[llvm-objcopy] Default --output-target to --input-target when unspecified

[MaskRay]

Summary

Fixes PR42171.

In GNU objcopy, if -O (--output-target) is not specified, the value is
copied from -I (--input-target).

objcopy -I binary -B i386:x86-64 a.txt b       # b is copied from a.txt
llvm-objcopy -I binary -B i386:x86-64 a.txt b  # b is an x86-64 object file

This patch changes our behavior to match GNU. GNU objcopy has another
strange behavior:

objcopy -I binary -O elf64-x86-64 a.txt b.o                # e_machine is EM_NONE
objcopy -I binary -B i386:x86-64 -O elf64-x86-64 a.txt b.o # e_machine is EM_X86_64

It is documented that -B is not used unless -I binary. Then, it is
unclear why the e_machine field cannot be inferred from -O. We just
fill in the e_machine field regardless of -B.

[jhenderson]

Would you mind explaining the purpose of the -B switch now? I'm just struggling to grasp it a little at the moment.

test/tools/llvm-objcopy/ELF/binary-input-and-output.test

11 ↗

(On Diff #218596)

Nit: are specified -> are both specified.

15 ↗

(On Diff #218596)

Would you also mind adding a second hash to the existing comments in this file for consistency throughout? Happy for that to be a separate NFC commit if you want, or in this one.

If you do that, please add the missing trailing full-stops too.

21 ↗

(On Diff #218596)

Nit: are specified -> are both specified.

Perhaps also worth adding "for non-binary output" or similar, to distinguish this case from the above similar one.

test/tools/llvm-objcopy/ELF/binary-input-error.test

3–4 ↗

(On Diff #218596)

This is valid behaviour now, right? We need a test case to show this works.

test/tools/llvm-objcopy/ELF/binary-input.test

4 ↗

(On Diff #218596)

I think you can remove the -B here, right?

test/tools/llvm-objcopy/ELF/new-symbol-visibility.test

4 ↗

(On Diff #218596)

Can you remove the -B here and elsewhere in this test too?

[MaskRay]

Address review comments

[MaskRay]

binutils-gdb/binutils/NEWS says

* New command-line switch to objcopy -B (or --binary-architecture) which sets
  the architecture of the output file to the given argument.  This option only
  makes sense, if the input target is binary.  Otherwise it is ignored.
  By Stefan Geuken.

I assume it means -B is only useful when -I binary is used. Now there are 2 cases:

• -I binary -O binary => it is a copy. -B is ignored.
• -I binary -O elf64-x86-64 => elf64-x86-64 makes the output little-endian and 64-bit, but e_machine is not set. -B i386:x86-64 is supposed to set the e_machine field. This means if -I binary is used and -O is not binary, -B should always be set. I don't know why it behaves that way. I guess we may be able to ignore -B entirely. I'll do that in the next patch.

% file b.o
b.o: ELF 64-bit LSB relocatable, no machine, version 1 (SYSV), not stripped

test/tools/llvm-objcopy/ELF/binary-input-error.test

3–4 ↗

(On Diff #218596)

Added a test to binary-input-and-output.test

## If -O is not specified, it defaults to -I, i.e. "binary".
# RUN: llvm-objcopy -I binary %t.txt %t.3.txt
# RUN: cmp %t.txt %t.3.txt

[jhenderson]

test/tools/llvm-objcopy/ELF/binary-input-and-output.test

22 ↗

(On Diff #218635)

%t.3.txt -> %t.5.txt here and the line below.

test/tools/llvm-objcopy/ELF/binary-output-target.test

18–19 ↗

(On Diff #218635)

Was this case just missing before? Maybe worth making this change independently.

[MaskRay]

%t.3.txt -> %t.5.txt

Delete a powerpc32 test which did not exist before

test/tools/llvm-objcopy/ELF/binary-output-target.test

18–19 ↗

(On Diff #218635)

It was missed before. Will add it separately.

[jakehehrlich]

So it seems like we've been preferring the architecture implied by -O instead of the one preferred by -B. My intuition is that it should be the opposite. Did we find a reason to prefer using -O?

[jakehehrlich]

If our intent is to prefer using -O over -B when possible (as seems to be the case) then this all looks good to me. I just want an explanation if one exists.

[MaskRay]

In D67144#1658101, @jakehehrlich wrote:

If our intent is to prefer using -O over -B when possible (as seems to be the case) then this all looks good to me. I just want an explanation if one exists.

I prefer -O to -B. -O seems more natural to describe the desired output format (O -> output, the opposite of -I -> input).

{llvm-,}objcopy -O binary a.o a.txt can extract the text file. There is no -B alternative, so -O has to be kept around.

After this change, we can essentially ignore -B, like what elftoolchain (used by FreeBSD) does:

case 'B':
  /* ignored */
  break;
  ...
case 'I':
case 's':
  set_input_target(ecp, optarg);
  break;
  ...
case 'O':
  set_output_target(ecp, optarg);
  break;

% elfcopy/elfcopy -I binary -O elf64-x86-64 /tmp/c/a.txt /tmp/c/a.o; file /tmp/c/a.o
/tmp/c/a.o: ELF 64-bit LSB relocatable, x86-64, version 1 (SYSV), not stripped

# I have tried a very old version of binutils (~2003). The weird behavior has been around for many years...
% /tmp/p/binutils-2.14/Release/binutils/objcopy -I binary -O elf64-x86-64 a.txt a.o; file a.o  
a.o: ELF 64-bit LSB relocatable, no machine, version 1 (SYSV), not stripped
% /tmp/p/binutils-2.14/Release/binutils/objcopy -I binary -B i386:x86-64 -O elf64-x86-64 a.txt a.o; file a.o
a.o: ELF 64-bit LSB relocatable, x86-64, version 1 (SYSV), not stripped

[MaskRay]

@jakehehrlich Ping :)

[jakehehrlich]

I'm happy with this since the votes seem to favor this behavior.

llvm/trunk/test/tools/llvm-objcopy/ELF/binary-input-and-output.test

 # RUN: echo abcd > %t.txt
 
 # Preserve input to verify it is not modified
 # RUN: cp %t.txt %t-copy.txt
 
-# -I binary -O binary preserves payload through in-memory representation
-# RUN: llvm-objcopy -I binary -B i386:x86-64 -O binary %t.txt %t.2.txt
+## -I binary -O binary preserves payload through in-memory representation.
+# RUN: llvm-objcopy -I binary -O binary %t.txt %t.2.txt
 # RUN: cmp %t.txt %t.2.txt
 # RUN: cmp %t.txt %t-copy.txt
 
-# -I binary -O binary preserves payload through an intermediate object file
-# RUN: llvm-objcopy -I binary -B i386:x86-64 %t.txt %t.o
-# RUN: llvm-objcopy -O binary %t.o %t.3.txt
+## If -O is not specified, it defaults to -I, i.e. "binary".
+# RUN: llvm-objcopy -I binary %t.txt %t.3.txt
 # RUN: cmp %t.txt %t.3.txt
-# RUN: cmp %t.txt %t-copy.txt
+
+## -I binary -O elf* creates an ELF. -O binary transforms it back to the
+## original file.
+# RUN: llvm-objcopy -I binary -O elf64-x86-64 %t.txt %t.o
+# RUN: llvm-objcopy -O binary %t.o %t.4.txt
+# RUN: cmp %t.txt %t.4.txt
+
+## -B is ignored.
+# RUN: llvm-objcopy -I binary -B i386:x86-64 -O binary %t.txt %t.5.txt
+# RUN: cmp %t.txt %t.5.txt

llvm/trunk/test/tools/llvm-objcopy/ELF/binary-input-arch.test

-# RUN: echo abcd > %t.txt
-
-# RUN: llvm-objcopy -I binary -B aarch64 %t.txt %t.aarch64.o
-# RUN: llvm-readobj --file-headers %t.aarch64.o | FileCheck %s --check-prefixes=CHECK,LE,AARCH64,64
-
-# RUN: llvm-objcopy -I binary -B arm %t.txt %t.arm.o
-# RUN: llvm-readobj --file-headers %t.arm.o | FileCheck %s --check-prefixes=CHECK,LE,ARM,32
-
-# RUN: llvm-objcopy -I binary -B i386 %t.txt %t.i386.o
-# RUN: llvm-readobj --file-headers %t.i386.o | FileCheck %s --check-prefixes=CHECK,LE,I386,32
-
-# RUN: llvm-objcopy -I binary -B i386:x86-64 %t.txt %t.i386_x86-64.o
-# RUN: llvm-readobj --file-headers %t.i386_x86-64.o | FileCheck %s --check-prefixes=CHECK,LE,X86-64,64
-
-# RUN: llvm-objcopy -I binary -B mips %t.txt %t.mips.o
-# RUN: llvm-readobj --file-headers %t.mips.o | FileCheck %s --check-prefixes=CHECK,BE,MIPS,32
-
-# RUN: llvm-objcopy -I binary -B powerpc:common64 %t.txt %t.powerpc_common64.o
-# RUN: llvm-readobj --file-headers %t.powerpc_common64.o | FileCheck %s --check-prefixes=CHECK,LE,PPC,64
-
-# RUN: llvm-objcopy -I binary -B riscv:rv32 %t.txt %t.rv32.o
-# RUN: llvm-readobj --file-headers %t.rv32.o | FileCheck %s --check-prefixes=CHECK,LE,RISCV32,32
-
-# RUN: llvm-objcopy -I binary -B riscv:rv64 %t.txt %t.rv64.o
-# RUN: llvm-readobj --file-headers %t.rv64.o | FileCheck %s --check-prefixes=CHECK,LE,RISCV64,64
-
-# RUN: llvm-objcopy -I binary -B sparc %t.txt %t.sparc.o
-# RUN: llvm-readobj --file-headers %t.sparc.o | FileCheck %s --check-prefixes=CHECK,BE,SPARC,32
-
-# RUN: llvm-objcopy -I binary -B sparcel %t.txt %t.sparcel.o
-# RUN: llvm-readobj --file-headers %t.sparcel.o | FileCheck %s --check-prefixes=CHECK,LE,SPARCEL,32
-
-# RUN: llvm-objcopy -I binary -B x86-64 %t.txt %t.x86-64.o
-# RUN: llvm-readobj --file-headers %t.x86-64.o | FileCheck %s --check-prefixes=CHECK,LE,X86-64,64
-
-# CHECK: Format:
-# 32-SAME:      ELF32-
-# 64-SAME:      ELF64-
-# AARCH64-SAME: aarch64-little
-# ARM-SAME:     arm-little
-# I386-SAME:    i386
-# MIPS-SAME:    mips{{$}}
-# RISCV32-SAME: riscv{{$}}
-# RISCV64-SAME: riscv{{$}}
-# PPC-SAME:     ppc64
-# SPARC-SAME:   sparc
-# SPARCEL-SAME: sparc
-# X86-64-SAME:  x86-64
-
-# AARCH64-NEXT: Arch: aarch64
-# ARM-NEXT:     Arch: arm
-# I386-NEXT:    Arch: i386
-# MIPS-NEXT:    Arch: mips{{$$}}
-# PPC-NEXT:     Arch: powerpc64le
-# RISCV32-NEXT: Arch: riscv32
-# RISCV64-NEXT: Arch: riscv64
-# SPARC-NEXT:   Arch: sparc{{$}}
-# SPARCEL-NEXT: Arch: sparcel
-# X86-64-NEXT:  Arch: x86_64
-
-# 32-NEXT:      AddressSize: 32bit
-# 64-NEXT:      AddressSize: 64bit
-
-# CHECK:        ElfHeader {
-# CHECK-NEXT:     Ident {
-# CHECK-NEXT:       Magic: (7F 45 4C 46)
-# 32-NEXT:          Class: 32-bit (0x1)
-# 64-NEXT:          Class: 64-bit (0x2)
-# LE-NEXT:          DataEncoding: LittleEndian (0x1)
-# BE-NEXT:          DataEncoding: BigEndian (0x2)
-# CHECK-NEXT:       FileVersion: 1
-# CHECK-NEXT:       OS/ABI: SystemV (0x0)
-# CHECK-NEXT:       ABIVersion: 0
-# CHECK-NEXT:       Unused: (00 00 00 00 00 00 00)
-# CHECK-NEXT:     }
-# CHECK-NEXT:     Type: Relocatable (0x1)
-# AARCH64-NEXT:   Machine: EM_AARCH64 (0xB7)
-# ARM-NEXT:       Machine: EM_ARM (0x28)
-# I386-NEXT:      Machine: EM_386 (0x3)
-# MIPS-NEXT:      Machine: EM_MIPS (0x8)
-# PPC-NEXT:       Machine: EM_PPC64 (0x15)
-# RISCV32-NEXT:   Machine: EM_RISCV (0xF3)
-# RISCV64-NEXT:   Machine: EM_RISCV (0xF3)
-# SPARC-NEXT:     Machine: EM_SPARC (0x2)
-# SPARCEL-NEXT:   Machine: EM_SPARC (0x2)
-# X86-64-NEXT:    Machine: EM_X86_64 (0x3E)
-# CHECK-NEXT:     Version: 1
-# CHECK-NEXT:     Entry: 0x0
-# CHECK-NEXT:     ProgramHeaderOffset:
-# CHECK-NEXT:     SectionHeaderOffset:
-# CHECK-NEXT:     Flags [ (0x0)
-# CHECK-NEXT:     ]
-# 32-NEXT:        HeaderSize: 52
-# 64-NEXT:        HeaderSize: 64
-# 32-NEXT:        ProgramHeaderEntrySize: 0
-# 64-NEXT:        ProgramHeaderEntrySize: 0
-# CHECK-NEXT:     ProgramHeaderCount: 0
-# 32-NEXT:        SectionHeaderEntrySize: 40
-# 64-NEXT:        SectionHeaderEntrySize: 64
-# CHECK-NEXT:     SectionHeaderCount: 4
-# CHECK-NEXT:     StringTableSectionIndex:
-# CHECK-NEXT:   }

llvm/trunk/test/tools/llvm-objcopy/ELF/binary-input-error.test

 # RUN: echo abcd > %t.txt
 
-# RUN: not llvm-objcopy -I binary %t.txt %t.o 2>&1 \
-# RUN:   | FileCheck %s --check-prefix=MISSING-BINARY-ARCH
-
 # RUN: not llvm-objcopy -I binary -B xyz %t.txt %t.o 2>&1 \
 # RUN:   | FileCheck %s --check-prefix=BAD-BINARY-ARCH
 
-# MISSING-BINARY-ARCH: specified binary input without specifiying an architecture
 # BAD-BINARY-ARCH: invalid architecture: 'xyz'

llvm/trunk/test/tools/llvm-objcopy/ELF/binary-input.test

 # RUN: echo -n abcd > %t.x-txt
 # Preserve input to verify it is not modified
 # RUN: cp %t.x-txt %t-copy.txt
-# RUN: llvm-objcopy -I binary -B i386:x86-64 %t.x-txt %t.o
+# RUN: llvm-objcopy -I binary -O elf64-x86-64 %t.x-txt %t.o
 # RUN: llvm-readobj --sections --symbols %t.o | FileCheck %s
 # RUN: cmp %t.x-txt %t-copy.txt
 
 # CHECK:      Sections [
 # CHECK-NEXT:   Section {
 # CHECK-NEXT:     Index: 0
 # CHECK-NEXT:     Name:  (0)
 # CHECK-NEXT:     Type: SHT_NULL (0x0)

llvm/trunk/test/tools/llvm-objcopy/ELF/binary-output-target.test

+# RUN: echo abcd > %t.txt
+
+# RUN: llvm-objcopy -I binary -O elf64-littleaarch64 %t.txt %t.aarch64.o
+# RUN: llvm-readobj --file-headers %t.aarch64.o | FileCheck %s --check-prefixes=CHECK,LE,AARCH64,64
+
+# RUN: llvm-objcopy -I binary -O elf32-littlearm %t.txt %t.arm.o
+# RUN: llvm-readobj --file-headers %t.arm.o | FileCheck %s --check-prefixes=CHECK,LE,ARM,32
+
+# RUN: llvm-objcopy -I binary -O elf32-i386 %t.txt %t.i386.o
+# RUN: llvm-readobj --file-headers %t.i386.o | FileCheck %s --check-prefixes=CHECK,LE,I386,32
+
+# RUN: llvm-objcopy -I binary -O elf64-x86-64 %t.txt %t.x86-64.o
+# RUN: llvm-readobj --file-headers %t.x86-64.o | FileCheck %s --check-prefixes=CHECK,LE,X86-64,64
+
+# RUN: llvm-objcopy -I binary -O elf32-bigmips %t.txt %t.mips.o
+# RUN: llvm-readobj --file-headers %t.mips.o | FileCheck %s --check-prefixes=CHECK,BE,MIPS,32
+
+# RUN: llvm-objcopy -I binary -O elf64-powerpcle %t.txt %t.ppc64le.o
+# RUN: llvm-readobj --file-headers %t.ppc64le.o | FileCheck %s --check-prefixes=CHECK,LE,PPC64,64
+
+# RUN: llvm-objcopy -I binary -O elf32-littleriscv %t.txt %t.rv32.o
+# RUN: llvm-readobj --file-headers %t.rv32.o | FileCheck %s --check-prefixes=CHECK,LE,RISCV32,32
+
+# RUN: llvm-objcopy -I binary -O elf64-littleriscv %t.txt %t.rv64.o
+# RUN: llvm-readobj --file-headers %t.rv64.o | FileCheck %s --check-prefixes=CHECK,LE,RISCV64,64
+
+# RUN: llvm-objcopy -I binary -O elf32-sparc %t.txt %t.sparc.o
+# RUN: llvm-readobj --file-headers %t.sparc.o | FileCheck %s --check-prefixes=CHECK,BE,SPARC,32
+
+# RUN: llvm-objcopy -I binary -O elf32-sparcel %t.txt %t.sparcel.o
+# RUN: llvm-readobj --file-headers %t.sparcel.o | FileCheck %s --check-prefixes=CHECK,LE,SPARCEL,32
+
+# CHECK: Format:
+# 32-SAME:      ELF32-
+# 64-SAME:      ELF64-
+# AARCH64-SAME: aarch64-little
+# ARM-SAME:     arm-little
+# I386-SAME:    i386
+# MIPS-SAME:    mips{{$}}
+# RISCV32-SAME: riscv{{$}}
+# RISCV64-SAME: riscv{{$}}
+# PPC-SAME:     ppc{{$}}
+# PPC64le-SAME: ppc64
+# SPARC-SAME:   sparc
+# SPARCEL-SAME: sparc
+# X86-64-SAME:  x86-64
+
+# AARCH64-NEXT: Arch: aarch64
+# ARM-NEXT:     Arch: arm
+# I386-NEXT:    Arch: i386
+# MIPS-NEXT:    Arch: mips{{$}}
+# PPC-NEXT:     Arch: powerpc{{$}}
+# PPC64-NEXT:   Arch: powerpc64le
+# RISCV32-NEXT: Arch: riscv32
+# RISCV64-NEXT: Arch: riscv64
+# SPARC-NEXT:   Arch: sparc{{$}}
+# SPARCEL-NEXT: Arch: sparcel
+# X86-64-NEXT:  Arch: x86_64
+
+# 32-NEXT:      AddressSize: 32bit
+# 64-NEXT:      AddressSize: 64bit
+
+# CHECK:        ElfHeader {
+# CHECK-NEXT:     Ident {
+# CHECK-NEXT:       Magic: (7F 45 4C 46)
+# 32-NEXT:          Class: 32-bit (0x1)
+# 64-NEXT:          Class: 64-bit (0x2)
+# LE-NEXT:          DataEncoding: LittleEndian (0x1)
+# BE-NEXT:          DataEncoding: BigEndian (0x2)
+# CHECK-NEXT:       FileVersion: 1
+# CHECK-NEXT:       OS/ABI: SystemV (0x0)
+# CHECK-NEXT:       ABIVersion: 0
+# CHECK-NEXT:       Unused: (00 00 00 00 00 00 00)
+# CHECK-NEXT:     }
+# CHECK-NEXT:     Type: Relocatable (0x1)
+
+## In GNU objcopy, e_machine is EM_NONE unless -B is specifid.
+## It is not very clear why it behaves this way.
+## We fill in the field regardless of -B.
+# AARCH64-NEXT:   Machine: EM_AARCH64 (0xB7)
+# ARM-NEXT:       Machine: EM_ARM (0x28)
+# I386-NEXT:      Machine: EM_386 (0x3)
+# MIPS-NEXT:      Machine: EM_MIPS (0x8)
+# PPC-NEXT:       Machine: EM_PPC (0x14)
+# PPC64-NEXT:     Machine: EM_PPC64 (0x15)
+# RISCV32-NEXT:   Machine: EM_RISCV (0xF3)
+# RISCV64-NEXT:   Machine: EM_RISCV (0xF3)
+# SPARC-NEXT:     Machine: EM_SPARC (0x2)
+# SPARCEL-NEXT:   Machine: EM_SPARC (0x2)
+# X86-64-NEXT:    Machine: EM_X86_64 (0x3E)
+
+# CHECK-NEXT:     Version: 1
+# CHECK-NEXT:     Entry: 0x0
+# CHECK-NEXT:     ProgramHeaderOffset:
+# CHECK-NEXT:     SectionHeaderOffset:
+# CHECK-NEXT:     Flags [ (0x0)
+# CHECK-NEXT:     ]
+# 32-NEXT:        HeaderSize: 52
+# 64-NEXT:        HeaderSize: 64
+# 32-NEXT:        ProgramHeaderEntrySize: 0
+# 64-NEXT:        ProgramHeaderEntrySize: 0
+# CHECK-NEXT:     ProgramHeaderCount: 0
+# 32-NEXT:        SectionHeaderEntrySize: 40
+# 64-NEXT:        SectionHeaderEntrySize: 64
+# CHECK-NEXT:     SectionHeaderCount: 4
+# CHECK-NEXT:     StringTableSectionIndex:
+# CHECK-NEXT:   }

llvm/trunk/test/tools/llvm-objcopy/ELF/new-symbol-visibility.test

 ## Ensure that the visibility of the start, end and size symbols created when
 ## using binary input can be specified with the --new-symbol-visibility switch.
 
-# RUN: llvm-objcopy -I binary -B i386:x86-64 %s %t.unspecified
+# RUN: llvm-objcopy -I binary -O elf64-x86-64 %s %t.unspecified
 # RUN: llvm-readelf -s %t.unspecified | FileCheck %s --check-prefix=BINARY -DVISIBILITY=DEFAULT
-# RUN: llvm-objcopy --new-symbol-visibility default -I binary -B i386:x86-64 %s %t.default
+# RUN: llvm-objcopy --new-symbol-visibility default -I binary -O elf64-x86-64 %s %t.default
 # RUN: llvm-readelf -s %t.default | FileCheck %s --check-prefix=BINARY -DVISIBILITY=DEFAULT
-# RUN: llvm-objcopy --new-symbol-visibility hidden -I binary -B i386:x86-64 %s %t.hidden
+# RUN: llvm-objcopy --new-symbol-visibility hidden -I binary -O elf64-x86-64 %s %t.hidden
 # RUN: llvm-readelf -s %t.hidden | FileCheck %s --check-prefix=BINARY -DVISIBILITY=HIDDEN
-# RUN: llvm-objcopy --new-symbol-visibility protected -I binary -B i386:x86-64 %s %t.protected
+# RUN: llvm-objcopy --new-symbol-visibility protected -I binary -O elf64-x86-64 %s %t.protected
 # RUN: llvm-readelf -s %t.protected | FileCheck %s --check-prefix=BINARY -DVISIBILITY=PROTECTED
-# RUN: llvm-objcopy --new-symbol-visibility internal -I binary -B i386:x86-64 %s %t.internal
+# RUN: llvm-objcopy --new-symbol-visibility internal -I binary -O elf64-x86-64 %s %t.internal
 # RUN: llvm-readelf -s %t.internal | FileCheck %s --check-prefix=BINARY -DVISIBILITY=INTERNAL
 
 # BINARY:      [[VISIBILITY]] {{.*}} _binary{{.*}}_start
 # BINARY-NEXT: [[VISIBILITY]] {{.*}} _binary{{.*}}_end
 # BINARY-NEXT: [[VISIBILITY]] {{.*}} _binary{{.*}}_size
 
-# RUN: not llvm-objcopy --new-symbol-visibility fluff -I binary -B i386:x86-64 %s %t.error 2>&1 \
-# RUN:   | FileCheck %s --check-prefix=ERR
+# RUN: not llvm-objcopy --new-symbol-visibility fluff -I binary -O elf64-x86-64 \
+# RUN:   %s %t.error 2>&1 | FileCheck %s --check-prefix=ERR
 
 # ERR: error: 'fluff' is not a valid symbol visibility

llvm/trunk/tools/llvm-objcopy/CopyConfig.cpp

Expected<DriverConfig> parseObjcopyOptions(ArrayRef<const char *> ArgsArr) {
   // explicitly specified by -I option (e.g. -Ielf32-x86-64) and guess the
   // format by llvm::object::createBinary regardless of the option value.
   Config.InputFormat = StringSwitch<FileFormat>(InputFormat)
                            .Case("binary", FileFormat::Binary)
                            .Case("ihex", FileFormat::IHex)
                            .Default(FileFormat::Unspecified);
   if (Config.InputFormat == FileFormat::Binary) {
     auto BinaryArch = InputArgs.getLastArgValue(OBJCOPY_binary_architecture);
-    if (BinaryArch.empty())
-      return createStringError(
-          errc::invalid_argument,
-          "specified binary input without specifiying an architecture");
+    if (!BinaryArch.empty()) {
       Expected<const MachineInfo &> MI = getMachineInfo(BinaryArch);
       if (!MI)
         return MI.takeError();
       Config.BinaryArch = *MI;
     }
+  }
 
   if (opt::Arg *A = InputArgs.getLastArg(OBJCOPY_new_symbol_visibility)) {
     const uint8_t Invalid = 0xff;
     Config.NewSymbolVisibility = StringSwitch<uint8_t>(A->getValue())
                                      .Case("default", ELF::STV_DEFAULT)
                                      .Case("hidden", ELF::STV_HIDDEN)
                                      .Case("internal", ELF::STV_INTERNAL)
                                      .Case("protected", ELF::STV_PROTECTED)
                                      .Default(Invalid);
 
     if (Config.NewSymbolVisibility == Invalid)
       return createStringError(
           errc::invalid_argument, "'%s' is not a valid symbol visibility",
           InputArgs.getLastArgValue(OBJCOPY_new_symbol_visibility).str().c_str());
   }
 
   Config.OutputFormat = StringSwitch<FileFormat>(OutputFormat)
                             .Case("binary", FileFormat::Binary)
                             .Case("ihex", FileFormat::IHex)
                             .Default(FileFormat::Unspecified);
-  if (Config.OutputFormat == FileFormat::Unspecified && !OutputFormat.empty()) {
-    Expected<TargetInfo> Target = getOutputTargetInfoByTargetName(OutputFormat);
+  if (Config.OutputFormat == FileFormat::Unspecified) {
+    if (OutputFormat.empty()) {
+      Config.OutputFormat = Config.InputFormat;
+    } else {
+      Expected<TargetInfo> Target =
+          getOutputTargetInfoByTargetName(OutputFormat);
       if (!Target)
         return Target.takeError();
       Config.OutputFormat = Target->Format;
       Config.OutputArch = Target->Machine;
     }
+  }
 
   if (auto Arg = InputArgs.getLastArg(OBJCOPY_compress_debug_sections,
                                       OBJCOPY_compress_debug_sections_eq)) {
     Config.CompressionType = DebugCompressionType::Z;
 
     if (Arg->getOption().getID() == OBJCOPY_compress_debug_sections_eq) {
       Config.CompressionType =
           StringSwitch<DebugCompressionType>(