[llvm-objcopy] Add --gap-fill and --pad-to options

llvm/docs/CommandGuide/llvm-objcopy.rst

@@ -324,6 +324,11 @@ them.
 
  Extract the named partition from the output.
 
+.. option:: --gap-fill <value>
+
+ For binary outputs, fill the gaps between sections with ``<value>`` instead
+ of zero. The value must be an unsigned 8-bit integer.
+
 .. option:: --globalize-symbol <symbol>
 
  Mark any defined symbols named ``<symbol>`` as global symbols in the output.
@@ -411,6 +416,11 @@ them.
  be the same as the value specified for :option:`--input-target` or the input
  file's format if that option is also unspecified.
 
+.. option:: --pad-to <address>
+
+ For binary outputs, pad the output to the load address ``<address>`` using a value
+ of zero or the value specified by :option:`--gap-fill`.
+
 .. option:: --prefix-alloc-sections <prefix>
 
  Add ``<prefix>`` to the front of the names of all allocatable sections in the

llvm/docs/ReleaseNotes.rst

@@ -259,6 +259,9 @@ Changes to the LLVM tools
 
 * llvm-symbolizer and llvm-addr2line now support addresses specified as symbol names.
 
+* llvm-objcopy now supports ``--gap-fill`` and ``--pad-to`` options, for
+  ELF input and binary output files only.
+
 Changes to LLDB
 ---------------------------------
 

llvm/include/llvm/ObjCopy/CommonConfig.h

@@ -214,6 +214,8 @@ struct CommonConfig {
   // Cached gnu_debuglink's target CRC
   uint32_t GnuDebugLinkCRC32;
   std::optional<StringRef> ExtractPartition;
+  uint8_t GapFill = 0;
+  uint64_t PadTo = 0;
   StringRef SplitDWO;
   StringRef SymbolsPrefix;
   StringRef AllocSectionsPrefix;

llvm/lib/ObjCopy/ConfigManager.cpp

@@ -23,7 +23,8 @@ Expected<const COFFConfig &> ConfigManager::getCOFFConfig() const {
       Common.ExtractDWO || Common.PreserveDates || Common.StripDWO ||
       Common.StripNonAlloc || Common.StripSections || Common.Weaken ||
       Common.DecompressDebugSections ||
-      Common.DiscardMode == DiscardType::Locals || !Common.SymbolsToAdd.empty())
+      Common.DiscardMode == DiscardType::Locals ||
+      !Common.SymbolsToAdd.empty() || Common.GapFill != 0 || Common.PadTo != 0)
     return createStringError(llvm::errc::invalid_argument,
                              "option is not supported for COFF");
 
@@ -42,7 +43,8 @@ Expected<const MachOConfig &> ConfigManager::getMachOConfig() const {
       Common.PreserveDates || Common.StripAllGNU || Common.StripDWO ||
       Common.StripNonAlloc || Common.StripSections ||
       Common.DecompressDebugSections || Common.StripUnneeded ||
-      Common.DiscardMode == DiscardType::Locals || !Common.SymbolsToAdd.empty())
+      Common.DiscardMode == DiscardType::Locals ||
+      !Common.SymbolsToAdd.empty() || Common.GapFill != 0 || Common.PadTo != 0)
     return createStringError(llvm::errc::invalid_argument,
                              "option is not supported for MachO");
 
@@ -60,7 +62,8 @@ Expected<const WasmConfig &> ConfigManager::getWasmConfig() const {
       !Common.SymbolsToWeaken.empty() || !Common.SymbolsToKeepGlobal.empty() ||
       !Common.SectionsToRename.empty() || !Common.SetSectionAlignment.empty() ||
       !Common.SetSectionFlags.empty() || !Common.SetSectionType.empty() ||
-      !Common.SymbolsToRename.empty())
+      !Common.SymbolsToRename.empty() || Common.GapFill != 0 ||
+      Common.PadTo != 0)
     return createStringError(llvm::errc::invalid_argument,
                              "only flags for section dumping, removal, and "
                              "addition are supported");
@@ -86,7 +89,8 @@ Expected<const XCOFFConfig &> ConfigManager::getXCOFFConfig() const {
       Common.ExtractMainPartition || Common.OnlyKeepDebug ||
       Common.PreserveDates || Common.StripAllGNU || Common.StripDWO ||
       Common.StripDebug || Common.StripNonAlloc || Common.StripSections ||
-      Common.Weaken || Common.StripUnneeded || Common.DecompressDebugSections) {
+      Common.Weaken || Common.StripUnneeded || Common.DecompressDebugSections ||
+      Common.GapFill != 0 || Common.PadTo != 0) {
     return createStringError(
         llvm::errc::invalid_argument,
         "no flags are supported yet, only basic copying is allowed");

llvm/lib/ObjCopy/ELF/ELFObjcopy.cpp

@@ -180,7 +180,7 @@ static std::unique_ptr<Writer> createWriter(const CommonConfig &Config,
                                             ElfType OutputElfType) {
   switch (Config.OutputFormat) {
   case FileFormat::Binary:
-    return std::make_unique<BinaryWriter>(Obj, Out);
+    return std::make_unique<BinaryWriter>(Obj, Out, Config);
   case FileFormat::IHex:
     return std::make_unique<IHexWriter>(Obj, Out);
   default:

llvm/lib/ObjCopy/ELF/ELFObject.cpp

@@ -2636,9 +2636,36 @@ template <class ELFT> Error ELFWriter<ELFT>::finalize() {
 }
 
 Error BinaryWriter::write() {
-  for (const SectionBase &Sec : Obj.allocSections())
+  SmallVector<const SectionBase *, 30> SectionsToWrite;
+  for (const SectionBase &Sec : Obj.allocSections()) {
+    if (Sec.Type != SHT_NOBITS)
+      SectionsToWrite.push_back(&Sec);
+  }
+
+  if (SectionsToWrite.empty())
+    return Error::success();
+
+  llvm::stable_sort(SectionsToWrite,
+                    [](const SectionBase *LHS, const SectionBase *RHS) {
+                      return LHS->Offset < RHS->Offset;
+                    });
+
+  assert(SectionsToWrite.front()->Offset == 0);
+
+  for (size_t i = 0; i != SectionsToWrite.size(); ++i) {
+    const SectionBase &Sec = *SectionsToWrite[i];
     if (Error Err = Sec.accept(*SecWriter))
       return Err;
+    if (GapFill == 0)
+      continue;
+    uint64_t PadOffset = (i < SectionsToWrite.size() - 1)
+                             ? SectionsToWrite[i + 1]->Offset
+                             : Buf->getBufferSize();
+    assert(PadOffset <= Buf->getBufferSize());
+    assert(Sec.Offset + Sec.Size <= PadOffset);
+    std::fill(Buf->getBufferStart() + Sec.Offset + Sec.Size,
+              Buf->getBufferStart() + PadOffset, GapFill);
+  }
 
   // TODO: Implement direct writing to the output stream (without intermediate
   // memory buffer Buf).
@@ -2664,7 +2691,7 @@ Error BinaryWriter::finalize() {
   // file size. This might not be the same as the offset returned by
   // layoutSections, because we want to truncate the last segment to the end of
   // its last non-empty section, to match GNU objcopy's behaviour.
-  TotalSize = 0;
+  TotalSize = PadTo > MinAddr ? PadTo - MinAddr : 0;
   for (SectionBase &Sec : Obj.allocSections())
     if (Sec.Type != SHT_NOBITS && Sec.Size > 0) {
       Sec.Offset = Sec.Addr - MinAddr;

llvm/lib/ObjCopy/ELF/ELFObject.h

@@ -357,6 +357,8 @@ template <class ELFT> class ELFWriter : public Writer {
 
 class BinaryWriter : public Writer {
 private:
+  const uint8_t GapFill;
+  const uint64_t PadTo;
   std::unique_ptr<BinarySectionWriter> SecWriter;
 
   uint64_t TotalSize = 0;
@@ -365,7 +367,8 @@ class BinaryWriter : public Writer {
   ~BinaryWriter() {}
   Error finalize() override;
   Error write() override;
-  BinaryWriter(Object &Obj, raw_ostream &Out) : Writer(Obj, Out) {}
+  BinaryWriter(Object &Obj, raw_ostream &Out, const CommonConfig &Config)
+      : Writer(Obj, Out), GapFill(Config.GapFill), PadTo(Config.PadTo) {}
 };
 
 class IHexWriter : public Writer {

llvm/test/tools/llvm-objcopy/ELF/gap-fill.test

@@ -0,0 +1,176 @@
+# RUN: yaml2obj --docnum=1 %s -o %t
+
+# RUN: not llvm-objcopy --gap-fill 1 %t 2>&1 | FileCheck %s --check-prefix=NOT-BINARY
+# NOT-BINARY: error: '--gap-fill' is only supported for binary output
+
+# RUN: not llvm-objcopy -O binary --gap-fill= %t %t.bin 2>&1 | FileCheck %s --check-prefix=BAD-FORMAT
+# BAD-FORMAT: error: --gap-fill: bad number:
+
+# RUN: not llvm-objcopy -O binary --gap-fill=x %t %t.bin 2>&1 | FileCheck %s --check-prefix=BAD-INPUT
+# BAD-INPUT: error: --gap-fill: bad number: x
+
+# RUN: not llvm-objcopy -O binary --gap-fill=0x %t %t.bin 2>&1 | FileCheck %s --check-prefix=BAD-INPUT2
+# BAD-INPUT2: error: --gap-fill: bad number: 0x
+
+# RUN: not llvm-objcopy -O binary --gap-fill=0x1G %t %t.bin 2>&1 | FileCheck %s --check-prefix=BAD-INPUT3
+# BAD-INPUT3: error: --gap-fill: bad number: 0x1G
+
+# RUN: not llvm-objcopy -O binary --gap-fill=ff %t %t.bin 2>&1 | FileCheck %s --check-prefix=BAD-INPUT4
+# BAD-INPUT4: error: --gap-fill: bad number: ff
+
+# RUN: not llvm-objcopy -O binary --gap-fill=0x1122 %t %t-val16 2>&1 | FileCheck %s --check-prefix=TRUNCATED-ERR
+# TRUNCATED-ERR: error: gap-fill value 0x1122 is out of range (0 to 0xff)
+
+## Test no gap fill with all allocatable output sections.
+# RUN: llvm-objcopy -O binary %t %t-default
+# RUN: od -v -Ax -t x1 %t-default | FileCheck %s --check-prefix=DEFAULT --match-full-lines
+# DEFAULT:      000000 ee ff 11 22 33 44 aa bb cc dd fe dc ba 00 a1 b2
+# DEFAULT-NEXT: 000010 c3 d4 00 00 00 00 00 00 00 00 00 00 00 00 00 00
+# DEFAULT-NEXT: 000020 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
+# DEFAULT-NEXT: 000030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
+# DEFAULT-NEXT: 000040 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
+# DEFAULT-NEXT: 000050 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
+# DEFAULT-NEXT: 000060 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
+# DEFAULT-NEXT: 000070 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
+# DEFAULT-NEXT: 000080 00 00 89 ab cd ef
+# DEFAULT-NEXT: 000086
+
+## Test gap fill with all allocatable output sections.
+# RUN: llvm-objcopy -O binary --gap-fill=0xe9 %t %t-filled
+# RUN: od -v -Ax -t x1 %t-filled | FileCheck %s --check-prefix=FULL --match-full-lines
+# FULL:      000000 ee ff 11 22 33 44 aa bb cc dd fe dc ba e9 a1 b2
+# FULL-NEXT: 000010 c3 d4 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# FULL-NEXT: 000020 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# FULL-NEXT: 000030 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# FULL-NEXT: 000040 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# FULL-NEXT: 000050 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# FULL-NEXT: 000060 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# FULL-NEXT: 000070 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# FULL-NEXT: 000080 e9 e9 89 ab cd ef
+# FULL-NEXT: 000086
+
+## Test gap fill with a decimal value.
+# RUN: llvm-objcopy -O binary --gap-fill=99 %t %t-filled-decimal
+# RUN: od -v -Ax -t x1 %t-filled-decimal | FileCheck %s --check-prefix=DEC --match-full-lines
+# DEC:      000000 ee ff 11 22 33 44 aa bb cc dd fe dc ba 63 a1 b2
+# DEC-NEXT: 000010 c3 d4 63 63 63 63 63 63 63 63 63 63 63 63 63 63
+# DEC-NEXT: 000020 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63
+# DEC-NEXT: 000030 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63
+# DEC-NEXT: 000040 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63
+# DEC-NEXT: 000050 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63
+# DEC-NEXT: 000060 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63
+# DEC-NEXT: 000070 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63 63
+# DEC-NEXT: 000080 63 63 89 ab cd ef
+# DEC-NEXT: 000086
+
+## Test gap fill with the last section removed, should be truncated.
+# RUN: llvm-objcopy -O binary --gap-fill=0xe9 --remove-section=.foo %t %t-filled
+# RUN: od -v -Ax -t x1 %t-filled | FileCheck %s --check-prefix=REMOVE-LAST-SECTION --match-full-lines
+# REMOVE-LAST-SECTION: 000000 ee ff 11 22 33 44 aa bb cc dd fe dc ba e9 a1 b2
+# REMOVE-LAST-SECTION-NEXT: 000010 c3 d4
+# REMOVE-LAST-SECTION-NEXT: 000012
+
+## Test gap fill with the middle section removed, should be filled.
+# RUN: llvm-objcopy -O binary --gap-fill=0xe9 --remove-section=.gap2 %t %t-filled
+# RUN: od -v -Ax -t x1 %t-filled | FileCheck %s --check-prefix=REMOVE-MIDDLE-SECTION --match-full-lines
+# REMOVE-MIDDLE-SECTION:      000000 ee ff 11 22 33 44 aa bb cc dd fe dc ba e9 e9 e9
+# REMOVE-MIDDLE-SECTION-NEXT: 000010 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# REMOVE-MIDDLE-SECTION-NEXT: 000020 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# REMOVE-MIDDLE-SECTION-NEXT: 000030 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# REMOVE-MIDDLE-SECTION-NEXT: 000040 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# REMOVE-MIDDLE-SECTION-NEXT: 000050 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# REMOVE-MIDDLE-SECTION-NEXT: 000060 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# REMOVE-MIDDLE-SECTION-NEXT: 000070 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9 e9
+# REMOVE-MIDDLE-SECTION-NEXT: 000080 e9 e9 89 ab cd ef
+# REMOVE-MIDDLE-SECTION-NEXT: 000086
+
+--- !ELF
+FileHeader:
+  Class:           ELFCLASS64
+  Data:            ELFDATA2LSB
+  Type:            ET_EXEC
+  Machine:         EM_X86_64
+Sections:
+  - Name:            .space1
+    Type:            Fill
+    Pattern:         'ABCD'
+    Size:            0x2
+  - Name:            .nogap
+    Type:            SHT_PROGBITS
+    Flags:           [ SHF_ALLOC ]
+    Address:         0x0102
+    Size:            0x6
+    Content:         'EEFF11223344'
+  - Name:            .gap1
+    Type:            SHT_PROGBITS
+    Flags:           [ SHF_ALLOC, SHF_EXECINSTR ]
+    Address:         0x0108
+    Content:         'AABBCCDDFEDCBA'
+  - Name:            .space2
+    Type:            Fill
+    Pattern:         'DC'
+    Size:            1
+  - Name:            .gap2
+    Type:            SHT_PROGBITS
+    Flags:           [ SHF_ALLOC ]
+    Address:         0x0110
+    Content:         'A1B2C3D4'
+  - Name:            .space3
+    Type:            Fill
+    Pattern:         'FE'
+    Size:            0x1
+  - Name:            .nobit_tbss
+    Type:            SHT_NOBITS
+    Flags:           [ SHF_WRITE, SHF_ALLOC, SHF_TLS ]
+    Address:         0x0180
+    Size:            0x0018
+  - Name:            .space4
+    Type:            Fill
+    Pattern:         '01234567'
+    Size:            0x4
+  - Name:            .foo
+    Type:            SHT_PROGBITS
+    Flags:           [ SHF_WRITE, SHF_ALLOC ]
+    Address:         0x0184
+    Content:         '89ABCDEF'
+  - Name:            .nobit_bss
+    Type:            SHT_NOBITS
+    Flags:           [ SHF_WRITE, SHF_ALLOC ]
+    Address:         0x018A
+    Size:            0x0008
+  - Name:            .comment
+    Type:            SHT_PROGBITS
+    Flags:           [ SHF_MERGE, SHF_STRINGS ]
+    EntSize:         0x0001
+    Content:         4743433A
+
+## In this test, output sections are defined out of order with respect to their
+## load addresses. Verify that gaps are still correctly filled.
+
+# RUN: yaml2obj --docnum=2 %s -o %t.2
+# RUN: llvm-objcopy -O binary --gap-fill=0xe9 %t.2 %t.2.filled
+# RUN: od -v -Ax -t x1 %t.2.filled | FileCheck --match-full-lines %s
+# CHECK: 000000 aa bb cc dd e9 e9 e9 e9 11 22 33 44
+
+--- !ELF
+FileHeader:
+  Class:           ELFCLASS64
+  Data:            ELFDATA2LSB
+  Type:            ET_EXEC
+  Machine:         EM_X86_64
+Sections:
+  - Name:            .bss
+    Type:            SHT_NOBITS
+    Flags:           [ SHF_ALLOC, SHF_WRITE ]
+    Address:         0x0104
+    Size:            4
+  - Name:            .section1
+    Type:            SHT_PROGBITS
+    Flags:           [ SHF_ALLOC, SHF_WRITE ]
+    Address:         0x0108
+    Content:         '11223344'
+  - Name:            .section3
+    Type:            SHT_PROGBITS
+    Flags:           [ SHF_ALLOC, SHF_WRITE ]
+    Address:         0x0100
+    Content:         'AABBCCDD'