Reland D61583 [ELF] Error on relocations to STT_SECTION symbols if the sections were discarded

This restores r361830 "[ELF] Error on relocations to STT_SECTION symbols if the sections were discarded"
and dependent commits (r362218, r362497) which were reverted by r364321, with a fix of a --gdb-index issue.

.rela.debug_ranges contains relocations of range list entries:

    // start address of a range list entry
    // old: 0; after r361830: 0
    00000000000033a0 R_X86_64_64 .text._ZN2v88internal7Isolate7factoryEv + 0
    // end address of a range list entry
    // old: 0xe; after r361830: 0
    00000000000033a8 R_X86_64_64 .text._ZN2v88internal7Isolate7factoryEv + e

If both start and end addresses of a range list entry resolve to 0,
DWARFDebugRangeList::isEndOfListEntry() will return true, then the
.debug_range decoding loop will terminate prematurely:

    while (true) {
      decode StartAddress
      decode EndAddress
      if (Entry.isEndOfListEntry()) // prematurely
        break;
      Entries.push_back(Entry);
    }

In lld/ELF/SyntheticSections.cpp, readAddressAreas() will read
incomplete address ranges and the resulting .gdb_index will be
incomplete. For files that gdb hasn't loaded their debug info, gdb uses
.gdb_index to map addresses to CUs. The absent entries make gdb fail to
symbolize some addresses.

To address this issue, we simply allow relocations to undefined symbols
in DWARF.cpp:findAux() and let RelocationResolver resolve them.

This patch should fix:

[1] http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190603/659848.html
[2] https://bugs.chromium.org/p/chromium/issues/detail?id=978067

llvm-svn: 364391

14 files changed, 212 insertions, 94 deletions

diff --git a/lld/ELF/DWARF.cpp b/lld/ELF/DWARF.cpp
index 6b90a038e87..9266320765a 100644
--- a/lld/ELF/DWARF.cpp
+++ b/lld/ELF/DWARF.cpp
@@ -92,21 +92,20 @@ LLDDwarfObj<ELFT>::findAux(const InputSectionBase &Sec, uint64_t Pos,
   const typename ELFT::Sym &Sym = File->template getELFSyms<ELFT>()[SymIndex];
   uint32_t SecIndex = File->getSectionIndex(Sym);
 
-  // Broken debug info can point to a non-Defined symbol.
-  auto *DR = dyn_cast<Defined>(&File->getRelocTargetSym(Rel));
-  if (!DR) {
-    RelType Type = Rel.getType(Config->IsMips64EL);
-    if (Type != Target->NoneRel)
-      error(toString(File) + ": relocation " + lld::toString(Type) + " at 0x" +
-            llvm::utohexstr(Rel.r_offset) + " has unsupported target");
-    return None;
-  }
-  uint64_t Val = DR->Value;
+  // An undefined symbol may be a symbol defined in a discarded section. We
+  // shall still resolve it. This is important for --gdb-index: the end address
+  // offset of an entry in .debug_ranges is relocated. If it is not resolved,
+  // its zero value will terminate the decoding of .debug_ranges prematurely.
+  Symbol &S = File->getRelocTargetSym(Rel);
+  uint64_t Val = 0;
+  if (auto *DR = dyn_cast<Defined>(&S)) {
+    Val = DR->Value;
 
-  // FIXME: We should be consistent about always adding the file
-  // offset or not.
-  if (DR->Section->Flags & ELF::SHF_ALLOC)
-    Val += cast<InputSection>(DR->Section)->getOffsetInFile();
+    // FIXME: We should be consistent about always adding the file
+    // offset or not.
+    if (DR->Section->Flags & ELF::SHF_ALLOC)
+      Val += cast<InputSection>(DR->Section)->getOffsetInFile();
+  }
 
   DataRefImpl D;
   D.p = getAddend<ELFT>(Rel);
diff --git a/lld/ELF/InputFiles.cpp b/lld/ELF/InputFiles.cpp
index 0c36ef141ae..d1a72f0adc4 100644
--- a/lld/ELF/InputFiles.cpp
+++ b/lld/ELF/InputFiles.cpp
@@ -1080,6 +1080,9 @@ template <class ELFT> void ObjFile<ELFT>::initializeSymbols() {
 
       if (ESym.st_shndx == SHN_UNDEF)
         this->Symbols[I] = make<Undefined>(this, Name, Binding, StOther, Type);
+      else if (Sec == &InputSection::Discarded)
+        this->Symbols[I] = make<Undefined>(this, Name, Binding, StOther, Type,
+                                           /*DiscardedSecIdx=*/SecIdx);
       else
         this->Symbols[I] =
             make<Defined>(this, Name, Binding, StOther, Type, Value, Size, Sec);
diff --git a/lld/ELF/InputSection.cpp b/lld/ELF/InputSection.cpp
index 3e8d093e6d2..d077c017ca7 100644
--- a/lld/ELF/InputSection.cpp
+++ b/lld/ELF/InputSection.cpp
@@ -412,7 +412,8 @@ void InputSection::copyRelocations(uint8_t *Buf, ArrayRef<RelTy> Rels) {
 
   for (const RelTy &Rel : Rels) {
     RelType Type = Rel.getType(Config->IsMips64EL);
-    Symbol &Sym = getFile<ELFT>()->getRelocTargetSym(Rel);
+    const ObjFile<ELFT> *File = getFile<ELFT>();
+    Symbol &Sym = File->getRelocTargetSym(Rel);
 
     auto *P = reinterpret_cast<typename ELFT::Rela *>(Buf);
     Buf += sizeof(RelTy);
@@ -435,10 +436,23 @@ void InputSection::copyRelocations(uint8_t *Buf, ArrayRef<RelTy> Rels) {
       // .eh_frame is horribly special and can reference discarded sections. To
       // avoid having to parse and recreate .eh_frame, we just replace any
       // relocation in it pointing to discarded sections with R_*_NONE, which
-      // hopefully creates a frame that is ignored at runtime.
+      // hopefully creates a frame that is ignored at runtime. Also, don't warn
+      // on .gcc_except_table and debug sections.
+      //
+      // See the comment in maybeReportUndefined for PPC64 .toc .
       auto *D = dyn_cast<Defined>(&Sym);
       if (!D) {
-        error("STT_SECTION symbol should be defined");
+        if (!Sec->Name.startswith(".debug") &&
+            !Sec->Name.startswith(".zdebug") && Sec->Name != ".eh_frame" &&
+            Sec->Name != ".gcc_except_table" && Sec->Name != ".toc") {
+          uint32_t SecIdx = cast<Undefined>(Sym).DiscardedSecIdx;
+          Elf_Shdr_Impl<ELFT> Sec =
+              CHECK(File->getObj().sections(), File)[SecIdx];
+          warn("relocation refers to a discarded section: " +
+               CHECK(File->getObj().getSectionName(&Sec), File) +
+               "\n>>> referenced by " + getObjMsg(P->r_offset));
+        }
+        P->setSymbolAndType(0, 0, false);
         continue;
       }
       SectionBase *Section = D->Section->Repl;
diff --git a/lld/ELF/Relocations.cpp b/lld/ELF/Relocations.cpp
index 639a3c5555e..8b8a3b1d4f9 100644
--- a/lld/ELF/Relocations.cpp
+++ b/lld/ELF/Relocations.cpp
@@ -653,9 +653,17 @@ static std::string maybeReportDiscarded(Undefined &Sym) {
     return "";
   ArrayRef<Elf_Shdr_Impl<ELFT>> ObjSections =
       CHECK(File->getObj().sections(), File);
-  std::string Msg =
-      "relocation refers to a symbol in a discarded section: " + toString(Sym) +
-      "\n>>> defined in " + toString(File);
+
+  std::string Msg;
+  if (Sym.Type == ELF::STT_SECTION) {
+    Msg = "relocation refers to a discarded section: ";
+    Msg += CHECK(
+        File->getObj().getSectionName(&ObjSections[Sym.DiscardedSecIdx]), File);
+  } else {
+    Msg = "relocation refers to a symbol in a discarded section: " +
+          toString(Sym);
+  }
+  Msg += "\n>>> defined in " + toString(File);
 
   Elf_Shdr_Impl<ELFT> ELFSec = ObjSections[Sym.DiscardedSecIdx - 1];
   if (ELFSec.sh_type != SHT_GROUP)
diff --git a/lld/ELF/Symbols.cpp b/lld/ELF/Symbols.cpp
index 5300aafaff8..9a4de4faf10 100644
--- a/lld/ELF/Symbols.cpp
+++ b/lld/ELF/Symbols.cpp
@@ -48,17 +48,11 @@ static uint64_t getSymVA(const Symbol &Sym, int64_t &Addend) {
     auto &D = cast<Defined>(Sym);
     SectionBase *IS = D.Section;
 
-    // According to the ELF spec reference to a local symbol from outside
-    // the group are not allowed. Unfortunately .eh_frame breaks that rule
-    // and must be treated specially. For now we just replace the symbol with
-    // 0.
-    if (IS == &InputSection::Discarded)
-      return 0;
-
     // This is an absolute symbol.
     if (!IS)
       return D.Value;
 
+    assert(IS != &InputSection::Discarded);
     IS = IS->Repl;
 
     uint64_t Offset = D.Value;
diff --git a/lld/test/ELF/Inputs/comdat-discarded-reloc.s b/lld/test/ELF/Inputs/comdat-discarded-reloc.s
index 9526f5ac95c..d89575f27e9 100644
--- a/lld/test/ELF/Inputs/comdat-discarded-reloc.s
+++ b/lld/test/ELF/Inputs/comdat-discarded-reloc.s
@@ -1,6 +1,8 @@
-.section .text.bar1,"aG",@progbits,group,comdat
+.global bar, _start
 
-.section .text.bar2
-.global bar
-bar:
- .quad .text.bar1
+.section .text.foo,"aG",@progbits,group,comdat
+
+.section .text
+_start:
+ .quad .text.foo
+ .quad bar
diff --git a/lld/test/ELF/comdat-discarded-error.s b/lld/test/ELF/comdat-discarded-error.s
index afefdaf02c8..d5aa4aa68df 100644
--- a/lld/test/ELF/comdat-discarded-error.s
+++ b/lld/test/ELF/comdat-discarded-error.s
@@ -5,7 +5,7 @@
 # RUN: echo '.section .text.foo,"axG",@progbits,foo,comdat; .globl bar; bar:' |\
 # RUN:   llvm-mc -filetype=obj -triple=x86_64 - -o %t3.o
 
-# RUN: not ld.lld %t1.o %t2.o %t3.o -o /dev/null 2>&1 | FileCheck %s
+# RUN: not ld.lld %t2.o %t3.o %t1.o -o /dev/null 2>&1 | FileCheck %s
 
 # CHECK:      error: relocation refers to a symbol in a discarded section: bar
 # CHECK-NEXT: >>> defined in {{.*}}3.o
@@ -13,6 +13,16 @@
 # CHECK-NEXT: >>> prevailing definition is in {{.*}}2.o
 # CHECK-NEXT: >>> referenced by {{.*}}1.o:(.text+0x1)
 
+# CHECK:      error: relocation refers to a discarded section: .text.foo
+# CHECK-NEXT: >>> defined in {{.*}}1.o
+# CHECK-NEXT: >>> section group signature: foo
+# CHECK-NEXT: >>> prevailing definition is in {{.*}}2.o
+# CHECK-NEXT: >>> referenced by {{.*}}1.o:(.data+0x0)
+
 .globl _start
 _start:
   jmp bar
+
+.section .text.foo,"axG",@progbits,foo,comdat
+.data
+  .quad .text.foo
diff --git a/lld/test/ELF/comdat-discarded-gdb-index.s b/lld/test/ELF/comdat-discarded-gdb-index.s
new file mode 100644
index 00000000000..43505960498
--- /dev/null
+++ b/lld/test/ELF/comdat-discarded-gdb-index.s
@@ -0,0 +1,63 @@
+# REQUIRES: x86
+# RUN: llvm-mc -filetype=obj -triple=x86_64 %s -o %t.o
+# RUN: ld.lld --gdb-index %t.o %t.o -o %t
+
+## .debug_info has a relocation to .text.foo . The second %t.o is discarded.
+## Check we don't error on the relocation.
+# CHECK: .rela.debug_info {
+# CHECK-NEXT: 0xC R_X86_64_64 .text.foo 0x0
+
+.section .text.foo,"axG",@progbits,foo,comdat
+.globl foo
+.Lfunc_begin0:
+foo:
+  ret
+.Lfunc_end0:
+
+.section .debug_abbrev,"",@progbits
+  .byte   1       # Abbreviation Code
+  .byte   17      # DW_TAG_compile_unit
+  .byte   1       # DW_CHILDREN_yes
+  .byte   17      # DW_AT_low_pc
+  .byte   1       # DW_FORM_addr
+  .byte   18      # DW_AT_high_pc
+  .byte   6       # DW_FORM_data4
+  .ascii  "\264B" # DW_AT_GNU_pubnames
+  .byte   25      # DW_FORM_flag_present
+  .byte   0       # EOM(1)
+  .byte   0       # EOM(2)
+  .byte   2       # Abbreviation Code
+  .byte   46      # DW_TAG_subprogram
+  .byte   0       # DW_CHILDREN_no
+  .byte   3       # DW_AT_name
+  .byte   8       # DW_FORM_string
+  .byte   0       # EOM(1)
+  .byte   0       # EOM(2)
+  .byte   0
+
+.section .debug_info,"",@progbits
+.Lcu_begin0:
+  .long   .Lcu_end0 - .Lcu_begin0 - 4
+  .short  4              # DWARF version number
+  .long   0              # Offset Into Abbrev. Section
+  .byte   4              # Address Size
+.Ldie0:
+  .byte   1              # Abbrev [1] DW_TAG_compile_unit
+  .quad   .Lfunc_begin0  # DW_AT_low_pc
+  .long   .Lfunc_end0 - .Lfunc_begin0 # DW_AT_high_pc
+  .byte   2              # Abbrev [2] DW_TAG_subprogram
+  .asciz  "foo"          # DW_AT_name
+  .byte   0
+.Lcu_end0:
+
+.section .debug_gnu_pubnames,"",@progbits
+  .long   .LpubNames_end0 - .LpubNames_begin0
+.LpubNames_begin0:
+  .short  2              # Version
+  .long   .Lcu_begin0    # CU Offset
+  .long   .Lcu_end0 - .Lcu_begin0
+  .long   .Ldie0 - .Lcu_begin0
+  .byte   48             # Attributes: FUNCTION, EXTERNAL
+  .asciz  "foo"          # External Name
+  .long   0
+.LpubNames_end0:
diff --git a/lld/test/ELF/comdat-discarded-reloc.s b/lld/test/ELF/comdat-discarded-reloc.s
index d23baf386e9..4d50ab10e51 100644
--- a/lld/test/ELF/comdat-discarded-reloc.s
+++ b/lld/test/ELF/comdat-discarded-reloc.s
@@ -1,17 +1,42 @@
 # REQUIRES: x86
 # RUN: llvm-mc -filetype=obj -triple=x86_64-pc-linux %s -o %t.o
 # RUN: llvm-mc -filetype=obj -triple=x86_64-pc-linux %p/Inputs/comdat-discarded-reloc.s -o %t2.o
-# RUN: ld.lld -gc-sections %t.o %t2.o -o %t
+# RUN: ld.lld -gc-sections --noinhibit-exec %t2.o %t.o -o /dev/null
+# RUN: ld.lld -r %t2.o %t.o -o %t 2>&1 | FileCheck --check-prefix=WARN %s
+# RUN: llvm-readobj -r %t | FileCheck --check-prefix=RELOC %s
 
 ## ELF spec doesn't allow a relocation to point to a deduplicated
 ## COMDAT section. Unfortunately this happens in practice (e.g. .eh_frame)
 ## Test case checks we do not crash.
 
-.global bar, _start
+# WARN: warning: relocation refers to a discarded section: .text.bar1
+# WARN-NEXT: >>> referenced by {{.*}}.o:(.rela.text.bar2+0x0)
+# WARN-NOT: warning
 
-.section .text.foo,"aG",@progbits,group,comdat
+# RELOC:      .rela.eh_frame {
+# RELOC-NEXT:   R_X86_64_NONE
+# RELOC-NEXT: }
+# RELOC-NEXT: .rela.debug_foo {
+# RELOC-NEXT:   R_X86_64_NONE
+# RELOC-NEXT: }
+# RELOC-NEXT: .rela.gcc_except_table {
+# RELOC-NEXT:   R_X86_64_NONE
+# RELOC-NEXT: }
 
-.section .text
-_start:
- .quad .text.foo
- .quad bar
+.section .text.bar1,"aG",@progbits,group,comdat
+
+## .text.bar1 in this file is discarded. Warn on the relocation.
+.section .text.bar2,"ax"
+.globl bar
+bar:
+  .quad .text.bar1
+
+## Don't warn on .eh_frame, .debug*, .zdebug*, or .gcc_except_table
+.section .eh_frame,"a"
+  .quad .text.bar1
+
+.section .debug_foo
+  .quad .text.bar1
+
+.section .gcc_except_table,"a"
+  .quad .text.bar1
diff --git a/lld/test/ELF/comdat.s b/lld/test/ELF/comdat.s
index 86103e5d9eb..9e3f5a81d30 100644
--- a/lld/test/ELF/comdat.s
+++ b/lld/test/ELF/comdat.s
@@ -1,7 +1,7 @@
 // REQUIRES: x86
 // RUN: llvm-mc -filetype=obj -triple=x86_64-pc-linux %s -o %t.o
 // RUN: llvm-mc -filetype=obj -triple=x86_64-pc-linux %p/Inputs/comdat.s -o %t2.o
-// RUN: ld.lld -shared %t.o %t.o %t2.o -o %t
+// RUN: ld.lld -shared %t.o %t2.o -o %t
 // RUN: llvm-objdump -d %t | FileCheck %s
 // RUN: llvm-readobj -S --symbols %t | FileCheck --check-prefix=READ %s
 
@@ -31,9 +31,7 @@ foo:
 // CHECK-EMPTY:
 // CHECK-NEXT: bar:
 // 0x1000 - 0x1001 - 5 = -6
-// 0      - 0x1006 - 5 = -4107
 // CHECK-NEXT:   1001:	{{.*}}  callq  -6
-// CHECK-NEXT:   1006:	{{.*}}  callq  -4107
 
         .section .text3,"axG",@progbits,zed,comdat,unique,0
 
diff --git a/lld/test/ELF/gdb-index-ranges-discarded.s b/lld/test/ELF/gdb-index-ranges-discarded.s
new file mode 100644
index 00000000000..2d5ce15736f
--- /dev/null
+++ b/lld/test/ELF/gdb-index-ranges-discarded.s
@@ -0,0 +1,51 @@
+# REQUIRES: x86
+# RUN: llvm-mc -filetype=obj -triple=x86_64 %s -o %t.o
+# RUN: ld.lld --gdb-index -e main %t.o -o %t
+# RUN: llvm-dwarfdump -gdb-index %t | FileCheck %s
+
+# CHECK:      .gdb_index contents:
+# CHECK:       Address area offset = 0x28, has 1 entries:
+# CHECK-NEXT:    Low/High address = [0x201000, 0x201001) (Size: 0x1), CU id = 0
+
+## .debug_ranges contains 2 entries. .Lfunc_end0 is defined in the discarded
+## .text.foo. Test we resolve it to a non-zero value, otherwise the address area
+## of .gdb_index will not included [.Lfunc_begin1, .Lfunc_end1).
+
+.section .text.foo,"aex",@progbits
+.Lfunc_begin0:
+  ret
+.Lfunc_end0:
+
+.section .text.bar,"ax",@progbits
+.Lfunc_begin1:
+  ret
+.Lfunc_end1:
+
+.section .debug_abbrev,"",@progbits
+  .byte 1         # Abbreviation Code
+  .byte 17        # DW_TAG_compile_unit
+  .byte 0         # DW_CHILDREN_no
+  .byte 85        # DW_AT_ranges
+  .byte 23        # DW_FORM_sec_offset
+  .byte 0         # EOM(1)
+  .byte 0         # EOM(2)
+  .byte 0
+
+.section .debug_info,"",@progbits
+.Lcu_begin0:
+  .long .Lcu_end0 - .Lcu_begin0 - 4
+  .short 4        # DWARF version number
+  .long  0        # Offset Into Abbrev. Section
+  .byte  8        # Address Size
+.Ldie0:
+  .byte  1        # Abbrev [1] DW_TAG_compile_unit
+  .long  0        # DW_AT_ranges
+.Lcu_end0:
+
+.section .debug_ranges,"",@progbits
+  .quad .Lfunc_begin0
+  .quad .Lfunc_end0
+  .quad .Lfunc_begin1
+  .quad .Lfunc_end1
+  .quad 0
+  .quad 0
diff --git a/lld/test/ELF/invalid-undef-section-symbol.test b/lld/test/ELF/invalid-undef-section-symbol.test
deleted file mode 100644
index 1d66885eadf..00000000000
--- a/lld/test/ELF/invalid-undef-section-symbol.test
+++ /dev/null
@@ -1,26 +0,0 @@
-# RUN: yaml2obj %s -o %t.o
-# RUN: not ld.lld -r %t.o -o /dev/null 2>&1 | FileCheck %s
-
-# We used to crash at this.
-# CHECK: STT_SECTION symbol should be defined
-
---- !ELF
-FileHeader:
-  Class:           ELFCLASS64
-  Data:            ELFDATA2LSB
-  Type:            ET_REL
-  Machine:         EM_X86_64
-Sections:
-  - Name:            .text
-    Type:            SHT_PROGBITS
-  - Name:            .rela.text
-    Type:            SHT_RELA
-    AddressAlign:    0x0000000000000008
-    Info:            .text
-    Relocations:
-      - Offset:          0x0000000000000000
-        Symbol:          .text
-        Type:            R_X86_64_NONE
-Symbols:
-  - Name:            .text
-    Type:            STT_SECTION
diff --git a/lld/test/ELF/relocatable-eh-frame.s b/lld/test/ELF/relocatable-eh-frame.s
deleted file mode 100644
index dee906acb87..00000000000
--- a/lld/test/ELF/relocatable-eh-frame.s
+++ /dev/null
@@ -1,19 +0,0 @@
-# REQUIRES: x86
-# RUN: llvm-mc -filetype=obj -triple=x86_64-pc-linux %s -o %t.o
-# RUN: ld.lld -r %t.o %t.o -o %t
-# RUN: llvm-readobj -r %t | FileCheck %s
-# RUN: ld.lld %t -o %t.so -shared
-# RUN: llvm-objdump -h %t.so | FileCheck --check-prefix=DSO %s
-
-# DSO: .eh_frame     00000034
-
-# CHECK:      Relocations [
-# CHECK-NEXT:   Section ({{.*}}) .rela.eh_frame {
-# CHECK-NEXT:     0x20 R_X86_64_PC32 .foo 0x0
-# CHECK-NEXT:     0x50 R_X86_64_NONE - 0x0
-# CHECK-NEXT:   }
-# CHECK-NEXT: ]
-
-.section .foo,"aG",@progbits,bar,comdat
-.cfi_startproc
-.cfi_endproc
diff --git a/lld/test/ELF/undef-broken-debug.test b/lld/test/ELF/undef-broken-debug.test
index 2ca8494a180..76449a59b73 100644
--- a/lld/test/ELF/undef-broken-debug.test
+++ b/lld/test/ELF/undef-broken-debug.test
@@ -1,12 +1,8 @@
 # REQUIRES: x86
 # RUN: yaml2obj %s -o %t.o
-# RUN: not ld.lld %t.o -o /dev/null 2>&1 | FileCheck %s
+# RUN: not ld.lld %t.o -o /dev/null
 
-# The debug info has a broken relocation. Check that we don't crash
-# and still report the undefined symbol.
-
-# CHECK: error: {{.*}}.o: relocation R_X86_64_64 at 0x29 has unsupported target
-# CHECK: error: undefined symbol: bar
+# The debug info has a broken relocation. Check that we don't crash.
 
 # We used to dereference null Target in DWARF.cpp:findAux while reporting a duplicate symbol error,
 # because Target wasn't initialized yet.