[BOLT] Add BinaryContext::IsStripped

[nhuhuan]

Summary

Determine stripped status of a binary based on .symtab

Test Plan:

ninja check-bolt

[Amir]

LGTM, but please drop "[NFC]" from the title.
@maksfb: can you please take a look if it makes sense to you?

[yota9]

LGTM, also simple test would be appreciated (e.g. by adding BOLT-INFO about the presumption that the binary was stripped).

[ayermolo]

Can you add a test?

[nhuhuan]

Print strip status. Add a test.

[Amir]

I'm requesting changes because the user-facing message needs to be updated. Please also update the test accordingly.

bolt/lib/Rewrite/RewriteInstance.cpp

1645

bolt/test/X86/is-strip.s

3

Please update the summary.

7

I would prefer reusing some existing input source that doesn't require system-linux to make this test as general as possible.

[nhuhuan]

Address feedbacks.

[Amir]

bolt/lib/Rewrite/RewriteInstance.cpp

1645

clang-format? (I can't make the right format from the inline comment box)

[nhuhuan]

Clang-formatted.

[thakis]

bolt/test/X86/is-strip.s

3

This doesn't pass on systems were the system linker isn't an ELF linker, e.g. macOS:

+ : 'RUN: at line 3'
+ /Users/thakis/src/llvm-build-runtimes2/bin/clang++ /Users/thakis/src/llvm-project/bolt/test/X86/Inputs/linenumber.cpp -o /Users/thakis/src/llvm-build-runtimes2/tools/bolt/test/X86/Output/is-strip.s.tmp -Wl,-q
ld: unknown option: -q
clang-16: error: linker command failed with exit code 1 (use -v to see invocation)

This seems to fix it:

-# RUN: %clang++ %p/Inputs/linenumber.cpp -o %t -Wl,-q
+# RUN: %clang++ %cflags %p/Inputs/linenumber.cpp -o %t -Wl,-q

Is that the right fix?

[thakis]

bolt/test/X86/is-strip.s

3

Many other tests pass %cflags too at least, so that's probably correct. Uploaded D138306 for this.

bolt/include/bolt/Core/BinaryContext.h

public:
   /// Symbolic disassembler.
   std::unique_ptr<MCDisassembler> SymbolicDisAsm;
 
   std::unique_ptr<MCAsmBackend> MAB;
 
   /// Indicates if relocations are available for usage.
   bool HasRelocations{false};
 
+  /// Indicates if the binary is stripped
+  bool IsStripped{false};
+
   /// Is the binary always loaded at a fixed address. Shared objects and
   /// position-independent executables (PIEs) are examples of binaries that
   /// will have HasFixedLoadAddress set to false.
   bool HasFixedLoadAddress{true};
 
   /// True if the binary has no dynamic dependencies, i.e., if it was statically
   /// linked.
   bool IsStaticExecutable{false};

bolt/lib/Rewrite/RewriteInstance.cpp

 
 uint64_t RewriteInstance::getLSDAAddress() { return LSDASection->getAddress(); }
 
 Error RewriteInstance::readSpecialSections() {
   NamedRegionTimer T("readSpecialSections", "read special sections",
                      TimerGroupName, TimerGroupDesc, opts::TimeRewrite);
 
   bool HasTextRelocations = false;
+  bool HasSymbolTable = false;
   bool HasDebugInfo = false;
 
   // Process special sections.
   for (const SectionRef &Section : InputFile->sections()) {
     Expected<StringRef> SectionNameOrErr = Section.getName();
     check_error(SectionNameOrErr.takeError(), "cannot get section name");
     StringRef SectionName = *SectionNameOrErr;
 
Error RewriteInstance::readSpecialSections() {
   }
 
   if (HasDebugInfo && !opts::UpdateDebugSections && !opts::AggregateOnly) {
     errs() << "BOLT-WARNING: debug info will be stripped from the binary. "
               "Use -update-debug-sections to keep it.\n";
   }
 
   HasTextRelocations = (bool)BC->getUniqueSectionByName(".rela.text");
+  HasSymbolTable = (bool)BC->getUniqueSectionByName(".symtab");
   LSDASection = BC->getUniqueSectionByName(".gcc_except_table");
   EHFrameSection = BC->getUniqueSectionByName(".eh_frame");
   GOTPLTSection = BC->getUniqueSectionByName(".got.plt");
   RelaPLTSection = BC->getUniqueSectionByName(".rela.plt");
   RelaDynSection = BC->getUniqueSectionByName(".rela.dyn");
   BuildIDSection = BC->getUniqueSectionByName(".note.gnu.build-id");
   SDTSection = BC->getUniqueSectionByName(".note.stapsdt");
   PseudoProbeDescSection = BC->getUniqueSectionByName(".pseudo_probe_desc");
if (opts::RelocationMode == cl::BOU_TRUE && !HasTextRelocations) {
     errs() << "BOLT-ERROR: relocations against code are missing from the input "
               "file. Cannot proceed in relocations mode (-relocs).\n";
     exit(1);
   }
 
   BC->HasRelocations =
       HasTextRelocations && (opts::RelocationMode != cl::BOU_FALSE);
 
+  BC->IsStripped = !HasSymbolTable;
+
   // Force non-relocation mode for heatmap generation
   if (opts::HeatmapMode)
     BC->HasRelocations = false;
 
   if (BC->HasRelocations)
     outs() << "BOLT-INFO: enabling " << (opts::StrictMode ? "strict " : "")
            << "relocation mode\n";
 
+  if (BC->IsStripped)
+    outs() << "BOLT-INFO: input binary is stripped. The support is limited and "
+           << "is considered experimental.\n";
+
   // Read EH frame for function boundaries info.
   Expected<const DWARFDebugFrame *> EHFrameOrError = BC->DwCtx->getEHFrame();
   if (!EHFrameOrError)
     report_error("expected valid eh_frame section", EHFrameOrError.takeError());
   CFIRdWrt.reset(new CFIReaderWriter(*EHFrameOrError.get()));
 
   // Parse build-id
   parseBuildID();

bolt/test/X86/is-strip.s

+# This test checks whether a binary is stripped or not.
+
+# RUN: %clang++ %p/Inputs/linenumber.cpp -o %t -Wl,-q
+# RUN: llvm-bolt %t -o %t.out 2>&1 | FileCheck %s -check-prefix=CHECK-NOSTRIP
+# RUN: cp %t %t.stripped
+# RUN: llvm-strip -s %t.stripped
+# RUN: llvm-bolt %t.stripped -o %t.out 2>&1 | FileCheck %s -check-prefix=CHECK-STRIP
+
+# CHECK-NOSTRIP-NOT: BOLT-INFO: input binary is stripped. The support is limited and is considered experimental.
+# CHECK-STRIP: BOLT-INFO: input binary is stripped. The support is limited and is considered experimental.