[RISCV] Add function that check extension name with version

[BeMg]

Summary

Check whether a extension string with version is valid, and get the targetfeature from it.

New functions be used in RISCVISAInfo for https://reviews.llvm.org/D151730.

[BeMg]

update

[kito-cheng]

Could you add some testcase in llvm/unittests/Support/RISCVISAInfoTest.cpp?

[craig.topper]

llvm/lib/Support/RISCVISAInfo.cpp

1200

This function is poorly named. Shouldn't it be findLastNonVersionCharacter?

1201

I think writing as an assignment is more common

StringRef Name = Ext.substr(0, Pos);
StringRef Vers = Ext.substr(Pos);

[craig.topper]

llvm/lib/Support/RISCVISAInfo.cpp

1200

Or it should be findFirstVersionCharacter and take care of the +1

[BeMg]

1. update function body
2. Add unit test
3. findFirstNonVersionCharacter -> findLastNonVersionCharacter

[craig.topper]

llvm/lib/Support/RISCVISAInfo.cpp

184

Can you make a separate patch for renaming this?

[BeMg]

llvm/lib/Support/RISCVISAInfo.cpp

184

OK

[BeMg]

rename findLastNonVersionCharacter with https://reviews.llvm.org/D152506

[BeMg]

rebase

[craig.topper]

llvm/lib/Support/RISCVISAInfo.cpp

1222

I don't think Vers needs to exist. It's only used for !Vers.empty() which is equivalent to Pos != Ext.size()

llvm/unittests/Support/RISCVISAInfoTest.cpp

490

Use EXPECT_TRUE and EXPECT_FALSE

[BeMg]

1. Remove Vers and use Pos != Ext.size()
2. Update testcase by using EXPECT_TRUE|EXPECT_FALSE

[BeMg]

1. Let isSupportedExtension support extension name with version
2. Remove isSupportedExtensionWithVersion

[BeMg]

Revert previous update

[BeMg]

And rename getTargetFeatureFromOneExt into getTargetFeatureForExtension

[BeMg]

rebase

[BeMg]

rebase and update testcase

[BeMg]

rebase and update testcase

[BeMg]

revert last patch because it's wrong one

[craig.topper]

LGTM

llvm/include/llvm/Support/RISCVISAInfo.h

public:
 
   bool hasExtension(StringRef Ext) const;
   std::string toString() const;
   std::vector<std::string> toFeatureVector() const;
   StringRef computeDefaultABI() const;
 
   static bool isSupportedExtensionFeature(StringRef Ext);
   static bool isSupportedExtension(StringRef Ext);
+  static bool isSupportedExtensionWithVersion(StringRef Ext);
   static bool isSupportedExtension(StringRef Ext, unsigned MajorVersion,
                                    unsigned MinorVersion);
   static llvm::Expected<std::unique_ptr<RISCVISAInfo>>
   postProcessAndChecking(std::unique_ptr<RISCVISAInfo> &&ISAInfo);
+  static std::string getTargetFeatureFromOneExt(StringRef Ext);
 
 private:
   RISCVISAInfo(unsigned XLen)
       : XLen(XLen), FLen(0), MinVLen(0), MaxELen(0), MaxELenFp(0) {}
 
   unsigned XLen;
   unsigned FLen;
   unsigned MinVLen;

llvm/lib/Support/RISCVISAInfo.cpp

 
 // This function finds the first character that doesn't belong to a version
 // (e.g. zba1p0 is extension 'zba' of version '1p0'). So the function will
 // consume [0-9]*p[0-9]* starting from the backward. An extension name will not
 // end with a digit or the letter 'p', so this function will parse correctly.
 // NOTE: This function is NOT able to take empty strings or strings that only
 // have version numbers and no extension name. It assumes the extension name
 // will be at least more than one character.
-static size_t findFirstNonVersionCharacter(StringRef Ext) {
+static size_t findLastNonVersionCharacter(StringRef Ext) {
   assert(!Ext.empty() &&
          "Already guarded by if-statement in ::parseArchString");
 
   int Pos = Ext.size() - 1;
   while (Pos > 0 && isDigit(Ext[Pos]))
     Pos--;
   if (Pos > 0 && Ext[Pos] == 'p' && isDigit(Ext[Pos - 1])) {
     Pos--;
RISCVISAInfo::parseArchString(StringRef Arch, bool EnableExperimentalExtension,
   if (Split.size() > 1 || Split[0] != "") {
     for (StringRef Ext : Split) {
       if (Ext.empty())
         return createStringError(errc::invalid_argument,
                                  "extension name missing after separator '_'");
 
       StringRef Type = getExtensionType(Ext);
       StringRef Desc = getExtensionTypeDesc(Ext);
-      auto Pos = findFirstNonVersionCharacter(Ext) + 1;
+      auto Pos = findLastNonVersionCharacter(Ext) + 1;
       StringRef Name(Ext.substr(0, Pos));
       StringRef Vers(Ext.substr(Pos));
 
       if (Type.empty()) {
         if (IgnoreUnknown)
           continue;
         return createStringError(errc::invalid_argument,
                                  "invalid extension prefix '" + Ext + "'");
if (XLen == 32) {
     if (hasExtension("d"))
       return "lp64d";
     if (hasExtension("e"))
       return "lp64e";
     return "lp64";
   }
   llvm_unreachable("Invalid XLEN");
 }
+
+bool RISCVISAInfo::isSupportedExtensionWithVersion(StringRef Ext) {
+  if (Ext.empty())
+    return false;
+
+  auto Pos = findLastNonVersionCharacter(Ext) + 1;
+  StringRef Name = Ext.substr(0, Pos);
+  StringRef Vers = Ext.substr(Pos);
+  if (Vers.empty())
+    return false;
+
+  unsigned Major, Minor, ConsumeLength;
+  if (auto E = getExtensionVersion(Name, Vers, Major, Minor, ConsumeLength,
+                                   true, true)) {
+    consumeError(std::move(E));
+    return false;
+  }
+
+  return true;
+}
+
+std::string RISCVISAInfo::getTargetFeatureFromOneExt(StringRef Ext) {
+  if (Ext.empty())
+    return std::string();
+
+  auto Pos = findLastNonVersionCharacter(Ext) + 1;
+  StringRef Name = Ext.substr(0, Pos);
+  StringRef Vers = Ext.substr(Pos);
+
+  if (!Vers.empty() && !isSupportedExtensionWithVersion(Ext))
+    return std::string();
+
+  if (!isSupportedExtension(Name))
+    return std::string();
+
+  return isExperimentalExtension(Name) ? "experimental-" + Name.str()
+                                       : Name.str();
+}

llvm/unittests/Support/RISCVISAInfoTest.cpp

TEST(OrderedExtensionMap, ExtensionsAreCorrectlyOrdered) {
   for (const auto &Ext : Exts)
     ExtNames.push_back(Ext.first);
 
   // FIXME: 'l' and 'y' should be ordered after 'i', 'm', 'c'.
   EXPECT_THAT(ExtNames,
               ElementsAre("i", "m", "l", "c", "y", "zicsr", "zmfoo", "zfinx",
                            "zzfoo", "sbar", "sfoo", "xbar", "xfoo"));
 }
+
+TEST(isSupportedExtensionWithVersion, AcceptsSingleExtensionWithVersion) {
+  EXPECT_EQ(RISCVISAInfo::isSupportedExtensionWithVersion("zbb1p0"), true);
+  EXPECT_EQ(RISCVISAInfo::isSupportedExtensionWithVersion("zbb"), false);
+  EXPECT_EQ(RISCVISAInfo::isSupportedExtensionWithVersion("zfoo1p0"), false);
+  EXPECT_EQ(RISCVISAInfo::isSupportedExtensionWithVersion("zfoo"), false);
+  EXPECT_EQ(RISCVISAInfo::isSupportedExtensionWithVersion(""), false);
+  EXPECT_EQ(RISCVISAInfo::isSupportedExtensionWithVersion("c2p0zbb1p0"), false);
+}
+
+TEST(getTargetFeatureFromOneExt, RetrieveTargetFeatureFromOneExt) {
+  EXPECT_EQ(RISCVISAInfo::getTargetFeatureFromOneExt("zbb"), "zbb");
+  EXPECT_EQ(RISCVISAInfo::getTargetFeatureFromOneExt("zihintntl0p2"),
+            "experimental-zihintntl");
+  EXPECT_EQ(RISCVISAInfo::getTargetFeatureFromOneExt("zihintntl"),
+            "experimental-zihintntl");
+  EXPECT_EQ(RISCVISAInfo::getTargetFeatureFromOneExt("zihintntl1234p4321"), "");
+  EXPECT_EQ(RISCVISAInfo::getTargetFeatureFromOneExt("zfoo"), "");
+  EXPECT_EQ(RISCVISAInfo::getTargetFeatureFromOneExt(""), "");
+  EXPECT_EQ(RISCVISAInfo::getTargetFeatureFromOneExt("zbbzihintntl"), "");
+}