llvm-project/llvm/lib/MC/MCAsmBackend.cpp
Fangrui Song dc3a4c0fcf
MC: Restructure MCFragment as a fixed part and a variable tail
Refactor the fragment representation of `push rax; jmp foo; nop; jmp foo`,
previously encoded as
`MCDataFragment(nop); MCRelaxableFragment(jmp foo); MCDataFragment(nop); MCRelaxableFragment(jmp foo)`,

to

```
MCFragment(fixed: push rax, variable: jmp foo)
MCFragment(fixed: nop, variable: jmp foo)
```

Changes:

* Eliminate MCEncodedFragment, moving content and fixup storage to MCFragment.
* The new MCFragment contains a fixed-size content (similar to previous
  MCDataFragment) and an optional variable-size tail.
* The variable-size tail supports FT_Relaxable, FT_LEB, FT_Dwarf, and
  FT_DwarfFrame, with plans to extend to other fragment types.
  dyn_cast/isa should be avoided for the converted fragment subclasses.
* In `setVarFixups`, source fixup offsets are relative to the variable part's start.
  Stored fixup (in `FixupStorage`) offsets are relative to the fixed part's start.
  A lot of code does `getFragmentOffset(Frag) + Fixup.getOffset()`,
  expecting the fixup offset to be relative to the fixed part's start.
* HexagonAsmBackend::fixupNeedsRelaxationAdvanced needs to know the
  associated instruction for a fixup. We have to add a `const MCFragment &` parameter.
* In MCObjectStreamer, extend `absoluteSymbolDiff` to apply to
  FT_Relaxable as otherwise there would be many more FT_DwarfFrame
  fragments in -g compilations.

https://llvm-compile-time-tracker.com/compare.php?from=28e1473e8e523150914e8c7ea50b44fb0d2a8d65&to=778d68ad1d48e7f111ea853dd249912c601bee89&stat=instructions:u

```
stage2-O0-g instructins:u geomeon (-0.07%)
stage1-ReleaseLTO-g (link only) max-rss geomean (-0.39%)
```

```
% /t/clang-old -g -c sqlite3.i -w -mllvm -debug-only=mc-dump &| awk '/^[0-9]+/{s[$2]++;tot++} END{print "Total",tot; n=asorti(s, si); for(i=1;i<=n;i++) print si[i],s[si[i]]}'
Total 59675
Align 2215
Data 29700
Dwarf 12044
DwarfCallFrame 4216
Fill 92
LEB 12
Relaxable 11396
% /t/clang-new -g -c sqlite3.i -w -mllvm -debug-only=mc-dump &| awk '/^[0-9]+/{s[$2]++;tot++} END{print "Total",tot; n=asorti(s, si); for(i=1;i<=n;i++) print si[i],s[si[i]]}'
Total 32287
Align 2215
Data 2312
Dwarf 12044
DwarfCallFrame 4216
Fill 92
LEB 12
Relaxable 11396
```

Pull Request: https://github.com/llvm/llvm-project/pull/148544
2025-07-15 21:56:55 -07:00

146 lines
5.1 KiB
C++

//===- MCAsmBackend.cpp - Target MC Assembly Backend ----------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
#include "llvm/MC/MCAsmBackend.h"
#include "llvm/MC/MCAssembler.h"
#include "llvm/MC/MCDXContainerWriter.h"
#include "llvm/MC/MCELFObjectWriter.h"
#include "llvm/MC/MCGOFFObjectWriter.h"
#include "llvm/MC/MCMachObjectWriter.h"
#include "llvm/MC/MCObjectWriter.h"
#include "llvm/MC/MCSPIRVObjectWriter.h"
#include "llvm/MC/MCWasmObjectWriter.h"
#include "llvm/MC/MCWinCOFFObjectWriter.h"
#include "llvm/MC/MCXCOFFObjectWriter.h"
#include <cassert>
#include <cstddef>
#include <cstdint>
using namespace llvm;
MCAsmBackend::~MCAsmBackend() = default;
MCContext &MCAsmBackend::getContext() const { return Asm->getContext(); }
std::unique_ptr<MCObjectWriter>
MCAsmBackend::createObjectWriter(raw_pwrite_stream &OS) const {
auto TW = createObjectTargetWriter();
bool IsLE = Endian == llvm::endianness::little;
switch (TW->getFormat()) {
case Triple::MachO:
return std::make_unique<MachObjectWriter>(
cast<MCMachObjectTargetWriter>(std::move(TW)), OS, IsLE);
case Triple::COFF:
return createWinCOFFObjectWriter(
cast<MCWinCOFFObjectTargetWriter>(std::move(TW)), OS);
case Triple::ELF:
return std::make_unique<ELFObjectWriter>(
cast<MCELFObjectTargetWriter>(std::move(TW)), OS, IsLE);
case Triple::SPIRV:
return createSPIRVObjectWriter(
cast<MCSPIRVObjectTargetWriter>(std::move(TW)), OS);
case Triple::Wasm:
return createWasmObjectWriter(cast<MCWasmObjectTargetWriter>(std::move(TW)),
OS);
case Triple::GOFF:
return createGOFFObjectWriter(cast<MCGOFFObjectTargetWriter>(std::move(TW)),
OS);
case Triple::XCOFF:
return createXCOFFObjectWriter(
cast<MCXCOFFObjectTargetWriter>(std::move(TW)), OS);
case Triple::DXContainer:
return std::make_unique<DXContainerObjectWriter>(
cast<MCDXContainerTargetWriter>(std::move(TW)), OS);
default:
llvm_unreachable("unexpected object format");
}
}
std::unique_ptr<MCObjectWriter>
MCAsmBackend::createDwoObjectWriter(raw_pwrite_stream &OS,
raw_pwrite_stream &DwoOS) const {
auto TW = createObjectTargetWriter();
switch (TW->getFormat()) {
case Triple::COFF:
return createWinCOFFDwoObjectWriter(
cast<MCWinCOFFObjectTargetWriter>(std::move(TW)), OS, DwoOS);
case Triple::ELF:
return std::make_unique<ELFObjectWriter>(
cast<MCELFObjectTargetWriter>(std::move(TW)), OS, DwoOS,
Endian == llvm::endianness::little);
case Triple::Wasm:
return createWasmDwoObjectWriter(
cast<MCWasmObjectTargetWriter>(std::move(TW)), OS, DwoOS);
default:
report_fatal_error("dwo only supported with COFF, ELF, and Wasm");
}
}
std::optional<MCFixupKind> MCAsmBackend::getFixupKind(StringRef Name) const {
return std::nullopt;
}
MCFixupKindInfo MCAsmBackend::getFixupKindInfo(MCFixupKind Kind) const {
// clang-format off
static const MCFixupKindInfo Builtins[] = {
{"FK_NONE", 0, 0, 0},
{"FK_Data_1", 0, 8, 0},
{"FK_Data_2", 0, 16, 0},
{"FK_Data_4", 0, 32, 0},
{"FK_Data_8", 0, 64, 0},
{"FK_Data_leb128", 0, 0, 0},
{"FK_SecRel_1", 0, 8, 0},
{"FK_SecRel_2", 0, 16, 0},
{"FK_SecRel_4", 0, 32, 0},
{"FK_SecRel_8", 0, 64, 0},
};
// clang-format on
assert(size_t(Kind - FK_NONE) < std::size(Builtins) && "Unknown fixup kind");
return Builtins[Kind - FK_NONE];
}
bool MCAsmBackend::fixupNeedsRelaxationAdvanced(const MCFragment &,
const MCFixup &Fixup,
const MCValue &, uint64_t Value,
bool Resolved) const {
if (!Resolved)
return true;
return fixupNeedsRelaxation(Fixup, Value);
}
void MCAsmBackend::maybeAddReloc(const MCFragment &F, const MCFixup &Fixup,
const MCValue &Target, uint64_t &Value,
bool IsResolved) {
if (!IsResolved)
Asm->getWriter().recordRelocation(F, Fixup, Target, Value);
}
bool MCAsmBackend::isDarwinCanonicalPersonality(const MCSymbol *Sym) const {
// Consider a NULL personality (ie., no personality encoding) to be canonical
// because it's always at 0.
if (!Sym)
return true;
if (!Sym->isMachO())
llvm_unreachable("Expected MachO symbols only");
StringRef name = Sym->getName();
// XXX: We intentionally leave out "___gcc_personality_v0" because, despite
// being system-defined like these two, it is not very commonly-used.
// Reserving an empty slot for it seems silly.
return name == "___gxx_personality_v0" || name == "___objc_personality_v0";
}
const MCSubtargetInfo *MCAsmBackend::getSubtargetInfo(const MCFragment &F) {
const MCSubtargetInfo *STI = nullptr;
STI = F.getSubtargetInfo();
assert(!F.hasInstructions() || STI != nullptr);
return STI;
}