[yaml2obj] - Add `ELFYAML::YAMLIntUInt` to fix how we parse a relocation `Addend` key.

[grimar]

Summary

This patch makes Relocation::Addend to be ELFYAML::YAMLInt and not int64_t.

1. For an 64-bit object any hex/decimal addends in the range [INT64_MIN, UINT64_MAX] is accepted.
2. For an 32-bit object any hex/decimal addends in range [INT32_MIN, UINT32_MAX] is accepted.
3. Negative hex numbers like -0xffffffff are not accepted.
4. It is printed as decimal. I.e. obj2yaml will print something like "Addend: 125", this matches the current behavior.

This fixes all FIXMEs in relocation-addend.yaml.

[MaskRay]

For decimal values: a) 64-bit case: accept any [INT64MIN, INT64MAX] value. b) 32-bit case: accept any [INT32MIN, INT32MAX] value.

It should probably accept the union of int32_t and uint32_t, i.e. [-2147483648, 4294967295]. Examples: R_AARCH64_ABS32, R_AARCH64_PREL32, R_PPC64_ADDR32.

Not all relocation types behave this way, but we can treat all relocations the same way and be permissive here.

See https://github.com/llvm/llvm-project/blob/master/lld/ELF/Target.h#L223 checkIntUInt

llvm/lib/ObjectYAML/ELFYAML.cpp

998

Not very necessary, I think.

[grimar]

In D75527#1903438, @MaskRay wrote:

It should probably accept the union of int32_t and uint32_t, i.e. [-2147483648, 4294967295]. Examples: R_AARCH64_ABS32, R_AARCH64_PREL32, R_PPC64_ADDR32.

I think your suggestion is fine. I'll go this way if there will be no objections.

But what should we print (obj2yaml)? We print int64_t decimal currently.
Should we switch to printing a hex form then?

llvm/lib/ObjectYAML/ELFYAML.cpp

998

This syntax is just strange.
Imagine you do Addend: 0xFFFFFFFF, i.e. you mean -1.
If you write Addend: -0xFFFFFFFF, the result will be 1, or 0x00000001 in hex. The hex sequence is
very different because of that minus. I think it is a bad practice to use -0xFFFFFFFF form for Addend value
or any other key probably.

We either need a test case for this, or to restrict it explicitly probably.
(Doesn't seem we can just ignore it? And I do not think we want to support it).
I selected to restrict it.

[MaskRay]

In D75527#1904856, @grimar wrote:

In D75527#1903438, @MaskRay wrote:

It should probably accept the union of int32_t and uint32_t, i.e. [-2147483648, 4294967295]. Examples: R_AARCH64_ABS32, R_AARCH64_PREL32, R_PPC64_ADDR32.

I think your suggestion is fine. I'll go this way if there will be no objections.

But what should we print (obj2yaml)? We print int64_t decimal currently.
Should we switch to printing a hex form then?

For most relocation types, the addend should be interpreted as a signed integer. We don't want obj2yaml to learn the semantics of relocation types, so letting obj2yaml unconditionally dump an addend as a signed integer should be fine. An addend (of a 32-bit object) greater than or equal to 0x80000000 is usually accepted as input, so yaml2obj should also accept them.

I just think that -0x1 and -0x5 don't look so strange (no need to reject), but I'd like to hear a third opinion.

[jhenderson]

My inclination is that yaml2obj should support both signed and unsigned variants of both hex and decimal numbers, if feasible. I think obj2yaml should print as unsigned hex numbers, since that is what is displayed by tools such as llvm-readelf. In other words, an addend of -4 would be printed by obj2yaml as "0xFFFFFFFFFFFFFFFC" in a 64-bit object, but could be specified as "-4", "-0x0000000000000004", or "0xFFFFFFFFFFFFFFFC" for yaml2obj.

[grimar]

I am not strongly against the idea to print them as unsigned hex numbers, but FTR:

In D75527#1907129, @jhenderson wrote:

I think obj2yaml should print as unsigned hex numbers, since that is what is displayed by tools such as llvm-readelf.

But not GNU readelf. See "FIXME" in D75671: llvm-readelf -r displays decimal addends.

[jhenderson]

In D75527#1907531, @grimar wrote:

I am not strongly against the idea to print them as unsigned hex numbers, but FTR:

In D75527#1907129, @jhenderson wrote:

I think obj2yaml should print as unsigned hex numbers, since that is what is displayed by tools such as llvm-readelf.

But not GNU readelf. See "FIXME" in D75671: llvm-readelf -r displays decimal addends.

Apparently my memory is playing tricks on me. llvm-readobj displays unsigned hex.

[grimar]

In D75527#1907129, @jhenderson wrote:

My inclination is that yaml2obj should support both signed and unsigned variants of both hex and decimal numbers, if feasible. I think obj2yaml should print as unsigned hex numbers, since that is what is displayed by tools such as llvm-readelf. In other words, an addend of -4 would be printed by obj2yaml as "0xFFFFFFFFFFFFFFFC" in a 64-bit object, but could be specified as "-4", "-0x0000000000000004", or "0xFFFFFFFFFFFFFFFC" for yaml2obj.

Seems @MaskRay's answer https://reviews.llvm.org/D75527#1907129 responds this.

I think we can start with fixing of hex values accepting. The current patch does not change the obj2yaml's output. We can change/not-change it later.

[grimar]

In D75527#1907129, @jhenderson wrote:

My inclination is that yaml2obj should support both signed and unsigned variants of both hex and decimal numbers, if feasible.

I'd like to clarify what you mean.

Are you ok to support Addend: -4294967295? (4294967295 == 0xFFFFFFFF) for 32 bit ELF
and Addend: -18446744073709551615 (18446744073709551615 == 0xFFFFFFFFFFFFFFFF) for 64-bit one?
(it is the same as Addend: 1)

Looks like that because we are not going to restrict
Addend: -0xFFFFFFFF or Addend: -0xFFFFFFFFFFFFFFFF it seems.

If so, then this patch can be a bit simpler, we can probably check - and always read value as unsigned int64.
I.e. both:

• [-18446744073709551615, 18446744073709551615]
• [ -0xFFFFFFFFFFFFFFFF, 0xFFFFFFFFFFFFFFFF]

StringRef ScalarTraits<ELFYAML::YAMLInt>::input(StringRef Scalar, void *Ctx,
                                                ELFYAML::YAMLInt &Val) {
  const bool Is64 = static_cast<ELFYAML::Object *>(Ctx)->Header.Class ==
                    ELFYAML::ELF_ELFCLASS(ELF::ELFCLASS64);
  StringRef ErrMsg = "invalid number";

  bool IsNegative = Scalar.front() == '-';
  if (IsNegative)
    Scalar = Scalar.drop_front();

  unsigned long long UInt;
  if (getAsUnsignedInteger(Scalar, /*Radix=*/0, UInt))
    return ErrMsg;

  // For a 32-bit target we allow values in a range of a uint32_t.
  if (!Is64 && (UInt > UINT32_MAX))
    return ErrMsg;

  Val = IsNegative ? (-UInt) : UInt;
  return "";
}

Did you mean something like that?

[jhenderson]

In D75527#1909495, @grimar wrote:

In D75527#1907129, @jhenderson wrote:

My inclination is that yaml2obj should support both signed and unsigned variants of both hex and decimal numbers, if feasible.

I'd like to clarify what you mean.

Are you ok to support Addend: -4294967295? (4294967295 == 0xFFFFFFFF) for 32 bit ELF
and Addend: -18446744073709551615 (18446744073709551615 == 0xFFFFFFFFFFFFFFFF) for 64-bit one?
(it is the same as Addend: 1)

Looks like that because we are not going to restrict
Addend: -0xFFFFFFFF or Addend: -0xFFFFFFFFFFFFFFFF it seems.

If so, then this patch can be a bit simpler, we can probably check - and always read value as unsigned int64.
I.e. both:

• [-18446744073709551615, 18446744073709551615]
• [ -0xFFFFFFFFFFFFFFFF, 0xFFFFFFFFFFFFFFFF]

StringRef ScalarTraits<ELFYAML::YAMLInt>::input(StringRef Scalar, void *Ctx,
                                                ELFYAML::YAMLInt &Val) {
  const bool Is64 = static_cast<ELFYAML::Object *>(Ctx)->Header.Class ==
                    ELFYAML::ELF_ELFCLASS(ELF::ELFCLASS64);
  StringRef ErrMsg = "invalid number";

  bool IsNegative = Scalar.front() == '-';
  if (IsNegative)
    Scalar = Scalar.drop_front();

  unsigned long long UInt;
  if (getAsUnsignedInteger(Scalar, /*Radix=*/0, UInt))
    return ErrMsg;

  // For a 32-bit target we allow values in a range of a uint32_t.
  if (!Is64 && (UInt > UINT32_MAX))
    return ErrMsg;

  Val = IsNegative ? (-UInt) : UInt;
  return "";
}

Did you mean something like that?

Not looked at the code too critically, but that more or less looks like what I'd expect. In hex land, I expect a given hex number to correspond identically to the bytes written. In other words 0xffffffffffffffff would be written with the byte values 0xff 0xff etc, and have an effective value of -1. I think therefore it's logical to cause a '-' sign to negate the number. In other words, -0xffffffffffffffff would end up as 1. If somebody instead writes a decimal number, I think it can just be written as you'd expect (i.e. it's positive or negative, depending on the presence of the '-' sign). The difficulty is what to do about overflow detection, I guess. To keep things simple, I think it's best to do what you suggest in the quoted post. If I'm not mistaken, it's what the "natural" overflow behaviour would be for C++, so probably isn't too surprising.

[grimar]

• Updated implementation as discussed.

@MaskRay, are you OK with this approach?

[jhenderson]

LGTM, with a couple of suggestions.

llvm/test/tools/yaml2obj/ELF/relocation-addend.yaml

135

Maybe these should use 'G' instead of 'Q', since that's the first invalid value?

138

Maybe also worth testing '-' on its own, and '--1234'.

[grimar]

• Addressed review comments.

llvm/test/tools/yaml2obj/ELF/relocation-addend.yaml

138

Maybe also worth testing '-'

This helped to catch an assert, thanks! It turned out that on such input we have an empty Scalar
(because of the error state) in ScalarTraits<ELFYAML::YAMLInt>::input and that was not handled properly
before.

[MaskRay]

For an 64-bit object any hex/decimal addends in range [-(2^64 - 1), 2^64 - 1)] is accepted.
For an 64-bit object any hex/decimal addends in range [-(2^32 - 1), 2^32 - 1)] is accepted.

My feeling is that the ranges should be:
[-(2^63 - 1), 2^64 - 1)]
[-(2^31 - 1), 2^32 - 1)]

It should probably accept the union of int32_t and uint32_t, i.e. [-2147483648, 4294967295]. Examples: R_AARCH64_ABS32, R_AARCH64_PREL32, R_PPC64_ADDR32.

For example, for an ELFCLASS32 object, an addend of -0xffffffff or -0x80000001 is invalid.

[grimar]

In D75527#1921825, @MaskRay wrote:

For an 64-bit object any hex/decimal addends in range [-(2^64 - 1), 2^64 - 1)] is accepted.
For an 64-bit object any hex/decimal addends in range [-(2^32 - 1), 2^32 - 1)] is accepted.

My feeling is that the ranges should be:
[-(2^63 - 1), 2^64 - 1)]
[-(2^31 - 1), 2^32 - 1)]

It should probably accept the union of int32_t and uint32_t, i.e. [-2147483648, 4294967295]. Examples: R_AARCH64_ABS32, R_AARCH64_PREL32, R_PPC64_ADDR32.

But it does not work good for other relocations, isn't? Different relocations allows different ranges.
We are anyways not 100% correct with such approach until we don't handle each possible relocation individually.

Currently (this diff), YAMLInt is kind of universal type. It can be used not only for addends, but as a
type or a base for type for other keys probably.

For example, for an ELFCLASS32 object, an addend of -0xffffffff or -0x80000001 is invalid.

I have to say about what I think about negative hex numbers again, sorry.
Negitive hex numbers in YAML is something I suggested to ban in the first version of this diff.

I can read -0xffffffff as --1 or as -4294967296 or as a -0x00000000ffffffff at the same time.
But why actually we want to allow users to do this and think about values that hard?
Who will use negative hex numbers in a YAML tests and for what? I think it accepts less readable
tests and probably gives no any benefits. I think we should not focus too much on them (or just ban).

Speaking about decimal version of -0xffffffff: I have a test in this diff:

## Addend == -(2^32 - 1)
# RUN: yaml2obj --docnum=2 %s -o %t32.decimal.min -DADDEND=-4294967295
# RUN: llvm-readobj -r %t32.decimal.min | FileCheck %s --check-prefix=TEST -DADDEND=0x1
# RUN: yaml2obj --docnum=2 %s -o %t32.hex.min -DADDEND=-0xFFFFFFFF
# RUN: llvm-readobj -r %t32.hex.min | FileCheck %s --check-prefix=TEST -DADDEND=0x1

-0xffffffff is accepted for 32-bits platform and reads the same as -4294967295, i.e. as 0x1.

While I understand why you think that -0xffffffff` should be restricted in this case,
the current behavior looks acceptable for me in general.

I need this patch for D75671 which wants to do Addend: 0xffffffffffffffff
for example and fails:

YAML:17:17: error: invalid number
        Addend: 0xffffffffffffffff
                ^~~~~~~~~~~~~~~~~~
yaml2obj: error: failed to parse YAML input: Invalid argument

This is a problem I am trying to solve first of all.

[MaskRay]

In D75527#1922848, @grimar wrote:

In D75527#1921825, @MaskRay wrote:

For an 64-bit object any hex/decimal addends in range [-(2^64 - 1), 2^64 - 1)] is accepted.
For an 64-bit object any hex/decimal addends in range [-(2^32 - 1), 2^32 - 1)] is accepted.

My feeling is that the ranges should be:
[-(2^63 - 1), 2^64 - 1)]
[-(2^31 - 1), 2^32 - 1)]

It should probably accept the union of int32_t and uint32_t, i.e. [-2147483648, 4294967295]. Examples: R_AARCH64_ABS32, R_AARCH64_PREL32, R_PPC64_ADDR32.

But it does not work good for other relocations, isn't? Different relocations allows different ranges.
We are anyways not 100% correct with such approach until we don't handle each possible relocation individually.

I agree. So we just take the union of the use cases of all relocation types, but the range [-0x80000000, 0xffffffff] does not have to be larger.

Currently (this diff), YAMLInt is kind of universal type. It can be used not only for addends, but as a
type or a base for type for other keys probably.

For example, for an ELFCLASS32 object, an addend of -0xffffffff or -0x80000001 is invalid.

I have to say about what I think about negative hex numbers again, sorry.
Negitive hex numbers in YAML is something I suggested to ban in the first version of this diff.

I can read -0xffffffff as --1 or as -4294967296 or as a -0x00000000ffffffff at the same time.
But why actually we want to allow users to do this and think about values that hard?
Who will use negative hex numbers in a YAML tests and for what? I think it accepts less readable
tests and probably gives no any benefits. I think we should not focus too much on them (or just ban).

Speaking about decimal version of -0xffffffff: I have a test in this diff:

## Addend == -(2^32 - 1)
# RUN: yaml2obj --docnum=2 %s -o %t32.decimal.min -DADDEND=-4294967295
# RUN: llvm-readobj -r %t32.decimal.min | FileCheck %s --check-prefix=TEST -DADDEND=0x1
# RUN: yaml2obj --docnum=2 %s -o %t32.hex.min -DADDEND=-0xFFFFFFFF
# RUN: llvm-readobj -r %t32.hex.min | FileCheck %s --check-prefix=TEST -DADDEND=0x1

-0xffffffff is accepted for 32-bits platform and reads the same as -4294967295, i.e. as 0x1.

While I understand why you think that -0xffffffff` should be restricted in this case,
the current behavior looks acceptable for me in general.

I need this patch for D75671 which wants to do Addend: 0xffffffffffffffff
for example and fails:

YAML:17:17: error: invalid number
        Addend: 0xffffffffffffffff
                ^~~~~~~~~~~~~~~~~~
yaml2obj: error: failed to parse YAML input: Invalid argument

This is a problem I am trying to solve first of all.

I do take inspiration from lld's checkIntUInt: D14957 D45051 D63690.

For example, for an ELFCLASS32 object, an addend of -0xffffffff or -0x80000001 is invalid.

I don't differentiate a decimal and a hexadecimal. Put it in another way, I think an addend of -4294967295 or -2147483649 is invalid. An addend is used to represent an address relative to a symbol or the load base by an offset. The offset can be as small as -2147483648 but it cannot be smaller than that.

[jhenderson]

Latest updates look good to me, but holding off another LGTM, given the range comments.

I think I can be persuaded that -0x12345678 should not be allowed. I guess very few people would write that, so emitting an error is probably sufficient. I've given it some more thought and I'm okay with rejecting ranges that can't be physically represented in the field too, although I'd allow the maximum/minimum possible values (i.e. INT64_MIN/UINT64_MAX, and possibly 32-bit equivalents). Basically, I'll defer to either of you two on this one. I don't have any strong preferences.

[grimar]

• Reimplemented in according to latest discussion.
• Renamed the new type to YAMLIntUInt.

[grimar]

• Removed 2 minor unrelated cleanup changes.

[grimar]

@jhenderson, does this version LGTY?

llvm/lib/ObjectYAML/ELFYAML.cpp

996

I'll change this to "might want to use them".

[jhenderson]

LGTM, with a couple of small nits.

llvm/lib/ObjectYAML/ELFYAML.cpp

996

Let's be more specific, since I think the real reason is that it is ambiguous what to do with them:

"We do not accept negative hex numbers because their meaning is ambiguous. For example, would -0xfffffffff mean 1 or INT32_MIN?"

llvm/test/tools/yaml2obj/ELF/relocation-addend.yaml

7–8

I don't think you need the i64/ui64 bits of the values.

Also, why not just -9223372036854775808?

72–73

Same as above.

[grimar]

llvm/test/tools/yaml2obj/ELF/relocation-addend.yaml

7–8

Also, why not just -9223372036854775808?

I've just took it from my stdint.h:

// These macros must exactly match those in the Windows SDK's intsafe.h.
#define INT8_MIN         (-127i8 - 1)
#define INT16_MIN        (-32767i16 - 1)
#define INT32_MIN        (-2147483647i32 - 1)
#define INT64_MIN        (-9223372036854775807i64 - 1)
#define INT8_MAX         127i8
#define INT16_MAX        32767i16
#define INT32_MAX        2147483647i32
#define INT64_MAX        9223372036854775807i64
#define UINT8_MAX        0xffui8
#define UINT16_MAX       0xffffui16
#define UINT32_MAX       0xffffffffui32
#define UINT64_MAX       0xffffffffffffffffui64

I do not mind to change it.

llvm/include/llvm/ObjectYAML/ELFYAML.h

 LLVM_YAML_STRONG_TYPEDEF(uint8_t, MIPS_AFL_REG)
 LLVM_YAML_STRONG_TYPEDEF(uint8_t, MIPS_ABI_FP)
 LLVM_YAML_STRONG_TYPEDEF(uint32_t, MIPS_AFL_EXT)
 LLVM_YAML_STRONG_TYPEDEF(uint32_t, MIPS_AFL_ASE)
 LLVM_YAML_STRONG_TYPEDEF(uint32_t, MIPS_AFL_FLAGS1)
 LLVM_YAML_STRONG_TYPEDEF(uint32_t, MIPS_ISA)
 
 LLVM_YAML_STRONG_TYPEDEF(StringRef, YAMLFlowString)
+LLVM_YAML_STRONG_TYPEDEF(int64_t, YAMLIntUInt)
 
 // For now, hardcode 64 bits everywhere that 32 or 64 would be needed
 // since 64-bit can hold 32-bit values too.
 struct FileHeader {
   ELF_ELFCLASS Class;
   ELF_ELFDATA Data;
   ELF_ELFOSABI OSABI;
   llvm::yaml::Hex8 ABIVersion;
struct Group : Section {
 
   Group() : Section(ChunkKind::Group) {}
 
   static bool classof(const Chunk *S) { return S->Kind == ChunkKind::Group; }
 };
 
 struct Relocation {
   llvm::yaml::Hex64 Offset;
-  int64_t Addend;
+  YAMLIntUInt Addend;
   ELF_REL Type;
   Optional<StringRef> Symbol;
 };
 
 struct RelocationSection : Section {
   std::vector<Relocation> Relocations;
   StringRef RelocatableSec; /* Info */
 

 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::ELFYAML::VerneedEntry)
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::ELFYAML::Relocation)
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::ELFYAML::SectionOrType)
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::ELFYAML::SectionName)
 
 namespace llvm {
 namespace yaml {
 
+template <> struct ScalarTraits<ELFYAML::YAMLIntUInt> {
+  static void output(const ELFYAML::YAMLIntUInt &Val, void *Ctx,
+                     raw_ostream &Out);
+  static StringRef input(StringRef Scalar, void *Ctx,
+                         ELFYAML::YAMLIntUInt &Val);
+  static QuotingType mustQuote(StringRef) { return QuotingType::None; }
+};
+
 template <>
 struct ScalarEnumerationTraits<ELFYAML::ELF_ET> {
   static void enumeration(IO &IO, ELFYAML::ELF_ET &Value);
 };
 
 template <> struct ScalarEnumerationTraits<ELFYAML::ELF_PT> {
   static void enumeration(IO &IO, ELFYAML::ELF_PT &Value);
 };

llvm/lib/ObjectYAML/ELFYAML.cpp

struct NormalizedOther {
 
   IO &YamlIO;
   Optional<std::vector<StOtherPiece>> Other;
   std::string UnknownFlagsHolder;
 };
 
 } // end anonymous namespace
 
+void ScalarTraits<ELFYAML::YAMLIntUInt>::output(const ELFYAML::YAMLIntUInt &Val,
+                                                void *Ctx, raw_ostream &Out) {
+  Out << Val;
+}
+
+StringRef ScalarTraits<ELFYAML::YAMLIntUInt>::input(StringRef Scalar, void *Ctx,
+                                                    ELFYAML::YAMLIntUInt &Val) {
+  const bool Is64 = static_cast<ELFYAML::Object *>(Ctx)->Header.Class ==
+                    ELFYAML::ELF_ELFCLASS(ELF::ELFCLASS64);
+  StringRef ErrMsg = "invalid number";
+  // We do not accept negative hex numbers for simplicity
+  // because perhaps nobody might want to them.
+  if (Scalar.empty() || Scalar.startswith("-0x"))
+    return ErrMsg;
+
+  if (Scalar.startswith("-")) {
+    const int64_t MinVal = Is64 ? INT64_MIN : INT32_MIN;
+    long long Int;
+    if (getAsSignedInteger(Scalar, /*Radix=*/0, Int) || (Int < MinVal))
+      return ErrMsg;
+    Val = Int;
+    return "";
+  }
+
+  const uint64_t MaxVal = Is64 ? UINT64_MAX : UINT32_MAX;
+  unsigned long long UInt;
+  if (getAsUnsignedInteger(Scalar, /*Radix=*/0, UInt) || (UInt > MaxVal))
+    return ErrMsg;
+  Val = UInt;
+  return "";
+}
+
 void MappingTraits<ELFYAML::Symbol>::mapping(IO &IO, ELFYAML::Symbol &Symbol) {
   IO.mapOptional("Name", Symbol.Name, StringRef());
   IO.mapOptional("StName", Symbol.StName);
   IO.mapOptional("Type", Symbol.Type, ELFYAML::ELF_STT(0));
   IO.mapOptional("Section", Symbol.Section, StringRef());
   IO.mapOptional("Index", Symbol.Index);
   IO.mapOptional("Binding", Symbol.Binding, ELFYAML::ELF_STB(0));
   IO.mapOptional("Value", Symbol.Value, Hex64(0));
MappingNormalization<NormalizedMips64RelType, ELFYAML::ELF_REL> Key(
         IO, Rel.Type);
     IO.mapRequired("Type", Key->Type);
     IO.mapOptional("Type2", Key->Type2, ELFYAML::ELF_REL(ELF::R_MIPS_NONE));
     IO.mapOptional("Type3", Key->Type3, ELFYAML::ELF_REL(ELF::R_MIPS_NONE));
     IO.mapOptional("SpecSym", Key->SpecSym, ELFYAML::ELF_RSS(ELF::RSS_UNDEF));
   } else
     IO.mapRequired("Type", Rel.Type);
 
-  IO.mapOptional("Addend", Rel.Addend, (int64_t)0);
+  IO.mapOptional("Addend", Rel.Addend, (ELFYAML::YAMLIntUInt)0);
 }
 
 void MappingTraits<ELFYAML::Object>::mapping(IO &IO, ELFYAML::Object &Object) {
   assert(!IO.getContext() && "The IO context is initialized already");
   IO.setContext(&Object);
   IO.mapTag("!ELF", true);
   IO.mapRequired("FileHeader", Object.Header);
   IO.mapOptional("ProgramHeaders", Object.ProgramHeaders);

llvm/test/tools/yaml2obj/ELF/relocation-addend.yaml

 ## Here we document how yaml2obj handles relocation addend descriptions.
 
 ## Case 1: Check a 64-bit object.
 
-## Case 1.1: Document we accept an addend with the
-## value INT64_MAX = 2^63-1 = 0x7FFFFFFFFFFFFFFF = 9223372036854775807.
+## Case 1.1: Document we accept any hex/decimal addends in [INT64_MIN, UINT64_MAX].
 
-# RUN: yaml2obj %s -o %t1 -D ADDEND=9223372036854775807
-# RUN: llvm-readobj -r %t1 | FileCheck %s --check-prefix=MAX64
-# RUN: yaml2obj %s -o %t2 -D ADDEND=0x7FFFFFFFFFFFFFFF
-# RUN: llvm-readobj -r %t2 | FileCheck %s --check-prefix=MAX64
+## INT64_MIN   == (-9223372036854775807i64 - 1)
+## UINT64_MAX  == 0xffffffffffffffffui64
 
-# MAX64: 0x0 R_X86_64_PC32 foo 0x7FFFFFFFFFFFFFFF
+## Addend == UINT64_MAX.
+# RUN: yaml2obj %s -o %t64.decimal.max -DADDEND=18446744073709551615
+# RUN: llvm-readobj -r %t64.decimal.max | FileCheck %s --check-prefix=TEST -DADDEND=0xFFFFFFFFFFFFFFFF
+# RUN: yaml2obj %s -o %t64.hex.max -DADDEND=0xFFFFFFFFFFFFFFFF
+# RUN: llvm-readobj -r %t64.hex.max | FileCheck %s --check-prefix=TEST -DADDEND=0xFFFFFFFFFFFFFFFF
+
+## Addend == first positive integer.
+# RUN: yaml2obj %s -o %t64.decimal.first.pos -DADDEND=1
+# RUN: llvm-readobj -r %t64.decimal.first.pos | FileCheck %s --check-prefix=TEST -DADDEND=0x1
+# RUN: yaml2obj %s -o %t64.hex.first.pos -DADDEND=0x1
+# RUN: llvm-readobj -r %t64.hex.first.pos | FileCheck %s --check-prefix=TEST -DADDEND=0x1
+
+## Addend == 0.
+# RUN: yaml2obj %s -o %t64.decimal.null -DADDEND=0
+# RUN: llvm-readobj -r %t64.decimal.null | FileCheck %s --check-prefix=TEST -DADDEND=0x0
+# RUN: yaml2obj %s -o %t64.hex.null -DADDEND=0x0
+# RUN: llvm-readobj -r %t64.hex.null | FileCheck %s --check-prefix=TEST -DADDEND=0x0
+
+## Addend == first negative integer.
+# RUN: yaml2obj %s -o %t64.decimal.first.neg -DADDEND=-1
+# RUN: llvm-readobj -r %t64.decimal.first.neg | FileCheck %s --check-prefix=TEST -DADDEND=0xFFFFFFFFFFFFFFFF
+## We do not accept negative hex addends.
+# RUN: not yaml2obj %s -o /dev/null -DADDEND=-0x1 2>&1 | FileCheck %s --check-prefix=ERR
+
+## Addend == INT64_MIN.
+# RUN: yaml2obj %s -o %t64.decimal.min -DADDEND=-9223372036854775808
+# RUN: llvm-readobj -r %t64.decimal.min | FileCheck %s --check-prefix=TEST -DADDEND=0x8000000000000000
+# TEST: 0x0 R_{{.*}}_PC32 foo [[ADDEND]]
+
+# Case 1.2: Document we do not accept any hex/decimal addends outside of the range specified.
+
+## Addend == 2^64.
+# RUN: not yaml2obj %s -o /dev/null -DADDEND=18446744073709551616 2>&1 | FileCheck %s --check-prefix=ERR
+# RUN: not yaml2obj %s -o /dev/null -DADDEND=0x10000000000000000 2>&1 | FileCheck %s --check-prefix=ERR
 
-## Case 1.2: Check we report an error when an addend is greater than INT64_MAX and
-## it is in decimal form. We use (INT64_MAX + 1).
-# RUN: not yaml2obj %s -o %t3 -D ADDEND=9223372036854775808 2>&1 | FileCheck %s --check-prefix=OVERFLOW64
+## Addend == INT64_MIN - 1.
+# RUN: not yaml2obj %s -o /dev/null -DADDEND=-9223372036854775809 2>&1 | FileCheck %s --check-prefix=ERR
 
-# OVERFLOW64: error: invalid number
-
-## Case 1.3: Document we accept an addend with the
-## value INT64_MIN = -2^63 = 0x8000000000000000 = -9223372036854775808.
-
-# RUN: yaml2obj %s -o %t3 -D ADDEND=-9223372036854775808
-# RUN: llvm-readobj -r %t3 | FileCheck %s --check-prefix=MIN64
-
-# MIN64: 0x0 R_X86_64_PC32 foo 0x8000000000000000
-
-## FIXME: We should support the following case instead.
-# RUN: not yaml2obj %s -o /dev/null -D ADDEND=0x8000000000000000 2>&1 | FileCheck %s --check-prefix=OVERFLOW64
-
-## Case 1.4: Check we report an error when an addend is less than INT64_MIN and
-## it is in decimal form. We use (INT64_MIN - 1).
-# RUN: not yaml2obj %s -o /dev/null -D ADDEND=-9223372036854775809 2>&1 | FileCheck %s --check-prefix=OVERFLOW64
+# ERR: invalid number
 
 --- !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
     Info: .text
     Link: .symtab
     Relocations:
       - Type:   R_X86_64_PC32
         Symbol: foo
         Addend: [[ADDEND]]
 Symbols:
   - Name: foo
 
 ## Case 2: Check a 32-bit object.
 
-## Case 2.1: Document we accept an addend with the
-## value INT32_MAX = 2^31-1 = 0x7FFFFFFF = 2,147,483,647.
-
-# RUN: yaml2obj --docnum=2 %s -o %t4 -D ADDEND=2147483647
-# RUN: llvm-readobj -r %t4 | FileCheck %s --check-prefix=MAX32
-# RUN: yaml2obj --docnum=2 %s -o %t5 -D ADDEND=0x7FFFFFFF
-# RUN: cmp %t4 %t5
-
-# MAX32: 0x0 R_386_PC32 foo 0x7FFFFFFF{{$}}
+## INT32_MIN   == (-2147483647i32 - 1)
+## UINT32_MAX  == 0xffffffffui32
 
-## Case 2.2: Check we report an error when an addend is greater than INT32_MAX and
-## it is specified in decimal form. We use (INT32_MAX + 1).
+## Case 2.1: Document we accept any hex/decimal addends in [INT32_MIN, UINT32_MAX].
 
-## FIXME: The following case should fail, see OVERFLOW64.
-# RUN: yaml2obj --docnum=2 %s -o %t6 -D ADDEND=2147483648
-# RUN: llvm-readobj -r %t6 | FileCheck %s --check-prefix=OVERFLOW32-1
+## Addend == UINT32_MAX.
+# RUN: yaml2obj --docnum=2 %s -o %t32.decimal.max -DADDEND=4294967295
+# RUN: llvm-readobj -r %t32.decimal.max | FileCheck %s --check-prefix=TEST -DADDEND=0xFFFFFFFF
+# RUN: yaml2obj --docnum=2 %s -o %t32.hex.max -DADDEND=0xFFFFFFFF
+# RUN: llvm-readobj -r %t32.hex.max | FileCheck %s --check-prefix=TEST -DADDEND=0xFFFFFFFF
+
+## Addend == first positive integer.
+# RUN: yaml2obj --docnum=2 %s -o %t32.decimal.first.pos -DADDEND=1
+# RUN: llvm-readobj -r %t32.decimal.first.pos | FileCheck %s --check-prefix=TEST -DADDEND=0x1
+# RUN: yaml2obj --docnum=2 %s -o %t32.hex.first.pos -DADDEND=0x1
+# RUN: llvm-readobj -r %t32.hex.first.pos | FileCheck %s --check-prefix=TEST -DADDEND=0x1
+
+## Addend == 0.
+# RUN: yaml2obj --docnum=2 %s -o %t32.decimal.null -DADDEND=0
+# RUN: llvm-readobj -r %t32.decimal.null | FileCheck %s --check-prefix=TEST -DADDEND=0x0
+# RUN: yaml2obj --docnum=2 %s -o %t32.hex.null -DADDEND=0x0
+# RUN: llvm-readobj -r %t32.hex.null | FileCheck %s --check-prefix=TEST -DADDEND=0x0
+
+## Addend == first negative integer.
+# RUN: yaml2obj --docnum=2 %s -o %t32.decimal.first.neg -DADDEND=-1
+# RUN: llvm-readobj -r %t32.decimal.first.neg | FileCheck %s --check-prefix=TEST -DADDEND=0xFFFFFFFF
+## We do not accept negative hex addends.
+# RUN: not yaml2obj --docnum=2 %s -o /dev/null -DADDEND=-0x1 2>&1 | FileCheck %s --check-prefix=ERR
+
+## Addend == INT32_MIN
+# RUN: yaml2obj --docnum=2 %s -o %t32.decimal.min -DADDEND=-2147483648
+# RUN: llvm-readobj -r %t32.decimal.min | FileCheck %s --check-prefix=TEST -DADDEND=0x80000000
+
+# Case 2.2: Document we do not accept any hex/decimal addends outside of the range specified.
+
+## Addend == 2^32.
+# RUN: not yaml2obj --docnum=2 %s -o /dev/null -DADDEND=4294967296 2>&1 | FileCheck %s --check-prefix=ERR
+# RUN: not yaml2obj --docnum=2 %s -o /dev/null -DADDEND=0x100000000 2>&1 | FileCheck %s --check-prefix=ERR
 
-# OVERFLOW32-1: 0x0 R_386_PC32 foo 0x80000000{{$}}
-
-## Case 2.3: Document we accept an addend with the
-## value INT32_MIN = -2^31 = 0x80000000 = -2,147,483,648.
-
-# RUN: yaml2obj --docnum=2 %s -o %t7 -D ADDEND=-2147483648
-# RUN: llvm-readobj -r %t7 | FileCheck %s --check-prefix=MIN32
-# RUN: yaml2obj --docnum=2 %s -o %t8 -D ADDEND=0x80000000
-# RUN: cmp %t7 %t8
-
-# MIN32: 0x0 R_386_PC32 foo 0x80000000{{$}}
-
-## Case 2.4: Check we report an error when an addend is less than INT32_MIN and
-## it is in decimal form. We use (INT32_MIN - 1).
-
-## FIXME: The following case should fail, see OVERFLOW64.
-# RUN: yaml2obj --docnum=2 %s -o %t9 -D ADDEND=-2147483649
-# RUN: llvm-readobj -r %t9 | FileCheck %s --check-prefix=OVERFLOW32-2
-
-# OVERFLOW32-2: 0x0 R_386_PC32 foo 0x7FFFFFFF{{$}}
+## Addend == INT32_MIN - 1.
+# RUN: not yaml2obj --docnum=2 %s -o /dev/null -DADDEND=-2147483649 2>&1 | FileCheck %s --check-prefix=ERR
 
 --- !ELF
 FileHeader:
   Class:   ELFCLASS32
   Data:    ELFDATA2LSB
   Type:    ET_REL
   Machine: EM_386
 Sections:
   - Name: .text
     Type: SHT_PROGBITS
   - Name: .rela.text
     Type: SHT_RELA
     Info: .text
     Link: .symtab
     Relocations:
       - Type:   R_386_PC32
         Symbol: foo
         Addend: [[ADDEND]]
 Symbols:
   - Name: foo
+
+## Case 3: Check we do not allow invalid values.
+# RUN: not yaml2obj %s -D ADDEND=0x1122GGEE 2>&1 | FileCheck %s --check-prefix=ERR
+# RUN: not yaml2obj %s -D ADDEND=-0x1122GGEE 2>&1 | FileCheck %s --check-prefix=ERR
+# RUN: not yaml2obj %s -D ADDEND=1234G5 2>&1 | FileCheck %s --check-prefix=ERR
+# RUN: not yaml2obj %s -D ADDEND=-1234G5 2>&1 | FileCheck %s --check-prefix=ERR
+# RUN: not yaml2obj %s -D ADDEND=foo 2>&1 | FileCheck %s --check-prefix=ERR
+# RUN: not yaml2obj %s -D ADDEND=- 2>&1 | FileCheck %s --check-prefix=ERR
+# RUN: not yaml2obj %s -D ADDEND=--1234 2>&1 | FileCheck %s --check-prefix=ERR