[llvm-dwarfdump] - Dump the older versions of .eh_frame/.debug_frame correctly.

[grimar]

Summary

The issue is the following.

DWARF 2 used version 1 for .debug_frame.
(Appendix G, p. 416 http://dwarfstd.org/doc/DWARF5.pdf)

lib/MC now always sets version 1 for .eh_frame (and sets 1-4 versions for .debug_frame correctly):
https://github.com/llvm-mirror/llvm/blob/master/lib/MC/MCDwarf.cpp#L1530
https://github.com/llvm-mirror/llvm/blob/master/lib/MC/MCDwarf.cpp#L1562
https://github.com/llvm-mirror/llvm/blob/master/lib/MC/MCDwarf.cpp#L1602

In version 1, return_address_register was defined as ubyte, while other versions
switched to uleb128.
(p 62, http://www.dwarfstd.org/doc/dwarf-2.0.0.pdf)

Patch teaches llvm-dwarfdump about this difference.

[grimar]

btw, I am not sure why do we still emit .eh_frame of version 1 in lib/MC.
For example, LLD supports versions 1 and 3 from the beginning I think, and it seems we could switch to 3.

Though at the same time I do not know if this "Return address register" can be > 7 bits,
so I guess having it as the ubyte is perhaps fine and then it is ok to have .eh_frame of version 1.

[dblaikie]

Seems plausible - though I'll leave it to @probinson for final sign off as I don't know too much about eh/debug_frame things.

[grimar]

Paul, what do you think? (Ping)

[probinson]

The indentation in the test file looks a little funny; please make sure there are no hard tabs in the file. Aside from that LGTM.

FTR, the format of .eh_frame is specified by the ELF spec, and has not been updated to reflect newer DWARF versions, that I'm aware of. If we can get the ELF spec updated, then we wouldn't need this fussy stuff. The format of .debug_frame of course is under the control of the DWARF spec, and so (even though it's really kind of stupid) the content does have to vary slightly depending on which section name we use.

lib/DebugInfo/DWARF/DWARFDebugFrame.cpp

if (IsCIE) {
       const char *Augmentation = Data.getCStr(&Offset);
       StringRef AugmentationString(Augmentation ? Augmentation : "");
       uint8_t AddressSize = Version < 4 ? Data.getAddressSize() :
                                           Data.getU8(&Offset);
       Data.setAddressSize(AddressSize);
       uint8_t SegmentDescriptorSize = Version < 4 ? 0 : Data.getU8(&Offset);
       uint64_t CodeAlignmentFactor = Data.getULEB128(&Offset);
       int64_t DataAlignmentFactor = Data.getSLEB128(&Offset);
-      uint64_t ReturnAddressRegister = Data.getULEB128(&Offset);
+      uint64_t ReturnAddressRegister =
+          Version == 1 ? Data.getU8(&Offset) : Data.getULEB128(&Offset);
 
       // Parse the augmentation data for EH CIEs
       StringRef AugmentationData("");
       uint32_t FDEPointerEncoding = DW_EH_PE_absptr;
       uint32_t LSDAPointerEncoding = DW_EH_PE_omit;
       Optional<uint64_t> Personality;
       Optional<uint32_t> PersonalityEncoding;
       if (IsEH) {

test/tools/llvm-dwarfdump/X86/eh-frame-return-address-reg.s

+# RUN: llvm-mc %s -filetype obj -triple x86_64-pc-linux -o %t.o
+# RUN: llvm-dwarfdump -v %t.o | FileCheck %s
+
+# The format of the .eh_frame section is similar in
+# format and purpose to the .debug_frame section.
+# Version 1 is often used for .eh_frame,
+# and also it was used for DWARF v2. For that case,
+# return address register should be encoded as ubyte,
+# while later versions use ULEB128. This test case
+# checks that we are able to dump it correctly.
+
+# CHECK:      .eh_frame contents:
+# CHECK:      00000000 00000010 ffffffff CIE
+# CHECK-NEXT:   Version:               1
+# CHECK-NEXT:   Augmentation:          "zR"
+# CHECK-NEXT:   Code alignment factor: 1
+# CHECK-NEXT:   Data alignment factor: 1
+# CHECK-NEXT:   Return address column: 240
+# CHECK-NEXT:   Augmentation data:     1A
+
+.text
+.global _start
+_start:
+ nop
+
+.section .eh_frame, "a"
+  .long 16   # Size
+  .long 0x00 # ID
+  .byte 0x01 # Version
+
+  .byte 0x7A # Augmentation string: "zR"
+  .byte 0x52
+  .byte 0x00
+
+  .byte 0x01 # Code alignment factor, ULEB128
+  .byte 0x01 # Data alignment factor, ULEB128
+  
+  .byte 0xF0 # Return address register, ubyte for version 1.
+
+  .byte 0x01 # LEB128
+  .byte 0x1A # DW_EH_PE_pcrel | DW_EH_PE_sdata2
+
+  .byte 0x00
+  .byte 0x00
+  .byte 0x00
+
+  .long 10   # Size
+  .long 24   # ID
+fde:
+  .long _start - fde
+  .word 0