[AMDGPU] Runtime metadata fixes:

  - Verify that runtime metadata is actually valid runtime metadata when assembling, otherwise we could accept the following when assembling, but ocl runtime will reject it:
    .amdgpu_runtime_metadata
    { amd.MDVersion: [ 2, 1 ], amd.RandomUnknownKey, amd.IsaInfo: ...
  - Make IsaInfo optional, and always emit it.

Differential Revision: https://reviews.llvm.org/D30349

llvm-svn: 296324

5 files changed, 79 insertions, 32 deletions

diff --git a/llvm/lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp b/llvm/lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp
index 0f0e0ab68c9..b2135d518d1 100644
--- a/llvm/lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp
+++ b/llvm/lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp
@@ -789,12 +789,12 @@ public:
       : MCTargetAsmParser(Options, STI), MII(MII), Parser(_Parser) {
     MCAsmParserExtension::Initialize(Parser);
 
-    if (getSTI().getFeatureBits().none()) {
+    if (getFeatureBits().none()) {
       // Set default features.
       copySTI().ToggleFeature("SOUTHERN_ISLANDS");
     }
 
-    setAvailableFeatures(ComputeAvailableFeatures(getSTI().getFeatureBits()));
+    setAvailableFeatures(ComputeAvailableFeatures(getFeatureBits()));
 
     {
       // TODO: make those pre-defined variables read-only.
@@ -802,7 +802,7 @@ public:
       // MCSymbol::isRedefinable is intended for another purpose, and
       // AsmParser::parseDirectiveSet() cannot be specialized for specific target.
       AMDGPU::IsaInfo::IsaVersion ISA =
-          AMDGPU::IsaInfo::getIsaVersion(getSTI().getFeatureBits());
+          AMDGPU::IsaInfo::getIsaVersion(getFeatureBits());
       MCContext &Ctx = getContext();
       MCSymbol *Sym =
           Ctx.getOrCreateSymbol(Twine(".option.machine_version_major"));
@@ -828,7 +828,7 @@ public:
   }
 
   bool hasInv2PiInlineImm() const {
-    return getSTI().getFeatureBits()[AMDGPU::FeatureInv2PiInlineImm];
+    return getFeatureBits()[AMDGPU::FeatureInv2PiInlineImm];
   }
 
   bool hasSGPR102_SGPR103() const {
@@ -850,6 +850,10 @@ public:
     return &MII;
   }
 
+  const FeatureBitset &getFeatureBits() const {
+    return getSTI().getFeatureBits();
+  }
+
   void setForcedEncodingSize(unsigned Size) { ForcedEncodingSize = Size; }
   void setForcedDPP(bool ForceDPP_) { ForcedDPP = ForceDPP_; }
   void setForcedSDWA(bool ForceSDWA_) { ForcedSDWA = ForceSDWA_; }
@@ -1870,7 +1874,7 @@ bool AMDGPUAsmParser::ParseDirectiveHSACodeObjectISA() {
   // targeted GPU.
   if (getLexer().is(AsmToken::EndOfStatement)) {
     AMDGPU::IsaInfo::IsaVersion ISA =
-        AMDGPU::IsaInfo::getIsaVersion(getSTI().getFeatureBits());
+        AMDGPU::IsaInfo::getIsaVersion(getFeatureBits());
     getTargetStreamer().EmitDirectiveHSACodeObjectISA(ISA.Major, ISA.Minor,
                                                       ISA.Stepping,
                                                       "AMD", "AMDGPU");
@@ -1947,7 +1951,8 @@ bool AMDGPUAsmParser::ParseDirectiveRuntimeMetadata() {
 
   MS.flush();
 
-  getTargetStreamer().EmitRuntimeMetadata(Metadata);
+  if (getTargetStreamer().EmitRuntimeMetadata(getFeatureBits(), Metadata))
+    return Error(getParser().getTok().getLoc(), "invalid runtime metadata");
 
   return false;
 }
@@ -1965,7 +1970,7 @@ bool AMDGPUAsmParser::ParseAMDKernelCodeTValue(StringRef ID,
 
 bool AMDGPUAsmParser::ParseDirectiveAMDKernelCodeT() {
   amd_kernel_code_t Header;
-  AMDGPU::initDefaultAMDKernelCodeT(Header, getSTI().getFeatureBits());
+  AMDGPU::initDefaultAMDKernelCodeT(Header, getFeatureBits());
 
   while (true) {
     // Lex EndOfStatement.  This is in a while loop, because lexing a comment
@@ -2459,7 +2464,7 @@ bool AMDGPUAsmParser::parseCnt(int64_t &IntVal) {
     Parser.Lex();
 
   AMDGPU::IsaInfo::IsaVersion ISA =
-      AMDGPU::IsaInfo::getIsaVersion(getSTI().getFeatureBits());
+      AMDGPU::IsaInfo::getIsaVersion(getFeatureBits());
   if (CntName == "vmcnt")
     IntVal = encodeVmcnt(ISA, IntVal, CntVal);
   else if (CntName == "expcnt")
@@ -2475,7 +2480,7 @@ bool AMDGPUAsmParser::parseCnt(int64_t &IntVal) {
 OperandMatchResultTy
 AMDGPUAsmParser::parseSWaitCntOps(OperandVector &Operands) {
   AMDGPU::IsaInfo::IsaVersion ISA =
-      AMDGPU::IsaInfo::getIsaVersion(getSTI().getFeatureBits());
+      AMDGPU::IsaInfo::getIsaVersion(getFeatureBits());
   int64_t Waitcnt = getWaitcntBitMask(ISA);
   SMLoc S = Parser.getTok().getLoc();
 
diff --git a/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPURuntimeMD.cpp b/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPURuntimeMD.cpp
index 9bd8eddb60a..0e87d750f5d 100644
--- a/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPURuntimeMD.cpp
+++ b/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPURuntimeMD.cpp
@@ -38,6 +38,7 @@
 #include <vector>
 
 using namespace llvm;
+using namespace llvm::AMDGPU::IsaInfo;
 using namespace ::AMDGPU::RuntimeMD;
 
 static cl::opt<bool>
@@ -88,7 +89,7 @@ template <> struct MappingTraits<Kernel::Metadata> {
         INVALID_KERNEL_INDEX);
     YamlIO.mapOptional(KeyName::NoPartialWorkGroups, K.NoPartialWorkGroups,
         uint8_t(0));
-    YamlIO.mapRequired(KeyName::Args, K.Args);
+    YamlIO.mapOptional(KeyName::Args, K.Args);
   }
   static const bool flow = true;
 };
@@ -116,7 +117,7 @@ template <> struct MappingTraits<IsaInfo::Metadata> {
 template <> struct MappingTraits<Program::Metadata> {
   static void mapping(IO &YamlIO, Program::Metadata &Prog) {
     YamlIO.mapRequired(KeyName::MDVersion, Prog.MDVersionSeq);
-    YamlIO.mapRequired(KeyName::IsaInfo, Prog.IsaInfo);
+    YamlIO.mapOptional(KeyName::IsaInfo, Prog.IsaInfo);
     YamlIO.mapOptional(KeyName::PrintfInfo, Prog.PrintfInfo);
     YamlIO.mapOptional(KeyName::Kernels, Prog.Kernels);
   }
@@ -375,6 +376,20 @@ static Kernel::Metadata getRuntimeMDForKernel(const Function &F) {
   return Kernel;
 }
 
+static void getIsaInfo(const FeatureBitset &Features, IsaInfo::Metadata &IIM) {
+  IIM.WavefrontSize = getWavefrontSize(Features);
+  IIM.LocalMemorySize = getLocalMemorySize(Features);
+  IIM.EUsPerCU = getEUsPerCU(Features);
+  IIM.MaxWavesPerEU = getMaxWavesPerEU(Features);
+  IIM.MaxFlatWorkGroupSize = getMaxFlatWorkGroupSize(Features);
+  IIM.SGPRAllocGranule = getSGPRAllocGranule(Features);
+  IIM.TotalNumSGPRs = getTotalNumSGPRs(Features);
+  IIM.AddressableNumSGPRs = getAddressableNumSGPRs(Features);
+  IIM.VGPRAllocGranule = getVGPRAllocGranule(Features);
+  IIM.TotalNumVGPRs = getTotalNumVGPRs(Features);
+  IIM.AddressableNumVGPRs = getAddressableNumVGPRs(Features);
+}
+
 Program::Metadata::Metadata(const std::string &YAML) {
   yaml::Input Input(YAML);
   Input >> *this;
@@ -411,18 +426,7 @@ std::string llvm::getRuntimeMDYAMLString(const FeatureBitset &Features,
   Prog.MDVersionSeq.push_back(MDVersion);
   Prog.MDVersionSeq.push_back(MDRevision);
 
-  IsaInfo::Metadata &IIM = Prog.IsaInfo;
-  IIM.WavefrontSize = AMDGPU::IsaInfo::getWavefrontSize(Features);
-  IIM.LocalMemorySize = AMDGPU::IsaInfo::getLocalMemorySize(Features);
-  IIM.EUsPerCU = AMDGPU::IsaInfo::getEUsPerCU(Features);
-  IIM.MaxWavesPerEU = AMDGPU::IsaInfo::getMaxWavesPerEU(Features);
-  IIM.MaxFlatWorkGroupSize = AMDGPU::IsaInfo::getMaxFlatWorkGroupSize(Features);
-  IIM.SGPRAllocGranule = AMDGPU::IsaInfo::getSGPRAllocGranule(Features);
-  IIM.TotalNumSGPRs = AMDGPU::IsaInfo::getTotalNumSGPRs(Features);
-  IIM.AddressableNumSGPRs = AMDGPU::IsaInfo::getAddressableNumSGPRs(Features);
-  IIM.VGPRAllocGranule = AMDGPU::IsaInfo::getVGPRAllocGranule(Features);
-  IIM.TotalNumVGPRs = AMDGPU::IsaInfo::getTotalNumVGPRs(Features);
-  IIM.AddressableNumVGPRs = AMDGPU::IsaInfo::getAddressableNumVGPRs(Features);
+  getIsaInfo(Features, Prog.IsaInfo);
 
   // Set PrintfInfo.
   if (auto MD = M.getNamedMetadata("llvm.printf.fmts")) {
@@ -451,3 +455,16 @@ std::string llvm::getRuntimeMDYAMLString(const FeatureBitset &Features,
 
   return YAML;
 }
+
+ErrorOr<std::string> llvm::getRuntimeMDYAMLString(const FeatureBitset &Features,
+                                                  StringRef YAML) {
+  Program::Metadata Prog;
+  yaml::Input Input(YAML);
+  Input >> Prog;
+
+  getIsaInfo(Features, Prog.IsaInfo);
+
+  if (Input.error())
+    return Input.error();
+  return Prog.toYAML();
+}
diff --git a/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPURuntimeMD.h b/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPURuntimeMD.h
index 8a73b3d6426..2c1acdca797 100644
--- a/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPURuntimeMD.h
+++ b/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPURuntimeMD.h
@@ -14,15 +14,20 @@
 #ifndef LLVM_LIB_TARGET_AMDGPU_MCTARGETDESC_AMDGPURUNTIMEMD_H
 #define LLVM_LIB_TARGET_AMDGPU_MCTARGETDESC_AMDGPURUNTIMEMD_H
 
+#include "llvm/Support/ErrorOr.h"
 #include <string>
 
 namespace llvm {
 class FeatureBitset;
 class Module;
 
-// Get runtime metadata as YAML string.
+/// \returns Runtime metadata as YAML string.
 std::string getRuntimeMDYAMLString(const FeatureBitset &Features,
                                    const Module &M);
 
+/// \returns \p YAML if \p YAML is valid runtime metadata, error otherwise.
+ErrorOr<std::string> getRuntimeMDYAMLString(const FeatureBitset &Features,
+                                            StringRef YAML);
+
 }
 #endif
diff --git a/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.cpp b/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.cpp
index 5935c6a49f0..fc07225eebe 100644
--- a/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.cpp
+++ b/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.cpp
@@ -100,10 +100,17 @@ void AMDGPUTargetAsmStreamer::EmitRuntimeMetadata(const FeatureBitset &Features,
   OS << "\n\t.end_amdgpu_runtime_metadata\n";
 }
 
-void AMDGPUTargetAsmStreamer::EmitRuntimeMetadata(StringRef Metadata) {
+bool AMDGPUTargetAsmStreamer::EmitRuntimeMetadata(const FeatureBitset &Features,
+                                                  StringRef Metadata) {
+  auto VerifiedMetadata = getRuntimeMDYAMLString(Features, Metadata);
+  if (!VerifiedMetadata)
+    return true;
+
   OS << "\t.amdgpu_runtime_metadata";
-  OS << Metadata;
+  OS << VerifiedMetadata.get();
   OS << "\t.end_amdgpu_runtime_metadata\n";
+
+  return false;
 }
 
 //===----------------------------------------------------------------------===//
@@ -216,7 +223,12 @@ void AMDGPUTargetELFStreamer::EmitAMDGPUHsaProgramScopeGlobal(
   Symbol->setBinding(ELF::STB_GLOBAL);
 }
 
-void AMDGPUTargetELFStreamer::EmitRuntimeMetadata(StringRef Metadata) {
+bool AMDGPUTargetELFStreamer::EmitRuntimeMetadata(const FeatureBitset &Features,
+                                                  StringRef Metadata) {
+  auto VerifiedMetadata = getRuntimeMDYAMLString(Features, Metadata);
+  if (!VerifiedMetadata)
+    return true;
+
   // Create two labels to mark the beginning and end of the desc field
   // and a MCExpr to calculate the size of the desc field.
   auto &Context = getContext();
@@ -231,13 +243,15 @@ void AMDGPUTargetELFStreamer::EmitRuntimeMetadata(StringRef Metadata) {
     PT_NOTE::NT_AMDGPU_HSA_RUNTIME_METADATA,
     [&](MCELFStreamer &OS) {
       OS.EmitLabel(DescBegin);
-      OS.EmitBytes(Metadata);
+      OS.EmitBytes(VerifiedMetadata.get());
       OS.EmitLabel(DescEnd);
     }
   );
+
+  return false;
 }
 
 void AMDGPUTargetELFStreamer::EmitRuntimeMetadata(const FeatureBitset &Features,
                                                   const Module &M) {
-  EmitRuntimeMetadata(getRuntimeMDYAMLString(Features, M));
+  EmitRuntimeMetadata(Features, getRuntimeMDYAMLString(Features, M));
 }
diff --git a/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.h b/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.h
index 8504d1a18cb..20d7bc113cf 100644
--- a/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.h
+++ b/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.h
@@ -50,7 +50,9 @@ public:
   virtual void EmitRuntimeMetadata(const FeatureBitset &Features,
                                    const Module &M) = 0;
 
-  virtual void EmitRuntimeMetadata(StringRef Metadata) = 0;
+  /// \returns False on success, true on failure.
+  virtual bool EmitRuntimeMetadata(const FeatureBitset &Features,
+                                   StringRef Metadata) = 0;
 };
 
 class AMDGPUTargetAsmStreamer : public AMDGPUTargetStreamer {
@@ -75,7 +77,9 @@ public:
   void EmitRuntimeMetadata(const FeatureBitset &Features,
                            const Module &M) override;
 
-  void EmitRuntimeMetadata(StringRef Metadata) override;
+  /// \returns False on success, true on failure.
+  bool EmitRuntimeMetadata(const FeatureBitset &Features,
+                           StringRef Metadata) override;
 };
 
 class AMDGPUTargetELFStreamer : public AMDGPUTargetStreamer {
@@ -107,7 +111,9 @@ public:
   void EmitRuntimeMetadata(const FeatureBitset &Features,
                            const Module &M) override;
 
-  void EmitRuntimeMetadata(StringRef Metadata) override;
+  /// \returns False on success, true on failure.
+  bool EmitRuntimeMetadata(const FeatureBitset &Features,
+                           StringRef Metadata) override;
 };
 
 }