[llvm-readobj] Construct relocation-aware DWARFDataExtractor to decode .eh_frame addresses correctly

[MaskRay]

Summary

In an object file, a "PC Begin" field in a FDE is usually relocated by a
PC-relative relocation. Use a relocation-aware DWARFDataExtractor overload (with
DWARFContext and a reference to its internal .eh_frame representation) to decode
addresses correctly. In an object file, most sections have addresses of zero. So
the displayed addresses are almost always offsets relative to the start of the
associated text section.

DWARFContext::create handles .eh_frame and .rela.eh_frame by itself, so if there
are more than one .eh_frame (technically possible, but almost always erronerous
in practice), this will only handle the first one. Supporting multiple
.eh_frame is beyond the scope of this patch.

[MaskRay]

Add a comment

[grimar]

The description seems to be confusing, it says: "Use a relocation-aware DWARFDataExtractor instead of plain DataExtractor",
though both the code on the left and on the right use DWARFDataExtractor type and the difference is in the internal logic.

The rest looks fine to me (with a nit about headers order), I'd like somebody else to look at this too though.

llvm/tools/llvm-readobj/DwarfCFIEHPrinter.h

26

I think you could just add "DWARFContext.h", no need to sort the headers in this patch.

[probinson]

I agree with @grimar as it looks like the change is from a DWARFDataExtractor without relocations to one with relocations.
Regarding header order, in cases like this it's usual to do an NFC commit to sort the headers that are there, and then the functional patch just adds the new one in the right place.

[MaskRay]

Fix description (Use a relocation-aware DWARFDataExtractor constructor)

[MaskRay]

@probinson Looks good now? :)

[jhenderson]

LGTM.

llvm/test/tools/llvm-readobj/ELF/AArch64/dwarf-cfi.s

 # RUN: llvm-mc -triple arm64-linux -filetype obj -o - %s | llvm-readobj -u - | FileCheck %s
 
 # CHECK:      [0x0] CIE length=16
 # CHECK-NEXT:   version: 1
 # CHECK-NEXT:   augmentation: zR
 # CHECK-NEXT:   code_alignment_factor: 1
 # CHECK-NEXT:   data_alignment_factor: -4
 # CHECK-NEXT:   return_address_register: 30
 
 # CHECK:        Program:
 # CHECK-NEXT: DW_CFA_def_cfa: reg31 +0
 
-## FIXME Use getEHFrameSection() so that the address is decoded correctly.
 # CHECK:      [0x14] FDE length=16 cie=[0x0]
-# CHECK-NEXT:   initial_location: 0x1c
-# CHECK-NEXT:   address_range: 0x4 (end : 0x20)
+# CHECK-NEXT:   initial_location: 0x0
+# CHECK-NEXT:   address_range: 0x4 (end : 0x4)
 
 # CHECK:        Program:
 # CHECK-NEXT: DW_CFA_nop:
 # CHECK-NEXT: DW_CFA_nop:
 # CHECK-NEXT: DW_CFA_nop:
 
 foo:
 .cfi_startproc
 ret
 .cfi_endproc

llvm/test/tools/llvm-readobj/ELF/ARM/dwarf-cfi.s

 # RUN: llvm-mc -triple arm-linux -filetype obj -o - %s | llvm-readobj -u - | FileCheck %s
 
 # CHECK:      [0x0] CIE length=16
 # CHECK-NEXT:   version: 1
 # CHECK-NEXT:   augmentation: zR
 # CHECK-NEXT:   code_alignment_factor: 1
 # CHECK-NEXT:   data_alignment_factor: -4
 # CHECK-NEXT:   return_address_register: 14
 
 # CHECK:        Program:
 # CHECK-NEXT: DW_CFA_def_cfa: reg13 +0
 
-## FIXME Use getEHFrameSection() so that the address is decoded correctly.
 # CHECK:      [0x14] FDE length=16 cie=[0x0]
-# CHECK-NEXT:   initial_location: 0x1c
-# CHECK-NEXT:   address_range: 0x4 (end : 0x20)
+# CHECK-NEXT:   initial_location: 0x0
+# CHECK-NEXT:   address_range: 0x4 (end : 0x4)
 
 # CHECK:        Program:
 # CHECK-NEXT: DW_CFA_nop:
 # CHECK-NEXT: DW_CFA_nop:
 # CHECK-NEXT: DW_CFA_nop:
 
 .cpu cortex-a8
 
 foo:
 .cfi_startproc
 bx lr
 .cfi_endproc

llvm/tools/llvm-readobj/DwarfCFIEHPrinter.h

 //===--- DwarfCFIEHPrinter.h - DWARF-based Unwind Information Printer -----===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_TOOLS_LLVM_READOBJ_DWARFCFIEHPRINTER_H
 #define LLVM_TOOLS_LLVM_READOBJ_DWARFCFIEHPRINTER_H
 
 #include "Error.h"
 #include "llvm-readobj.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/BinaryFormat/Dwarf.h"
+#include "llvm/DebugInfo/DWARF/DWARFContext.h"
 #include "llvm/DebugInfo/DWARF/DWARFDataExtractor.h"
 #include "llvm/DebugInfo/DWARF/DWARFDebugFrame.h"
 #include "llvm/Object/ELF.h"
 #include "llvm/Object/ELFObjectFile.h"
 #include "llvm/Object/ELFTypes.h"
 #include "llvm/Support/Casting.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/Endian.h"
 #include "llvm/Support/Format.h"
 #include "llvm/Support/ScopedPrinter.h"
 #include "llvm/Support/type_traits.h"
 
 namespace llvm {
 namespace DwarfCFIEH {
 
 template <typename ELFT> class PrinterContext {
   using Elf_Shdr = typename ELFT::Shdr;
   using Elf_Phdr = typename ELFT::Phdr;
W.startLine() << format(".eh_frame section at offset 0x%" PRIx64
                           ShOffset, Address);
   W.indent();
 
   Expected<ArrayRef<uint8_t>> DataOrErr =
       ObjF->getELFFile()->getSectionContents(EHFrameShdr);
   if (!DataOrErr)
     reportError(DataOrErr.takeError(), ObjF->getFileName());
 
-  DWARFDataExtractor DE(*DataOrErr,
+  // Construct DWARFDataExtractor to handle relocations ("PC Begin" fields).
+  std::unique_ptr<DWARFContext> DICtx = DWARFContext::create(*ObjF, nullptr);
+  DWARFDataExtractor DE(DICtx->getDWARFObj(),
+                        DICtx->getDWARFObj().getEHFrameSection(),
                         ELFT::TargetEndianness == support::endianness::little,
                         ELFT::Is64Bits ? 8 : 4);
   DWARFDebugFrame EHFrame(Triple::ArchType(ObjF->getArch()), /*IsEH=*/true,
                           /*EHFrameAddress=*/Address);
   if (Error E = EHFrame.parse(DE))
     reportError(std::move(E), ObjF->getFileName());
 
   for (const dwarf::FrameEntry &Entry : EHFrame) {