aboutsummaryrefslogtreecommitdiff
path: root/lib/MC
diff options
context:
space:
mode:
authorDimitry Andric <dim@FreeBSD.org>2011-07-17 15:36:56 +0000
committerDimitry Andric <dim@FreeBSD.org>2011-07-17 15:36:56 +0000
commit411bd29eea3c360d5b48a18a17b5e87f5671af0e (patch)
treec8086addb211fa670a9d2b1038d8c2e453229755 /lib/MC
parent56fe8f14099930935e3870e3e823c322a85c1c89 (diff)
downloadsrc-411bd29eea3c360d5b48a18a17b5e87f5671af0e.tar.gz
src-411bd29eea3c360d5b48a18a17b5e87f5671af0e.zip
Vendor import of llvm trunk r135360:vendor/llvm/llvm-r135360
Notes
Notes: svn path=/vendor/llvm/dist/; revision=224133 svn path=/vendor/llvm/llvm-r135360/; revision=224134; tag=vendor/llvm/llvm-r135360
Diffstat (limited to 'lib/MC')
-rw-r--r--lib/MC/CMakeLists.txt2
-rw-r--r--lib/MC/MCAsmInfo.cpp4
-rw-r--r--lib/MC/MCAsmStreamer.cpp31
-rw-r--r--lib/MC/MCDisassembler/Disassembler.cpp9
-rw-r--r--lib/MC/MCDisassembler/EDDisassembler.cpp22
-rw-r--r--lib/MC/MCDisassembler/EDDisassembler.h1
-rw-r--r--lib/MC/MCDisassembler/EDInfo.h19
-rw-r--r--lib/MC/MCDisassembler/EDOperand.cpp21
-rw-r--r--lib/MC/MCDwarf.cpp324
-rw-r--r--lib/MC/MCELFStreamer.cpp1
-rw-r--r--lib/MC/MCELFStreamer.h134
-rw-r--r--lib/MC/MCLoggingStreamer.cpp6
-rw-r--r--lib/MC/MCMachOStreamer.cpp1
-rw-r--r--lib/MC/MCNullStreamer.cpp3
-rw-r--r--lib/MC/MCObjectStreamer.cpp5
-rw-r--r--lib/MC/MCParser/AsmParser.cpp49
-rw-r--r--lib/MC/MCParser/COFFAsmParser.cpp6
-rw-r--r--lib/MC/MCParser/MCAsmParser.cpp6
-rw-r--r--lib/MC/MCParser/TargetAsmParser.cpp4
-rw-r--r--lib/MC/MCStreamer.cpp3
-rw-r--r--lib/MC/MCSubtargetInfo.cpp96
-rw-r--r--lib/MC/MCWin64EH.cpp10
-rw-r--r--lib/MC/MachObjectWriter.cpp2111
-rw-r--r--lib/MC/SubtargetFeature.cpp397
24 files changed, 1513 insertions, 1752 deletions
diff --git a/lib/MC/CMakeLists.txt b/lib/MC/CMakeLists.txt
index a77ecd3bd8ad..22afa7e91cbe 100644
--- a/lib/MC/CMakeLists.txt
+++ b/lib/MC/CMakeLists.txt
@@ -28,12 +28,14 @@ add_llvm_library(LLVMMC
MCSectionELF.cpp
MCSectionMachO.cpp
MCStreamer.cpp
+ MCSubtargetInfo.cpp
MCSymbol.cpp
MCValue.cpp
MCWin64EH.cpp
MachObjectWriter.cpp
WinCOFFStreamer.cpp
WinCOFFObjectWriter.cpp
+ SubtargetFeature.cpp
TargetAsmBackend.cpp
)
diff --git a/lib/MC/MCAsmInfo.cpp b/lib/MC/MCAsmInfo.cpp
index 73b259eaa0fe..502b60b0edf4 100644
--- a/lib/MC/MCAsmInfo.cpp
+++ b/lib/MC/MCAsmInfo.cpp
@@ -23,6 +23,9 @@
using namespace llvm;
MCAsmInfo::MCAsmInfo() {
+ PointerSize = 4;
+ IsLittleEndian = true;
+ StackGrowsUp = false;
HasSubsectionsViaSymbols = false;
HasMachoZeroFillDirective = false;
HasMachoTBSSDirective = false;
@@ -78,6 +81,7 @@ MCAsmInfo::MCAsmInfo() {
DwarfRequiresRelocationForSectionOffset = true;
DwarfSectionOffsetDirective = 0;
DwarfUsesLabelOffsetForRanges = true;
+ DwarfRegNumForCFI = false;
HasMicrosoftFastStdCallMangling = false;
AsmTransCBE = 0;
diff --git a/lib/MC/MCAsmStreamer.cpp b/lib/MC/MCAsmStreamer.cpp
index e8b09fcaced8..d5d08e8f69fb 100644
--- a/lib/MC/MCAsmStreamer.cpp
+++ b/lib/MC/MCAsmStreamer.cpp
@@ -19,6 +19,7 @@
#include "llvm/MC/MCSymbol.h"
#include "llvm/ADT/OwningPtr.h"
#include "llvm/ADT/SmallString.h"
+#include "llvm/ADT/StringExtras.h"
#include "llvm/ADT/Twine.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/MathExtras.h"
@@ -33,8 +34,10 @@ using namespace llvm;
namespace {
class MCAsmStreamer : public MCStreamer {
+protected:
formatted_raw_ostream &OS;
const MCAsmInfo &MAI;
+private:
OwningPtr<MCInstPrinter> InstPrinter;
OwningPtr<MCCodeEmitter> Emitter;
OwningPtr<TargetAsmBackend> AsmBackend;
@@ -134,7 +137,8 @@ public:
virtual void EmitWeakReference(MCSymbol *Alias, const MCSymbol *Symbol);
virtual void EmitDwarfAdvanceLineAddr(int64_t LineDelta,
const MCSymbol *LastLabel,
- const MCSymbol *Label);
+ const MCSymbol *Label,
+ unsigned PointerSize);
virtual void EmitDwarfAdvanceFrameAddr(const MCSymbol *LastLabel,
const MCSymbol *Label);
@@ -361,9 +365,9 @@ void MCAsmStreamer::EmitWeakReference(MCSymbol *Alias, const MCSymbol *Symbol) {
void MCAsmStreamer::EmitDwarfAdvanceLineAddr(int64_t LineDelta,
const MCSymbol *LastLabel,
- const MCSymbol *Label) {
- EmitDwarfSetLineAddr(LineDelta, Label,
- getContext().getTargetAsmInfo().getPointerSize());
+ const MCSymbol *Label,
+ unsigned PointerSize) {
+ EmitDwarfSetLineAddr(LineDelta, Label, PointerSize);
}
void MCAsmStreamer::EmitDwarfAdvanceFrameAddr(const MCSymbol *LastLabel,
@@ -600,7 +604,7 @@ void MCAsmStreamer::EmitValueImpl(const MCExpr *Value, unsigned Size,
int64_t IntValue;
if (!Value->EvaluateAsAbsolute(IntValue))
report_fatal_error("Don't know how to emit this value.");
- if (getContext().getTargetAsmInfo().isLittleEndian()) {
+ if (getContext().getAsmInfo().isLittleEndian()) {
EmitIntValue((uint32_t)(IntValue >> 0 ), 4, AddrSpace);
EmitIntValue((uint32_t)(IntValue >> 32), 4, AddrSpace);
} else {
@@ -822,9 +826,9 @@ void MCAsmStreamer::EmitCFIEndProc() {
}
void MCAsmStreamer::EmitRegisterName(int64_t Register) {
- if (InstPrinter) {
- const TargetAsmInfo &asmInfo = getContext().getTargetAsmInfo();
- unsigned LLVMRegister = asmInfo.getLLVMRegNum(Register, true);
+ if (InstPrinter && !MAI.useDwarfRegNumForCFI()) {
+ const TargetAsmInfo &TAI = getContext().getTargetAsmInfo();
+ unsigned LLVMRegister = TAI.getLLVMRegNum(Register, true);
InstPrinter->printRegName(OS, LLVMRegister);
} else {
OS << Register;
@@ -1085,7 +1089,7 @@ void MCAsmStreamer::AddEncodingComment(const MCInst &Inst) {
}
}
- // FIXME: Node the fixup comments for Thumb2 are completely bogus since the
+ // FIXME: Note the fixup comments for Thumb2 are completely bogus since the
// high order halfword of a 32-bit Thumb2 instruction is emitted first.
OS << "encoding: [";
for (unsigned i = 0, e = Code.size(); i != e; ++i) {
@@ -1120,7 +1124,7 @@ void MCAsmStreamer::AddEncodingComment(const MCInst &Inst) {
unsigned Bit = (Code[i] >> j) & 1;
unsigned FixupBit;
- if (getContext().getTargetAsmInfo().isLittleEndian())
+ if (getContext().getAsmInfo().isLittleEndian())
FixupBit = i * 8 + j;
else
FixupBit = i * 8 + (7-j);
@@ -1241,13 +1245,12 @@ void MCAsmStreamer::Finish() {
if (!UseCFI)
EmitFrames(false);
}
-
MCStreamer *llvm::createAsmStreamer(MCContext &Context,
formatted_raw_ostream &OS,
bool isVerboseAsm, bool useLoc,
- bool useCFI,
- MCInstPrinter *IP, MCCodeEmitter *CE,
- TargetAsmBackend *TAB, bool ShowInst) {
+ bool useCFI, MCInstPrinter *IP,
+ MCCodeEmitter *CE, TargetAsmBackend *TAB,
+ bool ShowInst) {
return new MCAsmStreamer(Context, OS, isVerboseAsm, useLoc, useCFI,
IP, CE, TAB, ShowInst);
}
diff --git a/lib/MC/MCDisassembler/Disassembler.cpp b/lib/MC/MCDisassembler/Disassembler.cpp
index 6e636f07f1d1..5480b4b12b2c 100644
--- a/lib/MC/MCDisassembler/Disassembler.cpp
+++ b/lib/MC/MCDisassembler/Disassembler.cpp
@@ -40,6 +40,7 @@ LLVMDisasmContextRef LLVMCreateDisasm(const char *TripleName, void *DisInfo,
llvm::InitializeAllTargetInfos();
// FIXME: We shouldn't need to initialize the Target(Machine)s.
llvm::InitializeAllTargets();
+ llvm::InitializeAllMCAsmInfos();
llvm::InitializeAllAsmPrinters();
llvm::InitializeAllAsmParsers();
llvm::InitializeAllDisassemblers();
@@ -50,16 +51,18 @@ LLVMDisasmContextRef LLVMCreateDisasm(const char *TripleName, void *DisInfo,
assert(TheTarget && "Unable to create target!");
// Get the assembler info needed to setup the MCContext.
- const MCAsmInfo *MAI = TheTarget->createAsmInfo(TripleName);
+ const MCAsmInfo *MAI = TheTarget->createMCAsmInfo(TripleName);
assert(MAI && "Unable to create target asm info!");
// Package up features to be passed to target/subtarget
std::string FeaturesStr;
+ std::string CPU;
// FIXME: We shouldn't need to do this (and link in codegen).
// When we split this out, we should do it in a way that makes
// it straightforward to switch subtargets on the fly.
- TargetMachine *TM = TheTarget->createTargetMachine(TripleName, FeaturesStr);
+ TargetMachine *TM = TheTarget->createTargetMachine(TripleName, CPU,
+ FeaturesStr);
assert(TM && "Unable to create target machine!");
// Get the target assembler info needed to setup the context.
@@ -77,7 +80,7 @@ LLVMDisasmContextRef LLVMCreateDisasm(const char *TripleName, void *DisInfo,
// Set up the instruction printer.
int AsmPrinterVariant = MAI->getAssemblerDialect();
- MCInstPrinter *IP = TheTarget->createMCInstPrinter(*TM, AsmPrinterVariant,
+ MCInstPrinter *IP = TheTarget->createMCInstPrinter(AsmPrinterVariant,
*MAI);
assert(IP && "Unable to create instruction printer!");
diff --git a/lib/MC/MCDisassembler/EDDisassembler.cpp b/lib/MC/MCDisassembler/EDDisassembler.cpp
index 91c5284892a5..bdd99afe1ae4 100644
--- a/lib/MC/MCDisassembler/EDDisassembler.cpp
+++ b/lib/MC/MCDisassembler/EDDisassembler.cpp
@@ -23,6 +23,7 @@
#include "llvm/MC/MCInst.h"
#include "llvm/MC/MCInstPrinter.h"
#include "llvm/MC/MCStreamer.h"
+#include "llvm/MC/MCSubtargetInfo.h"
#include "llvm/MC/MCParser/AsmLexer.h"
#include "llvm/MC/MCParser/MCAsmParser.h"
#include "llvm/MC/MCParser/MCParsedAsmOperand.h"
@@ -106,6 +107,7 @@ void EDDisassembler::initialize() {
InitializeAllTargetInfos();
InitializeAllTargets();
+ InitializeAllMCAsmInfos();
InitializeAllAsmPrinters();
InitializeAllAsmParsers();
InitializeAllDisassemblers();
@@ -167,11 +169,11 @@ EDDisassembler::EDDisassembler(CPUKey &key) :
if (!Tgt)
return;
+ std::string CPU;
std::string featureString;
-
- TargetMachine.reset(Tgt->createTargetMachine(tripleString,
+ TargetMachine.reset(Tgt->createTargetMachine(tripleString, CPU,
featureString));
-
+
const TargetRegisterInfo *registerInfo = TargetMachine->getRegisterInfo();
if (!registerInfo)
@@ -179,11 +181,11 @@ EDDisassembler::EDDisassembler(CPUKey &key) :
initMaps(*registerInfo);
- AsmInfo.reset(Tgt->createAsmInfo(tripleString));
+ AsmInfo.reset(Tgt->createMCAsmInfo(tripleString));
if (!AsmInfo)
return;
-
+
Disassembler.reset(Tgt->createMCDisassembler());
if (!Disassembler)
@@ -193,8 +195,7 @@ EDDisassembler::EDDisassembler(CPUKey &key) :
InstString.reset(new std::string);
InstStream.reset(new raw_string_ostream(*InstString));
- InstPrinter.reset(Tgt->createMCInstPrinter(*TargetMachine, LLVMSyntaxVariant,
- *AsmInfo));
+ InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo));
if (!InstPrinter)
return;
@@ -372,8 +373,11 @@ int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
OwningPtr<MCAsmParser> genericParser(createMCAsmParser(*Tgt, sourceMgr,
context, *streamer,
*AsmInfo));
- OwningPtr<TargetAsmParser> TargetParser(Tgt->createAsmParser(*genericParser,
- *TargetMachine));
+
+ StringRef triple = tripleFromArch(Key.Arch);
+ OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(triple, "", ""));
+ OwningPtr<TargetAsmParser> TargetParser(Tgt->createAsmParser(*STI,
+ *genericParser));
AsmToken OpcodeToken = genericParser->Lex();
AsmToken NextToken = genericParser->Lex(); // consume next token, because specificParser expects us to
diff --git a/lib/MC/MCDisassembler/EDDisassembler.h b/lib/MC/MCDisassembler/EDDisassembler.h
index 2fcc09d4bef0..11d69c151cf9 100644
--- a/lib/MC/MCDisassembler/EDDisassembler.h
+++ b/lib/MC/MCDisassembler/EDDisassembler.h
@@ -41,6 +41,7 @@ class MCInstPrinter;
class MCInst;
class MCParsedAsmOperand;
class MCStreamer;
+class MCSubtargetInfo;
template <typename T> class SmallVectorImpl;
class SourceMgr;
class Target;
diff --git a/lib/MC/MCDisassembler/EDInfo.h b/lib/MC/MCDisassembler/EDInfo.h
index ad5728263133..e43ad1635246 100644
--- a/lib/MC/MCDisassembler/EDInfo.h
+++ b/lib/MC/MCDisassembler/EDInfo.h
@@ -25,8 +25,11 @@ enum OperandTypes {
kOperandTypeARMBranchTarget,
kOperandTypeARMSoReg,
kOperandTypeARMSoImm,
+ kOperandTypeARMRotImm,
kOperandTypeARMSoImm2Part,
kOperandTypeARMPredicate,
+ kOperandTypeAddrModeImm12,
+ kOperandTypeLdStSOReg,
kOperandTypeARMAddrMode2,
kOperandTypeARMAddrMode2Offset,
kOperandTypeARMAddrMode3,
@@ -38,13 +41,20 @@ enum OperandTypes {
kOperandTypeARMAddrMode7,
kOperandTypeARMAddrModePC,
kOperandTypeARMRegisterList,
+ kOperandTypeARMDPRRegisterList,
+ kOperandTypeARMSPRRegisterList,
kOperandTypeARMTBAddrMode,
kOperandTypeThumbITMask,
- kOperandTypeThumbAddrModeS1,
- kOperandTypeThumbAddrModeS2,
- kOperandTypeThumbAddrModeS4,
+ kOperandTypeThumbAddrModeRegS1,
+ kOperandTypeThumbAddrModeRegS2,
+ kOperandTypeThumbAddrModeRegS4,
+ kOperandTypeThumbAddrModeImmS1,
+ kOperandTypeThumbAddrModeImmS2,
+ kOperandTypeThumbAddrModeImmS4,
kOperandTypeThumbAddrModeRR,
kOperandTypeThumbAddrModeSP,
+ kOperandTypeThumbAddrModePC,
+ kOperandTypeThumb2AddrModeReg,
kOperandTypeThumb2SoReg,
kOperandTypeThumb2SoImm,
kOperandTypeThumb2AddrModeImm8,
@@ -52,8 +62,7 @@ enum OperandTypes {
kOperandTypeThumb2AddrModeImm12,
kOperandTypeThumb2AddrModeSoReg,
kOperandTypeThumb2AddrModeImm8s4,
- kOperandTypeThumb2AddrModeImm8s4Offset,
- kOperandTypeThumb2AddrModeReg
+ kOperandTypeThumb2AddrModeImm8s4Offset
};
enum OperandFlags {
diff --git a/lib/MC/MCDisassembler/EDOperand.cpp b/lib/MC/MCDisassembler/EDOperand.cpp
index 492bb08f336a..6a4e56ff72c4 100644
--- a/lib/MC/MCDisassembler/EDOperand.cpp
+++ b/lib/MC/MCDisassembler/EDOperand.cpp
@@ -61,11 +61,14 @@ EDOperand::EDOperand(const EDDisassembler &disassembler,
switch (operandType) {
default:
case kOperandTypeARMRegisterList:
+ case kOperandTypeARMDPRRegisterList:
+ case kOperandTypeARMSPRRegisterList:
break;
case kOperandTypeImmediate:
case kOperandTypeRegister:
case kOperandTypeARMBranchTarget:
case kOperandTypeARMSoImm:
+ case kOperandTypeARMRotImm:
case kOperandTypeThumb2SoImm:
case kOperandTypeARMSoImm2Part:
case kOperandTypeARMPredicate:
@@ -78,6 +81,7 @@ EDOperand::EDOperand(const EDDisassembler &disassembler,
numMCOperands = 1;
break;
case kOperandTypeThumb2SoReg:
+ case kOperandTypeAddrModeImm12:
case kOperandTypeARMAddrMode2Offset:
case kOperandTypeARMAddrMode3Offset:
case kOperandTypeARMAddrMode4:
@@ -86,17 +90,22 @@ EDOperand::EDOperand(const EDDisassembler &disassembler,
case kOperandTypeThumb2AddrModeImm8:
case kOperandTypeThumb2AddrModeImm12:
case kOperandTypeThumb2AddrModeImm8s4:
+ case kOperandTypeThumbAddrModeImmS1:
+ case kOperandTypeThumbAddrModeImmS2:
+ case kOperandTypeThumbAddrModeImmS4:
case kOperandTypeThumbAddrModeRR:
case kOperandTypeThumbAddrModeSP:
+ case kOperandTypeThumbAddrModePC:
numMCOperands = 2;
break;
case kOperandTypeARMSoReg:
+ case kOperandTypeLdStSOReg:
case kOperandTypeARMAddrMode2:
case kOperandTypeARMAddrMode3:
case kOperandTypeThumb2AddrModeSoReg:
- case kOperandTypeThumbAddrModeS1:
- case kOperandTypeThumbAddrModeS2:
- case kOperandTypeThumbAddrModeS4:
+ case kOperandTypeThumbAddrModeRegS1:
+ case kOperandTypeThumbAddrModeRegS2:
+ case kOperandTypeThumbAddrModeRegS4:
case kOperandTypeARMAddrMode6Offset:
numMCOperands = 3;
break;
@@ -270,9 +279,9 @@ int EDOperand::isMemory() {
case kOperandTypeARMAddrMode7:
case kOperandTypeARMAddrModePC:
case kOperandTypeARMBranchTarget:
- case kOperandTypeThumbAddrModeS1:
- case kOperandTypeThumbAddrModeS2:
- case kOperandTypeThumbAddrModeS4:
+ case kOperandTypeThumbAddrModeRegS1:
+ case kOperandTypeThumbAddrModeRegS2:
+ case kOperandTypeThumbAddrModeRegS4:
case kOperandTypeThumbAddrModeRR:
case kOperandTypeThumbAddrModeSP:
case kOperandTypeThumb2SoImm:
diff --git a/lib/MC/MCDwarf.cpp b/lib/MC/MCDwarf.cpp
index 13cb81ab441b..ad86db13d510 100644
--- a/lib/MC/MCDwarf.cpp
+++ b/lib/MC/MCDwarf.cpp
@@ -7,22 +7,21 @@
//
//===----------------------------------------------------------------------===//
-#include "llvm/ADT/FoldingSet.h"
#include "llvm/MC/MCAsmInfo.h"
#include "llvm/MC/MCDwarf.h"
-#include "llvm/MC/MCAssembler.h"
#include "llvm/MC/MCStreamer.h"
#include "llvm/MC/MCSymbol.h"
#include "llvm/MC/MCExpr.h"
#include "llvm/MC/MCContext.h"
#include "llvm/MC/MCObjectWriter.h"
-#include "llvm/ADT/SmallString.h"
-#include "llvm/ADT/Twine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/raw_ostream.h"
-#include "llvm/Target/TargetAsmBackend.h"
#include "llvm/Target/TargetAsmInfo.h"
+#include "llvm/ADT/FoldingSet.h"
+#include "llvm/ADT/SmallString.h"
+#include "llvm/ADT/StringExtras.h"
+#include "llvm/ADT/Twine.h"
using namespace llvm;
// Given a special op, return the address skip amount (in units of
@@ -30,28 +29,27 @@ using namespace llvm;
#define SPECIAL_ADDR(op) (((op) - DWARF2_LINE_OPCODE_BASE)/DWARF2_LINE_RANGE)
// The maximum address skip amount that can be encoded with a special op.
-#define MAX_SPECIAL_ADDR_DELTA SPECIAL_ADDR(255)
+#define MAX_SPECIAL_ADDR_DELTA SPECIAL_ADDR(255)
// First special line opcode - leave room for the standard opcodes.
// Note: If you want to change this, you'll have to update the
// "standard_opcode_lengths" table that is emitted in DwarfFileTable::Emit().
-#define DWARF2_LINE_OPCODE_BASE 13
+#define DWARF2_LINE_OPCODE_BASE 13
// Minimum line offset in a special line info. opcode. This value
// was chosen to give a reasonable range of values.
-#define DWARF2_LINE_BASE -5
+#define DWARF2_LINE_BASE -5
// Range of line offsets in a special line info. opcode.
-# define DWARF2_LINE_RANGE 14
+#define DWARF2_LINE_RANGE 14
// Define the architecture-dependent minimum instruction length (in bytes).
// This value should be rather too small than too big.
-# define DWARF2_LINE_MIN_INSN_LENGTH 1
+#define DWARF2_LINE_MIN_INSN_LENGTH 1
// Note: when DWARF2_LINE_MIN_INSN_LENGTH == 1 which is the current setting,
// this routine is a nop and will be optimized away.
-static inline uint64_t ScaleAddrDelta(uint64_t AddrDelta)
-{
+static inline uint64_t ScaleAddrDelta(uint64_t AddrDelta) {
if (DWARF2_LINE_MIN_INSN_LENGTH == 1)
return AddrDelta;
if (AddrDelta % DWARF2_LINE_MIN_INSN_LENGTH != 0) {
@@ -174,7 +172,9 @@ static inline void EmitDwarfLineTable(MCStreamer *MCOS,
// At this point we want to emit/create the sequence to encode the delta in
// line numbers and the increment of the address from the previous Label
// and the current Label.
- MCOS->EmitDwarfAdvanceLineAddr(LineDelta, LastLabel, Label);
+ const MCAsmInfo &asmInfo = MCOS->getContext().getAsmInfo();
+ MCOS->EmitDwarfAdvanceLineAddr(LineDelta, LastLabel, Label,
+ asmInfo.getPointerSize());
LastLine = it->getLine();
LastLabel = Label;
@@ -198,7 +198,9 @@ static inline void EmitDwarfLineTable(MCStreamer *MCOS,
// Switch back the the dwarf line section.
MCOS->SwitchSection(context.getTargetAsmInfo().getDwarfLineSection());
- MCOS->EmitDwarfAdvanceLineAddr(INT64_MAX, LastLabel, SectionEnd);
+ const MCAsmInfo &asmInfo = MCOS->getContext().getAsmInfo();
+ MCOS->EmitDwarfAdvanceLineAddr(INT64_MAX, LastLabel, SectionEnd,
+ asmInfo.getPointerSize());
}
//
@@ -291,7 +293,7 @@ void MCDwarfFileTable::Emit(MCStreamer *MCOS) {
const std::vector<const MCSection *> &MCLineSectionOrder =
MCOS->getContext().getMCLineSectionOrder();
for (std::vector<const MCSection*>::const_iterator it =
- MCLineSectionOrder.begin(), ie = MCLineSectionOrder.end(); it != ie;
+ MCLineSectionOrder.begin(), ie = MCLineSectionOrder.end(); it != ie;
++it) {
const MCSection *Sec = *it;
const MCLineSection *Line = MCLineSections.lookup(Sec);
@@ -354,10 +356,7 @@ void MCDwarfLineAddr::Encode(int64_t LineDelta, uint64_t AddrDelta,
OS << char(dwarf::DW_LNS_const_add_pc);
else {
OS << char(dwarf::DW_LNS_advance_pc);
- SmallString<32> Tmp;
- raw_svector_ostream OSE(Tmp);
- MCObjectWriter::EncodeULEB128(AddrDelta, OSE);
- OS << OSE.str();
+ MCObjectWriter::EncodeULEB128(AddrDelta, OS);
}
OS << char(dwarf::DW_LNS_extended_op);
OS << char(1);
@@ -432,25 +431,24 @@ void MCDwarfFile::dump() const {
static int getDataAlignmentFactor(MCStreamer &streamer) {
MCContext &context = streamer.getContext();
- const TargetAsmInfo &asmInfo = context.getTargetAsmInfo();
+ const MCAsmInfo &asmInfo = context.getAsmInfo();
int size = asmInfo.getPointerSize();
- if (asmInfo.getStackGrowthDirection() == TargetFrameLowering::StackGrowsUp)
+ if (asmInfo.isStackGrowthDirectionUp())
return size;
- else
- return -size;
+ else
+ return -size;
}
static unsigned getSizeForEncoding(MCStreamer &streamer,
unsigned symbolEncoding) {
MCContext &context = streamer.getContext();
- const TargetAsmInfo &asmInfo = context.getTargetAsmInfo();
unsigned format = symbolEncoding & 0x0f;
switch (format) {
default:
assert(0 && "Unknown Encoding");
case dwarf::DW_EH_PE_absptr:
case dwarf::DW_EH_PE_signed:
- return asmInfo.getPointerSize();
+ return context.getAsmInfo().getPointerSize();
case dwarf::DW_EH_PE_udata2:
case dwarf::DW_EH_PE_sdata2:
return 2;
@@ -464,13 +462,14 @@ static unsigned getSizeForEncoding(MCStreamer &streamer,
}
static void EmitSymbol(MCStreamer &streamer, const MCSymbol &symbol,
- unsigned symbolEncoding) {
+ unsigned symbolEncoding, const char *comment = 0) {
MCContext &context = streamer.getContext();
const MCAsmInfo &asmInfo = context.getAsmInfo();
const MCExpr *v = asmInfo.getExprForFDESymbol(&symbol,
symbolEncoding,
streamer);
unsigned size = getSizeForEncoding(streamer, symbolEncoding);
+ if (streamer.isVerboseAsm() && comment) streamer.AddComment(comment);
streamer.EmitAbsValue(v, size);
}
@@ -486,11 +485,11 @@ static void EmitPersonality(MCStreamer &streamer, const MCSymbol &symbol,
}
static const MachineLocation TranslateMachineLocation(
- const TargetAsmInfo &AsmInfo,
+ const TargetAsmInfo &TAI,
const MachineLocation &Loc) {
unsigned Reg = Loc.getReg() == MachineLocation::VirtualFP ?
MachineLocation::VirtualFP :
- unsigned(AsmInfo.getDwarfRegNum(Loc.getReg(), true));
+ unsigned(TAI.getDwarfRegNum(Loc.getReg(), true));
const MachineLocation &NewLoc = Loc.isReg() ?
MachineLocation(Reg) : MachineLocation(Reg, Loc.getOffset());
return NewLoc;
@@ -503,13 +502,18 @@ namespace {
bool UsingCFI;
bool IsEH;
const MCSymbol *SectionStart;
-
public:
FrameEmitterImpl(bool usingCFI, bool isEH, const MCSymbol *sectionStart) :
CFAOffset(0), CIENum(0), UsingCFI(usingCFI), IsEH(isEH),
SectionStart(sectionStart) {
}
+ /// EmitCompactUnwind - Emit the unwind information in a compact way. If
+ /// we're successful, return 'true'. Otherwise, return 'false' and it will
+ /// emit the normal CIE and FDE.
+ bool EmitCompactUnwind(MCStreamer &streamer,
+ const MCDwarfFrameInfo &frame);
+
const MCSymbol &EmitCIE(MCStreamer &streamer,
const MCSymbol *personality,
unsigned personalityEncoding,
@@ -524,11 +528,46 @@ namespace {
void EmitCFIInstruction(MCStreamer &Streamer,
const MCCFIInstruction &Instr);
};
+
+} // end anonymous namespace
+
+static void EmitEncodingByte(MCStreamer &Streamer, unsigned Encoding,
+ StringRef Prefix) {
+ if (Streamer.isVerboseAsm()) {
+ const char *EncStr = 0;
+ switch (Encoding) {
+ default: EncStr = "<unknown encoding>";
+ case dwarf::DW_EH_PE_absptr: EncStr = "absptr";
+ case dwarf::DW_EH_PE_omit: EncStr = "omit";
+ case dwarf::DW_EH_PE_pcrel: EncStr = "pcrel";
+ case dwarf::DW_EH_PE_udata4: EncStr = "udata4";
+ case dwarf::DW_EH_PE_udata8: EncStr = "udata8";
+ case dwarf::DW_EH_PE_sdata4: EncStr = "sdata4";
+ case dwarf::DW_EH_PE_sdata8: EncStr = "sdata8";
+ case dwarf::DW_EH_PE_pcrel |dwarf::DW_EH_PE_udata4: EncStr = "pcrel udata4";
+ case dwarf::DW_EH_PE_pcrel |dwarf::DW_EH_PE_sdata4: EncStr = "pcrel sdata4";
+ case dwarf::DW_EH_PE_pcrel |dwarf::DW_EH_PE_udata8: EncStr = "pcrel udata8";
+ case dwarf::DW_EH_PE_pcrel |dwarf::DW_EH_PE_sdata8: EncStr = "pcrel sdata8";
+ case dwarf::DW_EH_PE_indirect |dwarf::DW_EH_PE_pcrel|dwarf::DW_EH_PE_udata4:
+ EncStr = "indirect pcrel udata4";
+ case dwarf::DW_EH_PE_indirect |dwarf::DW_EH_PE_pcrel|dwarf::DW_EH_PE_sdata4:
+ EncStr = "indirect pcrel sdata4";
+ case dwarf::DW_EH_PE_indirect |dwarf::DW_EH_PE_pcrel|dwarf::DW_EH_PE_udata8:
+ EncStr = "indirect pcrel udata8";
+ case dwarf::DW_EH_PE_indirect |dwarf::DW_EH_PE_pcrel|dwarf::DW_EH_PE_sdata8:
+ EncStr = "indirect pcrel sdata8";
+ }
+
+ Streamer.AddComment(Twine(Prefix) + " = " + EncStr);
+ }
+
+ Streamer.EmitIntValue(Encoding, 1);
}
void FrameEmitterImpl::EmitCFIInstruction(MCStreamer &Streamer,
const MCCFIInstruction &Instr) {
int dataAlignmentFactor = getDataAlignmentFactor(Streamer);
+ bool VerboseAsm = Streamer.isVerboseAsm();
switch (Instr.getOperation()) {
case MCCFIInstruction::Move:
@@ -540,9 +579,13 @@ void FrameEmitterImpl::EmitCFIInstruction(MCStreamer &Streamer,
// If advancing cfa.
if (Dst.isReg() && Dst.getReg() == MachineLocation::VirtualFP) {
if (Src.getReg() == MachineLocation::VirtualFP) {
+ if (VerboseAsm) Streamer.AddComment("DW_CFA_def_cfa_offset");
Streamer.EmitIntValue(dwarf::DW_CFA_def_cfa_offset, 1);
} else {
+ if (VerboseAsm) Streamer.AddComment("DW_CFA_def_cfa");
Streamer.EmitIntValue(dwarf::DW_CFA_def_cfa, 1);
+ if (VerboseAsm) Streamer.AddComment(Twine("Reg ") +
+ Twine(Src.getReg()));
Streamer.EmitULEB128IntValue(Src.getReg());
}
@@ -551,47 +594,62 @@ void FrameEmitterImpl::EmitCFIInstruction(MCStreamer &Streamer,
else
CFAOffset = -Src.getOffset();
+ if (VerboseAsm) Streamer.AddComment(Twine("Offset " + Twine(CFAOffset)));
Streamer.EmitULEB128IntValue(CFAOffset);
return;
}
if (Src.isReg() && Src.getReg() == MachineLocation::VirtualFP) {
assert(Dst.isReg() && "Machine move not supported yet.");
+ if (VerboseAsm) Streamer.AddComment("DW_CFA_def_cfa_register");
Streamer.EmitIntValue(dwarf::DW_CFA_def_cfa_register, 1);
+ if (VerboseAsm) Streamer.AddComment(Twine("Reg ") + Twine(Dst.getReg()));
Streamer.EmitULEB128IntValue(Dst.getReg());
return;
}
unsigned Reg = Src.getReg();
-
int Offset = Dst.getOffset();
if (IsRelative)
Offset -= CFAOffset;
Offset = Offset / dataAlignmentFactor;
if (Offset < 0) {
+ if (VerboseAsm) Streamer.AddComment("DW_CFA_offset_extended_sf");
Streamer.EmitIntValue(dwarf::DW_CFA_offset_extended_sf, 1);
+ if (VerboseAsm) Streamer.AddComment(Twine("Reg ") + Twine(Reg));
Streamer.EmitULEB128IntValue(Reg);
+ if (VerboseAsm) Streamer.AddComment(Twine("Offset ") + Twine(Offset));
Streamer.EmitSLEB128IntValue(Offset);
} else if (Reg < 64) {
+ if (VerboseAsm) Streamer.AddComment(Twine("DW_CFA_offset + Reg(") +
+ Twine(Reg) + ")");
Streamer.EmitIntValue(dwarf::DW_CFA_offset + Reg, 1);
+ if (VerboseAsm) Streamer.AddComment(Twine("Offset ") + Twine(Offset));
Streamer.EmitULEB128IntValue(Offset);
} else {
+ if (VerboseAsm) Streamer.AddComment("DW_CFA_offset_extended");
Streamer.EmitIntValue(dwarf::DW_CFA_offset_extended, 1);
+ if (VerboseAsm) Streamer.AddComment(Twine("Reg ") + Twine(Reg));
Streamer.EmitULEB128IntValue(Reg);
+ if (VerboseAsm) Streamer.AddComment(Twine("Offset ") + Twine(Offset));
Streamer.EmitULEB128IntValue(Offset);
}
return;
}
case MCCFIInstruction::Remember:
+ if (VerboseAsm) Streamer.AddComment("DW_CFA_remember_state");
Streamer.EmitIntValue(dwarf::DW_CFA_remember_state, 1);
return;
case MCCFIInstruction::Restore:
+ if (VerboseAsm) Streamer.AddComment("DW_CFA_restore_state");
Streamer.EmitIntValue(dwarf::DW_CFA_restore_state, 1);
return;
case MCCFIInstruction::SameValue: {
unsigned Reg = Instr.getDestination().getReg();
+ if (VerboseAsm) Streamer.AddComment("DW_CFA_same_value");
Streamer.EmitIntValue(dwarf::DW_CFA_same_value, 1);
+ if (VerboseAsm) Streamer.AddComment(Twine("Reg ") + Twine(Reg));
Streamer.EmitULEB128IntValue(Reg);
return;
}
@@ -614,6 +672,7 @@ void FrameEmitterImpl::EmitCFIInstructions(MCStreamer &streamer,
if (BaseLabel && Label) {
MCSymbol *ThisSym = Label;
if (ThisSym != BaseLabel) {
+ if (streamer.isVerboseAsm()) streamer.AddComment("DW_CFA_advance_loc4");
streamer.EmitDwarfAdvanceFrameAddr(BaseLabel, ThisSym);
BaseLabel = ThisSym;
}
@@ -623,40 +682,128 @@ void FrameEmitterImpl::EmitCFIInstructions(MCStreamer &streamer,
}
}
+/// EmitCompactUnwind - Emit the unwind information in a compact way. If we're
+/// successful, return 'true'. Otherwise, return 'false' and it will emit the
+/// normal CIE and FDE.
+bool FrameEmitterImpl::EmitCompactUnwind(MCStreamer &Streamer,
+ const MCDwarfFrameInfo &Frame) {
+#if 1
+ return false;
+#else
+ MCContext &Context = Streamer.getContext();
+ const TargetAsmInfo &TAI = Context.getTargetAsmInfo();
+ bool VerboseAsm = Streamer.isVerboseAsm();
+
+ // range-start range-length compact-unwind-enc personality-func lsda
+ // _foo LfooEnd-_foo 0x00000023 0 0
+ // _bar LbarEnd-_bar 0x00000025 __gxx_personality except_tab1
+ //
+ // .section __LD,__compact_unwind,regular,debug
+ //
+ // # compact unwind for _foo
+ // .quad _foo
+ // .set L1,LfooEnd-_foo
+ // .long L1
+ // .long 0x01010001
+ // .quad 0
+ // .quad 0
+ //
+ // # compact unwind for _bar
+ // .quad _bar
+ // .set L2,LbarEnd-_bar
+ // .long L2
+ // .long 0x01020011
+ // .quad __gxx_personality
+ // .quad except_tab1
+
+ uint32_t Encoding =
+ TAI.getCompactUnwindEncoding(Frame.Instructions,
+ getDataAlignmentFactor(Streamer), IsEH);
+ if (!Encoding) return false;
+
+ // The encoding needs to know we have an LSDA.
+ if (Frame.Lsda)
+ Encoding |= 0x40000000;
+
+ Streamer.SwitchSection(TAI.getCompactUnwindSection());
+
+ // Range Start
+ unsigned FDEEncoding = TAI.getFDEEncoding(UsingCFI);
+ unsigned Size = getSizeForEncoding(Streamer, FDEEncoding);
+ if (VerboseAsm) Streamer.AddComment("Range Start");
+ Streamer.EmitSymbolValue(Frame.Function, Size);
+
+ // Range Length
+ const MCExpr *Range = MakeStartMinusEndExpr(Streamer, *Frame.Begin,
+ *Frame.End, 0);
+ if (VerboseAsm) Streamer.AddComment("Range Length");
+ Streamer.EmitAbsValue(Range, 4);
+
+ // Compact Encoding
+ Size = getSizeForEncoding(Streamer, dwarf::DW_EH_PE_udata4);
+ if (VerboseAsm) Streamer.AddComment(Twine("Compact Unwind Encoding: 0x") +
+ Twine(llvm::utohexstr(Encoding)));
+ Streamer.EmitIntValue(Encoding, Size);
+
+ // Personality Function
+ Size = getSizeForEncoding(Streamer, dwarf::DW_EH_PE_absptr);
+ if (VerboseAsm) Streamer.AddComment("Personality Function");
+ if (Frame.Personality)
+ Streamer.EmitSymbolValue(Frame.Personality, Size);
+ else
+ Streamer.EmitIntValue(0, Size); // No personality fn
+
+ // LSDA
+ Size = getSizeForEncoding(Streamer, Frame.LsdaEncoding);
+ if (VerboseAsm) Streamer.AddComment("LSDA");
+ if (Frame.Lsda)
+ Streamer.EmitSymbolValue(Frame.Lsda, Size);
+ else
+ Streamer.EmitIntValue(0, Size); // No LSDA
+
+ return true;
+#endif
+}
+
const MCSymbol &FrameEmitterImpl::EmitCIE(MCStreamer &streamer,
const MCSymbol *personality,
unsigned personalityEncoding,
const MCSymbol *lsda,
unsigned lsdaEncoding) {
MCContext &context = streamer.getContext();
- const TargetAsmInfo &asmInfo = context.getTargetAsmInfo();
+ const TargetAsmInfo &TAI = context.getTargetAsmInfo();
+ bool verboseAsm = streamer.isVerboseAsm();
MCSymbol *sectionStart;
- if (asmInfo.isFunctionEHFrameSymbolPrivate() || !IsEH)
+ if (TAI.isFunctionEHFrameSymbolPrivate() || !IsEH)
sectionStart = context.CreateTempSymbol();
else
sectionStart = context.GetOrCreateSymbol(Twine("EH_frame") + Twine(CIENum));
+ streamer.EmitLabel(sectionStart);
CIENum++;
- MCSymbol *sectionEnd = streamer.getContext().CreateTempSymbol();
+ MCSymbol *sectionEnd = context.CreateTempSymbol();
// Length
const MCExpr *Length = MakeStartMinusEndExpr(streamer, *sectionStart,
*sectionEnd, 4);
- streamer.EmitLabel(sectionStart);
+ if (verboseAsm) streamer.AddComment("CIE Length");
streamer.EmitAbsValue(Length, 4);
// CIE ID
unsigned CIE_ID = IsEH ? 0 : -1;
+ if (verboseAsm) streamer.AddComment("CIE ID Tag");
streamer.EmitIntValue(CIE_ID, 4);
// Version
+ if (verboseAsm) streamer.AddComment("DW_CIE_VERSION");
streamer.EmitIntValue(dwarf::DW_CIE_VERSION, 1);
// Augmentation String
SmallString<8> Augmentation;
if (IsEH) {
+ if (verboseAsm) streamer.AddComment("CIE Augmentation");
Augmentation += "z";
if (personality)
Augmentation += "P";
@@ -668,13 +815,16 @@ const MCSymbol &FrameEmitterImpl::EmitCIE(MCStreamer &streamer,
streamer.EmitIntValue(0, 1);
// Code Alignment Factor
+ if (verboseAsm) streamer.AddComment("CIE Code Alignment Factor");
streamer.EmitULEB128IntValue(1);
// Data Alignment Factor
+ if (verboseAsm) streamer.AddComment("CIE Data Alignment Factor");
streamer.EmitSLEB128IntValue(getDataAlignmentFactor(streamer));
// Return Address Register
- streamer.EmitULEB128IntValue(asmInfo.getDwarfRARegNum(true));
+ if (verboseAsm) streamer.AddComment("CIE Return Address Column");
+ streamer.EmitULEB128IntValue(TAI.getDwarfRARegNum(true));
// Augmentation Data Length (optional)
@@ -691,32 +841,38 @@ const MCSymbol &FrameEmitterImpl::EmitCIE(MCStreamer &streamer,
// Encoding of the FDE pointers
augmentationLength += 1;
+ if (verboseAsm) streamer.AddComment("Augmentation Size");
streamer.EmitULEB128IntValue(augmentationLength);
// Augmentation Data (optional)
if (personality) {
// Personality Encoding
- streamer.EmitIntValue(personalityEncoding, 1);
+ EmitEncodingByte(streamer, personalityEncoding,
+ "Personality Encoding");
// Personality
+ if (verboseAsm) streamer.AddComment("Personality");
EmitPersonality(streamer, *personality, personalityEncoding);
}
+
if (lsda)
- streamer.EmitIntValue(lsdaEncoding, 1); // LSDA Encoding
+ EmitEncodingByte(streamer, lsdaEncoding, "LSDA Encoding");
+
// Encoding of the FDE pointers
- streamer.EmitIntValue(asmInfo.getFDEEncoding(UsingCFI), 1);
+ EmitEncodingByte(streamer, TAI.getFDEEncoding(UsingCFI),
+ "FDE Encoding");
}
// Initial Instructions
- const std::vector<MachineMove> Moves = asmInfo.getInitialFrameState();
+ const std::vector<MachineMove> &Moves = TAI.getInitialFrameState();
std::vector<MCCFIInstruction> Instructions;
for (int i = 0, n = Moves.size(); i != n; ++i) {
MCSymbol *Label = Moves[i].getLabel();
const MachineLocation &Dst =
- TranslateMachineLocation(asmInfo, Moves[i].getDestination());
+ TranslateMachineLocation(TAI, Moves[i].getDestination());
const MachineLocation &Src =
- TranslateMachineLocation(asmInfo, Moves[i].getSource());
+ TranslateMachineLocation(TAI, Moves[i].getSource());
MCCFIInstruction Inst(Label, Dst, Src);
Instructions.push_back(Inst);
}
@@ -724,7 +880,8 @@ const MCSymbol &FrameEmitterImpl::EmitCIE(MCStreamer &streamer,
EmitCFIInstructions(streamer, Instructions, NULL);
// Padding
- streamer.EmitValueToAlignment(IsEH ? 4 : asmInfo.getPointerSize());
+ streamer.EmitValueToAlignment(IsEH
+ ? 4 : context.getAsmInfo().getPointerSize());
streamer.EmitLabel(sectionEnd);
return *sectionStart;
@@ -736,17 +893,19 @@ MCSymbol *FrameEmitterImpl::EmitFDE(MCStreamer &streamer,
MCContext &context = streamer.getContext();
MCSymbol *fdeStart = context.CreateTempSymbol();
MCSymbol *fdeEnd = context.CreateTempSymbol();
- const TargetAsmInfo &TAsmInfo = context.getTargetAsmInfo();
+ const TargetAsmInfo &TAI = context.getTargetAsmInfo();
+ bool verboseAsm = streamer.isVerboseAsm();
- if (!TAsmInfo.isFunctionEHFrameSymbolPrivate() && IsEH) {
- MCSymbol *EHSym = context.GetOrCreateSymbol(
- frame.Function->getName() + Twine(".eh"));
+ if (!TAI.isFunctionEHFrameSymbolPrivate() && IsEH) {
+ MCSymbol *EHSym =
+ context.GetOrCreateSymbol(frame.Function->getName() + Twine(".eh"));
streamer.EmitEHSymAttributes(frame.Function, EHSym);
streamer.EmitLabel(EHSym);
}
// Length
const MCExpr *Length = MakeStartMinusEndExpr(streamer, *fdeStart, *fdeEnd, 0);
+ if (verboseAsm) streamer.AddComment("FDE Length");
streamer.EmitAbsValue(Length, 4);
streamer.EmitLabel(fdeStart);
@@ -756,6 +915,7 @@ MCSymbol *FrameEmitterImpl::EmitFDE(MCStreamer &streamer,
if (IsEH) {
const MCExpr *offset = MakeStartMinusEndExpr(streamer, cieStart, *fdeStart,
0);
+ if (verboseAsm) streamer.AddComment("FDE CIE Offset");
streamer.EmitAbsValue(offset, 4);
} else if (!asmInfo.doesDwarfRequireRelocationForSectionOffset()) {
const MCExpr *offset = MakeStartMinusEndExpr(streamer, *SectionStart,
@@ -764,18 +924,20 @@ MCSymbol *FrameEmitterImpl::EmitFDE(MCStreamer &streamer,
} else {
streamer.EmitSymbolValue(&cieStart, 4);
}
- unsigned fdeEncoding = TAsmInfo.getFDEEncoding(UsingCFI);
+
+ unsigned fdeEncoding = TAI.getFDEEncoding(UsingCFI);
unsigned size = getSizeForEncoding(streamer, fdeEncoding);
// PC Begin
unsigned PCBeginEncoding = IsEH ? fdeEncoding :
(unsigned)dwarf::DW_EH_PE_absptr;
unsigned PCBeginSize = getSizeForEncoding(streamer, PCBeginEncoding);
- EmitSymbol(streamer, *frame.Begin, PCBeginEncoding);
+ EmitSymbol(streamer, *frame.Begin, PCBeginEncoding, "FDE initial location");
// PC Range
const MCExpr *Range = MakeStartMinusEndExpr(streamer, *frame.Begin,
*frame.End, 0);
+ if (verboseAsm) streamer.AddComment("FDE address range");
streamer.EmitAbsValue(Range, size);
if (IsEH) {
@@ -785,11 +947,13 @@ MCSymbol *FrameEmitterImpl::EmitFDE(MCStreamer &streamer,
if (frame.Lsda)
augmentationLength += getSizeForEncoding(streamer, frame.LsdaEncoding);
+ if (verboseAsm) streamer.AddComment("Augmentation size");
streamer.EmitULEB128IntValue(augmentationLength);
// Augmentation Data
if (frame.Lsda)
- EmitSymbol(streamer, *frame.Lsda, frame.LsdaEncoding);
+ EmitSymbol(streamer, *frame.Lsda, frame.LsdaEncoding,
+ "Language Specific Data Area");
}
// Call Frame Instructions
@@ -843,39 +1007,47 @@ namespace llvm {
};
}
-void MCDwarfFrameEmitter::Emit(MCStreamer &streamer,
- bool usingCFI,
- bool isEH) {
- MCContext &context = streamer.getContext();
- const TargetAsmInfo &asmInfo = context.getTargetAsmInfo();
- const MCSection &section = isEH ?
- *asmInfo.getEHFrameSection() : *asmInfo.getDwarfFrameSection();
- streamer.SwitchSection(&section);
- MCSymbol *SectionStart = context.CreateTempSymbol();
- streamer.EmitLabel(SectionStart);
-
- MCSymbol *fdeEnd = NULL;
+void MCDwarfFrameEmitter::Emit(MCStreamer &Streamer,
+ bool UsingCFI,
+ bool IsEH) {
+ MCContext &Context = Streamer.getContext();
+ const TargetAsmInfo &TAI = Context.getTargetAsmInfo();
+ const MCSection &Section = IsEH ? *TAI.getEHFrameSection() :
+ *TAI.getDwarfFrameSection();
+ Streamer.SwitchSection(&Section);
+ MCSymbol *SectionStart = Context.CreateTempSymbol();
+ Streamer.EmitLabel(SectionStart);
+
+ MCSymbol *FDEEnd = NULL;
DenseMap<CIEKey, const MCSymbol*> CIEStarts;
- FrameEmitterImpl Emitter(usingCFI, isEH, SectionStart);
+ FrameEmitterImpl Emitter(UsingCFI, IsEH, SectionStart);
const MCSymbol *DummyDebugKey = NULL;
- for (unsigned i = 0, n = streamer.getNumFrameInfos(); i < n; ++i) {
- const MCDwarfFrameInfo &frame = streamer.getFrameInfo(i);
- CIEKey key(frame.Personality, frame.PersonalityEncoding,
- frame.LsdaEncoding);
- const MCSymbol *&cieStart = isEH ? CIEStarts[key] : DummyDebugKey;
- if (!cieStart)
- cieStart = &Emitter.EmitCIE(streamer, frame.Personality,
- frame.PersonalityEncoding, frame.Lsda,
- frame.LsdaEncoding);
- fdeEnd = Emitter.EmitFDE(streamer, *cieStart, frame);
+ for (unsigned i = 0, n = Streamer.getNumFrameInfos(); i < n; ++i) {
+ const MCDwarfFrameInfo &Frame = Streamer.getFrameInfo(i);
+ if (IsEH && TAI.getCompactUnwindSection() &&
+ Emitter.EmitCompactUnwind(Streamer, Frame)) {
+ FDEEnd = NULL;
+ continue;
+ }
+
+ CIEKey Key(Frame.Personality, Frame.PersonalityEncoding,
+ Frame.LsdaEncoding);
+ const MCSymbol *&CIEStart = IsEH ? CIEStarts[Key] : DummyDebugKey;
+ if (!CIEStart)
+ CIEStart = &Emitter.EmitCIE(Streamer, Frame.Personality,
+ Frame.PersonalityEncoding, Frame.Lsda,
+ Frame.LsdaEncoding);
+
+ FDEEnd = Emitter.EmitFDE(Streamer, *CIEStart, Frame);
+
if (i != n - 1)
- streamer.EmitLabel(fdeEnd);
+ Streamer.EmitLabel(FDEEnd);
}
- streamer.EmitValueToAlignment(asmInfo.getPointerSize());
- if (fdeEnd)
- streamer.EmitLabel(fdeEnd);
+ Streamer.EmitValueToAlignment(Context.getAsmInfo().getPointerSize());
+ if (FDEEnd)
+ Streamer.EmitLabel(FDEEnd);
}
void MCDwarfFrameEmitter::EmitAdvanceLoc(MCStreamer &Streamer,
diff --git a/lib/MC/MCELFStreamer.cpp b/lib/MC/MCELFStreamer.cpp
index bbb2789ea81c..49340edbed5e 100644
--- a/lib/MC/MCELFStreamer.cpp
+++ b/lib/MC/MCELFStreamer.cpp
@@ -26,7 +26,6 @@
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/Target/TargetAsmBackend.h"
-#include "llvm/Target/TargetAsmInfo.h"
using namespace llvm;
diff --git a/lib/MC/MCELFStreamer.h b/lib/MC/MCELFStreamer.h
index db34d58ec600..855e7e9ca60f 100644
--- a/lib/MC/MCELFStreamer.h
+++ b/lib/MC/MCELFStreamer.h
@@ -138,137 +138,3 @@ private:
} // end llvm namespace
#endif
-//===- lib/MC/MCELFStreamer.h - ELF Object Output -------------------------===//
-//
-// The LLVM Compiler Infrastructure
-//
-// This file is distributed under the University of Illinois Open Source
-// License. See LICENSE.TXT for details.
-//
-//===----------------------------------------------------------------------===//
-//
-// This file assembles .s files and emits ELF .o object files.
-//
-//===----------------------------------------------------------------------===//
-
-#ifndef LLVM_MC_MCELFSTREAMER_H
-#define LLVM_MC_MCELFSTREAMER_H
-
-#include "llvm/ADT/SmallPtrSet.h"
-#include "llvm/ADT/DenseMap.h"
-#include "llvm/MC/MCAssembler.h"
-#include "llvm/MC/MCContext.h"
-#include "llvm/MC/MCObjectStreamer.h"
-#include "llvm/MC/MCSectionELF.h"
-
-namespace llvm {
-
-class MCELFStreamer : public MCObjectStreamer {
-public:
- MCELFStreamer(MCContext &Context, TargetAsmBackend &TAB,
- raw_ostream &OS, MCCodeEmitter *Emitter)
- : MCObjectStreamer(Context, TAB, OS, Emitter) {}
-
- ~MCELFStreamer() {}
-
- /// @name MCStreamer Interface
- /// @{
-
- virtual void InitSections();
- virtual void ChangeSection(const MCSection *Section);
- virtual void EmitLabel(MCSymbol *Symbol);
- virtual void EmitAssemblerFlag(MCAssemblerFlag Flag);
- virtual void EmitThumbFunc(MCSymbol *Func);
- virtual void EmitAssignment(MCSymbol *Symbol, const MCExpr *Value);
- virtual void EmitWeakReference(MCSymbol *Alias, const MCSymbol *Symbol);
- virtual void EmitSymbolAttribute(MCSymbol *Symbol, MCSymbolAttr Attribute);
- virtual void EmitSymbolDesc(MCSymbol *Symbol, unsigned DescValue) {
- assert(0 && "ELF doesn't support this directive");
- }
- virtual void EmitCommonSymbol(MCSymbol *Symbol, uint64_t Size,
- unsigned ByteAlignment);
- virtual void BeginCOFFSymbolDef(const MCSymbol *Symbol) {
- assert(0 && "ELF doesn't support this directive");
- }
-
- virtual void EmitCOFFSymbolStorageClass(int StorageClass) {
- assert(0 && "ELF doesn't support this directive");
- }
-
- virtual void EmitCOFFSymbolType(int Type) {
- assert(0 && "ELF doesn't support this directive");
- }
-
- virtual void EndCOFFSymbolDef() {
- assert(0 && "ELF doesn't support this directive");
- }
-
- virtual void EmitELFSize(MCSymbol *Symbol, const MCExpr *Value) {
- MCSymbolData &SD = getAssembler().getOrCreateSymbolData(*Symbol);
- SD.setSize(Value);
- }
-
- virtual void EmitLocalCommonSymbol(MCSymbol *Symbol, uint64_t Size);
-
- virtual void EmitZerofill(const MCSection *Section, MCSymbol *Symbol = 0,
- unsigned Size = 0, unsigned ByteAlignment = 0) {
- assert(0 && "ELF doesn't support this directive");
- }
- virtual void EmitTBSSSymbol(const MCSection *Section, MCSymbol *Symbol,
- uint64_t Size, unsigned ByteAlignment = 0) {
- assert(0 && "ELF doesn't support this directive");
- }
- virtual void EmitBytes(StringRef Data, unsigned AddrSpace);
- virtual void EmitValueToAlignment(unsigned ByteAlignment, int64_t Value = 0,
- unsigned ValueSize = 1,
- unsigned MaxBytesToEmit = 0);
- virtual void EmitCodeAlignment(unsigned ByteAlignment,
- unsigned MaxBytesToEmit = 0);
-
- virtual void EmitFileDirective(StringRef Filename);
-
- virtual void Finish();
-
-private:
- virtual void EmitInstToFragment(const MCInst &Inst);
- virtual void EmitInstToData(const MCInst &Inst);
-
- void fixSymbolsInTLSFixups(const MCExpr *expr);
-
- struct LocalCommon {
- MCSymbolData *SD;
- uint64_t Size;
- unsigned ByteAlignment;
- };
- std::vector<LocalCommon> LocalCommons;
-
- SmallPtrSet<MCSymbol *, 16> BindingExplicitlySet;
- /// @}
- void SetSection(StringRef Section, unsigned Type, unsigned Flags,
- SectionKind Kind) {
- SwitchSection(getContext().getELFSection(Section, Type, Flags, Kind));
- }
-
- void SetSectionData() {
- SetSection(".data", ELF::SHT_PROGBITS,
- ELF::SHF_WRITE |ELF::SHF_ALLOC,
- SectionKind::getDataRel());
- EmitCodeAlignment(4, 0);
- }
- void SetSectionText() {
- SetSection(".text", ELF::SHT_PROGBITS,
- ELF::SHF_EXECINSTR |
- ELF::SHF_ALLOC, SectionKind::getText());
- EmitCodeAlignment(4, 0);
- }
- void SetSectionBss() {
- SetSection(".bss", ELF::SHT_NOBITS,
- ELF::SHF_WRITE |
- ELF::SHF_ALLOC, SectionKind::getBSS());
- EmitCodeAlignment(4, 0);
- }
-};
-
-} // end llvm namespace
-
-#endif
diff --git a/lib/MC/MCLoggingStreamer.cpp b/lib/MC/MCLoggingStreamer.cpp
index 46ea9b844a6a..309752ec5f02 100644
--- a/lib/MC/MCLoggingStreamer.cpp
+++ b/lib/MC/MCLoggingStreamer.cpp
@@ -85,9 +85,11 @@ public:
virtual void EmitDwarfAdvanceLineAddr(int64_t LineDelta,
const MCSymbol *LastLabel,
- const MCSymbol *Label) {
+ const MCSymbol *Label,
+ unsigned PointerSize) {
LogCall("EmitDwarfAdvanceLineAddr");
- return Child->EmitDwarfAdvanceLineAddr(LineDelta, LastLabel, Label);
+ return Child->EmitDwarfAdvanceLineAddr(LineDelta, LastLabel, Label,
+ PointerSize);
}
virtual void EmitSymbolAttribute(MCSymbol *Symbol, MCSymbolAttr Attribute) {
diff --git a/lib/MC/MCMachOStreamer.cpp b/lib/MC/MCMachOStreamer.cpp
index 12aeb4f48fda..1b21249ca321 100644
--- a/lib/MC/MCMachOStreamer.cpp
+++ b/lib/MC/MCMachOStreamer.cpp
@@ -24,7 +24,6 @@
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/Target/TargetAsmBackend.h"
-#include "llvm/Target/TargetAsmInfo.h"
using namespace llvm;
diff --git a/lib/MC/MCNullStreamer.cpp b/lib/MC/MCNullStreamer.cpp
index f38b82231207..9577af010205 100644
--- a/lib/MC/MCNullStreamer.cpp
+++ b/lib/MC/MCNullStreamer.cpp
@@ -44,7 +44,8 @@ namespace {
virtual void EmitWeakReference(MCSymbol *Alias, const MCSymbol *Symbol){}
virtual void EmitDwarfAdvanceLineAddr(int64_t LineDelta,
const MCSymbol *LastLabel,
- const MCSymbol *Label) {}
+ const MCSymbol *Label,
+ unsigned PointerSize) {}
virtual void EmitSymbolAttribute(MCSymbol *Symbol, MCSymbolAttr Attribute){}
diff --git a/lib/MC/MCObjectStreamer.cpp b/lib/MC/MCObjectStreamer.cpp
index e230c5330203..8635aac00302 100644
--- a/lib/MC/MCObjectStreamer.cpp
+++ b/lib/MC/MCObjectStreamer.cpp
@@ -18,7 +18,6 @@
#include "llvm/MC/MCExpr.h"
#include "llvm/MC/MCSymbol.h"
#include "llvm/Target/TargetAsmBackend.h"
-#include "llvm/Target/TargetAsmInfo.h"
using namespace llvm;
MCObjectStreamer::MCObjectStreamer(MCContext &Context, TargetAsmBackend &TAB,
@@ -197,9 +196,9 @@ void MCObjectStreamer::EmitInstToFragment(const MCInst &Inst) {
void MCObjectStreamer::EmitDwarfAdvanceLineAddr(int64_t LineDelta,
const MCSymbol *LastLabel,
- const MCSymbol *Label) {
+ const MCSymbol *Label,
+ unsigned PointerSize) {
if (!LastLabel) {
- int PointerSize = getContext().getTargetAsmInfo().getPointerSize();
EmitDwarfSetLineAddr(LineDelta, Label, PointerSize);
return;
}
diff --git a/lib/MC/MCParser/AsmParser.cpp b/lib/MC/MCParser/AsmParser.cpp
index 4f55cea7bc5e..0c181f39611e 100644
--- a/lib/MC/MCParser/AsmParser.cpp
+++ b/lib/MC/MCParser/AsmParser.cpp
@@ -28,6 +28,7 @@
#include "llvm/MC/MCSymbol.h"
#include "llvm/MC/MCDwarf.h"
#include "llvm/Support/CommandLine.h"
+#include "llvm/Support/MathExtras.h"
#include "llvm/Support/MemoryBuffer.h"
#include "llvm/Support/SourceMgr.h"
#include "llvm/Support/raw_ostream.h"
@@ -84,6 +85,7 @@ private:
AsmLexer Lexer;
MCContext &Ctx;
MCStreamer &Out;
+ const MCAsmInfo &MAI;
SourceMgr &SrcMgr;
MCAsmParserExtension *GenericParser;
MCAsmParserExtension *PlatformParser;
@@ -135,7 +137,7 @@ public:
virtual MCContext &getContext() { return Ctx; }
virtual MCStreamer &getStreamer() { return Out; }
- virtual bool Warning(SMLoc L, const Twine &Meg);
+ virtual bool Warning(SMLoc L, const Twine &Msg);
virtual bool Error(SMLoc L, const Twine &Msg);
const AsmToken &Lex();
@@ -160,8 +162,9 @@ private:
void HandleMacroExit();
void PrintMacroInstantiations();
- void PrintMessage(SMLoc Loc, const Twine &Msg, const char *Type) const {
- SrcMgr.PrintMessage(Loc, Msg, Type);
+ void PrintMessage(SMLoc Loc, const Twine &Msg, const char *Type,
+ bool ShowLine = true) const {
+ SrcMgr.PrintMessage(Loc, Msg, Type, ShowLine);
}
/// EnterIncludeFile - Enter the specified file. This returns true on failure.
@@ -337,7 +340,7 @@ enum { DEFAULT_ADDRSPACE = 0 };
AsmParser::AsmParser(const Target &T, SourceMgr &_SM, MCContext &_Ctx,
MCStreamer &_Out, const MCAsmInfo &_MAI)
- : Lexer(_MAI), Ctx(_Ctx), Out(_Out), SrcMgr(_SM),
+ : Lexer(_MAI), Ctx(_Ctx), Out(_Out), MAI(_MAI), SrcMgr(_SM),
GenericParser(new GenericAsmParser), PlatformParser(0),
CurBuffer(0), MacrosEnabled(true) {
Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer));
@@ -466,6 +469,29 @@ bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) {
TokError("unassigned file number: " + Twine(i) + " for .file directives");
}
+ // Check to see that all assembler local symbols were actually defined.
+ // Targets that don't do subsections via symbols may not want this, though,
+ // so conservatively exclude them. Only do this if we're finalizing, though,
+ // as otherwise we won't necessarilly have seen everything yet.
+ if (!NoFinalize && MAI.hasSubsectionsViaSymbols()) {
+ const MCContext::SymbolTable &Symbols = getContext().getSymbols();
+ for (MCContext::SymbolTable::const_iterator i = Symbols.begin(),
+ e = Symbols.end();
+ i != e; ++i) {
+ MCSymbol *Sym = i->getValue();
+ // Variable symbols may not be marked as defined, so check those
+ // explicitly. If we know it's a variable, we have a definition for
+ // the purposes of this check.
+ if (Sym->isTemporary() && !Sym->isVariable() && !Sym->isDefined())
+ // FIXME: We would really like to refer back to where the symbol was
+ // first referenced for a source location. We need to add something
+ // to track that. Currently, we just point to the end of the file.
+ PrintMessage(getLexer().getLoc(), "assembler local symbol '" +
+ Sym->getName() + "' not defined", "error", false);
+ }
+ }
+
+
// Finalize the output stream if there are no errors and if the client wants
// us to.
if (!HadError && !NoFinalize)
@@ -1121,7 +1147,7 @@ bool AsmParser::ParseStatement() {
if (IDVal == ".weak_def_can_be_hidden")
return ParseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate);
- if (IDVal == ".comm")
+ if (IDVal == ".comm" || IDVal == ".common")
return ParseDirectiveComm(/*IsLocal=*/false);
if (IDVal == ".lcomm")
return ParseDirectiveComm(/*IsLocal=*/true);
@@ -1168,7 +1194,7 @@ bool AsmParser::ParseStatement() {
for (unsigned i = 0; i != ParsedOperands.size(); ++i) {
if (i != 0)
OS << ", ";
- ParsedOperands[i]->dump(OS);
+ ParsedOperands[i]->print(OS);
}
OS << "]";
@@ -1587,13 +1613,18 @@ bool AsmParser::ParseDirectiveValue(unsigned Size) {
for (;;) {
const MCExpr *Value;
+ SMLoc ExprLoc = getLexer().getLoc();
if (ParseExpression(Value))
return true;
// Special case constant expressions to match code generator.
- if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value))
- getStreamer().EmitIntValue(MCE->getValue(), Size, DEFAULT_ADDRSPACE);
- else
+ if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) {
+ assert(Size <= 8 && "Invalid size");
+ uint64_t IntValue = MCE->getValue();
+ if (!isUIntN(8 * Size, IntValue) && !isIntN(8 * Size, IntValue))
+ return Error(ExprLoc, "literal value out of range for directive");
+ getStreamer().EmitIntValue(IntValue, Size, DEFAULT_ADDRSPACE);
+ } else
getStreamer().EmitValue(Value, Size, DEFAULT_ADDRSPACE);
if (getLexer().is(AsmToken::EndOfStatement))
diff --git a/lib/MC/MCParser/COFFAsmParser.cpp b/lib/MC/MCParser/COFFAsmParser.cpp
index 64f635517b11..66ad384c7db2 100644
--- a/lib/MC/MCParser/COFFAsmParser.cpp
+++ b/lib/MC/MCParser/COFFAsmParser.cpp
@@ -401,14 +401,14 @@ bool COFFAsmParser::ParseAtUnwindOrAtExcept(bool &unwind, bool &except) {
bool COFFAsmParser::ParseSEHRegisterNumber(unsigned &RegNo) {
SMLoc startLoc = getLexer().getLoc();
if (getLexer().is(AsmToken::Percent)) {
- const TargetAsmInfo &asmInfo = getContext().getTargetAsmInfo();
+ const TargetAsmInfo &TAI = getContext().getTargetAsmInfo();
SMLoc endLoc;
unsigned LLVMRegNo;
if (getParser().getTargetParser().ParseRegister(LLVMRegNo,startLoc,endLoc))
return true;
// Check that this is a non-volatile register.
- const unsigned *NVRegs = asmInfo.getCalleeSavedRegs();
+ const unsigned *NVRegs = TAI.getCalleeSavedRegs();
unsigned i;
for (i = 0; NVRegs[i] != 0; ++i)
if (NVRegs[i] == LLVMRegNo)
@@ -416,7 +416,7 @@ bool COFFAsmParser::ParseSEHRegisterNumber(unsigned &RegNo) {
if (NVRegs[i] == 0)
return Error(startLoc, "expected non-volatile register");
- int SEHRegNo = asmInfo.getSEHRegNum(LLVMRegNo);
+ int SEHRegNo = TAI.getSEHRegNum(LLVMRegNo);
if (SEHRegNo < 0)
return Error(startLoc,"register can't be represented in SEH unwind info");
RegNo = SEHRegNo;
diff --git a/lib/MC/MCParser/MCAsmParser.cpp b/lib/MC/MCParser/MCAsmParser.cpp
index 70295efc613c..4030e41036aa 100644
--- a/lib/MC/MCParser/MCAsmParser.cpp
+++ b/lib/MC/MCParser/MCAsmParser.cpp
@@ -12,6 +12,8 @@
#include "llvm/MC/MCParser/MCAsmLexer.h"
#include "llvm/MC/MCParser/MCParsedAsmOperand.h"
#include "llvm/Support/SourceMgr.h"
+#include "llvm/Support/raw_ostream.h"
+#include "llvm/Support/Debug.h"
#include "llvm/Target/TargetAsmParser.h"
using namespace llvm;
@@ -41,4 +43,6 @@ bool MCAsmParser::ParseExpression(const MCExpr *&Res) {
return ParseExpression(Res, L);
}
-
+void MCParsedAsmOperand::dump() const {
+ dbgs() << " " << *this;
+}
diff --git a/lib/MC/MCParser/TargetAsmParser.cpp b/lib/MC/MCParser/TargetAsmParser.cpp
index 8d43c21f4bc9..512f6b044911 100644
--- a/lib/MC/MCParser/TargetAsmParser.cpp
+++ b/lib/MC/MCParser/TargetAsmParser.cpp
@@ -10,8 +10,8 @@
#include "llvm/Target/TargetAsmParser.h"
using namespace llvm;
-TargetAsmParser::TargetAsmParser(const Target &T)
- : TheTarget(T), AvailableFeatures(0)
+TargetAsmParser::TargetAsmParser()
+ : AvailableFeatures(0)
{
}
diff --git a/lib/MC/MCStreamer.cpp b/lib/MC/MCStreamer.cpp
index ae3ed0f3f61a..6e96b78e315b 100644
--- a/lib/MC/MCStreamer.cpp
+++ b/lib/MC/MCStreamer.cpp
@@ -15,7 +15,6 @@
#include "llvm/MC/MCSymbol.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/raw_ostream.h"
-#include "llvm/Target/TargetAsmInfo.h"
#include "llvm/ADT/SmallString.h"
#include "llvm/ADT/Twine.h"
#include <cstdlib>
@@ -81,7 +80,7 @@ void MCStreamer::EmitIntValue(uint64_t Value, unsigned Size,
assert((isUIntN(8 * Size, Value) || isIntN(8 * Size, Value)) &&
"Invalid size");
char buf[8];
- const bool isLittleEndian = Context.getTargetAsmInfo().isLittleEndian();
+ const bool isLittleEndian = Context.getAsmInfo().isLittleEndian();
for (unsigned i = 0; i != Size; ++i) {
unsigned index = isLittleEndian ? i : (Size - i - 1);
buf[i] = uint8_t(Value >> (index * 8));
diff --git a/lib/MC/MCSubtargetInfo.cpp b/lib/MC/MCSubtargetInfo.cpp
new file mode 100644
index 000000000000..86dc1083cee9
--- /dev/null
+++ b/lib/MC/MCSubtargetInfo.cpp
@@ -0,0 +1,96 @@
+//===-- MCSubtargetInfo.cpp - Subtarget Information -----------------------===//
+//
+// The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/MC/MCSubtargetInfo.h"
+#include "llvm/MC/MCInstrItineraries.h"
+#include "llvm/MC/SubtargetFeature.h"
+#include "llvm/ADT/StringRef.h"
+#include "llvm/ADT/Triple.h"
+#include "llvm/Support/raw_ostream.h"
+#include <algorithm>
+
+using namespace llvm;
+
+void
+MCSubtargetInfo::InitMCSubtargetInfo(StringRef TT, StringRef CPU, StringRef FS,
+ const SubtargetFeatureKV *PF,
+ const SubtargetFeatureKV *PD,
+ const SubtargetInfoKV *PI,
+ const InstrStage *IS,
+ const unsigned *OC,
+ const unsigned *FP,
+ unsigned NF, unsigned NP) {
+ TargetTriple = TT;
+ ProcFeatures = PF;
+ ProcDesc = PD;
+ ProcItins = PI;
+ Stages = IS;
+ OperandCycles = OC;
+ ForwardingPathes = FP;
+ NumFeatures = NF;
+ NumProcs = NP;
+
+ SubtargetFeatures Features(FS);
+ FeatureBits = Features.getFeatureBits(CPU, ProcDesc, NumProcs,
+ ProcFeatures, NumFeatures);
+}
+
+
+/// ReInitMCSubtargetInfo - Change CPU (and optionally supplemented with
+/// feature string) and recompute feature bits.
+uint64_t MCSubtargetInfo::ReInitMCSubtargetInfo(StringRef CPU, StringRef FS) {
+ SubtargetFeatures Features(FS);
+ FeatureBits = Features.getFeatureBits(CPU, ProcDesc, NumProcs,
+ ProcFeatures, NumFeatures);
+ return FeatureBits;
+}
+
+/// ToggleFeature - Toggle a feature and returns the re-computed feature
+/// bits. This version does not change the implied bits.
+uint64_t MCSubtargetInfo::ToggleFeature(uint64_t FB) {
+ FeatureBits ^= FB;
+ return FeatureBits;
+}
+
+/// ToggleFeature - Toggle a feature and returns the re-computed feature
+/// bits. This version will also change all implied bits.
+uint64_t MCSubtargetInfo::ToggleFeature(StringRef FS) {
+ SubtargetFeatures Features;
+ FeatureBits = Features.ToggleFeature(FeatureBits, FS,
+ ProcFeatures, NumFeatures);
+ return FeatureBits;
+}
+
+
+InstrItineraryData
+MCSubtargetInfo::getInstrItineraryForCPU(StringRef CPU) const {
+ assert(ProcItins && "Instruction itineraries information not available!");
+
+#ifndef NDEBUG
+ for (size_t i = 1; i < NumProcs; i++) {
+ assert(strcmp(ProcItins[i - 1].Key, ProcItins[i].Key) < 0 &&
+ "Itineraries table is not sorted");
+ }
+#endif
+
+ // Find entry
+ SubtargetInfoKV KV;
+ KV.Key = CPU.data();
+ const SubtargetInfoKV *Found =
+ std::lower_bound(ProcItins, ProcItins+NumProcs, KV);
+ if (Found == ProcItins+NumProcs || StringRef(Found->Key) != CPU) {
+ errs() << "'" << CPU
+ << "' is not a recognized processor for this target"
+ << " (ignoring processor)\n";
+ return InstrItineraryData();
+ }
+
+ return InstrItineraryData(Stages, OperandCycles, ForwardingPathes,
+ (InstrItinerary *)Found->Value);
+}
diff --git a/lib/MC/MCWin64EH.cpp b/lib/MC/MCWin64EH.cpp
index 9453f5c2a963..e698384a49f1 100644
--- a/lib/MC/MCWin64EH.cpp
+++ b/lib/MC/MCWin64EH.cpp
@@ -225,9 +225,9 @@ void MCWin64EHUnwindEmitter::EmitUnwindInfo(MCStreamer &streamer,
// Switch sections (the static function above is meant to be called from
// here and from Emit().
MCContext &context = streamer.getContext();
- const TargetAsmInfo &asmInfo = context.getTargetAsmInfo();
+ const TargetAsmInfo &TAI = context.getTargetAsmInfo();
const MCSection *xdataSect =
- asmInfo.getWin64EHTableSection(GetSectionSuffix(info->Function));
+ TAI.getWin64EHTableSection(GetSectionSuffix(info->Function));
streamer.SwitchSection(xdataSect);
llvm::EmitUnwindInfo(streamer, info);
@@ -236,11 +236,11 @@ void MCWin64EHUnwindEmitter::EmitUnwindInfo(MCStreamer &streamer,
void MCWin64EHUnwindEmitter::Emit(MCStreamer &streamer) {
MCContext &context = streamer.getContext();
// Emit the unwind info structs first.
- const TargetAsmInfo &asmInfo = context.getTargetAsmInfo();
+ const TargetAsmInfo &TAI = context.getTargetAsmInfo();
for (unsigned i = 0; i < streamer.getNumW64UnwindInfos(); ++i) {
MCWin64EHUnwindInfo &info = streamer.getW64UnwindInfo(i);
const MCSection *xdataSect =
- asmInfo.getWin64EHTableSection(GetSectionSuffix(info.Function));
+ TAI.getWin64EHTableSection(GetSectionSuffix(info.Function));
streamer.SwitchSection(xdataSect);
llvm::EmitUnwindInfo(streamer, &info);
}
@@ -248,7 +248,7 @@ void MCWin64EHUnwindEmitter::Emit(MCStreamer &streamer) {
for (unsigned i = 0; i < streamer.getNumW64UnwindInfos(); ++i) {
MCWin64EHUnwindInfo &info = streamer.getW64UnwindInfo(i);
const MCSection *pdataSect =
- asmInfo.getWin64EHFuncTableSection(GetSectionSuffix(info.Function));
+ TAI.getWin64EHFuncTableSection(GetSectionSuffix(info.Function));
streamer.SwitchSection(pdataSect);
EmitRuntimeFunction(streamer, &info);
}
diff --git a/lib/MC/MachObjectWriter.cpp b/lib/MC/MachObjectWriter.cpp
index f049b1c6e2a4..69efe231ad6e 100644
--- a/lib/MC/MachObjectWriter.cpp
+++ b/lib/MC/MachObjectWriter.cpp
@@ -23,34 +23,12 @@
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Target/TargetAsmBackend.h"
-// FIXME: Gross.
-#include "../Target/ARM/ARMFixupKinds.h"
-#include "../Target/X86/X86FixupKinds.h"
-
#include <vector>
using namespace llvm;
using namespace llvm::object;
-// FIXME: this has been copied from (or to) X86AsmBackend.cpp
-static unsigned getFixupKindLog2Size(unsigned Kind) {
- switch (Kind) {
- default:
- llvm_unreachable("invalid fixup kind!");
- case FK_PCRel_1:
- case FK_Data_1: return 0;
- case FK_PCRel_2:
- case FK_Data_2: return 1;
- case FK_PCRel_4:
- // FIXME: Remove these!!!
- case X86::reloc_riprel_4byte:
- case X86::reloc_riprel_4byte_movq_load:
- case X86::reloc_signed_4byte:
- case FK_Data_4: return 2;
- case FK_Data_8: return 3;
- }
-}
-
-static bool doesSymbolRequireExternRelocation(MCSymbolData *SD) {
+bool MachObjectWriter::
+doesSymbolRequireExternRelocation(const MCSymbolData *SD) {
// Undefined symbols are always extern.
if (SD->Symbol->isUndefined())
return true;
@@ -64,1557 +42,740 @@ static bool doesSymbolRequireExternRelocation(MCSymbolData *SD) {
return false;
}
-namespace {
-
-class MachObjectWriter : public MCObjectWriter {
- /// MachSymbolData - Helper struct for containing some precomputed information
- /// on symbols.
- struct MachSymbolData {
- MCSymbolData *SymbolData;
- uint64_t StringIndex;
- uint8_t SectionIndex;
+bool MachObjectWriter::
+MachSymbolData::operator<(const MachSymbolData &RHS) const {
+ return SymbolData->getSymbol().getName() <
+ RHS.SymbolData->getSymbol().getName();
+}
- // Support lexicographic sorting.
- bool operator<(const MachSymbolData &RHS) const {
- return SymbolData->getSymbol().getName() <
- RHS.SymbolData->getSymbol().getName();
- }
- };
+bool MachObjectWriter::isFixupKindPCRel(const MCAssembler &Asm, unsigned Kind) {
+ const MCFixupKindInfo &FKI = Asm.getBackend().getFixupKindInfo(
+ (MCFixupKind) Kind);
- /// The target specific Mach-O writer instance.
- llvm::OwningPtr<MCMachObjectTargetWriter> TargetObjectWriter;
+ return FKI.Flags & MCFixupKindInfo::FKF_IsPCRel;
+}
- /// @name Relocation Data
- /// @{
+uint64_t MachObjectWriter::getFragmentAddress(const MCFragment *Fragment,
+ const MCAsmLayout &Layout) const {
+ return getSectionAddress(Fragment->getParent()) +
+ Layout.getFragmentOffset(Fragment);
+}
- llvm::DenseMap<const MCSectionData*,
- std::vector<macho::RelocationEntry> > Relocations;
- llvm::DenseMap<const MCSectionData*, unsigned> IndirectSymBase;
+uint64_t MachObjectWriter::getSymbolAddress(const MCSymbolData* SD,
+ const MCAsmLayout &Layout) const {
+ const MCSymbol &S = SD->getSymbol();
+
+ // If this is a variable, then recursively evaluate now.
+ if (S.isVariable()) {
+ MCValue Target;
+ if (!S.getVariableValue()->EvaluateAsRelocatable(Target, Layout))
+ report_fatal_error("unable to evaluate offset for variable '" +
+ S.getName() + "'");
+
+ // Verify that any used symbols are defined.
+ if (Target.getSymA() && Target.getSymA()->getSymbol().isUndefined())
+ report_fatal_error("unable to evaluate offset to undefined symbol '" +
+ Target.getSymA()->getSymbol().getName() + "'");
+ if (Target.getSymB() && Target.getSymB()->getSymbol().isUndefined())
+ report_fatal_error("unable to evaluate offset to undefined symbol '" +
+ Target.getSymB()->getSymbol().getName() + "'");
+
+ uint64_t Address = Target.getConstant();
+ if (Target.getSymA())
+ Address += getSymbolAddress(&Layout.getAssembler().getSymbolData(
+ Target.getSymA()->getSymbol()), Layout);
+ if (Target.getSymB())
+ Address += getSymbolAddress(&Layout.getAssembler().getSymbolData(
+ Target.getSymB()->getSymbol()), Layout);
+ return Address;
+ }
- /// @}
- /// @name Symbol Table Data
- /// @{
+ return getSectionAddress(SD->getFragment()->getParent()) +
+ Layout.getSymbolOffset(SD);
+}
- SmallString<256> StringTable;
- std::vector<MachSymbolData> LocalSymbolData;
- std::vector<MachSymbolData> ExternalSymbolData;
- std::vector<MachSymbolData> UndefinedSymbolData;
+uint64_t MachObjectWriter::getPaddingSize(const MCSectionData *SD,
+ const MCAsmLayout &Layout) const {
+ uint64_t EndAddr = getSectionAddress(SD) + Layout.getSectionAddressSize(SD);
+ unsigned Next = SD->getLayoutOrder() + 1;
+ if (Next >= Layout.getSectionOrder().size())
+ return 0;
+
+ const MCSectionData &NextSD = *Layout.getSectionOrder()[Next];
+ if (NextSD.getSection().isVirtualSection())
+ return 0;
+ return OffsetToAlignment(EndAddr, NextSD.getAlignment());
+}
- /// @}
+void MachObjectWriter::WriteHeader(unsigned NumLoadCommands,
+ unsigned LoadCommandsSize,
+ bool SubsectionsViaSymbols) {
+ uint32_t Flags = 0;
-private:
- /// @name Utility Methods
- /// @{
+ if (SubsectionsViaSymbols)
+ Flags |= macho::HF_SubsectionsViaSymbols;
- bool isFixupKindPCRel(const MCAssembler &Asm, unsigned Kind) {
- const MCFixupKindInfo &FKI = Asm.getBackend().getFixupKindInfo(
- (MCFixupKind) Kind);
+ // struct mach_header (28 bytes) or
+ // struct mach_header_64 (32 bytes)
- return FKI.Flags & MCFixupKindInfo::FKF_IsPCRel;
- }
+ uint64_t Start = OS.tell();
+ (void) Start;
- /// @}
+ Write32(is64Bit() ? macho::HM_Object64 : macho::HM_Object32);
- SectionAddrMap SectionAddress;
- uint64_t getSectionAddress(const MCSectionData* SD) const {
- return SectionAddress.lookup(SD);
- }
- uint64_t getSymbolAddress(const MCSymbolData* SD,
- const MCAsmLayout &Layout) const {
- const MCSymbol &S = SD->getSymbol();
-
- // If this is a variable, then recursively evaluate now.
- if (S.isVariable()) {
- MCValue Target;
- if (!S.getVariableValue()->EvaluateAsRelocatable(Target, Layout))
- report_fatal_error("unable to evaluate offset for variable '" +
- S.getName() + "'");
-
- // Verify that any used symbols are defined.
- if (Target.getSymA() && Target.getSymA()->getSymbol().isUndefined())
- report_fatal_error("unable to evaluate offset to undefined symbol '" +
- Target.getSymA()->getSymbol().getName() + "'");
- if (Target.getSymB() && Target.getSymB()->getSymbol().isUndefined())
- report_fatal_error("unable to evaluate offset to undefined symbol '" +
- Target.getSymB()->getSymbol().getName() + "'");
-
- uint64_t Address = Target.getConstant();
- if (Target.getSymA())
- Address += getSymbolAddress(&Layout.getAssembler().getSymbolData(
- Target.getSymA()->getSymbol()), Layout);
- if (Target.getSymB())
- Address += getSymbolAddress(&Layout.getAssembler().getSymbolData(
- Target.getSymB()->getSymbol()), Layout);
- return Address;
- }
+ Write32(TargetObjectWriter->getCPUType());
+ Write32(TargetObjectWriter->getCPUSubtype());
- return getSectionAddress(SD->getFragment()->getParent()) +
- Layout.getSymbolOffset(SD);
- }
- uint64_t getFragmentAddress(const MCFragment *Fragment,
- const MCAsmLayout &Layout) const {
- return getSectionAddress(Fragment->getParent()) +
- Layout.getFragmentOffset(Fragment);
- }
+ Write32(macho::HFT_Object);
+ Write32(NumLoadCommands);
+ Write32(LoadCommandsSize);
+ Write32(Flags);
+ if (is64Bit())
+ Write32(0); // reserved
- uint64_t getPaddingSize(const MCSectionData *SD,
- const MCAsmLayout &Layout) const {
- uint64_t EndAddr = getSectionAddress(SD) + Layout.getSectionAddressSize(SD);
- unsigned Next = SD->getLayoutOrder() + 1;
- if (Next >= Layout.getSectionOrder().size())
- return 0;
-
- const MCSectionData &NextSD = *Layout.getSectionOrder()[Next];
- if (NextSD.getSection().isVirtualSection())
- return 0;
- return OffsetToAlignment(EndAddr, NextSD.getAlignment());
- }
+ assert(OS.tell() - Start ==
+ (is64Bit() ? macho::Header64Size : macho::Header32Size));
+}
-public:
- MachObjectWriter(MCMachObjectTargetWriter *MOTW, raw_ostream &_OS,
- bool _IsLittleEndian)
- : MCObjectWriter(_OS, _IsLittleEndian), TargetObjectWriter(MOTW) {
+/// WriteSegmentLoadCommand - Write a segment load command.
+///
+/// \arg NumSections - The number of sections in this segment.
+/// \arg SectionDataSize - The total size of the sections.
+void MachObjectWriter::WriteSegmentLoadCommand(unsigned NumSections,
+ uint64_t VMSize,
+ uint64_t SectionDataStartOffset,
+ uint64_t SectionDataSize) {
+ // struct segment_command (56 bytes) or
+ // struct segment_command_64 (72 bytes)
+
+ uint64_t Start = OS.tell();
+ (void) Start;
+
+ unsigned SegmentLoadCommandSize =
+ is64Bit() ? macho::SegmentLoadCommand64Size:
+ macho::SegmentLoadCommand32Size;
+ Write32(is64Bit() ? macho::LCT_Segment64 : macho::LCT_Segment);
+ Write32(SegmentLoadCommandSize +
+ NumSections * (is64Bit() ? macho::Section64Size :
+ macho::Section32Size));
+
+ WriteBytes("", 16);
+ if (is64Bit()) {
+ Write64(0); // vmaddr
+ Write64(VMSize); // vmsize
+ Write64(SectionDataStartOffset); // file offset
+ Write64(SectionDataSize); // file size
+ } else {
+ Write32(0); // vmaddr
+ Write32(VMSize); // vmsize
+ Write32(SectionDataStartOffset); // file offset
+ Write32(SectionDataSize); // file size
}
+ Write32(0x7); // maxprot
+ Write32(0x7); // initprot
+ Write32(NumSections);
+ Write32(0); // flags
- /// @name Target Writer Proxy Accessors
- /// @{
+ assert(OS.tell() - Start == SegmentLoadCommandSize);
+}
- bool is64Bit() const { return TargetObjectWriter->is64Bit(); }
- bool isARM() const {
- uint32_t CPUType = TargetObjectWriter->getCPUType() & ~mach::CTFM_ArchMask;
- return CPUType == mach::CTM_ARM;
+void MachObjectWriter::WriteSection(const MCAssembler &Asm,
+ const MCAsmLayout &Layout,
+ const MCSectionData &SD,
+ uint64_t FileOffset,
+ uint64_t RelocationsStart,
+ unsigned NumRelocations) {
+ uint64_t SectionSize = Layout.getSectionAddressSize(&SD);
+
+ // The offset is unused for virtual sections.
+ if (SD.getSection().isVirtualSection()) {
+ assert(Layout.getSectionFileSize(&SD) == 0 && "Invalid file size!");
+ FileOffset = 0;
}
- /// @}
-
- void WriteHeader(unsigned NumLoadCommands, unsigned LoadCommandsSize,
- bool SubsectionsViaSymbols) {
- uint32_t Flags = 0;
-
- if (SubsectionsViaSymbols)
- Flags |= macho::HF_SubsectionsViaSymbols;
-
- // struct mach_header (28 bytes) or
- // struct mach_header_64 (32 bytes)
-
- uint64_t Start = OS.tell();
- (void) Start;
-
- Write32(is64Bit() ? macho::HM_Object64 : macho::HM_Object32);
-
- Write32(TargetObjectWriter->getCPUType());
- Write32(TargetObjectWriter->getCPUSubtype());
-
- Write32(macho::HFT_Object);
- Write32(NumLoadCommands);
- Write32(LoadCommandsSize);
- Write32(Flags);
- if (is64Bit())
- Write32(0); // reserved
-
- assert(OS.tell() - Start ==
- (is64Bit() ? macho::Header64Size : macho::Header32Size));
+ // struct section (68 bytes) or
+ // struct section_64 (80 bytes)
+
+ uint64_t Start = OS.tell();
+ (void) Start;
+
+ const MCSectionMachO &Section = cast<MCSectionMachO>(SD.getSection());
+ WriteBytes(Section.getSectionName(), 16);
+ WriteBytes(Section.getSegmentName(), 16);
+ if (is64Bit()) {
+ Write64(getSectionAddress(&SD)); // address
+ Write64(SectionSize); // size
+ } else {
+ Write32(getSectionAddress(&SD)); // address
+ Write32(SectionSize); // size
}
+ Write32(FileOffset);
+
+ unsigned Flags = Section.getTypeAndAttributes();
+ if (SD.hasInstructions())
+ Flags |= MCSectionMachO::S_ATTR_SOME_INSTRUCTIONS;
+
+ assert(isPowerOf2_32(SD.getAlignment()) && "Invalid alignment!");
+ Write32(Log2_32(SD.getAlignment()));
+ Write32(NumRelocations ? RelocationsStart : 0);
+ Write32(NumRelocations);
+ Write32(Flags);
+ Write32(IndirectSymBase.lookup(&SD)); // reserved1
+ Write32(Section.getStubSize()); // reserved2
+ if (is64Bit())
+ Write32(0); // reserved3
+
+ assert(OS.tell() - Start == (is64Bit() ? macho::Section64Size :
+ macho::Section32Size));
+}
- /// WriteSegmentLoadCommand - Write a segment load command.
- ///
- /// \arg NumSections - The number of sections in this segment.
- /// \arg SectionDataSize - The total size of the sections.
- void WriteSegmentLoadCommand(unsigned NumSections,
- uint64_t VMSize,
- uint64_t SectionDataStartOffset,
- uint64_t SectionDataSize) {
- // struct segment_command (56 bytes) or
- // struct segment_command_64 (72 bytes)
-
- uint64_t Start = OS.tell();
- (void) Start;
-
- unsigned SegmentLoadCommandSize =
- is64Bit() ? macho::SegmentLoadCommand64Size:
- macho::SegmentLoadCommand32Size;
- Write32(is64Bit() ? macho::LCT_Segment64 : macho::LCT_Segment);
- Write32(SegmentLoadCommandSize +
- NumSections * (is64Bit() ? macho::Section64Size :
- macho::Section32Size));
-
- WriteBytes("", 16);
- if (is64Bit()) {
- Write64(0); // vmaddr
- Write64(VMSize); // vmsize
- Write64(SectionDataStartOffset); // file offset
- Write64(SectionDataSize); // file size
- } else {
- Write32(0); // vmaddr
- Write32(VMSize); // vmsize
- Write32(SectionDataStartOffset); // file offset
- Write32(SectionDataSize); // file size
- }
- Write32(0x7); // maxprot
- Write32(0x7); // initprot
- Write32(NumSections);
- Write32(0); // flags
+void MachObjectWriter::WriteSymtabLoadCommand(uint32_t SymbolOffset,
+ uint32_t NumSymbols,
+ uint32_t StringTableOffset,
+ uint32_t StringTableSize) {
+ // struct symtab_command (24 bytes)
- assert(OS.tell() - Start == SegmentLoadCommandSize);
- }
+ uint64_t Start = OS.tell();
+ (void) Start;
- void WriteSection(const MCAssembler &Asm, const MCAsmLayout &Layout,
- const MCSectionData &SD, uint64_t FileOffset,
- uint64_t RelocationsStart, unsigned NumRelocations) {
- uint64_t SectionSize = Layout.getSectionAddressSize(&SD);
+ Write32(macho::LCT_Symtab);
+ Write32(macho::SymtabLoadCommandSize);
+ Write32(SymbolOffset);
+ Write32(NumSymbols);
+ Write32(StringTableOffset);
+ Write32(StringTableSize);
- // The offset is unused for virtual sections.
- if (SD.getSection().isVirtualSection()) {
- assert(Layout.getSectionFileSize(&SD) == 0 && "Invalid file size!");
- FileOffset = 0;
- }
+ assert(OS.tell() - Start == macho::SymtabLoadCommandSize);
+}
- // struct section (68 bytes) or
- // struct section_64 (80 bytes)
+void MachObjectWriter::WriteDysymtabLoadCommand(uint32_t FirstLocalSymbol,
+ uint32_t NumLocalSymbols,
+ uint32_t FirstExternalSymbol,
+ uint32_t NumExternalSymbols,
+ uint32_t FirstUndefinedSymbol,
+ uint32_t NumUndefinedSymbols,
+ uint32_t IndirectSymbolOffset,
+ uint32_t NumIndirectSymbols) {
+ // struct dysymtab_command (80 bytes)
+
+ uint64_t Start = OS.tell();
+ (void) Start;
+
+ Write32(macho::LCT_Dysymtab);
+ Write32(macho::DysymtabLoadCommandSize);
+ Write32(FirstLocalSymbol);
+ Write32(NumLocalSymbols);
+ Write32(FirstExternalSymbol);
+ Write32(NumExternalSymbols);
+ Write32(FirstUndefinedSymbol);
+ Write32(NumUndefinedSymbols);
+ Write32(0); // tocoff
+ Write32(0); // ntoc
+ Write32(0); // modtaboff
+ Write32(0); // nmodtab
+ Write32(0); // extrefsymoff
+ Write32(0); // nextrefsyms
+ Write32(IndirectSymbolOffset);
+ Write32(NumIndirectSymbols);
+ Write32(0); // extreloff
+ Write32(0); // nextrel
+ Write32(0); // locreloff
+ Write32(0); // nlocrel
+
+ assert(OS.tell() - Start == macho::DysymtabLoadCommandSize);
+}
- uint64_t Start = OS.tell();
- (void) Start;
+void MachObjectWriter::WriteNlist(MachSymbolData &MSD,
+ const MCAsmLayout &Layout) {
+ MCSymbolData &Data = *MSD.SymbolData;
+ const MCSymbol &Symbol = Data.getSymbol();
+ uint8_t Type = 0;
+ uint16_t Flags = Data.getFlags();
+ uint32_t Address = 0;
- const MCSectionMachO &Section = cast<MCSectionMachO>(SD.getSection());
- WriteBytes(Section.getSectionName(), 16);
- WriteBytes(Section.getSegmentName(), 16);
- if (is64Bit()) {
- Write64(getSectionAddress(&SD)); // address
- Write64(SectionSize); // size
+ // Set the N_TYPE bits. See <mach-o/nlist.h>.
+ //
+ // FIXME: Are the prebound or indirect fields possible here?
+ if (Symbol.isUndefined())
+ Type = macho::STT_Undefined;
+ else if (Symbol.isAbsolute())
+ Type = macho::STT_Absolute;
+ else
+ Type = macho::STT_Section;
+
+ // FIXME: Set STAB bits.
+
+ if (Data.isPrivateExtern())
+ Type |= macho::STF_PrivateExtern;
+
+ // Set external bit.
+ if (Data.isExternal() || Symbol.isUndefined())
+ Type |= macho::STF_External;
+
+ // Compute the symbol address.
+ if (Symbol.isDefined()) {
+ if (Symbol.isAbsolute()) {
+ Address = cast<MCConstantExpr>(Symbol.getVariableValue())->getValue();
} else {
- Write32(getSectionAddress(&SD)); // address
- Write32(SectionSize); // size
+ Address = getSymbolAddress(&Data, Layout);
+ }
+ } else if (Data.isCommon()) {
+ // Common symbols are encoded with the size in the address
+ // field, and their alignment in the flags.
+ Address = Data.getCommonSize();
+
+ // Common alignment is packed into the 'desc' bits.
+ if (unsigned Align = Data.getCommonAlignment()) {
+ unsigned Log2Size = Log2_32(Align);
+ assert((1U << Log2Size) == Align && "Invalid 'common' alignment!");
+ if (Log2Size > 15)
+ report_fatal_error("invalid 'common' alignment '" +
+ Twine(Align) + "'");
+ // FIXME: Keep this mask with the SymbolFlags enumeration.
+ Flags = (Flags & 0xF0FF) | (Log2Size << 8);
}
- Write32(FileOffset);
-
- unsigned Flags = Section.getTypeAndAttributes();
- if (SD.hasInstructions())
- Flags |= MCSectionMachO::S_ATTR_SOME_INSTRUCTIONS;
-
- assert(isPowerOf2_32(SD.getAlignment()) && "Invalid alignment!");
- Write32(Log2_32(SD.getAlignment()));
- Write32(NumRelocations ? RelocationsStart : 0);
- Write32(NumRelocations);
- Write32(Flags);
- Write32(IndirectSymBase.lookup(&SD)); // reserved1
- Write32(Section.getStubSize()); // reserved2
- if (is64Bit())
- Write32(0); // reserved3
-
- assert(OS.tell() - Start == (is64Bit() ? macho::Section64Size :
- macho::Section32Size));
}
- void WriteSymtabLoadCommand(uint32_t SymbolOffset, uint32_t NumSymbols,
- uint32_t StringTableOffset,
- uint32_t StringTableSize) {
- // struct symtab_command (24 bytes)
+ // struct nlist (12 bytes)
- uint64_t Start = OS.tell();
- (void) Start;
+ Write32(MSD.StringIndex);
+ Write8(Type);
+ Write8(MSD.SectionIndex);
- Write32(macho::LCT_Symtab);
- Write32(macho::SymtabLoadCommandSize);
- Write32(SymbolOffset);
- Write32(NumSymbols);
- Write32(StringTableOffset);
- Write32(StringTableSize);
-
- assert(OS.tell() - Start == macho::SymtabLoadCommandSize);
- }
-
- void WriteDysymtabLoadCommand(uint32_t FirstLocalSymbol,
- uint32_t NumLocalSymbols,
- uint32_t FirstExternalSymbol,
- uint32_t NumExternalSymbols,
- uint32_t FirstUndefinedSymbol,
- uint32_t NumUndefinedSymbols,
- uint32_t IndirectSymbolOffset,
- uint32_t NumIndirectSymbols) {
- // struct dysymtab_command (80 bytes)
-
- uint64_t Start = OS.tell();
- (void) Start;
-
- Write32(macho::LCT_Dysymtab);
- Write32(macho::DysymtabLoadCommandSize);
- Write32(FirstLocalSymbol);
- Write32(NumLocalSymbols);
- Write32(FirstExternalSymbol);
- Write32(NumExternalSymbols);
- Write32(FirstUndefinedSymbol);
- Write32(NumUndefinedSymbols);
- Write32(0); // tocoff
- Write32(0); // ntoc
- Write32(0); // modtaboff
- Write32(0); // nmodtab
- Write32(0); // extrefsymoff
- Write32(0); // nextrefsyms
- Write32(IndirectSymbolOffset);
- Write32(NumIndirectSymbols);
- Write32(0); // extreloff
- Write32(0); // nextrel
- Write32(0); // locreloff
- Write32(0); // nlocrel
-
- assert(OS.tell() - Start == macho::DysymtabLoadCommandSize);
- }
-
- void WriteNlist(MachSymbolData &MSD, const MCAsmLayout &Layout) {
- MCSymbolData &Data = *MSD.SymbolData;
- const MCSymbol &Symbol = Data.getSymbol();
- uint8_t Type = 0;
- uint16_t Flags = Data.getFlags();
- uint32_t Address = 0;
+ // The Mach-O streamer uses the lowest 16-bits of the flags for the 'desc'
+ // value.
+ Write16(Flags);
+ if (is64Bit())
+ Write64(Address);
+ else
+ Write32(Address);
+}
- // Set the N_TYPE bits. See <mach-o/nlist.h>.
- //
- // FIXME: Are the prebound or indirect fields possible here?
- if (Symbol.isUndefined())
- Type = macho::STT_Undefined;
- else if (Symbol.isAbsolute())
- Type = macho::STT_Absolute;
- else
- Type = macho::STT_Section;
-
- // FIXME: Set STAB bits.
-
- if (Data.isPrivateExtern())
- Type |= macho::STF_PrivateExtern;
-
- // Set external bit.
- if (Data.isExternal() || Symbol.isUndefined())
- Type |= macho::STF_External;
-
- // Compute the symbol address.
- if (Symbol.isDefined()) {
- if (Symbol.isAbsolute()) {
- Address = cast<MCConstantExpr>(Symbol.getVariableValue())->getValue();
- } else {
- Address = getSymbolAddress(&Data, Layout);
- }
- } else if (Data.isCommon()) {
- // Common symbols are encoded with the size in the address
- // field, and their alignment in the flags.
- Address = Data.getCommonSize();
-
- // Common alignment is packed into the 'desc' bits.
- if (unsigned Align = Data.getCommonAlignment()) {
- unsigned Log2Size = Log2_32(Align);
- assert((1U << Log2Size) == Align && "Invalid 'common' alignment!");
- if (Log2Size > 15)
- report_fatal_error("invalid 'common' alignment '" +
- Twine(Align) + "'");
- // FIXME: Keep this mask with the SymbolFlags enumeration.
- Flags = (Flags & 0xF0FF) | (Log2Size << 8);
- }
- }
+void MachObjectWriter::RecordRelocation(const MCAssembler &Asm,
+ const MCAsmLayout &Layout,
+ const MCFragment *Fragment,
+ const MCFixup &Fixup,
+ MCValue Target,
+ uint64_t &FixedValue) {
+ TargetObjectWriter->RecordRelocation(this, Asm, Layout, Fragment, Fixup,
+ Target, FixedValue);
+}
- // struct nlist (12 bytes)
+void MachObjectWriter::BindIndirectSymbols(MCAssembler &Asm) {
+ // This is the point where 'as' creates actual symbols for indirect symbols
+ // (in the following two passes). It would be easier for us to do this sooner
+ // when we see the attribute, but that makes getting the order in the symbol
+ // table much more complicated than it is worth.
+ //
+ // FIXME: Revisit this when the dust settles.
- Write32(MSD.StringIndex);
- Write8(Type);
- Write8(MSD.SectionIndex);
+ // Bind non lazy symbol pointers first.
+ unsigned IndirectIndex = 0;
+ for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
+ ie = Asm.indirect_symbol_end(); it != ie; ++it, ++IndirectIndex) {
+ const MCSectionMachO &Section =
+ cast<MCSectionMachO>(it->SectionData->getSection());
- // The Mach-O streamer uses the lowest 16-bits of the flags for the 'desc'
- // value.
- Write16(Flags);
- if (is64Bit())
- Write64(Address);
- else
- Write32(Address);
- }
+ if (Section.getType() != MCSectionMachO::S_NON_LAZY_SYMBOL_POINTERS)
+ continue;
- // FIXME: We really need to improve the relocation validation. Basically, we
- // want to implement a separate computation which evaluates the relocation
- // entry as the linker would, and verifies that the resultant fixup value is
- // exactly what the encoder wanted. This will catch several classes of
- // problems:
- //
- // - Relocation entry bugs, the two algorithms are unlikely to have the same
- // exact bug.
- //
- // - Relaxation issues, where we forget to relax something.
- //
- // - Input errors, where something cannot be correctly encoded. 'as' allows
- // these through in many cases.
+ // Initialize the section indirect symbol base, if necessary.
+ if (!IndirectSymBase.count(it->SectionData))
+ IndirectSymBase[it->SectionData] = IndirectIndex;
- static bool isFixupKindRIPRel(unsigned Kind) {
- return Kind == X86::reloc_riprel_4byte ||
- Kind == X86::reloc_riprel_4byte_movq_load;
+ Asm.getOrCreateSymbolData(*it->Symbol);
}
- void RecordX86_64Relocation(const MCAssembler &Asm, const MCAsmLayout &Layout,
- const MCFragment *Fragment,
- const MCFixup &Fixup, MCValue Target,
- uint64_t &FixedValue) {
- unsigned IsPCRel = isFixupKindPCRel(Asm, Fixup.getKind());
- unsigned IsRIPRel = isFixupKindRIPRel(Fixup.getKind());
- unsigned Log2Size = getFixupKindLog2Size(Fixup.getKind());
-
- // See <reloc.h>.
- uint32_t FixupOffset =
- Layout.getFragmentOffset(Fragment) + Fixup.getOffset();
- uint32_t FixupAddress =
- getFragmentAddress(Fragment, Layout) + Fixup.getOffset();
- int64_t Value = 0;
- unsigned Index = 0;
- unsigned IsExtern = 0;
- unsigned Type = 0;
-
- Value = Target.getConstant();
-
- if (IsPCRel) {
- // Compensate for the relocation offset, Darwin x86_64 relocations only
- // have the addend and appear to have attempted to define it to be the
- // actual expression addend without the PCrel bias. However, instructions
- // with data following the relocation are not accommodated for (see comment
- // below regarding SIGNED{1,2,4}), so it isn't exactly that either.
- Value += 1LL << Log2Size;
- }
- if (Target.isAbsolute()) { // constant
- // SymbolNum of 0 indicates the absolute section.
- Type = macho::RIT_X86_64_Unsigned;
- Index = 0;
-
- // FIXME: I believe this is broken, I don't think the linker can
- // understand it. I think it would require a local relocation, but I'm not
- // sure if that would work either. The official way to get an absolute
- // PCrel relocation is to use an absolute symbol (which we don't support
- // yet).
- if (IsPCRel) {
- IsExtern = 1;
- Type = macho::RIT_X86_64_Branch;
- }
- } else if (Target.getSymB()) { // A - B + constant
- const MCSymbol *A = &Target.getSymA()->getSymbol();
- MCSymbolData &A_SD = Asm.getSymbolData(*A);
- const MCSymbolData *A_Base = Asm.getAtom(&A_SD);
-
- const MCSymbol *B = &Target.getSymB()->getSymbol();
- MCSymbolData &B_SD = Asm.getSymbolData(*B);
- const MCSymbolData *B_Base = Asm.getAtom(&B_SD);
-
- // Neither symbol can be modified.
- if (Target.getSymA()->getKind() != MCSymbolRefExpr::VK_None ||
- Target.getSymB()->getKind() != MCSymbolRefExpr::VK_None)
- report_fatal_error("unsupported relocation of modified symbol");
-
- // We don't support PCrel relocations of differences. Darwin 'as' doesn't
- // implement most of these correctly.
- if (IsPCRel)
- report_fatal_error("unsupported pc-relative relocation of difference");
-
- // The support for the situation where one or both of the symbols would
- // require a local relocation is handled just like if the symbols were
- // external. This is certainly used in the case of debug sections where
- // the section has only temporary symbols and thus the symbols don't have
- // base symbols. This is encoded using the section ordinal and
- // non-extern relocation entries.
-
- // Darwin 'as' doesn't emit correct relocations for this (it ends up with
- // a single SIGNED relocation); reject it for now. Except the case where
- // both symbols don't have a base, equal but both NULL.
- if (A_Base == B_Base && A_Base)
- report_fatal_error("unsupported relocation with identical base");
-
- Value += getSymbolAddress(&A_SD, Layout) -
- (A_Base == NULL ? 0 : getSymbolAddress(A_Base, Layout));
- Value -= getSymbolAddress(&B_SD, Layout) -
- (B_Base == NULL ? 0 : getSymbolAddress(B_Base, Layout));
-
- if (A_Base) {
- Index = A_Base->getIndex();
- IsExtern = 1;
- }
- else {
- Index = A_SD.getFragment()->getParent()->getOrdinal() + 1;
- IsExtern = 0;
- }
- Type = macho::RIT_X86_64_Unsigned;
-
- macho::RelocationEntry MRE;
- MRE.Word0 = FixupOffset;
- MRE.Word1 = ((Index << 0) |
- (IsPCRel << 24) |
- (Log2Size << 25) |
- (IsExtern << 27) |
- (Type << 28));
- Relocations[Fragment->getParent()].push_back(MRE);
-
- if (B_Base) {
- Index = B_Base->getIndex();
- IsExtern = 1;
- }
- else {
- Index = B_SD.getFragment()->getParent()->getOrdinal() + 1;
- IsExtern = 0;
- }
- Type = macho::RIT_X86_64_Subtractor;
- } else {
- const MCSymbol *Symbol = &Target.getSymA()->getSymbol();
- MCSymbolData &SD = Asm.getSymbolData(*Symbol);
- const MCSymbolData *Base = Asm.getAtom(&SD);
-
- // Relocations inside debug sections always use local relocations when
- // possible. This seems to be done because the debugger doesn't fully
- // understand x86_64 relocation entries, and expects to find values that
- // have already been fixed up.
- if (Symbol->isInSection()) {
- const MCSectionMachO &Section = static_cast<const MCSectionMachO&>(
- Fragment->getParent()->getSection());
- if (Section.hasAttribute(MCSectionMachO::S_ATTR_DEBUG))
- Base = 0;
- }
+ // Then lazy symbol pointers and symbol stubs.
+ IndirectIndex = 0;
+ for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
+ ie = Asm.indirect_symbol_end(); it != ie; ++it, ++IndirectIndex) {
+ const MCSectionMachO &Section =
+ cast<MCSectionMachO>(it->SectionData->getSection());
- // x86_64 almost always uses external relocations, except when there is no
- // symbol to use as a base address (a local symbol with no preceding
- // non-local symbol).
- if (Base) {
- Index = Base->getIndex();
- IsExtern = 1;
-
- // Add the local offset, if needed.
- if (Base != &SD)
- Value += Layout.getSymbolOffset(&SD) - Layout.getSymbolOffset(Base);
- } else if (Symbol->isInSection() && !Symbol->isVariable()) {
- // The index is the section ordinal (1-based).
- Index = SD.getFragment()->getParent()->getOrdinal() + 1;
- IsExtern = 0;
- Value += getSymbolAddress(&SD, Layout);
-
- if (IsPCRel)
- Value -= FixupAddress + (1 << Log2Size);
- } else if (Symbol->isVariable()) {
- const MCExpr *Value = Symbol->getVariableValue();
- int64_t Res;
- bool isAbs = Value->EvaluateAsAbsolute(Res, Layout, SectionAddress);
- if (isAbs) {
- FixedValue = Res;
- return;
- } else {
- report_fatal_error("unsupported relocation of variable '" +
- Symbol->getName() + "'");
- }
- } else {
- report_fatal_error("unsupported relocation of undefined symbol '" +
- Symbol->getName() + "'");
- }
+ if (Section.getType() != MCSectionMachO::S_LAZY_SYMBOL_POINTERS &&
+ Section.getType() != MCSectionMachO::S_SYMBOL_STUBS)
+ continue;
- MCSymbolRefExpr::VariantKind Modifier = Target.getSymA()->getKind();
- if (IsPCRel) {
- if (IsRIPRel) {
- if (Modifier == MCSymbolRefExpr::VK_GOTPCREL) {
- // x86_64 distinguishes movq foo@GOTPCREL so that the linker can
- // rewrite the movq to an leaq at link time if the symbol ends up in
- // the same linkage unit.
- if (unsigned(Fixup.getKind()) == X86::reloc_riprel_4byte_movq_load)
- Type = macho::RIT_X86_64_GOTLoad;
- else
- Type = macho::RIT_X86_64_GOT;
- } else if (Modifier == MCSymbolRefExpr::VK_TLVP) {
- Type = macho::RIT_X86_64_TLV;
- } else if (Modifier != MCSymbolRefExpr::VK_None) {
- report_fatal_error("unsupported symbol modifier in relocation");
- } else {
- Type = macho::RIT_X86_64_Signed;
-
- // The Darwin x86_64 relocation format has a problem where it cannot
- // encode an address (L<foo> + <constant>) which is outside the atom
- // containing L<foo>. Generally, this shouldn't occur but it does
- // happen when we have a RIPrel instruction with data following the
- // relocation entry (e.g., movb $012, L0(%rip)). Even with the PCrel
- // adjustment Darwin x86_64 uses, the offset is still negative and
- // the linker has no way to recognize this.
- //
- // To work around this, Darwin uses several special relocation types
- // to indicate the offsets. However, the specification or
- // implementation of these seems to also be incomplete; they should
- // adjust the addend as well based on the actual encoded instruction
- // (the additional bias), but instead appear to just look at the
- // final offset.
- switch (-(Target.getConstant() + (1LL << Log2Size))) {
- case 1: Type = macho::RIT_X86_64_Signed1; break;
- case 2: Type = macho::RIT_X86_64_Signed2; break;
- case 4: Type = macho::RIT_X86_64_Signed4; break;
- }
- }
- } else {
- if (Modifier != MCSymbolRefExpr::VK_None)
- report_fatal_error("unsupported symbol modifier in branch "
- "relocation");
-
- Type = macho::RIT_X86_64_Branch;
- }
- } else {
- if (Modifier == MCSymbolRefExpr::VK_GOT) {
- Type = macho::RIT_X86_64_GOT;
- } else if (Modifier == MCSymbolRefExpr::VK_GOTPCREL) {
- // GOTPCREL is allowed as a modifier on non-PCrel instructions, in
- // which case all we do is set the PCrel bit in the relocation entry;
- // this is used with exception handling, for example. The source is
- // required to include any necessary offset directly.
- Type = macho::RIT_X86_64_GOT;
- IsPCRel = 1;
- } else if (Modifier == MCSymbolRefExpr::VK_TLVP) {
- report_fatal_error("TLVP symbol modifier should have been rip-rel");
- } else if (Modifier != MCSymbolRefExpr::VK_None)
- report_fatal_error("unsupported symbol modifier in relocation");
- else
- Type = macho::RIT_X86_64_Unsigned;
- }
- }
+ // Initialize the section indirect symbol base, if necessary.
+ if (!IndirectSymBase.count(it->SectionData))
+ IndirectSymBase[it->SectionData] = IndirectIndex;
- // x86_64 always writes custom values into the fixups.
- FixedValue = Value;
-
- // struct relocation_info (8 bytes)
- macho::RelocationEntry MRE;
- MRE.Word0 = FixupOffset;
- MRE.Word1 = ((Index << 0) |
- (IsPCRel << 24) |
- (Log2Size << 25) |
- (IsExtern << 27) |
- (Type << 28));
- Relocations[Fragment->getParent()].push_back(MRE);
+ // Set the symbol type to undefined lazy, but only on construction.
+ //
+ // FIXME: Do not hardcode.
+ bool Created;
+ MCSymbolData &Entry = Asm.getOrCreateSymbolData(*it->Symbol, &Created);
+ if (Created)
+ Entry.setFlags(Entry.getFlags() | 0x0001);
}
+}
- void RecordScatteredRelocation(const MCAssembler &Asm,
- const MCAsmLayout &Layout,
- const MCFragment *Fragment,
- const MCFixup &Fixup, MCValue Target,
- unsigned Log2Size,
- uint64_t &FixedValue) {
- uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset();
- unsigned IsPCRel = isFixupKindPCRel(Asm, Fixup.getKind());
- unsigned Type = macho::RIT_Vanilla;
-
- // See <reloc.h>.
- const MCSymbol *A = &Target.getSymA()->getSymbol();
- MCSymbolData *A_SD = &Asm.getSymbolData(*A);
-
- if (!A_SD->getFragment())
- report_fatal_error("symbol '" + A->getName() +
- "' can not be undefined in a subtraction expression");
-
- uint32_t Value = getSymbolAddress(A_SD, Layout);
- uint64_t SecAddr = getSectionAddress(A_SD->getFragment()->getParent());
- FixedValue += SecAddr;
- uint32_t Value2 = 0;
-
- if (const MCSymbolRefExpr *B = Target.getSymB()) {
- MCSymbolData *B_SD = &Asm.getSymbolData(B->getSymbol());
-
- if (!B_SD->getFragment())
- report_fatal_error("symbol '" + B->getSymbol().getName() +
- "' can not be undefined in a subtraction expression");
-
- // Select the appropriate difference relocation type.
- //
- // Note that there is no longer any semantic difference between these two
- // relocation types from the linkers point of view, this is done solely
- // for pedantic compatibility with 'as'.
- Type = A_SD->isExternal() ? (unsigned)macho::RIT_Difference :
- (unsigned)macho::RIT_Generic_LocalDifference;
- Value2 = getSymbolAddress(B_SD, Layout);
- FixedValue -= getSectionAddress(B_SD->getFragment()->getParent());
- }
-
- // Relocations are written out in reverse order, so the PAIR comes first.
- if (Type == macho::RIT_Difference ||
- Type == macho::RIT_Generic_LocalDifference) {
- macho::RelocationEntry MRE;
- MRE.Word0 = ((0 << 0) |
- (macho::RIT_Pair << 24) |
- (Log2Size << 28) |
- (IsPCRel << 30) |
- macho::RF_Scattered);
- MRE.Word1 = Value2;
- Relocations[Fragment->getParent()].push_back(MRE);
+/// ComputeSymbolTable - Compute the symbol table data
+///
+/// \param StringTable [out] - The string table data.
+/// \param StringIndexMap [out] - Map from symbol names to offsets in the
+/// string table.
+void MachObjectWriter::
+ComputeSymbolTable(MCAssembler &Asm, SmallString<256> &StringTable,
+ std::vector<MachSymbolData> &LocalSymbolData,
+ std::vector<MachSymbolData> &ExternalSymbolData,
+ std::vector<MachSymbolData> &UndefinedSymbolData) {
+ // Build section lookup table.
+ DenseMap<const MCSection*, uint8_t> SectionIndexMap;
+ unsigned Index = 1;
+ for (MCAssembler::iterator it = Asm.begin(),
+ ie = Asm.end(); it != ie; ++it, ++Index)
+ SectionIndexMap[&it->getSection()] = Index;
+ assert(Index <= 256 && "Too many sections!");
+
+ // Index 0 is always the empty string.
+ StringMap<uint64_t> StringIndexMap;
+ StringTable += '\x00';
+
+ // Build the symbol arrays and the string table, but only for non-local
+ // symbols.
+ //
+ // The particular order that we collect the symbols and create the string
+ // table, then sort the symbols is chosen to match 'as'. Even though it
+ // doesn't matter for correctness, this is important for letting us diff .o
+ // files.
+ for (MCAssembler::symbol_iterator it = Asm.symbol_begin(),
+ ie = Asm.symbol_end(); it != ie; ++it) {
+ const MCSymbol &Symbol = it->getSymbol();
+
+ // Ignore non-linker visible symbols.
+ if (!Asm.isSymbolLinkerVisible(it->getSymbol()))
+ continue;
+
+ if (!it->isExternal() && !Symbol.isUndefined())
+ continue;
+
+ uint64_t &Entry = StringIndexMap[Symbol.getName()];
+ if (!Entry) {
+ Entry = StringTable.size();
+ StringTable += Symbol.getName();
+ StringTable += '\x00';
}
- macho::RelocationEntry MRE;
- MRE.Word0 = ((FixupOffset << 0) |
- (Type << 24) |
- (Log2Size << 28) |
- (IsPCRel << 30) |
- macho::RF_Scattered);
- MRE.Word1 = Value;
- Relocations[Fragment->getParent()].push_back(MRE);
- }
-
- void RecordARMScatteredRelocation(const MCAssembler &Asm,
- const MCAsmLayout &Layout,
- const MCFragment *Fragment,
- const MCFixup &Fixup, MCValue Target,
- unsigned Log2Size,
- uint64_t &FixedValue) {
- uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset();
- unsigned IsPCRel = isFixupKindPCRel(Asm, Fixup.getKind());
- unsigned Type = macho::RIT_Vanilla;
-
- // See <reloc.h>.
- const MCSymbol *A = &Target.getSymA()->getSymbol();
- MCSymbolData *A_SD = &Asm.getSymbolData(*A);
-
- if (!A_SD->getFragment())
- report_fatal_error("symbol '" + A->getName() +
- "' can not be undefined in a subtraction expression");
-
- uint32_t Value = getSymbolAddress(A_SD, Layout);
- uint64_t SecAddr = getSectionAddress(A_SD->getFragment()->getParent());
- FixedValue += SecAddr;
- uint32_t Value2 = 0;
-
- if (const MCSymbolRefExpr *B = Target.getSymB()) {
- MCSymbolData *B_SD = &Asm.getSymbolData(B->getSymbol());
-
- if (!B_SD->getFragment())
- report_fatal_error("symbol '" + B->getSymbol().getName() +
- "' can not be undefined in a subtraction expression");
-
- // Select the appropriate difference relocation type.
- Type = macho::RIT_Difference;
- Value2 = getSymbolAddress(B_SD, Layout);
- FixedValue -= getSectionAddress(B_SD->getFragment()->getParent());
- }
+ MachSymbolData MSD;
+ MSD.SymbolData = it;
+ MSD.StringIndex = Entry;
- // Relocations are written out in reverse order, so the PAIR comes first.
- if (Type == macho::RIT_Difference ||
- Type == macho::RIT_Generic_LocalDifference) {
- macho::RelocationEntry MRE;
- MRE.Word0 = ((0 << 0) |
- (macho::RIT_Pair << 24) |
- (Log2Size << 28) |
- (IsPCRel << 30) |
- macho::RF_Scattered);
- MRE.Word1 = Value2;
- Relocations[Fragment->getParent()].push_back(MRE);
+ if (Symbol.isUndefined()) {
+ MSD.SectionIndex = 0;
+ UndefinedSymbolData.push_back(MSD);
+ } else if (Symbol.isAbsolute()) {
+ MSD.SectionIndex = 0;
+ ExternalSymbolData.push_back(MSD);
+ } else {
+ MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
+ assert(MSD.SectionIndex && "Invalid section index!");
+ ExternalSymbolData.push_back(MSD);
}
-
- macho::RelocationEntry MRE;
- MRE.Word0 = ((FixupOffset << 0) |
- (Type << 24) |
- (Log2Size << 28) |
- (IsPCRel << 30) |
- macho::RF_Scattered);
- MRE.Word1 = Value;
- Relocations[Fragment->getParent()].push_back(MRE);
}
- void RecordARMMovwMovtRelocation(const MCAssembler &Asm,
- const MCAsmLayout &Layout,
- const MCFragment *Fragment,
- const MCFixup &Fixup, MCValue Target,
- uint64_t &FixedValue) {
- uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset();
- unsigned IsPCRel = isFixupKindPCRel(Asm, Fixup.getKind());
- unsigned Type = macho::RIT_ARM_Half;
-
- // See <reloc.h>.
- const MCSymbol *A = &Target.getSymA()->getSymbol();
- MCSymbolData *A_SD = &Asm.getSymbolData(*A);
-
- if (!A_SD->getFragment())
- report_fatal_error("symbol '" + A->getName() +
- "' can not be undefined in a subtraction expression");
-
- uint32_t Value = getSymbolAddress(A_SD, Layout);
- uint32_t Value2 = 0;
- uint64_t SecAddr = getSectionAddress(A_SD->getFragment()->getParent());
- FixedValue += SecAddr;
-
- if (const MCSymbolRefExpr *B = Target.getSymB()) {
- MCSymbolData *B_SD = &Asm.getSymbolData(B->getSymbol());
-
- if (!B_SD->getFragment())
- report_fatal_error("symbol '" + B->getSymbol().getName() +
- "' can not be undefined in a subtraction expression");
-
- // Select the appropriate difference relocation type.
- Type = macho::RIT_ARM_HalfDifference;
- Value2 = getSymbolAddress(B_SD, Layout);
- FixedValue -= getSectionAddress(B_SD->getFragment()->getParent());
- }
-
- // Relocations are written out in reverse order, so the PAIR comes first.
- // ARM_RELOC_HALF and ARM_RELOC_HALF_SECTDIFF abuse the r_length field:
- //
- // For these two r_type relocations they always have a pair following them
- // and the r_length bits are used differently. The encoding of the
- // r_length is as follows:
- // low bit of r_length:
- // 0 - :lower16: for movw instructions
- // 1 - :upper16: for movt instructions
- // high bit of r_length:
- // 0 - arm instructions
- // 1 - thumb instructions
- // the other half of the relocated expression is in the following pair
- // relocation entry in the the low 16 bits of r_address field.
- unsigned ThumbBit = 0;
- unsigned MovtBit = 0;
- switch ((unsigned)Fixup.getKind()) {
- default: break;
- case ARM::fixup_arm_movt_hi16:
- case ARM::fixup_arm_movt_hi16_pcrel:
- MovtBit = 1;
- break;
- case ARM::fixup_t2_movt_hi16:
- case ARM::fixup_t2_movt_hi16_pcrel:
- MovtBit = 1;
- // Fallthrough
- case ARM::fixup_t2_movw_lo16:
- case ARM::fixup_t2_movw_lo16_pcrel:
- ThumbBit = 1;
- break;
- }
+ // Now add the data for local symbols.
+ for (MCAssembler::symbol_iterator it = Asm.symbol_begin(),
+ ie = Asm.symbol_end(); it != ie; ++it) {
+ const MCSymbol &Symbol = it->getSymbol();
+ // Ignore non-linker visible symbols.
+ if (!Asm.isSymbolLinkerVisible(it->getSymbol()))
+ continue;
- if (Type == macho::RIT_ARM_HalfDifference) {
- uint32_t OtherHalf = MovtBit
- ? (FixedValue & 0xffff) : ((FixedValue & 0xffff0000) >> 16);
+ if (it->isExternal() || Symbol.isUndefined())
+ continue;
- macho::RelocationEntry MRE;
- MRE.Word0 = ((OtherHalf << 0) |
- (macho::RIT_Pair << 24) |
- (MovtBit << 28) |
- (ThumbBit << 29) |
- (IsPCRel << 30) |
- macho::RF_Scattered);
- MRE.Word1 = Value2;
- Relocations[Fragment->getParent()].push_back(MRE);
+ uint64_t &Entry = StringIndexMap[Symbol.getName()];
+ if (!Entry) {
+ Entry = StringTable.size();
+ StringTable += Symbol.getName();
+ StringTable += '\x00';
}
- macho::RelocationEntry MRE;
- MRE.Word0 = ((FixupOffset << 0) |
- (Type << 24) |
- (MovtBit << 28) |
- (ThumbBit << 29) |
- (IsPCRel << 30) |
- macho::RF_Scattered);
- MRE.Word1 = Value;
- Relocations[Fragment->getParent()].push_back(MRE);
- }
+ MachSymbolData MSD;
+ MSD.SymbolData = it;
+ MSD.StringIndex = Entry;
- void RecordTLVPRelocation(const MCAssembler &Asm,
- const MCAsmLayout &Layout,
- const MCFragment *Fragment,
- const MCFixup &Fixup, MCValue Target,
- uint64_t &FixedValue) {
- assert(Target.getSymA()->getKind() == MCSymbolRefExpr::VK_TLVP &&
- !is64Bit() &&
- "Should only be called with a 32-bit TLVP relocation!");
-
- unsigned Log2Size = getFixupKindLog2Size(Fixup.getKind());
- uint32_t Value = Layout.getFragmentOffset(Fragment)+Fixup.getOffset();
- unsigned IsPCRel = 0;
-
- // Get the symbol data.
- MCSymbolData *SD_A = &Asm.getSymbolData(Target.getSymA()->getSymbol());
- unsigned Index = SD_A->getIndex();
-
- // We're only going to have a second symbol in pic mode and it'll be a
- // subtraction from the picbase. For 32-bit pic the addend is the difference
- // between the picbase and the next address. For 32-bit static the addend
- // is zero.
- if (Target.getSymB()) {
- // If this is a subtraction then we're pcrel.
- uint32_t FixupAddress =
- getFragmentAddress(Fragment, Layout) + Fixup.getOffset();
- MCSymbolData *SD_B = &Asm.getSymbolData(Target.getSymB()->getSymbol());
- IsPCRel = 1;
- FixedValue = (FixupAddress - getSymbolAddress(SD_B, Layout) +
- Target.getConstant());
- FixedValue += 1ULL << Log2Size;
+ if (Symbol.isAbsolute()) {
+ MSD.SectionIndex = 0;
+ LocalSymbolData.push_back(MSD);
} else {
- FixedValue = 0;
+ MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
+ assert(MSD.SectionIndex && "Invalid section index!");
+ LocalSymbolData.push_back(MSD);
}
-
- // struct relocation_info (8 bytes)
- macho::RelocationEntry MRE;
- MRE.Word0 = Value;
- MRE.Word1 = ((Index << 0) |
- (IsPCRel << 24) |
- (Log2Size << 25) |
- (1 << 27) | // Extern
- (macho::RIT_Generic_TLV << 28)); // Type
- Relocations[Fragment->getParent()].push_back(MRE);
}
- static bool getARMFixupKindMachOInfo(unsigned Kind, unsigned &RelocType,
- unsigned &Log2Size) {
- RelocType = unsigned(macho::RIT_Vanilla);
- Log2Size = ~0U;
+ // External and undefined symbols are required to be in lexicographic order.
+ std::sort(ExternalSymbolData.begin(), ExternalSymbolData.end());
+ std::sort(UndefinedSymbolData.begin(), UndefinedSymbolData.end());
+
+ // Set the symbol indices.
+ Index = 0;
+ for (unsigned i = 0, e = LocalSymbolData.size(); i != e; ++i)
+ LocalSymbolData[i].SymbolData->setIndex(Index++);
+ for (unsigned i = 0, e = ExternalSymbolData.size(); i != e; ++i)
+ ExternalSymbolData[i].SymbolData->setIndex(Index++);
+ for (unsigned i = 0, e = UndefinedSymbolData.size(); i != e; ++i)
+ UndefinedSymbolData[i].SymbolData->setIndex(Index++);
+
+ // The string table is padded to a multiple of 4.
+ while (StringTable.size() % 4)
+ StringTable += '\x00';
+}
- switch (Kind) {
- default:
- return false;
+void MachObjectWriter::computeSectionAddresses(const MCAssembler &Asm,
+ const MCAsmLayout &Layout) {
+ uint64_t StartAddress = 0;
+ const SmallVectorImpl<MCSectionData*> &Order = Layout.getSectionOrder();
+ for (int i = 0, n = Order.size(); i != n ; ++i) {
+ const MCSectionData *SD = Order[i];
+ StartAddress = RoundUpToAlignment(StartAddress, SD->getAlignment());
+ SectionAddress[SD] = StartAddress;
+ StartAddress += Layout.getSectionAddressSize(SD);
+
+ // Explicitly pad the section to match the alignment requirements of the
+ // following one. This is for 'gas' compatibility, it shouldn't
+ /// strictly be necessary.
+ StartAddress += getPaddingSize(SD, Layout);
+ }
+}
- case FK_Data_1:
- Log2Size = llvm::Log2_32(1);
- return true;
- case FK_Data_2:
- Log2Size = llvm::Log2_32(2);
- return true;
- case FK_Data_4:
- Log2Size = llvm::Log2_32(4);
- return true;
- case FK_Data_8:
- Log2Size = llvm::Log2_32(8);
- return true;
+void MachObjectWriter::ExecutePostLayoutBinding(MCAssembler &Asm,
+ const MCAsmLayout &Layout) {
+ computeSectionAddresses(Asm, Layout);
- // Handle 24-bit branch kinds.
- case ARM::fixup_arm_ldst_pcrel_12:
- case ARM::fixup_arm_pcrel_10:
- case ARM::fixup_arm_adr_pcrel_12:
- case ARM::fixup_arm_condbranch:
- case ARM::fixup_arm_uncondbranch:
- RelocType = unsigned(macho::RIT_ARM_Branch24Bit);
- // Report as 'long', even though that is not quite accurate.
- Log2Size = llvm::Log2_32(4);
- return true;
+ // Create symbol data for any indirect symbols.
+ BindIndirectSymbols(Asm);
- // Handle Thumb branches.
- case ARM::fixup_arm_thumb_br:
- RelocType = unsigned(macho::RIT_ARM_ThumbBranch22Bit);
- Log2Size = llvm::Log2_32(2);
- return true;
-
- case ARM::fixup_arm_thumb_bl:
- case ARM::fixup_arm_thumb_blx:
- RelocType = unsigned(macho::RIT_ARM_ThumbBranch22Bit);
- Log2Size = llvm::Log2_32(4);
- return true;
+ // Compute symbol table information and bind symbol indices.
+ ComputeSymbolTable(Asm, StringTable, LocalSymbolData, ExternalSymbolData,
+ UndefinedSymbolData);
+}
- case ARM::fixup_arm_movt_hi16:
- case ARM::fixup_arm_movt_hi16_pcrel:
- case ARM::fixup_t2_movt_hi16:
- case ARM::fixup_t2_movt_hi16_pcrel:
- RelocType = unsigned(macho::RIT_ARM_HalfDifference);
- // Report as 'long', even though that is not quite accurate.
- Log2Size = llvm::Log2_32(4);
- return true;
+bool MachObjectWriter::
+IsSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm,
+ const MCSymbolData &DataA,
+ const MCFragment &FB,
+ bool InSet,
+ bool IsPCRel) const {
+ if (InSet)
+ return true;
- case ARM::fixup_arm_movw_lo16:
- case ARM::fixup_arm_movw_lo16_pcrel:
- case ARM::fixup_t2_movw_lo16:
- case ARM::fixup_t2_movw_lo16_pcrel:
- RelocType = unsigned(macho::RIT_ARM_Half);
- // Report as 'long', even though that is not quite accurate.
- Log2Size = llvm::Log2_32(4);
+ // The effective address is
+ // addr(atom(A)) + offset(A)
+ // - addr(atom(B)) - offset(B)
+ // and the offsets are not relocatable, so the fixup is fully resolved when
+ // addr(atom(A)) - addr(atom(B)) == 0.
+ const MCSymbolData *A_Base = 0, *B_Base = 0;
+
+ const MCSymbol &SA = DataA.getSymbol().AliasedSymbol();
+ const MCSection &SecA = SA.getSection();
+ const MCSection &SecB = FB.getParent()->getSection();
+
+ if (IsPCRel) {
+ // The simple (Darwin, except on x86_64) way of dealing with this was to
+ // assume that any reference to a temporary symbol *must* be a temporary
+ // symbol in the same atom, unless the sections differ. Therefore, any PCrel
+ // relocation to a temporary symbol (in the same section) is fully
+ // resolved. This also works in conjunction with absolutized .set, which
+ // requires the compiler to use .set to absolutize the differences between
+ // symbols which the compiler knows to be assembly time constants, so we
+ // don't need to worry about considering symbol differences fully resolved.
+
+ if (!Asm.getBackend().hasReliableSymbolDifference()) {
+ if (!SA.isTemporary() || !SA.isInSection() || &SecA != &SecB)
+ return false;
return true;
}
+ } else {
+ if (!TargetObjectWriter->useAggressiveSymbolFolding())
+ return false;
}
- void RecordARMRelocation(const MCAssembler &Asm, const MCAsmLayout &Layout,
- const MCFragment *Fragment, const MCFixup &Fixup,
- MCValue Target, uint64_t &FixedValue) {
- unsigned IsPCRel = isFixupKindPCRel(Asm, Fixup.getKind());
- unsigned Log2Size;
- unsigned RelocType = macho::RIT_Vanilla;
- if (!getARMFixupKindMachOInfo(Fixup.getKind(), RelocType, Log2Size)) {
- report_fatal_error("unknown ARM fixup kind!");
- return;
- }
- // If this is a difference or a defined symbol plus an offset, then we need
- // a scattered relocation entry. Differences always require scattered
- // relocations.
- if (Target.getSymB()) {
- if (RelocType == macho::RIT_ARM_Half ||
- RelocType == macho::RIT_ARM_HalfDifference)
- return RecordARMMovwMovtRelocation(Asm, Layout, Fragment, Fixup,
- Target, FixedValue);
- return RecordARMScatteredRelocation(Asm, Layout, Fragment, Fixup,
- Target, Log2Size, FixedValue);
- }
+ const MCFragment &FA = *Asm.getSymbolData(SA).getFragment();
- // Get the symbol data, if any.
- MCSymbolData *SD = 0;
- if (Target.getSymA())
- SD = &Asm.getSymbolData(Target.getSymA()->getSymbol());
+ A_Base = FA.getAtom();
+ if (!A_Base)
+ return false;
- // FIXME: For other platforms, we need to use scattered relocations for
- // internal relocations with offsets. If this is an internal relocation
- // with an offset, it also needs a scattered relocation entry.
- //
- // Is this right for ARM?
- uint32_t Offset = Target.getConstant();
- if (IsPCRel && RelocType == macho::RIT_Vanilla)
- Offset += 1 << Log2Size;
- if (Offset && SD && !doesSymbolRequireExternRelocation(SD))
- return RecordARMScatteredRelocation(Asm, Layout, Fragment, Fixup, Target,
- Log2Size, FixedValue);
-
- // See <reloc.h>.
- uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset();
- unsigned Index = 0;
- unsigned IsExtern = 0;
- unsigned Type = 0;
-
- if (Target.isAbsolute()) { // constant
- // FIXME!
- report_fatal_error("FIXME: relocations to absolute targets "
- "not yet implemented");
- } else {
- // Resolve constant variables.
- if (SD->getSymbol().isVariable()) {
- int64_t Res;
- if (SD->getSymbol().getVariableValue()->EvaluateAsAbsolute(
- Res, Layout, SectionAddress)) {
- FixedValue = Res;
- return;
- }
- }
+ B_Base = FB.getAtom();
+ if (!B_Base)
+ return false;
- // Check whether we need an external or internal relocation.
- if (doesSymbolRequireExternRelocation(SD)) {
- IsExtern = 1;
- Index = SD->getIndex();
- // For external relocations, make sure to offset the fixup value to
- // compensate for the addend of the symbol address, if it was
- // undefined. This occurs with weak definitions, for example.
- if (!SD->Symbol->isUndefined())
- FixedValue -= Layout.getSymbolOffset(SD);
- } else {
- // The index is the section ordinal (1-based).
- const MCSectionData &SymSD = Asm.getSectionData(
- SD->getSymbol().getSection());
- Index = SymSD.getOrdinal() + 1;
- FixedValue += getSectionAddress(&SymSD);
- }
- if (IsPCRel)
- FixedValue -= getSectionAddress(Fragment->getParent());
+ // If the atoms are the same, they are guaranteed to have the same address.
+ if (A_Base == B_Base)
+ return true;
- // The type is determined by the fixup kind.
- Type = RelocType;
- }
+ // Otherwise, we can't prove this is fully resolved.
+ return false;
+}
- // struct relocation_info (8 bytes)
- macho::RelocationEntry MRE;
- MRE.Word0 = FixupOffset;
- MRE.Word1 = ((Index << 0) |
- (IsPCRel << 24) |
- (Log2Size << 25) |
- (IsExtern << 27) |
- (Type << 28));
- Relocations[Fragment->getParent()].push_back(MRE);
+void MachObjectWriter::WriteObject(MCAssembler &Asm, const MCAsmLayout &Layout) {
+ unsigned NumSections = Asm.size();
+
+ // The section data starts after the header, the segment load command (and
+ // section headers) and the symbol table.
+ unsigned NumLoadCommands = 1;
+ uint64_t LoadCommandsSize = is64Bit() ?
+ macho::SegmentLoadCommand64Size + NumSections * macho::Section64Size :
+ macho::SegmentLoadCommand32Size + NumSections * macho::Section32Size;
+
+ // Add the symbol table load command sizes, if used.
+ unsigned NumSymbols = LocalSymbolData.size() + ExternalSymbolData.size() +
+ UndefinedSymbolData.size();
+ if (NumSymbols) {
+ NumLoadCommands += 2;
+ LoadCommandsSize += (macho::SymtabLoadCommandSize +
+ macho::DysymtabLoadCommandSize);
}
- void RecordRelocation(const MCAssembler &Asm, const MCAsmLayout &Layout,
- const MCFragment *Fragment, const MCFixup &Fixup,
- MCValue Target, uint64_t &FixedValue) {
- // FIXME: These needs to be factored into the target Mach-O writer.
- if (isARM()) {
- RecordARMRelocation(Asm, Layout, Fragment, Fixup, Target, FixedValue);
- return;
- }
- if (is64Bit()) {
- RecordX86_64Relocation(Asm, Layout, Fragment, Fixup, Target, FixedValue);
- return;
- }
-
- unsigned IsPCRel = isFixupKindPCRel(Asm, Fixup.getKind());
- unsigned Log2Size = getFixupKindLog2Size(Fixup.getKind());
-
- // If this is a 32-bit TLVP reloc it's handled a bit differently.
- if (Target.getSymA() &&
- Target.getSymA()->getKind() == MCSymbolRefExpr::VK_TLVP) {
- RecordTLVPRelocation(Asm, Layout, Fragment, Fixup, Target, FixedValue);
- return;
- }
-
- // If this is a difference or a defined symbol plus an offset, then we need
- // a scattered relocation entry.
- // Differences always require scattered relocations.
- if (Target.getSymB())
- return RecordScatteredRelocation(Asm, Layout, Fragment, Fixup,
- Target, Log2Size, FixedValue);
-
- // Get the symbol data, if any.
- MCSymbolData *SD = 0;
- if (Target.getSymA())
- SD = &Asm.getSymbolData(Target.getSymA()->getSymbol());
-
- // If this is an internal relocation with an offset, it also needs a
- // scattered relocation entry.
- uint32_t Offset = Target.getConstant();
- if (IsPCRel)
- Offset += 1 << Log2Size;
- if (Offset && SD && !doesSymbolRequireExternRelocation(SD))
- return RecordScatteredRelocation(Asm, Layout, Fragment, Fixup,
- Target, Log2Size, FixedValue);
-
- // See <reloc.h>.
- uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset();
- unsigned Index = 0;
- unsigned IsExtern = 0;
- unsigned Type = 0;
-
- if (Target.isAbsolute()) { // constant
- // SymbolNum of 0 indicates the absolute section.
- //
- // FIXME: Currently, these are never generated (see code below). I cannot
- // find a case where they are actually emitted.
- Type = macho::RIT_Vanilla;
- } else {
- // Resolve constant variables.
- if (SD->getSymbol().isVariable()) {
- int64_t Res;
- if (SD->getSymbol().getVariableValue()->EvaluateAsAbsolute(
- Res, Layout, SectionAddress)) {
- FixedValue = Res;
- return;
- }
- }
-
- // Check whether we need an external or internal relocation.
- if (doesSymbolRequireExternRelocation(SD)) {
- IsExtern = 1;
- Index = SD->getIndex();
- // For external relocations, make sure to offset the fixup value to
- // compensate for the addend of the symbol address, if it was
- // undefined. This occurs with weak definitions, for example.
- if (!SD->Symbol->isUndefined())
- FixedValue -= Layout.getSymbolOffset(SD);
- } else {
- // The index is the section ordinal (1-based).
- const MCSectionData &SymSD = Asm.getSectionData(
- SD->getSymbol().getSection());
- Index = SymSD.getOrdinal() + 1;
- FixedValue += getSectionAddress(&SymSD);
- }
- if (IsPCRel)
- FixedValue -= getSectionAddress(Fragment->getParent());
-
- Type = macho::RIT_Vanilla;
- }
-
- // struct relocation_info (8 bytes)
- macho::RelocationEntry MRE;
- MRE.Word0 = FixupOffset;
- MRE.Word1 = ((Index << 0) |
- (IsPCRel << 24) |
- (Log2Size << 25) |
- (IsExtern << 27) |
- (Type << 28));
- Relocations[Fragment->getParent()].push_back(MRE);
+ // Compute the total size of the section data, as well as its file size and vm
+ // size.
+ uint64_t SectionDataStart = (is64Bit() ? macho::Header64Size :
+ macho::Header32Size) + LoadCommandsSize;
+ uint64_t SectionDataSize = 0;
+ uint64_t SectionDataFileSize = 0;
+ uint64_t VMSize = 0;
+ for (MCAssembler::const_iterator it = Asm.begin(),
+ ie = Asm.end(); it != ie; ++it) {
+ const MCSectionData &SD = *it;
+ uint64_t Address = getSectionAddress(&SD);
+ uint64_t Size = Layout.getSectionAddressSize(&SD);
+ uint64_t FileSize = Layout.getSectionFileSize(&SD);
+ FileSize += getPaddingSize(&SD, Layout);
+
+ VMSize = std::max(VMSize, Address + Size);
+
+ if (SD.getSection().isVirtualSection())
+ continue;
+
+ SectionDataSize = std::max(SectionDataSize, Address + Size);
+ SectionDataFileSize = std::max(SectionDataFileSize, Address + FileSize);
}
- void BindIndirectSymbols(MCAssembler &Asm) {
- // This is the point where 'as' creates actual symbols for indirect symbols
- // (in the following two passes). It would be easier for us to do this
- // sooner when we see the attribute, but that makes getting the order in the
- // symbol table much more complicated than it is worth.
- //
- // FIXME: Revisit this when the dust settles.
+ // The section data is padded to 4 bytes.
+ //
+ // FIXME: Is this machine dependent?
+ unsigned SectionDataPadding = OffsetToAlignment(SectionDataFileSize, 4);
+ SectionDataFileSize += SectionDataPadding;
+
+ // Write the prolog, starting with the header and load command...
+ WriteHeader(NumLoadCommands, LoadCommandsSize,
+ Asm.getSubsectionsViaSymbols());
+ WriteSegmentLoadCommand(NumSections, VMSize,
+ SectionDataStart, SectionDataSize);
+
+ // ... and then the section headers.
+ uint64_t RelocTableEnd = SectionDataStart + SectionDataFileSize;
+ for (MCAssembler::const_iterator it = Asm.begin(),
+ ie = Asm.end(); it != ie; ++it) {
+ std::vector<macho::RelocationEntry> &Relocs = Relocations[it];
+ unsigned NumRelocs = Relocs.size();
+ uint64_t SectionStart = SectionDataStart + getSectionAddress(it);
+ WriteSection(Asm, Layout, *it, SectionStart, RelocTableEnd, NumRelocs);
+ RelocTableEnd += NumRelocs * macho::RelocationInfoSize;
+ }
- // Bind non lazy symbol pointers first.
- unsigned IndirectIndex = 0;
- for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
- ie = Asm.indirect_symbol_end(); it != ie; ++it, ++IndirectIndex) {
- const MCSectionMachO &Section =
- cast<MCSectionMachO>(it->SectionData->getSection());
+ // Write the symbol table load command, if used.
+ if (NumSymbols) {
+ unsigned FirstLocalSymbol = 0;
+ unsigned NumLocalSymbols = LocalSymbolData.size();
+ unsigned FirstExternalSymbol = FirstLocalSymbol + NumLocalSymbols;
+ unsigned NumExternalSymbols = ExternalSymbolData.size();
+ unsigned FirstUndefinedSymbol = FirstExternalSymbol + NumExternalSymbols;
+ unsigned NumUndefinedSymbols = UndefinedSymbolData.size();
+ unsigned NumIndirectSymbols = Asm.indirect_symbol_size();
+ unsigned NumSymTabSymbols =
+ NumLocalSymbols + NumExternalSymbols + NumUndefinedSymbols;
+ uint64_t IndirectSymbolSize = NumIndirectSymbols * 4;
+ uint64_t IndirectSymbolOffset = 0;
+
+ // If used, the indirect symbols are written after the section data.
+ if (NumIndirectSymbols)
+ IndirectSymbolOffset = RelocTableEnd;
+
+ // The symbol table is written after the indirect symbol data.
+ uint64_t SymbolTableOffset = RelocTableEnd + IndirectSymbolSize;
+
+ // The string table is written after symbol table.
+ uint64_t StringTableOffset =
+ SymbolTableOffset + NumSymTabSymbols * (is64Bit() ? macho::Nlist64Size :
+ macho::Nlist32Size);
+ WriteSymtabLoadCommand(SymbolTableOffset, NumSymTabSymbols,
+ StringTableOffset, StringTable.size());
+
+ WriteDysymtabLoadCommand(FirstLocalSymbol, NumLocalSymbols,
+ FirstExternalSymbol, NumExternalSymbols,
+ FirstUndefinedSymbol, NumUndefinedSymbols,
+ IndirectSymbolOffset, NumIndirectSymbols);
+ }
- if (Section.getType() != MCSectionMachO::S_NON_LAZY_SYMBOL_POINTERS)
- continue;
+ // Write the actual section data.
+ for (MCAssembler::const_iterator it = Asm.begin(),
+ ie = Asm.end(); it != ie; ++it) {
+ Asm.WriteSectionData(it, Layout);
- // Initialize the section indirect symbol base, if necessary.
- if (!IndirectSymBase.count(it->SectionData))
- IndirectSymBase[it->SectionData] = IndirectIndex;
+ uint64_t Pad = getPaddingSize(it, Layout);
+ for (unsigned int i = 0; i < Pad; ++i)
+ Write8(0);
+ }
- Asm.getOrCreateSymbolData(*it->Symbol);
- }
+ // Write the extra padding.
+ WriteZeros(SectionDataPadding);
- // Then lazy symbol pointers and symbol stubs.
- IndirectIndex = 0;
- for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
- ie = Asm.indirect_symbol_end(); it != ie; ++it, ++IndirectIndex) {
- const MCSectionMachO &Section =
- cast<MCSectionMachO>(it->SectionData->getSection());
-
- if (Section.getType() != MCSectionMachO::S_LAZY_SYMBOL_POINTERS &&
- Section.getType() != MCSectionMachO::S_SYMBOL_STUBS)
- continue;
-
- // Initialize the section indirect symbol base, if necessary.
- if (!IndirectSymBase.count(it->SectionData))
- IndirectSymBase[it->SectionData] = IndirectIndex;
-
- // Set the symbol type to undefined lazy, but only on construction.
- //
- // FIXME: Do not hardcode.
- bool Created;
- MCSymbolData &Entry = Asm.getOrCreateSymbolData(*it->Symbol, &Created);
- if (Created)
- Entry.setFlags(Entry.getFlags() | 0x0001);
+ // Write the relocation entries.
+ for (MCAssembler::const_iterator it = Asm.begin(),
+ ie = Asm.end(); it != ie; ++it) {
+ // Write the section relocation entries, in reverse order to match 'as'
+ // (approximately, the exact algorithm is more complicated than this).
+ std::vector<macho::RelocationEntry> &Relocs = Relocations[it];
+ for (unsigned i = 0, e = Relocs.size(); i != e; ++i) {
+ Write32(Relocs[e - i - 1].Word0);
+ Write32(Relocs[e - i - 1].Word1);
}
}
- /// ComputeSymbolTable - Compute the symbol table data
- ///
- /// \param StringTable [out] - The string table data.
- /// \param StringIndexMap [out] - Map from symbol names to offsets in the
- /// string table.
- void ComputeSymbolTable(MCAssembler &Asm, SmallString<256> &StringTable,
- std::vector<MachSymbolData> &LocalSymbolData,
- std::vector<MachSymbolData> &ExternalSymbolData,
- std::vector<MachSymbolData> &UndefinedSymbolData) {
- // Build section lookup table.
- DenseMap<const MCSection*, uint8_t> SectionIndexMap;
- unsigned Index = 1;
- for (MCAssembler::iterator it = Asm.begin(),
- ie = Asm.end(); it != ie; ++it, ++Index)
- SectionIndexMap[&it->getSection()] = Index;
- assert(Index <= 256 && "Too many sections!");
-
- // Index 0 is always the empty string.
- StringMap<uint64_t> StringIndexMap;
- StringTable += '\x00';
-
- // Build the symbol arrays and the string table, but only for non-local
- // symbols.
- //
- // The particular order that we collect the symbols and create the string
- // table, then sort the symbols is chosen to match 'as'. Even though it
- // doesn't matter for correctness, this is important for letting us diff .o
- // files.
- for (MCAssembler::symbol_iterator it = Asm.symbol_begin(),
- ie = Asm.symbol_end(); it != ie; ++it) {
- const MCSymbol &Symbol = it->getSymbol();
-
- // Ignore non-linker visible symbols.
- if (!Asm.isSymbolLinkerVisible(it->getSymbol()))
- continue;
-
- if (!it->isExternal() && !Symbol.isUndefined())
- continue;
-
- uint64_t &Entry = StringIndexMap[Symbol.getName()];
- if (!Entry) {
- Entry = StringTable.size();
- StringTable += Symbol.getName();
- StringTable += '\x00';
+ // Write the symbol table data, if used.
+ if (NumSymbols) {
+ // Write the indirect symbol entries.
+ for (MCAssembler::const_indirect_symbol_iterator
+ it = Asm.indirect_symbol_begin(),
+ ie = Asm.indirect_symbol_end(); it != ie; ++it) {
+ // Indirect symbols in the non lazy symbol pointer section have some
+ // special handling.
+ const MCSectionMachO &Section =
+ static_cast<const MCSectionMachO&>(it->SectionData->getSection());
+ if (Section.getType() == MCSectionMachO::S_NON_LAZY_SYMBOL_POINTERS) {
+ // If this symbol is defined and internal, mark it as such.
+ if (it->Symbol->isDefined() &&
+ !Asm.getSymbolData(*it->Symbol).isExternal()) {
+ uint32_t Flags = macho::ISF_Local;
+ if (it->Symbol->isAbsolute())
+ Flags |= macho::ISF_Absolute;
+ Write32(Flags);
+ continue;
+ }
}
- MachSymbolData MSD;
- MSD.SymbolData = it;
- MSD.StringIndex = Entry;
-
- if (Symbol.isUndefined()) {
- MSD.SectionIndex = 0;
- UndefinedSymbolData.push_back(MSD);
- } else if (Symbol.isAbsolute()) {
- MSD.SectionIndex = 0;
- ExternalSymbolData.push_back(MSD);
- } else {
- MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
- assert(MSD.SectionIndex && "Invalid section index!");
- ExternalSymbolData.push_back(MSD);
- }
+ Write32(Asm.getSymbolData(*it->Symbol).getIndex());
}
- // Now add the data for local symbols.
- for (MCAssembler::symbol_iterator it = Asm.symbol_begin(),
- ie = Asm.symbol_end(); it != ie; ++it) {
- const MCSymbol &Symbol = it->getSymbol();
-
- // Ignore non-linker visible symbols.
- if (!Asm.isSymbolLinkerVisible(it->getSymbol()))
- continue;
-
- if (it->isExternal() || Symbol.isUndefined())
- continue;
-
- uint64_t &Entry = StringIndexMap[Symbol.getName()];
- if (!Entry) {
- Entry = StringTable.size();
- StringTable += Symbol.getName();
- StringTable += '\x00';
- }
-
- MachSymbolData MSD;
- MSD.SymbolData = it;
- MSD.StringIndex = Entry;
-
- if (Symbol.isAbsolute()) {
- MSD.SectionIndex = 0;
- LocalSymbolData.push_back(MSD);
- } else {
- MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
- assert(MSD.SectionIndex && "Invalid section index!");
- LocalSymbolData.push_back(MSD);
- }
- }
+ // FIXME: Check that offsets match computed ones.
- // External and undefined symbols are required to be in lexicographic order.
- std::sort(ExternalSymbolData.begin(), ExternalSymbolData.end());
- std::sort(UndefinedSymbolData.begin(), UndefinedSymbolData.end());
-
- // Set the symbol indices.
- Index = 0;
+ // Write the symbol table entries.
for (unsigned i = 0, e = LocalSymbolData.size(); i != e; ++i)
- LocalSymbolData[i].SymbolData->setIndex(Index++);
+ WriteNlist(LocalSymbolData[i], Layout);
for (unsigned i = 0, e = ExternalSymbolData.size(); i != e; ++i)
- ExternalSymbolData[i].SymbolData->setIndex(Index++);
+ WriteNlist(ExternalSymbolData[i], Layout);
for (unsigned i = 0, e = UndefinedSymbolData.size(); i != e; ++i)
- UndefinedSymbolData[i].SymbolData->setIndex(Index++);
-
- // The string table is padded to a multiple of 4.
- while (StringTable.size() % 4)
- StringTable += '\x00';
- }
-
- void computeSectionAddresses(const MCAssembler &Asm,
- const MCAsmLayout &Layout) {
- uint64_t StartAddress = 0;
- const SmallVectorImpl<MCSectionData*> &Order = Layout.getSectionOrder();
- for (int i = 0, n = Order.size(); i != n ; ++i) {
- const MCSectionData *SD = Order[i];
- StartAddress = RoundUpToAlignment(StartAddress, SD->getAlignment());
- SectionAddress[SD] = StartAddress;
- StartAddress += Layout.getSectionAddressSize(SD);
- // Explicitly pad the section to match the alignment requirements of the
- // following one. This is for 'gas' compatibility, it shouldn't
- /// strictly be necessary.
- StartAddress += getPaddingSize(SD, Layout);
- }
- }
-
- void ExecutePostLayoutBinding(MCAssembler &Asm, const MCAsmLayout &Layout) {
- computeSectionAddresses(Asm, Layout);
-
- // Create symbol data for any indirect symbols.
- BindIndirectSymbols(Asm);
-
- // Compute symbol table information and bind symbol indices.
- ComputeSymbolTable(Asm, StringTable, LocalSymbolData, ExternalSymbolData,
- UndefinedSymbolData);
- }
-
- virtual bool IsSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm,
- const MCSymbolData &DataA,
- const MCFragment &FB,
- bool InSet,
- bool IsPCRel) const {
- if (InSet)
- return true;
+ WriteNlist(UndefinedSymbolData[i], Layout);
- // The effective address is
- // addr(atom(A)) + offset(A)
- // - addr(atom(B)) - offset(B)
- // and the offsets are not relocatable, so the fixup is fully resolved when
- // addr(atom(A)) - addr(atom(B)) == 0.
- const MCSymbolData *A_Base = 0, *B_Base = 0;
-
- const MCSymbol &SA = DataA.getSymbol().AliasedSymbol();
- const MCSection &SecA = SA.getSection();
- const MCSection &SecB = FB.getParent()->getSection();
-
- if (IsPCRel) {
- // The simple (Darwin, except on x86_64) way of dealing with this was to
- // assume that any reference to a temporary symbol *must* be a temporary
- // symbol in the same atom, unless the sections differ. Therefore, any
- // PCrel relocation to a temporary symbol (in the same section) is fully
- // resolved. This also works in conjunction with absolutized .set, which
- // requires the compiler to use .set to absolutize the differences between
- // symbols which the compiler knows to be assembly time constants, so we
- // don't need to worry about considering symbol differences fully
- // resolved.
-
- if (!Asm.getBackend().hasReliableSymbolDifference()) {
- if (!SA.isTemporary() || !SA.isInSection() || &SecA != &SecB)
- return false;
- return true;
- }
- } else {
- if (!TargetObjectWriter->useAggressiveSymbolFolding())
- return false;
- }
-
- const MCFragment &FA = *Asm.getSymbolData(SA).getFragment();
-
- A_Base = FA.getAtom();
- if (!A_Base)
- return false;
-
- B_Base = FB.getAtom();
- if (!B_Base)
- return false;
-
- // If the atoms are the same, they are guaranteed to have the same address.
- if (A_Base == B_Base)
- return true;
-
- // Otherwise, we can't prove this is fully resolved.
- return false;
+ // Write the string table.
+ OS << StringTable.str();
}
-
- void WriteObject(MCAssembler &Asm, const MCAsmLayout &Layout) {
- unsigned NumSections = Asm.size();
-
- // The section data starts after the header, the segment load command (and
- // section headers) and the symbol table.
- unsigned NumLoadCommands = 1;
- uint64_t LoadCommandsSize = is64Bit() ?
- macho::SegmentLoadCommand64Size + NumSections * macho::Section64Size :
- macho::SegmentLoadCommand32Size + NumSections * macho::Section32Size;
-
- // Add the symbol table load command sizes, if used.
- unsigned NumSymbols = LocalSymbolData.size() + ExternalSymbolData.size() +
- UndefinedSymbolData.size();
- if (NumSymbols) {
- NumLoadCommands += 2;
- LoadCommandsSize += (macho::SymtabLoadCommandSize +
- macho::DysymtabLoadCommandSize);
- }
-
- // Compute the total size of the section data, as well as its file size and
- // vm size.
- uint64_t SectionDataStart = (is64Bit() ? macho::Header64Size :
- macho::Header32Size) + LoadCommandsSize;
- uint64_t SectionDataSize = 0;
- uint64_t SectionDataFileSize = 0;
- uint64_t VMSize = 0;
- for (MCAssembler::const_iterator it = Asm.begin(),
- ie = Asm.end(); it != ie; ++it) {
- const MCSectionData &SD = *it;
- uint64_t Address = getSectionAddress(&SD);
- uint64_t Size = Layout.getSectionAddressSize(&SD);
- uint64_t FileSize = Layout.getSectionFileSize(&SD);
- FileSize += getPaddingSize(&SD, Layout);
-
- VMSize = std::max(VMSize, Address + Size);
-
- if (SD.getSection().isVirtualSection())
- continue;
-
- SectionDataSize = std::max(SectionDataSize, Address + Size);
- SectionDataFileSize = std::max(SectionDataFileSize, Address + FileSize);
- }
-
- // The section data is padded to 4 bytes.
- //
- // FIXME: Is this machine dependent?
- unsigned SectionDataPadding = OffsetToAlignment(SectionDataFileSize, 4);
- SectionDataFileSize += SectionDataPadding;
-
- // Write the prolog, starting with the header and load command...
- WriteHeader(NumLoadCommands, LoadCommandsSize,
- Asm.getSubsectionsViaSymbols());
- WriteSegmentLoadCommand(NumSections, VMSize,
- SectionDataStart, SectionDataSize);
-
- // ... and then the section headers.
- uint64_t RelocTableEnd = SectionDataStart + SectionDataFileSize;
- for (MCAssembler::const_iterator it = Asm.begin(),
- ie = Asm.end(); it != ie; ++it) {
- std::vector<macho::RelocationEntry> &Relocs = Relocations[it];
- unsigned NumRelocs = Relocs.size();
- uint64_t SectionStart = SectionDataStart + getSectionAddress(it);
- WriteSection(Asm, Layout, *it, SectionStart, RelocTableEnd, NumRelocs);
- RelocTableEnd += NumRelocs * macho::RelocationInfoSize;
- }
-
- // Write the symbol table load command, if used.
- if (NumSymbols) {
- unsigned FirstLocalSymbol = 0;
- unsigned NumLocalSymbols = LocalSymbolData.size();
- unsigned FirstExternalSymbol = FirstLocalSymbol + NumLocalSymbols;
- unsigned NumExternalSymbols = ExternalSymbolData.size();
- unsigned FirstUndefinedSymbol = FirstExternalSymbol + NumExternalSymbols;
- unsigned NumUndefinedSymbols = UndefinedSymbolData.size();
- unsigned NumIndirectSymbols = Asm.indirect_symbol_size();
- unsigned NumSymTabSymbols =
- NumLocalSymbols + NumExternalSymbols + NumUndefinedSymbols;
- uint64_t IndirectSymbolSize = NumIndirectSymbols * 4;
- uint64_t IndirectSymbolOffset = 0;
-
- // If used, the indirect symbols are written after the section data.
- if (NumIndirectSymbols)
- IndirectSymbolOffset = RelocTableEnd;
-
- // The symbol table is written after the indirect symbol data.
- uint64_t SymbolTableOffset = RelocTableEnd + IndirectSymbolSize;
-
- // The string table is written after symbol table.
- uint64_t StringTableOffset =
- SymbolTableOffset + NumSymTabSymbols * (is64Bit() ? macho::Nlist64Size :
- macho::Nlist32Size);
- WriteSymtabLoadCommand(SymbolTableOffset, NumSymTabSymbols,
- StringTableOffset, StringTable.size());
-
- WriteDysymtabLoadCommand(FirstLocalSymbol, NumLocalSymbols,
- FirstExternalSymbol, NumExternalSymbols,
- FirstUndefinedSymbol, NumUndefinedSymbols,
- IndirectSymbolOffset, NumIndirectSymbols);
- }
-
- // Write the actual section data.
- for (MCAssembler::const_iterator it = Asm.begin(),
- ie = Asm.end(); it != ie; ++it) {
- Asm.WriteSectionData(it, Layout);
-
- uint64_t Pad = getPaddingSize(it, Layout);
- for (unsigned int i = 0; i < Pad; ++i)
- Write8(0);
- }
-
- // Write the extra padding.
- WriteZeros(SectionDataPadding);
-
- // Write the relocation entries.
- for (MCAssembler::const_iterator it = Asm.begin(),
- ie = Asm.end(); it != ie; ++it) {
- // Write the section relocation entries, in reverse order to match 'as'
- // (approximately, the exact algorithm is more complicated than this).
- std::vector<macho::RelocationEntry> &Relocs = Relocations[it];
- for (unsigned i = 0, e = Relocs.size(); i != e; ++i) {
- Write32(Relocs[e - i - 1].Word0);
- Write32(Relocs[e - i - 1].Word1);
- }
- }
-
- // Write the symbol table data, if used.
- if (NumSymbols) {
- // Write the indirect symbol entries.
- for (MCAssembler::const_indirect_symbol_iterator
- it = Asm.indirect_symbol_begin(),
- ie = Asm.indirect_symbol_end(); it != ie; ++it) {
- // Indirect symbols in the non lazy symbol pointer section have some
- // special handling.
- const MCSectionMachO &Section =
- static_cast<const MCSectionMachO&>(it->SectionData->getSection());
- if (Section.getType() == MCSectionMachO::S_NON_LAZY_SYMBOL_POINTERS) {
- // If this symbol is defined and internal, mark it as such.
- if (it->Symbol->isDefined() &&
- !Asm.getSymbolData(*it->Symbol).isExternal()) {
- uint32_t Flags = macho::ISF_Local;
- if (it->Symbol->isAbsolute())
- Flags |= macho::ISF_Absolute;
- Write32(Flags);
- continue;
- }
- }
-
- Write32(Asm.getSymbolData(*it->Symbol).getIndex());
- }
-
- // FIXME: Check that offsets match computed ones.
-
- // Write the symbol table entries.
- for (unsigned i = 0, e = LocalSymbolData.size(); i != e; ++i)
- WriteNlist(LocalSymbolData[i], Layout);
- for (unsigned i = 0, e = ExternalSymbolData.size(); i != e; ++i)
- WriteNlist(ExternalSymbolData[i], Layout);
- for (unsigned i = 0, e = UndefinedSymbolData.size(); i != e; ++i)
- WriteNlist(UndefinedSymbolData[i], Layout);
-
- // Write the string table.
- OS << StringTable.str();
- }
- }
-};
-
}
MCObjectWriter *llvm::createMachObjectWriter(MCMachObjectTargetWriter *MOTW,
diff --git a/lib/MC/SubtargetFeature.cpp b/lib/MC/SubtargetFeature.cpp
new file mode 100644
index 000000000000..348cd4c9ab1b
--- /dev/null
+++ b/lib/MC/SubtargetFeature.cpp
@@ -0,0 +1,397 @@
+//===- SubtargetFeature.cpp - CPU characteristics Implementation ----------===//
+//
+// The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This file implements the SubtargetFeature interface.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/MC/SubtargetFeature.h"
+#include "llvm/Support/Debug.h"
+#include "llvm/Support/raw_ostream.h"
+#include "llvm/ADT/StringExtras.h"
+#include <algorithm>
+#include <cassert>
+#include <cctype>
+#include <cstdlib>
+using namespace llvm;
+
+//===----------------------------------------------------------------------===//
+// Static Helper Functions
+//===----------------------------------------------------------------------===//
+
+/// hasFlag - Determine if a feature has a flag; '+' or '-'
+///
+static inline bool hasFlag(const StringRef Feature) {
+ assert(!Feature.empty() && "Empty string");
+ // Get first character
+ char Ch = Feature[0];
+ // Check if first character is '+' or '-' flag
+ return Ch == '+' || Ch =='-';
+}
+
+/// StripFlag - Return string stripped of flag.
+///
+static inline std::string StripFlag(const StringRef Feature) {
+ return hasFlag(Feature) ? Feature.substr(1) : Feature;
+}
+
+/// isEnabled - Return true if enable flag; '+'.
+///
+static inline bool isEnabled(const StringRef Feature) {
+ assert(!Feature.empty() && "Empty string");
+ // Get first character
+ char Ch = Feature[0];
+ // Check if first character is '+' for enabled
+ return Ch == '+';
+}
+
+/// PrependFlag - Return a string with a prepended flag; '+' or '-'.
+///
+static inline std::string PrependFlag(const StringRef Feature,
+ bool IsEnabled) {
+ assert(!Feature.empty() && "Empty string");
+ if (hasFlag(Feature))
+ return Feature;
+ std::string Prefix = IsEnabled ? "+" : "-";
+ Prefix += Feature;
+ return Prefix;
+}
+
+/// Split - Splits a string of comma separated items in to a vector of strings.
+///
+static void Split(std::vector<std::string> &V, const StringRef S) {
+ if (S.empty())
+ return;
+
+ // Start at beginning of string.
+ size_t Pos = 0;
+ while (true) {
+ // Find the next comma
+ size_t Comma = S.find(',', Pos);
+ // If no comma found then the rest of the string is used
+ if (Comma == std::string::npos) {
+ // Add string to vector
+ V.push_back(S.substr(Pos));
+ break;
+ }
+ // Otherwise add substring to vector
+ V.push_back(S.substr(Pos, Comma - Pos));
+ // Advance to next item
+ Pos = Comma + 1;
+ }
+}
+
+/// Join a vector of strings to a string with a comma separating each element.
+///
+static std::string Join(const std::vector<std::string> &V) {
+ // Start with empty string.
+ std::string Result;
+ // If the vector is not empty
+ if (!V.empty()) {
+ // Start with the first feature
+ Result = V[0];
+ // For each successive feature
+ for (size_t i = 1; i < V.size(); i++) {
+ // Add a comma
+ Result += ",";
+ // Add the feature
+ Result += V[i];
+ }
+ }
+ // Return the features string
+ return Result;
+}
+
+/// Adding features.
+void SubtargetFeatures::AddFeature(const StringRef String,
+ bool IsEnabled) {
+ // Don't add empty features
+ if (!String.empty()) {
+ // Convert to lowercase, prepend flag and add to vector
+ Features.push_back(PrependFlag(LowercaseString(String), IsEnabled));
+ }
+}
+
+/// Find KV in array using binary search.
+template<typename T> const T *Find(const StringRef S, const T *A, size_t L) {
+ // Make the lower bound element we're looking for
+ T KV;
+ KV.Key = S.data();
+ // Determine the end of the array
+ const T *Hi = A + L;
+ // Binary search the array
+ const T *F = std::lower_bound(A, Hi, KV);
+ // If not found then return NULL
+ if (F == Hi || StringRef(F->Key) != S) return NULL;
+ // Return the found array item
+ return F;
+}
+
+/// getLongestEntryLength - Return the length of the longest entry in the table.
+///
+static size_t getLongestEntryLength(const SubtargetFeatureKV *Table,
+ size_t Size) {
+ size_t MaxLen = 0;
+ for (size_t i = 0; i < Size; i++)
+ MaxLen = std::max(MaxLen, std::strlen(Table[i].Key));
+ return MaxLen;
+}
+
+/// Display help for feature choices.
+///
+static void Help(const SubtargetFeatureKV *CPUTable, size_t CPUTableSize,
+ const SubtargetFeatureKV *FeatTable, size_t FeatTableSize) {
+ // Determine the length of the longest CPU and Feature entries.
+ unsigned MaxCPULen = getLongestEntryLength(CPUTable, CPUTableSize);
+ unsigned MaxFeatLen = getLongestEntryLength(FeatTable, FeatTableSize);
+
+ // Print the CPU table.
+ errs() << "Available CPUs for this target:\n\n";
+ for (size_t i = 0; i != CPUTableSize; i++)
+ errs() << " " << CPUTable[i].Key
+ << std::string(MaxCPULen - std::strlen(CPUTable[i].Key), ' ')
+ << " - " << CPUTable[i].Desc << ".\n";
+ errs() << "\n";
+
+ // Print the Feature table.
+ errs() << "Available features for this target:\n\n";
+ for (size_t i = 0; i != FeatTableSize; i++)
+ errs() << " " << FeatTable[i].Key
+ << std::string(MaxFeatLen - std::strlen(FeatTable[i].Key), ' ')
+ << " - " << FeatTable[i].Desc << ".\n";
+ errs() << "\n";
+
+ errs() << "Use +feature to enable a feature, or -feature to disable it.\n"
+ << "For example, llc -mcpu=mycpu -mattr=+feature1,-feature2\n";
+ std::exit(1);
+}
+
+//===----------------------------------------------------------------------===//
+// SubtargetFeatures Implementation
+//===----------------------------------------------------------------------===//
+
+SubtargetFeatures::SubtargetFeatures(const StringRef Initial) {
+ // Break up string into separate features
+ Split(Features, Initial);
+}
+
+
+std::string SubtargetFeatures::getString() const {
+ return Join(Features);
+}
+
+/// SetImpliedBits - For each feature that is (transitively) implied by this
+/// feature, set it.
+///
+static
+void SetImpliedBits(uint64_t &Bits, const SubtargetFeatureKV *FeatureEntry,
+ const SubtargetFeatureKV *FeatureTable,
+ size_t FeatureTableSize) {
+ for (size_t i = 0; i < FeatureTableSize; ++i) {
+ const SubtargetFeatureKV &FE = FeatureTable[i];
+
+ if (FeatureEntry->Value == FE.Value) continue;
+
+ if (FeatureEntry->Implies & FE.Value) {
+ Bits |= FE.Value;
+ SetImpliedBits(Bits, &FE, FeatureTable, FeatureTableSize);
+ }
+ }
+}
+
+/// ClearImpliedBits - For each feature that (transitively) implies this
+/// feature, clear it.
+///
+static
+void ClearImpliedBits(uint64_t &Bits, const SubtargetFeatureKV *FeatureEntry,
+ const SubtargetFeatureKV *FeatureTable,
+ size_t FeatureTableSize) {
+ for (size_t i = 0; i < FeatureTableSize; ++i) {
+ const SubtargetFeatureKV &FE = FeatureTable[i];
+
+ if (FeatureEntry->Value == FE.Value) continue;
+
+ if (FE.Implies & FeatureEntry->Value) {
+ Bits &= ~FE.Value;
+ ClearImpliedBits(Bits, &FE, FeatureTable, FeatureTableSize);
+ }
+ }
+}
+
+/// ToggleFeature - Toggle a feature and returns the newly updated feature
+/// bits.
+uint64_t
+SubtargetFeatures::ToggleFeature(uint64_t Bits, const StringRef Feature,
+ const SubtargetFeatureKV *FeatureTable,
+ size_t FeatureTableSize) {
+ // Find feature in table.
+ const SubtargetFeatureKV *FeatureEntry =
+ Find(StripFlag(Feature), FeatureTable, FeatureTableSize);
+ // If there is a match
+ if (FeatureEntry) {
+ if ((Bits & FeatureEntry->Value) == FeatureEntry->Value) {
+ Bits &= ~FeatureEntry->Value;
+
+ // For each feature that implies this, clear it.
+ ClearImpliedBits(Bits, FeatureEntry, FeatureTable, FeatureTableSize);
+ } else {
+ Bits |= FeatureEntry->Value;
+
+ // For each feature that this implies, set it.
+ SetImpliedBits(Bits, FeatureEntry, FeatureTable, FeatureTableSize);
+ }
+ } else {
+ errs() << "'" << Feature
+ << "' is not a recognized feature for this target"
+ << " (ignoring feature)\n";
+ }
+
+ return Bits;
+}
+
+
+/// getFeatureBits - Get feature bits a CPU.
+///
+uint64_t SubtargetFeatures::getFeatureBits(const StringRef CPU,
+ const SubtargetFeatureKV *CPUTable,
+ size_t CPUTableSize,
+ const SubtargetFeatureKV *FeatureTable,
+ size_t FeatureTableSize) {
+ if (!FeatureTableSize || !CPUTableSize)
+ return 0;
+
+#ifndef NDEBUG
+ for (size_t i = 1; i < CPUTableSize; i++) {
+ assert(strcmp(CPUTable[i - 1].Key, CPUTable[i].Key) < 0 &&
+ "CPU table is not sorted");
+ }
+ for (size_t i = 1; i < FeatureTableSize; i++) {
+ assert(strcmp(FeatureTable[i - 1].Key, FeatureTable[i].Key) < 0 &&
+ "CPU features table is not sorted");
+ }
+#endif
+ uint64_t Bits = 0; // Resulting bits
+
+ // Check if help is needed
+ if (CPU == "help")
+ Help(CPUTable, CPUTableSize, FeatureTable, FeatureTableSize);
+
+ // Find CPU entry if CPU name is specified.
+ if (!CPU.empty()) {
+ const SubtargetFeatureKV *CPUEntry = Find(CPU, CPUTable, CPUTableSize);
+ // If there is a match
+ if (CPUEntry) {
+ // Set base feature bits
+ Bits = CPUEntry->Value;
+
+ // Set the feature implied by this CPU feature, if any.
+ for (size_t i = 0; i < FeatureTableSize; ++i) {
+ const SubtargetFeatureKV &FE = FeatureTable[i];
+ if (CPUEntry->Value & FE.Value)
+ SetImpliedBits(Bits, &FE, FeatureTable, FeatureTableSize);
+ }
+ } else {
+ errs() << "'" << CPU
+ << "' is not a recognized processor for this target"
+ << " (ignoring processor)\n";
+ }
+ }
+
+ // Iterate through each feature
+ for (size_t i = 0, E = Features.size(); i < E; i++) {
+ const StringRef Feature = Features[i];
+
+ // Check for help
+ if (Feature == "+help")
+ Help(CPUTable, CPUTableSize, FeatureTable, FeatureTableSize);
+
+ // Find feature in table.
+ const SubtargetFeatureKV *FeatureEntry =
+ Find(StripFlag(Feature), FeatureTable, FeatureTableSize);
+ // If there is a match
+ if (FeatureEntry) {
+ // Enable/disable feature in bits
+ if (isEnabled(Feature)) {
+ Bits |= FeatureEntry->Value;
+
+ // For each feature that this implies, set it.
+ SetImpliedBits(Bits, FeatureEntry, FeatureTable, FeatureTableSize);
+ } else {
+ Bits &= ~FeatureEntry->Value;
+
+ // For each feature that implies this, clear it.
+ ClearImpliedBits(Bits, FeatureEntry, FeatureTable, FeatureTableSize);
+ }
+ } else {
+ errs() << "'" << Feature
+ << "' is not a recognized feature for this target"
+ << " (ignoring feature)\n";
+ }
+ }
+
+ return Bits;
+}
+
+/// Get scheduling itinerary of a CPU.
+void *SubtargetFeatures::getItinerary(const StringRef CPU,
+ const SubtargetInfoKV *Table,
+ size_t TableSize) {
+ assert(Table && "missing table");
+#ifndef NDEBUG
+ for (size_t i = 1; i < TableSize; i++) {
+ assert(strcmp(Table[i - 1].Key, Table[i].Key) < 0 && "Table is not sorted");
+ }
+#endif
+
+ // Find entry
+ const SubtargetInfoKV *Entry = Find(CPU, Table, TableSize);
+
+ if (Entry) {
+ return Entry->Value;
+ } else {
+ errs() << "'" << CPU
+ << "' is not a recognized processor for this target"
+ << " (ignoring processor)\n";
+ return NULL;
+ }
+}
+
+/// print - Print feature string.
+///
+void SubtargetFeatures::print(raw_ostream &OS) const {
+ for (size_t i = 0, e = Features.size(); i != e; ++i)
+ OS << Features[i] << " ";
+ OS << "\n";
+}
+
+/// dump - Dump feature info.
+///
+void SubtargetFeatures::dump() const {
+ print(dbgs());
+}
+
+/// getDefaultSubtargetFeatures - Return a string listing the features
+/// associated with the target triple.
+///
+/// FIXME: This is an inelegant way of specifying the features of a
+/// subtarget. It would be better if we could encode this information
+/// into the IR. See <rdar://5972456>.
+///
+void SubtargetFeatures::getDefaultSubtargetFeatures(const Triple& Triple) {
+ if (Triple.getVendor() == Triple::Apple) {
+ if (Triple.getArch() == Triple::ppc) {
+ // powerpc-apple-*
+ AddFeature("altivec");
+ } else if (Triple.getArch() == Triple::ppc64) {
+ // powerpc64-apple-*
+ AddFeature("64bit");
+ AddFeature("altivec");
+ }
+ }
+}