Emit root module block and blockinfo subblock

* It looks like LLVM (at DXC's version anyway) hardcodes a series of
  abbreviations, so we do the same
This commit is contained in:
baldurk
2021-08-31 18:01:34 +01:00
parent c77c9e8814
commit 9e3385d18d
10 changed files with 596 additions and 237 deletions
+17 -14
View File
@@ -29,12 +29,15 @@
#include "common/common.h"
#include "common/formatting.h"
#include "os/os_specific.h"
#include "llvm_common.h"
#include "llvm_decoder.h"
#define IS_KNOWN(val, KnownID) (decltype(KnownID)(val) == KnownID)
namespace DXIL
{
using namespace LLVMBC;
void ParseConstant(const LLVMBC::BlockOrRecord &constant, const Type *&curType,
std::function<const Type *(uint64_t)> getType,
std::function<const Type *(const Type *, Type::PointerAddrSpace)> getPtrType,
@@ -302,7 +305,7 @@ Program::Program(const byte *bytes, size_t length)
LLVMBC::BlockOrRecord root = reader.ReadToplevelBlock();
// the top-level block should be MODULE_BLOCK
RDCASSERT(KnownBlocks(root.id) == KnownBlocks::MODULE_BLOCK);
RDCASSERT(KnownBlock(root.id) == KnownBlock::MODULE_BLOCK);
// we should have consumed all bits, only one top-level block
RDCASSERT(reader.AtEndOfStream());
@@ -497,11 +500,11 @@ Program::Program(const byte *bytes, size_t length)
}
else if(rootchild.IsBlock())
{
if(IS_KNOWN(rootchild.id, KnownBlocks::BLOCKINFO))
if(IS_KNOWN(rootchild.id, KnownBlock::BLOCKINFO))
{
// do nothing, this is internal parse data
}
else if(IS_KNOWN(rootchild.id, KnownBlocks::PARAMATTR_GROUP_BLOCK))
else if(IS_KNOWN(rootchild.id, KnownBlock::PARAMATTR_GROUP_BLOCK))
{
for(const LLVMBC::BlockOrRecord &attrgroup : rootchild.children)
{
@@ -562,7 +565,7 @@ Program::Program(const byte *bytes, size_t length)
m_AttributeGroups[id] = group;
}
}
else if(IS_KNOWN(rootchild.id, KnownBlocks::PARAMATTR_BLOCK))
else if(IS_KNOWN(rootchild.id, KnownBlock::PARAMATTR_BLOCK))
{
for(const LLVMBC::BlockOrRecord &paramattr : rootchild.children)
{
@@ -602,7 +605,7 @@ Program::Program(const byte *bytes, size_t length)
m_Attributes.push_back(attrs);
}
}
else if(IS_KNOWN(rootchild.id, KnownBlocks::TYPE_BLOCK))
else if(IS_KNOWN(rootchild.id, KnownBlock::TYPE_BLOCK))
{
rdcstr structname;
@@ -763,7 +766,7 @@ Program::Program(const byte *bytes, size_t length)
}
}
}
else if(IS_KNOWN(rootchild.id, KnownBlocks::CONSTANTS_BLOCK))
else if(IS_KNOWN(rootchild.id, KnownBlock::CONSTANTS_BLOCK))
{
const Type *t = NULL;
m_Constants.reserve(m_Constants.size() + rootchild.children.size());
@@ -820,7 +823,7 @@ Program::Program(const byte *bytes, size_t length)
}
}
}
else if(IS_KNOWN(rootchild.id, KnownBlocks::VALUE_SYMTAB_BLOCK))
else if(IS_KNOWN(rootchild.id, KnownBlock::VALUE_SYMTAB_BLOCK))
{
for(const LLVMBC::BlockOrRecord &symtab : rootchild.children)
{
@@ -868,7 +871,7 @@ Program::Program(const byte *bytes, size_t length)
}
}
}
else if(IS_KNOWN(rootchild.id, KnownBlocks::METADATA_BLOCK))
else if(IS_KNOWN(rootchild.id, KnownBlock::METADATA_BLOCK))
{
m_Metadata.reserve(rootchild.children.size());
for(size_t i = 0; i < rootchild.children.size(); i++)
@@ -945,7 +948,7 @@ Program::Program(const byte *bytes, size_t length)
}
}
}
else if(IS_KNOWN(rootchild.id, KnownBlocks::FUNCTION_BLOCK))
else if(IS_KNOWN(rootchild.id, KnownBlock::FUNCTION_BLOCK))
{
Function &f = m_Functions[functionDecls[0]];
functionDecls.erase(0);
@@ -980,7 +983,7 @@ Program::Program(const byte *bytes, size_t length)
{
if(funcChild.IsBlock())
{
if(IS_KNOWN(funcChild.id, KnownBlocks::CONSTANTS_BLOCK))
if(IS_KNOWN(funcChild.id, KnownBlock::CONSTANTS_BLOCK))
{
f.constants.reserve(funcChild.children.size());
@@ -1030,7 +1033,7 @@ Program::Program(const byte *bytes, size_t length)
instrSymbolStart = m_Symbols.size();
}
else if(IS_KNOWN(funcChild.id, KnownBlocks::METADATA_BLOCK))
else if(IS_KNOWN(funcChild.id, KnownBlock::METADATA_BLOCK))
{
f.metadata.resize(funcChild.children.size());
@@ -1080,7 +1083,7 @@ Program::Program(const byte *bytes, size_t length)
m++;
}
}
else if(IS_KNOWN(funcChild.id, KnownBlocks::VALUE_SYMTAB_BLOCK))
else if(IS_KNOWN(funcChild.id, KnownBlock::VALUE_SYMTAB_BLOCK))
{
for(const LLVMBC::BlockOrRecord &symtab : funcChild.children)
{
@@ -1141,7 +1144,7 @@ Program::Program(const byte *bytes, size_t length)
}
}
}
else if(IS_KNOWN(funcChild.id, KnownBlocks::METADATA_ATTACHMENT))
else if(IS_KNOWN(funcChild.id, KnownBlock::METADATA_ATTACHMENT))
{
for(const LLVMBC::BlockOrRecord &meta : funcChild.children)
{
@@ -1173,7 +1176,7 @@ Program::Program(const byte *bytes, size_t length)
f.instructions[(size_t)meta.ops[0]].attachedMeta.swap(attach);
}
}
else if(IS_KNOWN(funcChild.id, KnownBlocks::USELIST_BLOCK))
else if(IS_KNOWN(funcChild.id, KnownBlock::USELIST_BLOCK))
{
RDCDEBUG("Ignoring uselist block");
}
@@ -31,12 +31,6 @@
#include "driver/dx/official/d3dcommon.h"
#include "driver/shaders/dxbc/dxbc_common.h"
// undef some annoying defines that might come from OS headers
#undef VOID
#undef FLOAT
#undef LABEL
#undef OPAQUE
namespace LLVMBC
{
struct BlockOrRecord;
@@ -55,179 +49,6 @@ struct ProgramHeader
uint32_t BitcodeSize; // Size of LLVM bitcode.
};
enum class KnownBlocks : uint32_t
{
BLOCKINFO = 0,
// 1-7 reserved,
MODULE_BLOCK = 8,
PARAMATTR_BLOCK = 9,
PARAMATTR_GROUP_BLOCK = 10,
CONSTANTS_BLOCK = 11,
FUNCTION_BLOCK = 12,
TYPE_SYMTAB_BLOCK = 13,
VALUE_SYMTAB_BLOCK = 14,
METADATA_BLOCK = 15,
METADATA_ATTACHMENT = 16,
TYPE_BLOCK = 17,
USELIST_BLOCK = 18,
};
enum class ModuleRecord : uint32_t
{
VERSION = 1,
TRIPLE = 2,
DATALAYOUT = 3,
SECTIONNAME = 5,
GLOBALVAR = 7,
FUNCTION = 8,
ALIAS = 14,
};
enum class ConstantsRecord : uint32_t
{
SETTYPE = 1,
CONST_NULL = 2,
UNDEF = 3,
INTEGER = 4,
FLOAT = 6,
AGGREGATE = 7,
STRING = 8,
CSTRING = 9,
EVAL_CAST = 11,
EVAL_GEP = 20,
DATA = 22,
};
enum class FunctionRecord : uint32_t
{
DECLAREBLOCKS = 1,
INST_BINOP = 2,
INST_CAST = 3,
INST_GEP_OLD = 4,
INST_SELECT = 5,
INST_EXTRACTELT = 6,
INST_INSERTELT = 7,
INST_SHUFFLEVEC = 8,
INST_CMP = 9,
INST_RET = 10,
INST_BR = 11,
INST_SWITCH = 12,
INST_INVOKE = 13,
INST_UNREACHABLE = 15,
INST_PHI = 16,
INST_ALLOCA = 19,
INST_LOAD = 20,
INST_VAARG = 23,
INST_STORE_OLD = 24,
INST_EXTRACTVAL = 26,
INST_INSERTVAL = 27,
INST_CMP2 = 28,
INST_VSELECT = 29,
INST_INBOUNDS_GEP_OLD = 30,
INST_INDIRECTBR = 31,
DEBUG_LOC_AGAIN = 33,
INST_CALL = 34,
DEBUG_LOC = 35,
INST_FENCE = 36,
INST_CMPXCHG_OLD = 37,
INST_ATOMICRMW = 38,
INST_RESUME = 39,
INST_LANDINGPAD_OLD = 40,
INST_LOADATOMIC = 41,
INST_STOREATOMIC_OLD = 42,
INST_GEP = 43,
INST_STORE = 44,
INST_STOREATOMIC = 45,
INST_CMPXCHG = 46,
INST_LANDINGPAD = 47,
};
enum class ParamAttrRecord : uint32_t
{
ENTRY = 2,
};
enum class ParamAttrGroupRecord : uint32_t
{
ENTRY = 3,
};
enum class ValueSymtabRecord : uint32_t
{
ENTRY = 1,
BBENTRY = 2,
FNENTRY = 3,
COMBINED_ENTRY = 5,
};
enum class MetaDataRecord : uint32_t
{
STRING_OLD = 1,
VALUE = 2,
NODE = 3,
NAME = 4,
DISTINCT_NODE = 5,
KIND = 6,
LOCATION = 7,
OLD_NODE = 8,
OLD_FN_NODE = 9,
NAMED_NODE = 10,
ATTACHMENT = 11,
GENERIC_DEBUG = 12,
SUBRANGE = 13,
ENUMERATOR = 14,
BASIC_TYPE = 15,
FILE = 16,
DERIVED_TYPE = 17,
COMPOSITE_TYPE = 18,
SUBROUTINE_TYPE = 19,
COMPILE_UNIT = 20,
SUBPROGRAM = 21,
LEXICAL_BLOCK = 22,
LEXICAL_BLOCK_FILE = 23,
NAMESPACE = 24,
TEMPLATE_TYPE = 25,
TEMPLATE_VALUE = 26,
GLOBAL_VAR = 27,
LOCAL_VAR = 28,
EXPRESSION = 29,
OBJC_PROPERTY = 30,
IMPORTED_ENTITY = 31,
MODULE = 32,
MACRO = 33,
MACRO_FILE = 34,
STRINGS = 35,
GLOBAL_DECL_ATTACHMENT = 36,
GLOBAL_VAR_EXPR = 37,
INDEX_OFFSET = 38,
INDEX = 39,
LABEL = 40,
COMMON_BLOCK = 44,
};
enum class TypeRecord : uint32_t
{
NUMENTRY = 1,
VOID = 2,
FLOAT = 3,
DOUBLE = 4,
LABEL = 5,
OPAQUE = 6,
INTEGER = 7,
POINTER = 8,
FUNCTION_OLD = 9,
HALF = 10,
ARRAY = 11,
VECTOR = 12,
METADATA = 16,
STRUCT_ANON = 18,
STRUCT_NAME = 19,
STRUCT_NAMED = 20,
FUNCTION = 21,
};
struct Type
{
enum TypeKind
@@ -40,12 +40,20 @@ DXIL::ProgramEditor::~ProgramEditor()
DXBC::DXBCContainer::ReplaceDXILBytecode(m_OutBlob, EncodeProgram());
}
bytebuf DXIL::ProgramEditor::EncodeProgram()
bytebuf DXIL::ProgramEditor::EncodeProgram() const
{
bytebuf ret;
LLVMBC::BitcodeWriter writer(ret);
writer.BeginBlock(LLVMBC::KnownBlock::MODULE_BLOCK);
writer.Unabbrev((uint32_t)LLVMBC::ModuleRecord::VERSION, 1U);
writer.ModuleBlockInfo((uint32_t)m_Types.size());
writer.EndBlock();
ProgramHeader header;
header.ProgramVersion = ((m_Major & 0xf) << 4) | (m_Minor & 0xf);
@@ -42,7 +42,7 @@ public:
private:
bytebuf &m_OutBlob;
bytebuf EncodeProgram();
bytebuf EncodeProgram() const;
};
}; // namespace DXIL
@@ -24,26 +24,27 @@
#include "dxil_debuginfo.h"
#include "common/formatting.h"
#include "llvm_common.h"
#include "llvm_decoder.h"
namespace DXIL
{
bool Program::ParseDebugMetaRecord(const LLVMBC::BlockOrRecord &metaRecord, Metadata &meta)
{
MetaDataRecord id = (MetaDataRecord)metaRecord.id;
LLVMBC::MetaDataRecord id = (LLVMBC::MetaDataRecord)metaRecord.id;
auto getNonNullMeta = [this](uint64_t id) { return &m_Metadata[size_t(id)]; };
auto getMeta = [this](uint64_t id) { return id ? &m_Metadata[size_t(id - 1)] : NULL; };
auto getMetaString = [this](uint64_t id) { return id ? &m_Metadata[size_t(id - 1)].str : NULL; };
if(id == MetaDataRecord::FILE)
if(id == LLVMBC::MetaDataRecord::FILE)
{
meta.isDistinct = (metaRecord.ops[0] & 0x1);
meta.dwarf = new DIFile(getMeta(metaRecord.ops[1]), getMeta(metaRecord.ops[2]));
meta.children = {getMeta(metaRecord.ops[1]), getMeta(metaRecord.ops[2])};
}
else if(id == MetaDataRecord::COMPILE_UNIT)
else if(id == LLVMBC::MetaDataRecord::COMPILE_UNIT)
{
// should be at least 14 parameters
RDCASSERT(metaRecord.ops.size() >= 14);
@@ -62,7 +63,7 @@ bool Program::ParseDebugMetaRecord(const LLVMBC::BlockOrRecord &metaRecord, Meta
getMeta(metaRecord.ops[10]), getMeta(metaRecord.ops[11]),
getMeta(metaRecord.ops[12]), getMeta(metaRecord.ops[13])};
}
else if(id == MetaDataRecord::BASIC_TYPE)
else if(id == LLVMBC::MetaDataRecord::BASIC_TYPE)
{
meta.isDistinct = (metaRecord.ops[0] & 0x1);
@@ -70,7 +71,7 @@ bool Program::ParseDebugMetaRecord(const LLVMBC::BlockOrRecord &metaRecord, Meta
new DIBasicType(DW_TAG(metaRecord.ops[1]), getMetaString(metaRecord.ops[2]),
metaRecord.ops[3], metaRecord.ops[4], DW_ENCODING(metaRecord.ops[5]));
}
else if(id == MetaDataRecord::DERIVED_TYPE)
else if(id == LLVMBC::MetaDataRecord::DERIVED_TYPE)
{
meta.isDistinct = (metaRecord.ops[0] & 0x1);
@@ -83,7 +84,7 @@ bool Program::ParseDebugMetaRecord(const LLVMBC::BlockOrRecord &metaRecord, Meta
meta.children = {getMeta(metaRecord.ops[3]), getMeta(metaRecord.ops[5]),
getMeta(metaRecord.ops[6]), getMeta(metaRecord.ops[11])};
}
else if(id == MetaDataRecord::COMPOSITE_TYPE)
else if(id == LLVMBC::MetaDataRecord::COMPOSITE_TYPE)
{
meta.isDistinct = (metaRecord.ops[0] & 0x1);
@@ -98,7 +99,7 @@ bool Program::ParseDebugMetaRecord(const LLVMBC::BlockOrRecord &metaRecord, Meta
getMeta(metaRecord.ops[6]), getMeta(metaRecord.ops[11]),
getMeta(metaRecord.ops[14])};
}
else if(id == MetaDataRecord::TEMPLATE_TYPE)
else if(id == LLVMBC::MetaDataRecord::TEMPLATE_TYPE)
{
meta.isDistinct = (metaRecord.ops[0] & 0x1);
@@ -107,7 +108,7 @@ bool Program::ParseDebugMetaRecord(const LLVMBC::BlockOrRecord &metaRecord, Meta
meta.children = {getMeta(metaRecord.ops[2])};
}
else if(id == MetaDataRecord::TEMPLATE_VALUE)
else if(id == LLVMBC::MetaDataRecord::TEMPLATE_VALUE)
{
meta.isDistinct = (metaRecord.ops[0] & 0x1);
@@ -117,7 +118,7 @@ bool Program::ParseDebugMetaRecord(const LLVMBC::BlockOrRecord &metaRecord, Meta
meta.children = {getMeta(metaRecord.ops[3]), getMeta(metaRecord.ops[4])};
}
else if(id == MetaDataRecord::SUBPROGRAM)
else if(id == LLVMBC::MetaDataRecord::SUBPROGRAM)
{
meta.isDistinct = (metaRecord.ops[0] & 0x1);
@@ -134,7 +135,7 @@ bool Program::ParseDebugMetaRecord(const LLVMBC::BlockOrRecord &metaRecord, Meta
getMeta(metaRecord.ops[14]), getMeta(metaRecord.ops[15]),
getMeta(metaRecord.ops[16]), getMeta(metaRecord.ops[17])};
}
else if(id == MetaDataRecord::SUBROUTINE_TYPE)
else if(id == LLVMBC::MetaDataRecord::SUBROUTINE_TYPE)
{
meta.isDistinct = (metaRecord.ops[0] & 0x1);
@@ -142,7 +143,7 @@ bool Program::ParseDebugMetaRecord(const LLVMBC::BlockOrRecord &metaRecord, Meta
meta.children = {getMeta(metaRecord.ops[2])};
}
else if(id == MetaDataRecord::GLOBAL_VAR)
else if(id == LLVMBC::MetaDataRecord::GLOBAL_VAR)
{
meta.isDistinct = (metaRecord.ops[0] & 0x1);
@@ -165,7 +166,7 @@ bool Program::ParseDebugMetaRecord(const LLVMBC::BlockOrRecord &metaRecord, Meta
RDCERR("Unsupported version of global variable metadata");
}
}
else if(id == MetaDataRecord::LOCATION)
else if(id == LLVMBC::MetaDataRecord::LOCATION)
{
meta.isDistinct = (metaRecord.ops[0] & 0x1);
@@ -177,7 +178,7 @@ bool Program::ParseDebugMetaRecord(const LLVMBC::BlockOrRecord &metaRecord, Meta
meta.children = {getNonNullMeta(metaRecord.ops[3]), getMeta(metaRecord.ops[4])};
}
else if(id == MetaDataRecord::LOCAL_VAR)
else if(id == LLVMBC::MetaDataRecord::LOCAL_VAR)
{
meta.isDistinct = (metaRecord.ops[0] & 0x1);
@@ -189,7 +190,7 @@ bool Program::ParseDebugMetaRecord(const LLVMBC::BlockOrRecord &metaRecord, Meta
meta.children = {getMeta(metaRecord.ops[2]), getMeta(metaRecord.ops[4]),
getMeta(metaRecord.ops[6])};
}
else if(id == MetaDataRecord::LEXICAL_BLOCK)
else if(id == LLVMBC::MetaDataRecord::LEXICAL_BLOCK)
{
meta.isDistinct = (metaRecord.ops[0] & 0x1);
@@ -198,13 +199,13 @@ bool Program::ParseDebugMetaRecord(const LLVMBC::BlockOrRecord &metaRecord, Meta
meta.children = {getMeta(metaRecord.ops[1]), getMeta(metaRecord.ops[2])};
}
else if(id == MetaDataRecord::SUBRANGE)
else if(id == LLVMBC::MetaDataRecord::SUBRANGE)
{
meta.isDistinct = (metaRecord.ops[0] & 0x1);
meta.dwarf = new DISubrange(metaRecord.ops[1], LLVMBC::BitReader::svbr(metaRecord.ops[2]));
}
else if(id == MetaDataRecord::EXPRESSION)
else if(id == LLVMBC::MetaDataRecord::EXPRESSION)
{
DIExpression *expr = new DIExpression;
+197 -1
View File
@@ -24,10 +24,22 @@
#pragma once
#include "api/replay/rdcarray.h"
#include "common/common.h"
// undef some annoying defines that might come from OS headers
#undef VOID
#undef FLOAT
#undef LABEL
#undef OPAQUE
namespace LLVMBC
{
static const uint32_t BitcodeMagic = MAKE_FOURCC('B', 'C', 0xC0, 0xDE);
enum class AbbrevEncoding : uint8_t
{
Unknown = 0,
Fixed = 1,
VBR = 2,
Array = 3,
@@ -54,5 +66,189 @@ enum class BlockInfoRecord
SETRECORDNAME = 3,
};
static const uint32_t BitcodeMagic = MAKE_FOURCC('B', 'C', 0xC0, 0xDE);
enum class KnownBlock : uint32_t
{
BLOCKINFO = 0,
// 1-7 reserved,
MODULE_BLOCK = 8,
PARAMATTR_BLOCK = 9,
PARAMATTR_GROUP_BLOCK = 10,
CONSTANTS_BLOCK = 11,
FUNCTION_BLOCK = 12,
// TYPE_SYMTAB_BLOCK deprecated?
VALUE_SYMTAB_BLOCK = 14,
METADATA_BLOCK = 15,
METADATA_ATTACHMENT = 16,
TYPE_BLOCK = 17,
USELIST_BLOCK = 18,
Count,
};
enum class ModuleRecord : uint32_t
{
VERSION = 1,
TRIPLE = 2,
DATALAYOUT = 3,
SECTIONNAME = 5,
GLOBALVAR = 7,
FUNCTION = 8,
ALIAS = 14,
};
enum class ConstantsRecord : uint32_t
{
SETTYPE = 1,
CONST_NULL = 2,
UNDEF = 3,
INTEGER = 4,
FLOAT = 6,
AGGREGATE = 7,
STRING = 8,
CSTRING = 9,
EVAL_CAST = 11,
EVAL_GEP = 20,
DATA = 22,
};
enum class FunctionRecord : uint32_t
{
DECLAREBLOCKS = 1,
INST_BINOP = 2,
INST_CAST = 3,
INST_GEP_OLD = 4,
INST_SELECT = 5,
INST_EXTRACTELT = 6,
INST_INSERTELT = 7,
INST_SHUFFLEVEC = 8,
INST_CMP = 9,
INST_RET = 10,
INST_BR = 11,
INST_SWITCH = 12,
INST_INVOKE = 13,
INST_UNREACHABLE = 15,
INST_PHI = 16,
INST_ALLOCA = 19,
INST_LOAD = 20,
INST_VAARG = 23,
INST_STORE_OLD = 24,
INST_EXTRACTVAL = 26,
INST_INSERTVAL = 27,
INST_CMP2 = 28,
INST_VSELECT = 29,
INST_INBOUNDS_GEP_OLD = 30,
INST_INDIRECTBR = 31,
DEBUG_LOC_AGAIN = 33,
INST_CALL = 34,
DEBUG_LOC = 35,
INST_FENCE = 36,
INST_CMPXCHG_OLD = 37,
INST_ATOMICRMW = 38,
INST_RESUME = 39,
INST_LANDINGPAD_OLD = 40,
INST_LOADATOMIC = 41,
INST_STOREATOMIC_OLD = 42,
INST_GEP = 43,
INST_STORE = 44,
INST_STOREATOMIC = 45,
INST_CMPXCHG = 46,
INST_LANDINGPAD = 47,
};
enum class ParamAttrRecord : uint32_t
{
ENTRY = 2,
};
enum class ParamAttrGroupRecord : uint32_t
{
ENTRY = 3,
};
enum class ValueSymtabRecord : uint32_t
{
ENTRY = 1,
BBENTRY = 2,
FNENTRY = 3,
COMBINED_ENTRY = 5,
};
enum class MetaDataRecord : uint32_t
{
STRING_OLD = 1,
VALUE = 2,
NODE = 3,
NAME = 4,
DISTINCT_NODE = 5,
KIND = 6,
LOCATION = 7,
OLD_NODE = 8,
OLD_FN_NODE = 9,
NAMED_NODE = 10,
ATTACHMENT = 11,
GENERIC_DEBUG = 12,
SUBRANGE = 13,
ENUMERATOR = 14,
BASIC_TYPE = 15,
FILE = 16,
DERIVED_TYPE = 17,
COMPOSITE_TYPE = 18,
SUBROUTINE_TYPE = 19,
COMPILE_UNIT = 20,
SUBPROGRAM = 21,
LEXICAL_BLOCK = 22,
LEXICAL_BLOCK_FILE = 23,
NAMESPACE = 24,
TEMPLATE_TYPE = 25,
TEMPLATE_VALUE = 26,
GLOBAL_VAR = 27,
LOCAL_VAR = 28,
EXPRESSION = 29,
OBJC_PROPERTY = 30,
IMPORTED_ENTITY = 31,
MODULE = 32,
MACRO = 33,
MACRO_FILE = 34,
STRINGS = 35,
GLOBAL_DECL_ATTACHMENT = 36,
GLOBAL_VAR_EXPR = 37,
INDEX_OFFSET = 38,
INDEX = 39,
LABEL = 40,
COMMON_BLOCK = 44,
};
enum class TypeRecord : uint32_t
{
NUMENTRY = 1,
VOID = 2,
FLOAT = 3,
DOUBLE = 4,
LABEL = 5,
OPAQUE = 6,
INTEGER = 7,
POINTER = 8,
FUNCTION_OLD = 9,
HALF = 10,
ARRAY = 11,
VECTOR = 12,
METADATA = 16,
STRUCT_ANON = 18,
STRUCT_NAME = 19,
STRUCT_NAMED = 20,
FUNCTION = 21,
};
struct AbbrevParam
{
AbbrevEncoding encoding;
uint64_t value; // this is also the bitwidth for Fixed/VBR
};
struct AbbrevDesc
{
rdcarray<AbbrevParam> params;
};
};
+12 -23
View File
@@ -28,22 +28,12 @@
namespace LLVMBC
{
struct AbbrevParam
{
AbbrevEncoding encoding;
uint64_t value; // this is also the bitwidth for Fixed/VBR
};
struct AbbrevDesc
{
rdcarray<AbbrevParam> params;
};
// the temporary context while pushing/popping blocks
struct BlockContext
{
BlockContext(size_t size = 2) : abbrevSize(size) {}
BlockContext(size_t size, size_t offs = 0) : abbrevSize(size), lengthOffset(offs) {}
size_t abbrevSize;
size_t lengthOffset;
rdcarray<AbbrevDesc> abbrevs;
};
@@ -62,6 +52,8 @@ bool BitcodeReader::Valid(const byte *bitcode, size_t length)
BitcodeReader::BitcodeReader(const byte *bitcode, size_t length) : b(bitcode, length)
{
abbrevSize = 2;
uint32_t magic = b.Read<uint32_t>();
RDCASSERT(magic == BitcodeMagic);
@@ -78,7 +70,7 @@ BlockOrRecord BitcodeReader::ReadToplevelBlock()
BlockOrRecord ret;
// should hit ENTER_SUBBLOCK first for top-level block
uint32_t abbrevID = b.fixed<uint32_t>(abbrevSize());
uint32_t abbrevID = b.fixed<uint32_t>(abbrevSize);
RDCASSERT(abbrevID == ENTER_SUBBLOCK);
ReadBlockContents(ret);
@@ -95,7 +87,8 @@ void BitcodeReader::ReadBlockContents(BlockOrRecord &block)
{
block.id = b.vbr<uint32_t>(8);
blockStack.push_back(new BlockContext(b.vbr<size_t>(4)));
abbrevSize = b.vbr<size_t>(4);
blockStack.push_back(new BlockContext(abbrevSize));
b.align32bits();
block.blockDwordLength = b.Read<uint32_t>();
@@ -106,7 +99,7 @@ void BitcodeReader::ReadBlockContents(BlockOrRecord &block)
uint32_t abbrevID = ~0U;
do
{
abbrevID = b.fixed<uint32_t>(abbrevSize());
abbrevID = b.fixed<uint32_t>(abbrevSize);
if(abbrevID == END_BLOCK)
{
@@ -254,6 +247,8 @@ void BitcodeReader::ReadBlockContents(BlockOrRecord &block)
delete blockStack.back();
blockStack.erase(blockStack.size() - 1);
abbrevSize = blockStack.empty() ? 2 : blockStack.back()->abbrevSize;
}
uint64_t BitcodeReader::decodeAbbrevParam(const AbbrevParam &param)
@@ -267,19 +262,13 @@ uint64_t BitcodeReader::decodeAbbrevParam(const AbbrevParam &param)
case AbbrevEncoding::Char6: return b.c6();
case AbbrevEncoding::Literal: return param.value;
case AbbrevEncoding::Array:
case AbbrevEncoding::Blob: RDCERR("Array and blob types must be decoded specially");
case AbbrevEncoding::Blob: RDCERR("Array and blob types must be decoded specially"); break;
case AbbrevEncoding::Unknown: RDCERR("Unexpected abbrev encoding"); break;
}
return 0;
}
size_t BitcodeReader::abbrevSize() const
{
if(blockStack.empty())
return 2;
return blockStack.back()->abbrevSize;
}
const AbbrevDesc &BitcodeReader::getAbbrev(uint32_t blockId, uint32_t abbrevID)
{
const BlockInfo *info = blockInfo[blockId];
+1 -1
View File
@@ -68,10 +68,10 @@ public:
private:
BitReader b;
size_t abbrevSize;
void ReadBlockContents(BlockOrRecord &block);
const AbbrevDesc &getAbbrev(uint32_t blockId, uint32_t abbrevID);
size_t abbrevSize() const;
uint64_t decodeAbbrevParam(const AbbrevParam &param);
rdcarray<BlockContext *> blockStack;
+325 -1
View File
@@ -24,18 +24,342 @@
#include "llvm_encoder.h"
#include "os/os_specific.h"
#include "llvm_common.h"
namespace LLVMBC
{
static uint32_t GetBlockAbbrevSize(KnownBlock block)
{
uint32_t ret = 0;
// the abbrev sizes seem to be hardcoded in llvm? At least this matches dxc's llvm
switch(block)
{
case KnownBlock::BLOCKINFO: ret = 2; break;
case KnownBlock::MODULE_BLOCK: ret = 3; break;
case KnownBlock::PARAMATTR_BLOCK: ret = 3; break;
case KnownBlock::PARAMATTR_GROUP_BLOCK: ret = 3; break;
case KnownBlock::CONSTANTS_BLOCK: ret = 4; break;
case KnownBlock::FUNCTION_BLOCK: ret = 4; break;
case KnownBlock::VALUE_SYMTAB_BLOCK: ret = 4; break;
case KnownBlock::METADATA_BLOCK: ret = 3; break;
case KnownBlock::METADATA_ATTACHMENT: ret = 3; break;
case KnownBlock::TYPE_BLOCK: ret = 4; break;
case KnownBlock::USELIST_BLOCK: ret = 3; break;
case KnownBlock::Count: break;
}
return ret;
}
#define MagicFixedSizeNumTypes 99
#define AbbFixed(n) \
{ \
AbbrevEncoding::Fixed, n \
}
#define AbbVBR(n) \
{ \
AbbrevEncoding::VBR, n \
}
#define AbbArray() \
{ \
AbbrevEncoding::Array, 0 \
}
#define AbbLiteral(lit) \
{ \
AbbrevEncoding::Literal, uint64_t(lit) \
}
#define AbbChar6() \
{ \
AbbrevEncoding::Char6, 0 \
}
#define AbbFixedTypes() AbbFixed(MagicFixedSizeNumTypes)
using AbbrevDefinition = AbbrevParam[8];
// known abbreviations. Encoded as an array of abbrevs, with each one being an array of params (the
// last param will have AbbrevEncoding::Unknown == 0)
enum class ValueSymtabAbbrev
{
Entry8,
Entry7,
Entry6,
BBEntry6,
};
AbbrevDefinition ValueSymtabAbbrevDefs[] = {
// Entry8
{
AbbFixed(3), AbbVBR(8), AbbArray(), AbbFixed(8),
},
// Entry7
{
AbbLiteral(ValueSymtabRecord::ENTRY), AbbVBR(8), AbbArray(), AbbFixed(7),
},
// Entry6
{
AbbLiteral(ValueSymtabRecord::ENTRY), AbbVBR(8), AbbArray(), AbbChar6(),
},
// BBEntry6
{
AbbLiteral(ValueSymtabRecord::BBENTRY), AbbVBR(8), AbbArray(), AbbChar6(),
},
};
enum class ConstantsAbbrev
{
SetType,
Integer,
EvalCast,
Null,
};
AbbrevDefinition ConstantsAbbrevDefs[] = {
// SetType
{
AbbLiteral(ConstantsRecord::SETTYPE), AbbFixedTypes(),
},
// Integer
{
AbbLiteral(ConstantsRecord::INTEGER), AbbVBR(8),
},
// EvalCast
{
AbbLiteral(ConstantsRecord::EVAL_CAST), AbbFixed(4), AbbFixedTypes(), AbbVBR(8),
},
// Null
{
AbbLiteral(ConstantsRecord::CONST_NULL),
},
};
enum class FunctionAbbrev
{
Load,
BinOp,
BinOpFlags,
Cast,
RetVoid,
RetValue,
Unreachable,
GEP,
};
AbbrevDefinition FunctionAbbrevDefs[] = {
// Load
{
AbbLiteral(FunctionRecord::INST_LOAD), AbbVBR(6), AbbFixedTypes(), AbbVBR(4), AbbFixed(1),
},
// BinOp
{
AbbLiteral(FunctionRecord::INST_BINOP), AbbVBR(6), AbbVBR(6), AbbFixed(4),
},
// BinOpFlags
{
AbbLiteral(FunctionRecord::INST_BINOP), AbbVBR(6), AbbVBR(6), AbbFixed(4), AbbFixed(7),
},
// Cast
{
AbbLiteral(FunctionRecord::INST_CAST), AbbVBR(6), AbbFixedTypes(), AbbFixed(4),
},
// RetVoid
{
AbbLiteral(FunctionRecord::INST_RET),
},
// RetValue
{
AbbLiteral(FunctionRecord::INST_RET), AbbVBR(6),
},
// Unreachable
{
AbbLiteral(FunctionRecord::INST_UNREACHABLE),
},
// GEP
{
AbbLiteral(FunctionRecord::INST_GEP), AbbFixed(1), AbbFixedTypes(), AbbArray(), AbbVBR(6),
},
};
static AbbrevDefinition *GetAbbrevs(KnownBlock block)
{
AbbrevDefinition *ret = NULL;
switch(block)
{
case KnownBlock::VALUE_SYMTAB_BLOCK: ret = ValueSymtabAbbrevDefs; break;
case KnownBlock::CONSTANTS_BLOCK: ret = ConstantsAbbrevDefs; break;
case KnownBlock::FUNCTION_BLOCK: ret = FunctionAbbrevDefs; break;
default: break;
}
return ret;
}
static uint32_t GetNumAbbrevs(KnownBlock block)
{
uint32_t ret = 0;
switch(block)
{
case KnownBlock::VALUE_SYMTAB_BLOCK: ret = ARRAY_COUNT(ValueSymtabAbbrevDefs); break;
case KnownBlock::CONSTANTS_BLOCK: ret = ARRAY_COUNT(ConstantsAbbrevDefs); break;
case KnownBlock::FUNCTION_BLOCK: ret = ARRAY_COUNT(FunctionAbbrevDefs); break;
default: break;
}
return ret;
}
BitcodeWriter::BitcodeWriter(bytebuf &buf) : b(buf)
{
b.Write(LLVMBC::BitcodeMagic);
curBlock = KnownBlock::Count;
abbrevSize = 2;
}
BitcodeWriter::~BitcodeWriter()
{
}
void BitcodeWriter::BeginBlock(KnownBlock block)
{
uint32_t newAbbrevSize = GetBlockAbbrevSize(block);
if(newAbbrevSize == 0)
{
RDCERR("Encoding error: unrecognised block %u", block);
return;
}
b.fixed(abbrevSize, ENTER_SUBBLOCK);
b.vbr(8, block);
b.vbr(4, newAbbrevSize);
b.align32bits();
size_t offs = b.GetByteOffset();
// write a placeholder length
b.Write<uint32_t>(0U);
curBlock = block;
abbrevSize = newAbbrevSize;
blockStack.push_back({block, offs});
}
void BitcodeWriter::EndBlock()
{
b.vbr(abbrevSize, END_BLOCK);
b.align32bits();
size_t offs = blockStack.back().second;
// -4 because we don't include the word with the length itself
size_t lengthInBytes = b.GetByteOffset() - offs - 4;
b.PatchLengthWord(offs, uint32_t(lengthInBytes / 4));
blockStack.pop_back();
if(blockStack.empty())
{
curBlock = KnownBlock::Count;
abbrevSize = 2;
}
else
{
curBlock = blockStack.back().first;
abbrevSize = GetBlockAbbrevSize(curBlock);
}
}
void BitcodeWriter::ModuleBlockInfo(uint32_t numTypes)
{
// these abbrevs are hardcoded in llvm, at least at dxc's version
BeginBlock(KnownBlock::BLOCKINFO);
// the module-level blockinfo contains abbrevs for these block types that can be repeated
// subblocks
for(KnownBlock block :
{KnownBlock::VALUE_SYMTAB_BLOCK, KnownBlock::CONSTANTS_BLOCK, KnownBlock::FUNCTION_BLOCK})
{
Unabbrev((uint32_t)BlockInfoRecord::SETBID, (uint32_t)block);
AbbrevDefinition *abbrevs = GetAbbrevs(block);
uint32_t numAbbrevs = GetNumAbbrevs(block);
for(uint32_t i = 0; i < numAbbrevs; i++)
{
b.fixed(abbrevSize, DEFINE_ABBREV);
AbbrevParam *abbrev = abbrevs[i];
uint32_t numParams = 0;
while(abbrev[numParams].encoding != AbbrevEncoding::Unknown)
numParams++;
b.vbr(5, numParams);
for(uint32_t p = 0; p < numParams; p++)
{
AbbrevParam param = abbrev[p];
if(param.value == MagicFixedSizeNumTypes)
{
param.value = 32 - Bits::CountLeadingZeroes(numTypes);
}
const bool lit = param.encoding == AbbrevEncoding::Literal;
b.fixed<bool>(1, lit);
if(lit)
{
b.vbr(8, param.value);
}
else
{
b.fixed(3, param.encoding);
if(param.encoding == AbbrevEncoding::VBR || param.encoding == AbbrevEncoding::Fixed)
b.vbr(5, param.value);
}
}
}
}
EndBlock();
}
void BitcodeWriter::Unabbrev(uint32_t record, uint32_t val)
{
b.fixed(abbrevSize, UNABBREV_RECORD);
b.vbr(6, record);
b.vbr(6, 1U); // num parameters
b.vbr(6, val);
}
void BitcodeWriter::Unabbrev(uint32_t record, uint64_t val)
{
b.fixed(abbrevSize, UNABBREV_RECORD);
b.vbr(6, record);
b.vbr(6, 1U); // num parameters
b.vbr(6, val);
}
void BitcodeWriter::Unabbrev(uint32_t record, const rdcarray<uint32_t> &vals)
{
b.fixed(abbrevSize, UNABBREV_RECORD);
b.vbr(6, record);
b.vbr(6, vals.size());
for(uint32_t v : vals)
b.vbr(6, v);
}
void BitcodeWriter::Unabbrev(uint32_t record, const rdcarray<uint64_t> &vals)
{
b.fixed(abbrevSize, UNABBREV_RECORD);
b.vbr(6, record);
b.vbr(6, vals.size());
for(uint64_t v : vals)
b.vbr(6, v);
}
};
#if ENABLED(ENABLE_UNIT_TESTS)
@@ -27,6 +27,7 @@
#include "api/replay/rdcarray.h"
#include "api/replay/rdcstr.h"
#include "llvm_bitwriter.h"
#include "llvm_common.h"
namespace LLVMBC
{
@@ -36,8 +37,24 @@ public:
BitcodeWriter(bytebuf &buf);
~BitcodeWriter();
void BeginBlock(KnownBlock block);
void EndBlock();
void ModuleBlockInfo(uint32_t numTypes);
void Unabbrev(uint32_t record, uint32_t val);
void Unabbrev(uint32_t record, uint64_t val);
void Unabbrev(uint32_t record, const rdcarray<uint32_t> &vals);
void Unabbrev(uint32_t record, const rdcarray<uint64_t> &vals);
private:
BitWriter b;
size_t abbrevSize;
KnownBlock curBlock;
rdcarray<rdcpair<KnownBlock, size_t>> blockStack;
};
}; // namespace LLVMBC