[LLD][ELF] Correct sh_info for static symbol table

[peter.smith]

Summary

The sh_info field of the SHT_SYMTAB section holds the index for the first non-local symbol. When there are global symbols that are output with STB_LOCAL binding due to having hidden visibility or matching the local version from a version script, the calculated value of NumLocals + 1 does not account for them. This change adds in the number of local symbols that aren't covered by NumLocals.

I think that this property is of marginal use in an Executable/Shared object but there could be an ELF processing tool that could get confused if it skips all the local symbols in the file using the sh_info field.

The test updates are where we have a global symbol output with local binding and a check of a sh_info field.

I noticed this while working on support for synthetic local symbols. It feels like a bit of a hack, but if NumLocals is interpreted as number of locals stored outside of the SymbolTableSection::Symbols in Files it makes a bit more sense.

[ruiu]

ELF/SyntheticSections.cpp

1071–1072

Maybe we should stop using NumLocals at all here? I wonder if something like this works.

auto It = std::stable_partition(Symbols.begin(), Symbols.end(), [&](SymbolTableEntry &S) { returns true if S is STB_LOCAL; });
this->Info = It - Symbols.begin();
this->OutSec->Info = It - Symbols.begin();

[peter.smith]

Thank you for the comments. Updated to use std::stable_partition and rebased on top of D29021.

[ruiu]

LGTM

ELF/SyntheticSections.cpp

 template <class ELFT> void SymbolTableSection<ELFT>::finalize() {
   this->OutSec->Link = this->Link = StrTabSec.OutSec->SectionIndex;
   this->OutSec->Info = this->Info = NumLocals + 1;
   this->OutSec->Entsize = this->Entsize;
 
   if (Config->Relocatable)
     return;
 
   if (!StrTabSec.isDynamic()) {
     auto GlobBegin = Symbols.begin() + NumLocals;
-    std::stable_sort(GlobBegin, Symbols.end(), [](const SymbolTableEntry &L,
-                                                  const SymbolTableEntry &R) {
-      return L.Symbol->symbol()->computeBinding() == STB_LOCAL &&
-             R.Symbol->symbol()->computeBinding() != STB_LOCAL;
+    auto It = std::stable_partition(
+        GlobBegin, Symbols.end(), [](const SymbolTableEntry &S) {
+          return S.Symbol->symbol()->computeBinding() == STB_LOCAL;
         });
+    // update sh_info with number of Global symbols output with computed
+    // binding of STB_LOCAL
+    this->OutSec->Info = this->Info = 1 + It - Symbols.begin();
     return;
   }
 
   if (In<ELFT>::GnuHashTab)
     // NB: It also sorts Symbols to meet the GNU hash table requirements.
     In<ELFT>::GnuHashTab->addSymbols(Symbols);
   else if (Config->EMachine == EM_MIPS)
     std::stable_sort(Symbols.begin(), Symbols.end(),

test/ELF/basic-mips.s

 # CHECK-NEXT:     Name: .symtab
 # CHECK-NEXT:     Type: SHT_SYMTAB (0x2)
 # CHECK-NEXT:     Flags [ (0x0)
 # CHECK-NEXT:     ]
 # CHECK-NEXT:     Address: 0x0
 # CHECK-NEXT:     Offset: 0x20010
 # CHECK-NEXT:     Size: 48
 # CHECK-NEXT:     Link: 10
-# CHECK-NEXT:     Info: 1
+# CHECK-NEXT:     Info: 2
 # CHECK-NEXT:     AddressAlignment: 4
 # CHECK-NEXT:     EntrySize: 16
 # CHECK-NEXT:   }
 # CHECK-NEXT:   Section {
 # CHECK-NEXT:     Index: 9
 # CHECK-NEXT:     Name: .shstrtab
 # CHECK-NEXT:     Type: SHT_STRTAB (0x3)
 # CHECK-NEXT:     Flags [ (0x0)

test/ELF/basic-ppc.s

 // CHECK-NEXT:     Name: .symtab
 // CHECK-NEXT:     Type: SHT_SYMTAB (0x2)
 // CHECK-NEXT:     Flags [ (0x0)
 // CHECK-NEXT:     ]
 // CHECK-NEXT:     Address: 0x0
 // CHECK-NEXT:     Offset: 0x2038
 // CHECK-NEXT:     Size: 32
 // CHECK-NEXT:     Link: 9
-// CHECK-NEXT:     Info: 1
+// CHECK-NEXT:     Info: 2
 // CHECK-NEXT:     AddressAlignment: 4
 // CHECK-NEXT:     EntrySize: 16
 // CHECK-NEXT:     SectionData (
 // CHECK-NEXT:       0000: 00000000 00000000 00000000 00000000  |................|
 // CHECK-NEXT:       0010: 00000001 00002000 00000000 00020005  |...... .........|
 // CHECK-NEXT:     )
 // CHECK-NEXT:   }
 // CHECK-NEXT:   Section {