| 1 | //===- NVVMProperties.cpp - NVVM annotation utilities ---------------------===// |
| 2 | // |
| 3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
| 4 | // See https://llvm.org/LICENSE.txt for license information. |
| 5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
| 6 | // |
| 7 | //===----------------------------------------------------------------------===// |
| 8 | // |
| 9 | // This file contains NVVM attribute and metadata query utilities. |
| 10 | // |
| 11 | //===----------------------------------------------------------------------===// |
| 12 | |
| 13 | #include "NVVMProperties.h" |
| 14 | #include "llvm/ADT/ArrayRef.h" |
| 15 | #include "llvm/ADT/STLExtras.h" |
| 16 | #include "llvm/ADT/StringRef.h" |
| 17 | #include "llvm/IR/Argument.h" |
| 18 | #include "llvm/IR/Constants.h" |
| 19 | #include "llvm/IR/Function.h" |
| 20 | #include "llvm/IR/GlobalValue.h" |
| 21 | #include "llvm/IR/GlobalVariable.h" |
| 22 | #include "llvm/IR/Instructions.h" |
| 23 | #include "llvm/IR/Metadata.h" |
| 24 | #include "llvm/IR/Module.h" |
| 25 | #include "llvm/Support/ModRef.h" |
| 26 | #include "llvm/Support/Mutex.h" |
| 27 | #include "llvm/Support/NVVMAttributes.h" |
| 28 | #include <functional> |
| 29 | #include <map> |
| 30 | #include <mutex> |
| 31 | #include <numeric> |
| 32 | #include <string> |
| 33 | #include <vector> |
| 34 | |
| 35 | namespace llvm { |
| 36 | |
| 37 | namespace { |
| 38 | using AnnotationValues = std::map<std::string, std::vector<unsigned>>; |
| 39 | using AnnotationMap = std::map<const GlobalValue *, AnnotationValues>; |
| 40 | |
| 41 | struct AnnotationCache { |
| 42 | sys::Mutex Lock; |
| 43 | std::map<const Module *, AnnotationMap> Cache; |
| 44 | }; |
| 45 | |
| 46 | AnnotationCache &getAnnotationCache() { |
| 47 | static AnnotationCache AC; |
| 48 | return AC; |
| 49 | } |
| 50 | } // namespace |
| 51 | |
| 52 | // TODO: Replace these legacy nvvm.annotations metadata names with proper |
| 53 | // function/parameter attributes (like the NVVMAttr:: constants). |
| 54 | namespace NVVMMetadata { |
| 55 | constexpr StringLiteral Texture("texture" ); |
| 56 | constexpr StringLiteral Surface("surface" ); |
| 57 | constexpr StringLiteral Sampler("sampler" ); |
| 58 | constexpr StringLiteral ReadOnlyImage("rdoimage" ); |
| 59 | constexpr StringLiteral WriteOnlyImage("wroimage" ); |
| 60 | constexpr StringLiteral ReadWriteImage("rdwrimage" ); |
| 61 | constexpr StringLiteral Managed("managed" ); |
| 62 | } // namespace NVVMMetadata |
| 63 | |
| 64 | void clearAnnotationCache(const Module *Mod) { |
| 65 | auto &AC = getAnnotationCache(); |
| 66 | std::lock_guard<sys::Mutex> Guard(AC.Lock); |
| 67 | AC.Cache.erase(x: Mod); |
| 68 | } |
| 69 | |
| 70 | static void cacheAnnotationFromMD(const MDNode *MetadataNode, |
| 71 | AnnotationValues &RetVal) { |
| 72 | auto &AC = getAnnotationCache(); |
| 73 | std::lock_guard<sys::Mutex> Guard(AC.Lock); |
| 74 | assert(MetadataNode && "Invalid mdnode for annotation" ); |
| 75 | assert((MetadataNode->getNumOperands() % 2) == 1 && |
| 76 | "Invalid number of operands" ); |
| 77 | // start index = 1, to skip the global variable key |
| 78 | // increment = 2, to skip the value for each property-value pairs |
| 79 | for (unsigned I = 1, E = MetadataNode->getNumOperands(); I != E; I += 2) { |
| 80 | const MDString *Prop = dyn_cast<MDString>(Val: MetadataNode->getOperand(I)); |
| 81 | assert(Prop && "Annotation property not a string" ); |
| 82 | std::string Key = Prop->getString().str(); |
| 83 | |
| 84 | if (ConstantInt *Val = mdconst::dyn_extract<ConstantInt>( |
| 85 | MD: MetadataNode->getOperand(I: I + 1))) { |
| 86 | RetVal[Key].push_back(x: Val->getZExtValue()); |
| 87 | } else { |
| 88 | llvm_unreachable("Value operand not a constant int" ); |
| 89 | } |
| 90 | } |
| 91 | } |
| 92 | |
| 93 | static void cacheAnnotationFromMD(const Module *M, const GlobalValue *GV) { |
| 94 | auto &AC = getAnnotationCache(); |
| 95 | std::lock_guard<sys::Mutex> Guard(AC.Lock); |
| 96 | NamedMDNode *NMD = M->getNamedMetadata(Name: "nvvm.annotations" ); |
| 97 | if (!NMD) |
| 98 | return; |
| 99 | |
| 100 | AnnotationValues Tmp; |
| 101 | for (unsigned I = 0, E = NMD->getNumOperands(); I != E; ++I) { |
| 102 | const MDNode *Elem = NMD->getOperand(i: I); |
| 103 | GlobalValue *Entity = |
| 104 | mdconst::dyn_extract_or_null<GlobalValue>(MD: Elem->getOperand(I: 0)); |
| 105 | // entity may be null due to DCE |
| 106 | if (!Entity || Entity != GV) |
| 107 | continue; |
| 108 | |
| 109 | cacheAnnotationFromMD(MetadataNode: Elem, RetVal&: Tmp); |
| 110 | } |
| 111 | |
| 112 | if (Tmp.empty()) |
| 113 | return; |
| 114 | |
| 115 | AC.Cache[M][GV] = std::move(Tmp); |
| 116 | } |
| 117 | |
| 118 | static std::optional<unsigned> findOneNVVMAnnotation(const GlobalValue *GV, |
| 119 | StringRef Prop) { |
| 120 | auto &AC = getAnnotationCache(); |
| 121 | std::lock_guard<sys::Mutex> Guard(AC.Lock); |
| 122 | const Module *M = GV->getParent(); |
| 123 | auto ACIt = AC.Cache.find(x: M); |
| 124 | if (ACIt == AC.Cache.end()) |
| 125 | cacheAnnotationFromMD(M, GV); |
| 126 | else if (ACIt->second.find(x: GV) == ACIt->second.end()) |
| 127 | cacheAnnotationFromMD(M, GV); |
| 128 | |
| 129 | auto &KVP = AC.Cache[M][GV]; |
| 130 | auto It = KVP.find(x: Prop.str()); |
| 131 | if (It == KVP.end()) |
| 132 | return std::nullopt; |
| 133 | return It->second[0]; |
| 134 | } |
| 135 | |
| 136 | static bool findAllNVVMAnnotation(const GlobalValue *GV, StringRef Prop, |
| 137 | std::vector<unsigned> &RetVal) { |
| 138 | auto &AC = getAnnotationCache(); |
| 139 | std::lock_guard<sys::Mutex> Guard(AC.Lock); |
| 140 | const Module *M = GV->getParent(); |
| 141 | auto ACIt = AC.Cache.find(x: M); |
| 142 | if (ACIt == AC.Cache.end()) |
| 143 | cacheAnnotationFromMD(M, GV); |
| 144 | else if (ACIt->second.find(x: GV) == ACIt->second.end()) |
| 145 | cacheAnnotationFromMD(M, GV); |
| 146 | |
| 147 | auto &KVP = AC.Cache[M][GV]; |
| 148 | auto It = KVP.find(x: Prop.str()); |
| 149 | if (It == KVP.end()) |
| 150 | return false; |
| 151 | RetVal = It->second; |
| 152 | return true; |
| 153 | } |
| 154 | |
| 155 | static bool globalHasNVVMAnnotation(const Value &V, StringRef Prop) { |
| 156 | if (const auto *GV = dyn_cast<GlobalValue>(Val: &V)) |
| 157 | if (const auto Annot = findOneNVVMAnnotation(GV, Prop)) { |
| 158 | assert((*Annot == 1) && "Unexpected annotation on a symbol" ); |
| 159 | return true; |
| 160 | } |
| 161 | |
| 162 | return false; |
| 163 | } |
| 164 | |
| 165 | static bool argHasNVVMAnnotation(const Value &Val, StringRef Annotation) { |
| 166 | if (const auto *Arg = dyn_cast<Argument>(Val: &Val)) { |
| 167 | std::vector<unsigned> Annot; |
| 168 | if (findAllNVVMAnnotation(GV: Arg->getParent(), Prop: Annotation, RetVal&: Annot) && |
| 169 | is_contained(Range&: Annot, Element: Arg->getArgNo())) |
| 170 | return true; |
| 171 | } |
| 172 | return false; |
| 173 | } |
| 174 | |
| 175 | static std::optional<unsigned> getFnAttrParsedInt(const Function &F, |
| 176 | StringRef Attr) { |
| 177 | return F.hasFnAttribute(Kind: Attr) |
| 178 | ? std::optional(F.getFnAttributeAsParsedInteger(Kind: Attr)) |
| 179 | : std::nullopt; |
| 180 | } |
| 181 | |
| 182 | static SmallVector<unsigned, 3> getFnAttrParsedVector(const Function &F, |
| 183 | StringRef Attr) { |
| 184 | SmallVector<unsigned, 3> V; |
| 185 | auto &Ctx = F.getContext(); |
| 186 | |
| 187 | if (F.hasFnAttribute(Kind: Attr)) { |
| 188 | // We expect the attribute value to be of the form "x[,y[,z]]", where x, y, |
| 189 | // and z are unsigned values. |
| 190 | StringRef S = F.getFnAttribute(Kind: Attr).getValueAsString(); |
| 191 | for (unsigned I = 0; I < 3 && !S.empty(); I++) { |
| 192 | auto [First, Rest] = S.split(Separator: "," ); |
| 193 | unsigned IntVal; |
| 194 | if (First.trim().getAsInteger(Radix: 0, Result&: IntVal)) |
| 195 | Ctx.emitError(ErrorStr: "can't parse integer attribute " + First + " in " + Attr); |
| 196 | |
| 197 | V.push_back(Elt: IntVal); |
| 198 | S = Rest; |
| 199 | } |
| 200 | } |
| 201 | return V; |
| 202 | } |
| 203 | |
| 204 | static std::optional<uint64_t> getVectorProduct(ArrayRef<unsigned> V) { |
| 205 | if (V.empty()) |
| 206 | return std::nullopt; |
| 207 | |
| 208 | return std::accumulate(first: V.begin(), last: V.end(), init: uint64_t(1), |
| 209 | binary_op: std::multiplies<uint64_t>{}); |
| 210 | } |
| 211 | |
| 212 | PTXOpaqueType getPTXOpaqueType(const GlobalVariable &GV) { |
| 213 | if (findOneNVVMAnnotation(GV: &GV, Prop: NVVMMetadata::Texture)) |
| 214 | return PTXOpaqueType::Texture; |
| 215 | if (findOneNVVMAnnotation(GV: &GV, Prop: NVVMMetadata::Surface)) |
| 216 | return PTXOpaqueType::Surface; |
| 217 | if (findOneNVVMAnnotation(GV: &GV, Prop: NVVMMetadata::Sampler)) |
| 218 | return PTXOpaqueType::Sampler; |
| 219 | return PTXOpaqueType::None; |
| 220 | } |
| 221 | |
| 222 | PTXOpaqueType getPTXOpaqueType(const Argument &Arg) { |
| 223 | if (argHasNVVMAnnotation(Val: Arg, Annotation: NVVMMetadata::Sampler)) |
| 224 | return PTXOpaqueType::Sampler; |
| 225 | if (argHasNVVMAnnotation(Val: Arg, Annotation: NVVMMetadata::ReadOnlyImage)) |
| 226 | return PTXOpaqueType::Texture; |
| 227 | if (argHasNVVMAnnotation(Val: Arg, Annotation: NVVMMetadata::WriteOnlyImage) || |
| 228 | argHasNVVMAnnotation(Val: Arg, Annotation: NVVMMetadata::ReadWriteImage)) |
| 229 | return PTXOpaqueType::Surface; |
| 230 | return PTXOpaqueType::None; |
| 231 | } |
| 232 | |
| 233 | PTXOpaqueType getPTXOpaqueType(const Value &V) { |
| 234 | if (const auto *GV = dyn_cast<GlobalVariable>(Val: &V)) |
| 235 | return getPTXOpaqueType(GV: *GV); |
| 236 | if (const auto *Arg = dyn_cast<Argument>(Val: &V)) |
| 237 | return getPTXOpaqueType(Arg: *Arg); |
| 238 | return PTXOpaqueType::None; |
| 239 | } |
| 240 | |
| 241 | bool isManaged(const Value &V) { |
| 242 | return globalHasNVVMAnnotation(V, Prop: NVVMMetadata::Managed); |
| 243 | } |
| 244 | |
| 245 | SmallVector<unsigned, 3> getMaxNTID(const Function &F) { |
| 246 | return getFnAttrParsedVector(F, Attr: NVVMAttr::MaxNTID); |
| 247 | } |
| 248 | |
| 249 | SmallVector<unsigned, 3> getReqNTID(const Function &F) { |
| 250 | return getFnAttrParsedVector(F, Attr: NVVMAttr::ReqNTID); |
| 251 | } |
| 252 | |
| 253 | SmallVector<unsigned, 3> getClusterDim(const Function &F) { |
| 254 | return getFnAttrParsedVector(F, Attr: NVVMAttr::ClusterDim); |
| 255 | } |
| 256 | |
| 257 | std::optional<uint64_t> getOverallMaxNTID(const Function &F) { |
| 258 | // Note: The semantics here are a bit strange. The PTX ISA states the |
| 259 | // following (11.4.2. Performance-Tuning Directives: .maxntid): |
| 260 | // |
| 261 | // Note that this directive guarantees that the total number of threads does |
| 262 | // not exceed the maximum, but does not guarantee that the limit in any |
| 263 | // particular dimension is not exceeded. |
| 264 | return getVectorProduct(V: getMaxNTID(F)); |
| 265 | } |
| 266 | |
| 267 | std::optional<uint64_t> getOverallReqNTID(const Function &F) { |
| 268 | // Note: The semantics here are a bit strange. See getOverallMaxNTID. |
| 269 | return getVectorProduct(V: getReqNTID(F)); |
| 270 | } |
| 271 | |
| 272 | std::optional<uint64_t> getOverallClusterRank(const Function &F) { |
| 273 | // maxclusterrank and cluster_dim are mutually exclusive. |
| 274 | if (const auto ClusterRank = getMaxClusterRank(F)) |
| 275 | return ClusterRank; |
| 276 | |
| 277 | // Note: The semantics here are a bit strange. See getOverallMaxNTID. |
| 278 | return getVectorProduct(V: getClusterDim(F)); |
| 279 | } |
| 280 | |
| 281 | std::optional<unsigned> getMaxClusterRank(const Function &F) { |
| 282 | return getFnAttrParsedInt(F, Attr: NVVMAttr::MaxClusterRank); |
| 283 | } |
| 284 | |
| 285 | std::optional<unsigned> getMinCTASm(const Function &F) { |
| 286 | return getFnAttrParsedInt(F, Attr: NVVMAttr::MinCTASm); |
| 287 | } |
| 288 | |
| 289 | std::optional<unsigned> getMaxNReg(const Function &F) { |
| 290 | return getFnAttrParsedInt(F, Attr: NVVMAttr::MaxNReg); |
| 291 | } |
| 292 | |
| 293 | bool hasBlocksAreClusters(const Function &F) { |
| 294 | return F.hasFnAttribute(Kind: NVVMAttr::BlocksAreClusters); |
| 295 | } |
| 296 | |
| 297 | bool isParamGridConstant(const Argument &Arg) { |
| 298 | assert(isKernelFunction(*Arg.getParent()) && |
| 299 | "only kernel arguments can be grid_constant" ); |
| 300 | |
| 301 | if (!Arg.hasByValAttr()) |
| 302 | return false; |
| 303 | |
| 304 | // Lowering an argument as a grid_constant violates the byval semantics (and |
| 305 | // the C++ API) by reusing the same memory location for the argument across |
| 306 | // multiple threads. If an argument doesn't read memory and its address is not |
| 307 | // captured (its address is not compared with any value), then the tweak of |
| 308 | // the C++ API and byval semantics is unobservable by the program and we can |
| 309 | // lower the arg as a grid_constant. |
| 310 | if (Arg.onlyReadsMemory()) { |
| 311 | const auto CI = Arg.getAttributes().getCaptureInfo(); |
| 312 | if (!capturesAddress(CC: CI) && !capturesFullProvenance(CC: CI)) |
| 313 | return true; |
| 314 | } |
| 315 | |
| 316 | // "grid_constant" counts argument indices starting from 1 |
| 317 | return Arg.hasAttribute(Kind: NVVMAttr::GridConstant); |
| 318 | } |
| 319 | |
| 320 | MaybeAlign getAlign(const CallInst &I, unsigned Index) { |
| 321 | // First check the alignstack metadata. |
| 322 | if (MaybeAlign StackAlign = |
| 323 | I.getAttributes().getAttributes(Index).getStackAlignment()) |
| 324 | return StackAlign; |
| 325 | |
| 326 | // If that is missing, check the legacy nvvm metadata. |
| 327 | if (MDNode *AlignNode = I.getMetadata(Kind: "callalign" )) { |
| 328 | for (int I = 0, N = AlignNode->getNumOperands(); I < N; I++) { |
| 329 | if (const auto *CI = |
| 330 | mdconst::dyn_extract<ConstantInt>(MD: AlignNode->getOperand(I))) { |
| 331 | unsigned V = CI->getZExtValue(); |
| 332 | if ((V >> 16) == Index) |
| 333 | return Align(V & 0xFFFF); |
| 334 | if ((V >> 16) > Index) |
| 335 | return std::nullopt; |
| 336 | } |
| 337 | } |
| 338 | } |
| 339 | return std::nullopt; |
| 340 | } |
| 341 | |
| 342 | } // namespace llvm |
| 343 | |