[LLD] Add terminator to .eh_frame sections

[kettenis]

Summary

Currenlty ld.lld does not add a terminator (a CIE with its length field set to zero) to the .eh_frame sections it generates.
While the relevant standards (the AMD64 SysV ABI and the Linux LSB) are not explicit about this, such a terminator
is expected by some unwinder implementations and seems to be always emitted by ld.bfd. In addition to that, the
Linux LSB

https://refspecs.linuxfoundation.org/LSB_5.0.0/LSB-Core-generic/LSB-Core-generic/ehframechpt.html#EHFRAME

explicitly says that

The .eh_frame section shall contain 1 or more Call Frame Information (CFI) records.

Currently, if the .eh_frame sections of the input files only contain terminators, ld.lld emits a zero=sized .eh_frame section
which clearly doesn't meet that requirement.

The diff makes sure a terminator gets added to each .eh_frame section and adjusts all the relevant tests to account
for that. An additional test isn't needed as these adjustments mean that the existence of the terminator is tested for by
several tests already.

[ruiu]

tools/lld/ELF/OutputSections.cpp

413

This needs to be a bit more friendly for future software archaeologist. Please also mention that

• .eh_frame can optionally end with an empty CIE which is a 4 byte zero,
• adding a terminator is not required by the standard, but
• we add that for the sake of compatibility with the GNU linker because they always add that (and some tools seem to depend on that).

[kettenis]

Thanks for the review. This new version expands the comment to explain why we're adding the terminator.

[ruiu]

LGTM. Thanks!

[kettenis]

Could you commit this for me?

[emaste]

some unwinder implementations

Out of curiosity, do you have a link to a specific example?

tools/lld/ELF/OutputSections.cpp

for (CieRecord *Cie : Cies) {
     Cie->Piece->OutputOff = Off;
     Off += alignTo(Cie->Piece->size(), sizeof(uintX_t));
 
     for (EhSectionPiece *Fde : Cie->FdePieces) {
       Fde->OutputOff = Off;
       Off += alignTo(Fde->size(), sizeof(uintX_t));
     }
   }
-  this->Size = Off;
+  // Add terminator.
+  this->Size = Off + 4;
 }
 
 template <class ELFT> static uint64_t readFdeAddr(uint8_t *Buf, int Size) {
   const endianness E = ELFT::TargetEndianness;
   switch (Size) {
   case DW_EH_PE_udata2:
     return read16<E>(Buf);
   case DW_EH_PE_udata4:

tools/lld/test/ELF/eh-frame-hdr.s

 // HDR-NEXT:    Index:
 // HDR-NEXT:    Name: .eh_frame
 // HDR-NEXT:    Type: SHT_PROGBITS
 // HDR-NEXT:    Flags [
 // HDR-NEXT:      SHF_ALLOC
 // HDR-NEXT:    ]
 // HDR-NEXT:    Address: 0x200180
 // HDR-NEXT:    Offset: 0x180
-// HDR-NEXT:    Size: 96
+// HDR-NEXT:    Size: 100
 // HDR-NEXT:    Link: 0
 // HDR-NEXT:    Info: 0
 // HDR-NEXT:    AddressAlignment: 8
 // HDR-NEXT:    EntrySize: 0
 // HDR-NEXT:    SectionData (
 // HDR-NEXT:      0000: 14000000 00000000 017A5200 01781001
 // HDR-NEXT:      0010: 1B0C0708 90010000 14000000 1C000000
 // HDR-NEXT:      0020: 600E0000 01000000 00000000 00000000
 // HDR-NEXT:      0030: 14000000 34000000 490E0000 01000000
 // HDR-NEXT:      0040: 00000000 00000000 14000000 4C000000
 // HDR-NEXT:      0050: 320E0000 01000000 00000000 00000000
+// HDR-NEXT:      0060: 00000000
 // HDR-NEXT:    )
 //            CIE: 14000000 00000000 017A5200 01781001 1B0C0708 90010000
 //            FDE(1): 14000000 1C000000 600E0000 01000000 00000000 00000000
 //                    address of data (starts with 0x600E0000) = 0x200180 + 0x0020 = 0x2001A0
 //                    The starting address to which this FDE applies = 0xE60 + 0x2001A0 = 0x201000
 //                    The number of bytes after the start address to which this FDE applies = 0x01000000 = 1
 //            FDE(2): 14000000 34000000 490E0000 01000000 00000000 00000000
 //                    address of data (starts with 0x490E0000) = 0x200180 + 0x0038 = 0x2001B8

tools/lld/test/ELF/eh-frame-merge.s

 
 // CHECK:      Name: .eh_frame
 // CHECK-NEXT: Type: SHT_PROGBITS
 // CHECK-NEXT: Flags [
 // CHECK-NEXT:   SHF_ALLOC
 // CHECK-NEXT: ]
 // CHECK-NEXT: Address:
 // CHECK-NEXT: Offset:
-// CHECK-NEXT: Size: 96
+// CHECK-NEXT: Size: 100
 // CHECK-NEXT: Link: 0
 // CHECK-NEXT: Info: 0
 // CHECK-NEXT: AddressAlignment: 8
 // CHECK-NEXT: EntrySize: 0
 // CHECK-NEXT: SectionData (
 // CHECK-NEXT: 0000: 14000000 00000000 017A5200 01781001  |
 // CHECK-NEXT: 0010: 1B0C0708 90010000 14000000 1C000000  |
 // CHECK-NEXT: 0020: E80D0000 01000000 00000000 00000000  |
 // CHECK-NEXT: 0030: 14000000 34000000 D20D0000 02000000  |
 // CHECK-NEXT: 0040: 00000000 00000000 14000000 4C000000  |
 // CHECK-NEXT: 0050: B90D0000 01000000 00000000 00000000  |
+// CHECK-NEXT: 0060: 00000000                             |
 // CHECK-NEXT: )
 
 // CHECK:      Name: foo
 // CHECK-NEXT: Type: SHT_PROGBITS
 // CHECK-NEXT: Flags [
 // CHECK-NEXT:   SHF_ALLOC
 // CHECK-NEXT:   SHF_EXECINSTR
 // CHECK-NEXT: ]

tools/lld/test/ELF/ehframe-relocation.s

 // 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/ehframe-relocation.s  -o %t2.o
 // RUN: ld.lld %t.o %t2.o -o %t
 // RUN: llvm-readobj -s %t | FileCheck %s
 // RUN: llvm-objdump -d %t | FileCheck --check-prefix=DISASM %s
 
 // CHECK:      Name: .eh_frame
 // CHECK-NEXT: Type: SHT_PROGBITS
 // CHECK-NEXT: Flags [
 // CHECK-NEXT:   SHF_ALLOC
 // CHECK-NEXT: ]
 // CHECK-NEXT: Address: 0x200120
 // CHECK-NEXT: Offset:
-// CHECK-NEXT: Size: 48
+// CHECK-NEXT: Size: 52
 // CHECK-NOT: .eh_frame
 
 // 0x200120 = 2097440
 // 0x200120 + 5 = 2097445
 // DISASM:      Disassembly of section .text:
 // DISASM-NEXT: _start:
 // DISASM-NEXT:   201000: {{.*}} movq 2097440, %rax
 // DISASM-NEXT:   201008: {{.*}} movq 2097445, %rax
 
 .section .eh_frame,"ax",@unwind
 
 .section .text
 .globl _start
 _start:
  movq .eh_frame, %rax
  movq .eh_frame + 5, %rax

tools/lld/test/ELF/invalid-fde-rel.s

// the fde points to.
 
         .long 0x14
         .long 0x1c
         .long 0x0
         .long 0x0
         .long 0x0
         .long 0x0
 
-// CHECK:  1 .eh_frame     00000018
+// CHECK:  1 .eh_frame     0000001c

tools/lld/test/ELF/linkerscript/eh-frame-hdr.s

 # REQUIRES: x86
 # RUN: llvm-mc -filetype=obj -triple=x86_64-unknown-linux %s -o %t
 # RUN: echo "SECTIONS { \
 # RUN:          .eh_frame_hdr : {} \
 # RUN:          .eh_frame : {} \
 # RUN:       }" > %t.script
 # RUN: ld.lld -o %t1 --eh-frame-hdr --script %t.script %t
 # RUN: llvm-objdump -s -section=".eh_frame_hdr" %t1 | FileCheck %s
 
-# CHECK:      011b033b 14000000 01000000 49000000
+# CHECK:      011b033b 14000000 01000000 4d000000
 # CHECK-NEXT: 30000000
 
 .global _start
 _start:
  nop
 
 .section .dah,"ax",@progbits
 .cfi_startproc
  nop
 .cfi_endproc

tools/lld/test/ELF/linkerscript/symbols-synthetic.s

 # RUN:        PROVIDE_HIDDEN(_begin_sec = ADDR(.text)); \
 # RUN:        PROVIDE_HIDDEN(_end_sec = ADDR(.text) + SIZEOF(.text)); }" > %t.script
 # RUN: ld.lld -o %t1 --eh-frame-hdr --script %t.script %t
 # RUN: llvm-objdump -t %t1 | FileCheck --check-prefix=IN-SEC %s
 
 # SIMPLE:      0000000000000128         .foo    00000000 .hidden _end_sec
 # SIMPLE-NEXT: 0000000000000120         .foo    00000000 _begin_sec
 # SIMPLE-NEXT: 0000000000000128         *ABS*   00000000 _end_sec_abs
-# SIMPLE-NEXT: 0000000000001048         .text   00000000 _start
+# SIMPLE-NEXT: 000000000000104c         .text   00000000 _start
 # SIMPLE-NEXT: 0000000000000120         .foo    00000000 begin_foo
 # SIMPLE-NEXT: 0000000000000128         .foo    00000000 end_foo
 # SIMPLE-NEXT: 0000000000000008         *ABS*   00000000 size_foo_1
 # SIMPLE-NEXT: 0000000000000008         *ABS*   00000000 size_foo_1_abs
 # SIMPLE-NEXT: 0000000000001000         .foo    00000000 begin_bar
 # SIMPLE-NEXT: 0000000000001004         .foo    00000000 end_bar
 # SIMPLE-NEXT: 0000000000000ee4         *ABS*   00000000 size_foo_2
 # SIMPLE-NEXT: 0000000000000ee4         *ABS*   00000000 size_foo_3

tools/lld/test/ELF/map-file.s

 .weak bar
 bar:
 .long bar - .
 .long zed - .
 local:
 .comm   common,4,16
 
 // CHECK:      Address          Size             Align Out     In      File    Symbol
-// CHECK-NEXT: 0000000000200158 0000000000000030     8 .eh_frame
+// CHECK-NEXT: 0000000000200158 0000000000000034     8 .eh_frame
 // CHECK-NEXT: 0000000000201000 0000000000000015     4 .text
 // CHECK-NEXT: 0000000000201000 000000000000000e     4         .text
 // CHECK-NEXT: 0000000000201000 000000000000000e     4                 {{.*}}{{/|\\}}map-file.s.tmp1.o
 // CHECK-NEXT: 000000000020100e 0000000000000000     0                         local
 // CHECK-NEXT: 0000000000201005 0000000000000000     0                         f(int)
 // CHECK-NEXT: 0000000000201000 0000000000000000     0                         _start
 // CHECK-NEXT: 0000000000201010 0000000000000002     4                 {{.*}}{{/|\\}}map-file.s.tmp2.o
 // CHECK-NEXT: 0000000000201010 0000000000000000     0                         foo

tools/lld/test/ELF/relocatable-eh-frame.s

 # 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     00000030
+# 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