate(func_body): if func_line.strip() == "": - is_blank_line = True + # We should distinguish if the line is a 'fake' blank line generated by + # generalize_check_lines removing comments. + # Fortunately, generalize_check_lines does not change the index of each line, + # we can record the indices of blank lines preemptively. + if idx in blank_line_indices: + output_lines.append( + "{} {}-EMPTY:".format(comment_marker, checkprefix) + ) + else: + is_blank_line = True continue if not check_inst_comments: # Do not waste time checking IR comments unless necessary. |