[ELF] - Linkerscript: support VERSION command.

[grimar]

Summary

From PR28926 description:

Symbol version information can be provided via the --version-script command line option, or can be embedded into a VERSION block in a linker script:

VERSION { version-script-commands }

FreeBSD uses this for the Linux ABI vDSO support, presumably from the same use in the Linux kernel.

Reference: https://www.sourceware.org/binutils/docs/ld/VERSION.html#VERSION

Patch implements this feature.

[ruiu]

I don't see a reason to keep LinkerScript and SymbolListFile being separated. I'd think we should move code from SymbolListFile to LinkerScript.

[grimar]

In D23609#518431, @ruiu wrote:

I don't see a reason to keep LinkerScript and SymbolListFile being separated. I'd think we should move code from SymbolListFile to LinkerScript.

D23660 do that change.

[grimar]

• Reimplemented to use head changes.

[ruiu]

ELF/LinkerScript.cpp

716–719

Don't check this error inside this function. Instead, check this error outside of this function. IsCommand should be removed.

[grimar]

• Addressed review comments.

[ruiu]

What I was trying to say is to remove IsCommand parameter and do not rely on return value to determine what we are parsing. We should know what we are doing without that value. If you need to read X and Y, define readX and readY, instead of readZ(bool IsX) or something like that.

[grimar]

• Addressed review comments.

[grimar]

In D23609#530418, @ruiu wrote:

What I was trying to say is to remove IsCommand parameter and do not rely on return value to determine what we are parsing. We should know what we are doing without that value. If you need to read X and Y, define readX and readY, instead of readZ(bool IsX) or something like that.

Ok, I see. I updated the diff. Updated code looks close to what we had for parsing script file, but it has at least 2 differences atm, so I am not sure it is possible to extract common part (and if we want it here).

[ruiu]

Obviously, this patch contains duplicate code which does not look good. Let me create a patch based on this to describe what I was trying to say.

[grimar]

In D23609#530513, @ruiu wrote:

Obviously, this patch contains duplicate code which does not look good. Let me create a patch based on this to describe what I was trying to say.

Actually I think I completely understand what you want to see. Something like below probably.
There are still duplications of messages, and only a part of code can be extracted here with no pain.
I can continue working on that tomorrow if you think it is the right direction.

void ScriptParser::readVersionScript() {
  StringRef Msg = "anonymous version definition is used in "
                  "combination with other version definitions";
  if (skip("{")) {
    readVersionDeclaration("");
    if (!atEOF())
      setError(Msg);
    return;
  }

  while (!atEOF() && !Error)
    readNamedVersion();
}

void ScriptParser::readNamedVersion() {
  StringRef Msg = "anonymous version definition is used in "
                  "combination with other version definitions";
  StringRef VerStr = next();
  if (VerStr == "{") {
    setError(Msg);
    return;
  }
  expect("{");
  readVersionDeclaration(VerStr);
}

void ScriptParser::readVersion() {
  expect("{");
  if (skip("{")) {
    readVersionDeclaration("");
    expect("}");
    return;
  }

  while (!Error) {
    readNamedVersion();
    if (peek() == "}")
      break;
  }
  expect("}");
}

[grimar]

Closing this as r280284 was committed instead.

ELF/LinkerScript.cpp

private:
   void readInclude();
   void readNothing() {}
   void readOutput();
   void readOutputArch();
   void readOutputFormat();
   void readPhdrs();
   void readSearchDir();
   void readSections();
+  void readVersion();
 
   SymbolAssignment *readAssignment(StringRef Name);
   OutputSectionCommand *readOutputSectionDescription(StringRef OutSec);
   std::vector<uint8_t> readOutputSectionFiller();
   std::vector<StringRef> readOutputSectionPhdrs();
   InputSectionDescription *readInputSectionDescription(StringRef Tok);
   std::vector<StringRef> readInputFilePatterns();
   InputSectionDescription *readInputSectionRules(StringRef FilePattern);
   unsigned readPhdrType();
   SortKind readSortKind();
   SymbolAssignment *readProvideHidden(bool Provide, bool Hidden);
   SymbolAssignment *readProvideOrAssignment(StringRef Tok);
   void readSort();
   Expr readAssert();
 
   Expr readExpr();
   Expr readExpr1(Expr Lhs, int MinPrec);
   Expr readPrimary();
   Expr readTernary(Expr Cond);
   Expr readParenExpr();
 
   // For parsing version script.
   void readExtern(std::vector<SymbolVersion> *Globals);
-  void readVersion(StringRef VerStr);
+  void readVersionDeclaration(StringRef VerStr);
   void readGlobal(StringRef VerStr);
   void readLocal();
 
   const static StringMap<Handler> Cmd;
   ScriptConfiguration &Opt = *ScriptConfig;
   StringSaver Saver = {ScriptConfig->Alloc};
   bool IsUnderSysroot;
 };
 
 const StringMap<elf::ScriptParser::Handler> elf::ScriptParser::Cmd = {
     {"ENTRY", &ScriptParser::readEntry},
     {"EXTERN", &ScriptParser::readExtern},
     {"GROUP", &ScriptParser::readGroup},
     {"INCLUDE", &ScriptParser::readInclude},
     {"INPUT", &ScriptParser::readGroup},
     {"OUTPUT", &ScriptParser::readOutput},
     {"OUTPUT_ARCH", &ScriptParser::readOutputArch},
     {"OUTPUT_FORMAT", &ScriptParser::readOutputFormat},
     {"PHDRS", &ScriptParser::readPhdrs},
     {"SEARCH_DIR", &ScriptParser::readSearchDir},
     {"SECTIONS", &ScriptParser::readSections},
+    {"VERSION", &ScriptParser::readVersion},
     {";", &ScriptParser::readNothing}};
 
 void ScriptParser::readVersionScript() {
   StringRef Msg = "anonymous version definition is used in "
                   "combination with other version definitions";
   if (skip("{")) {
-    readVersion("");
+    readVersionDeclaration("");
     if (!atEOF())
       setError(Msg);
     return;
   }
 
   while (!atEOF() && !Error) {
     StringRef VerStr = next();
     if (VerStr == "{") {
       setError(Msg);
       return;
     }
     expect("{");
-    readVersion(VerStr);
+    readVersionDeclaration(VerStr);
   }
 }
 
+void ScriptParser::readVersion() {
+  StringRef Msg = "anonymous version definition is used in "
+                  "combination with other version definitions";
+  expect("{");
+  if (skip("{")) {
+    readVersionDeclaration("");
+    expect("}");
+    return;
+  }
+
+  while (!Error) {
+    if (peek() == "}")
+      break;
+    StringRef VerStr = next();
+    if (VerStr == "{") {
+      setError(Msg);
+      break;
+    }
+    expect("{");
+    readVersionDeclaration(VerStr);
+  }
+  expect("}");
+}
+
 void ScriptParser::readLinkerScript() {
   while (!atEOF()) {
     StringRef Tok = next();
     if (Handler Fn = Cmd.lookup(Tok)) {
       (this->*Fn)();
     } else if (SymbolAssignment *Cmd = readProvideOrAssignment(Tok)) {
       if (Opt.HasContents)
         Opt.Commands.emplace_back(Cmd);
unsigned ScriptParser::readPhdrType() {
 
   if (Ret == (unsigned)-1) {
     setError("invalid program header type: " + Tok);
     return PT_NULL;
   }
   return Ret;
 }
 
-void ScriptParser::readVersion(StringRef VerStr) {
+void ScriptParser::readVersionDeclaration(StringRef VerStr) {
   // Identifiers start at 2 because 0 and 1 are reserved
   // for VER_NDX_LOCAL and VER_NDX_GLOBAL constants.
   size_t VersionId = Config->VersionDefinitions.size() + 2;
   Config->VersionDefinitions.push_back({VerStr, VersionId});
 
   if (skip("global:") || peek() != "local:")
     readGlobal(VerStr);
   if (skip("local:"))

test/ELF/verdef.s

 # MAIN-NEXT:       Version: 4
 # MAIN-NEXT:       Name: c@LIBSAMPLE_3.0
 # MAIN-NEXT:     }
 # MAIN-NEXT:   ]
 # MAIN-NEXT: }
 # MAIN-NEXT: SHT_GNU_verdef {
 # MAIN-NEXT: }
 
+# RUN: echo "VERSION {       \
+# RUN:       LIBSAMPLE_1.0{  \
+# RUN:          global: a;   \
+# RUN:          local: *; }; \
+# RUN:       LIBSAMPLE_2.0{  \
+# RUN:          global: b;   \
+# RUN:          local: *; }; \
+# RUN:       LIBSAMPLE_3.0{  \
+# RUN:          global: c;   \
+# RUN:          local: *; }; }" > %t.script
+# RUN: ld.lld --script %t.script -shared -soname shared %t.o -o %t2.so
+# RUN: llvm-readobj -V -dyn-symbols %t2.so | FileCheck --check-prefix=DSO %s
+
 .globl a
 .type  a,@function
 a:
 retq
 
 .globl b
 .type  b,@function
 b:
 retq
 
 .globl c
 .type  c,@function
 c:
 retq

test/ELF/version-script.s

 # REQUIRES: x86
 
 # RUN: llvm-mc -filetype=obj -triple=x86_64-unknown-linux %p/Inputs/shared.s -o %t2.o
 # RUN: ld.lld -shared %t2.o -soname shared -o %t2.so
 
 # RUN: echo "{ global: foo1; foo3; local: *; };" > %t.script
 # RUN: llvm-mc -filetype=obj -triple=x86_64-unknown-linux %s -o %t.o
 # RUN: ld.lld --version-script %t.script -shared %t.o %t2.so -o %t.so
 # RUN: llvm-readobj -dyn-symbols %t.so | FileCheck --check-prefix=DSO %s
 
+# RUN: echo "VERSION { { global: foo1; foo3; local: *; }; }" > %t2.script
+# RUN: ld.lld --script %t2.script -shared %t.o %t2.so -o %t2_1.so
+# RUN: llvm-readobj -dyn-symbols %t2_1.so | FileCheck --check-prefix=DSO %s
+
 # RUN: echo "# comment" > %t3.script
 # RUN: echo "{ local: *; # comment" >> %t3.script
 # RUN: echo -n "}; # comment" >> %t3.script
 # RUN: ld.lld --version-script %t3.script -shared %t.o %t2.so -o %t3.so
 # RUN: llvm-readobj -dyn-symbols %t3.so | FileCheck --check-prefix=DSO2 %s
 
 # --version-script filters --dynamic-list.
 # RUN: echo "{ foo1; foo2; };" > %t.list