6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2025-02-25 03:30:37 +00:00
Code Issues Projects Releases Wiki Activity
llvm-6502/lib/Target/PowerPC/Disassembler/PPCDisassembler.cpp

370 lines
13 KiB
C++
Raw Normal View History

Add a disassembler to the PowerPC backend The tests for the disassembler were adapted from the encoder tests, and for the most part, the output from the disassembler matches that encoder-test inputs. There are some places where more-informative mnemonics could be produced (notably for the branch instructions), and those cases are noted in the tests with FIXMEs. Future work includes: - Generating more-informative mnemonics when possible (this may also be done in the printer). - Remove the dependence on positional "numbered" operand-to-variable mapping (for both encoding and decoding). - Internally using 64-bit instruction variants in 64-bit mode (if this turns out to matter). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197693 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-19 16:13:01 +00:00
//===------ PPCDisassembler.cpp - Disassembler for PowerPC ------*- C++ -*-===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
#include "PPC.h"
#include "llvm/MC/MCDisassembler.h"
#include "llvm/MC/MCFixedLenDisassembler.h"
#include "llvm/MC/MCInst.h"
#include "llvm/MC/MCSubtargetInfo.h"
#include "llvm/Support/TargetRegistry.h"
using namespace llvm;
[Modules] Make Support/Debug.h modular. This requires it to not change behavior based on other files defining DEBUG_TYPE, which means it cannot define DEBUG_TYPE at all. This is actually better IMO as it forces folks to define relevant DEBUG_TYPEs for their files. However, it requires all files that currently use DEBUG(...) to define a DEBUG_TYPE if they don't already. I've updated all such files in LLVM and will do the same for other upstream projects. This still leaves one important change in how LLVM uses the DEBUG_TYPE macro going forward: we need to only define the macro *after* header files have been #include-ed. Previously, this wasn't possible because Debug.h required the macro to be pre-defined. This commit removes that. By defining DEBUG_TYPE after the includes two things are fixed: - Header files that need to provide a DEBUG_TYPE for some inline code can do so by defining the macro before their inline code and undef-ing it afterward so the macro does not escape. - We no longer have rampant ODR violations due to including headers with different DEBUG_TYPE definitions. This may be mostly an academic violation today, but with modules these types of violations are easy to check for and potentially very relevant. Where necessary to suppor headers with DEBUG_TYPE, I have moved the definitions below the includes in this commit. I plan to move the rest of the DEBUG_TYPE macros in LLVM in subsequent commits; this one is big enough. The comments in Debug.h, which were hilariously out of date already, have been updated to reflect the recommended practice going forward. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206822 91177308-0d34-0410-b5e6-96231b3b80d8
2014-04-21 22:55:11 +00:00
#define DEBUG_TYPE "ppc-disassembler"
Add a disassembler to the PowerPC backend The tests for the disassembler were adapted from the encoder tests, and for the most part, the output from the disassembler matches that encoder-test inputs. There are some places where more-informative mnemonics could be produced (notably for the branch instructions), and those cases are noted in the tests with FIXMEs. Future work includes: - Generating more-informative mnemonics when possible (this may also be done in the printer). - Remove the dependence on positional "numbered" operand-to-variable mapping (for both encoding and decoding). - Internally using 64-bit instruction variants in 64-bit mode (if this turns out to matter). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197693 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-19 16:13:01 +00:00
typedef MCDisassembler::DecodeStatus DecodeStatus;
namespace {
class PPCDisassembler : public MCDisassembler {
public:
[MC] Require an MCContext when constructing an MCDisassembler. This patch re-introduces the MCContext member that was removed from MCDisassembler in r206063, and requires that an MCContext be passed in at MCDisassembler construction time. (Previously the MCContext member had been initialized in an ad-hoc fashion after construction). The MCCContext member can be used by MCDisassembler sub-classes to construct constant or target-specific MCExprs. This patch updates disassemblers for in-tree targets, and provides the MCRegisterInfo instance that some disassemblers were using through the MCContext (previously those backends were constructing their own MCRegisterInfo instances). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206241 91177308-0d34-0410-b5e6-96231b3b80d8
2014-04-15 04:40:56 +00:00
PPCDisassembler(const MCSubtargetInfo &STI, MCContext &Ctx)
: MCDisassembler(STI, Ctx) {}
Add a disassembler to the PowerPC backend The tests for the disassembler were adapted from the encoder tests, and for the most part, the output from the disassembler matches that encoder-test inputs. There are some places where more-informative mnemonics could be produced (notably for the branch instructions), and those cases are noted in the tests with FIXMEs. Future work includes: - Generating more-informative mnemonics when possible (this may also be done in the printer). - Remove the dependence on positional "numbered" operand-to-variable mapping (for both encoding and decoding). - Internally using 64-bit instruction variants in 64-bit mode (if this turns out to matter). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197693 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-19 16:13:01 +00:00
virtual ~PPCDisassembler() {}
Misc style fixes. NFC. This fixes a few cases of: * Wrong variable name style. * Lines longer than 80 columns. * Repeated names in comments. * clang-format of the above. This make the next patch a lot easier to read. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221615 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-10 18:11:10 +00:00
DecodeStatus getInstruction(MCInst &Instr, uint64_t &Size,
Pass an ArrayRef to MCDisassembler::getInstruction. With this patch MCDisassembler::getInstruction takes an ArrayRef<uint8_t> instead of a MemoryObject. Even on X86 there is a maximum size an instruction can have. Given that, it seems way simpler and more efficient to just pass an ArrayRef to the disassembler instead of a MemoryObject and have it do a virtual call every time it wants some extra bytes. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221751 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-12 02:04:27 +00:00
ArrayRef<uint8_t> Bytes, uint64_t Address,
Misc style fixes. NFC. This fixes a few cases of: * Wrong variable name style. * Lines longer than 80 columns. * Repeated names in comments. * clang-format of the above. This make the next patch a lot easier to read. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221615 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-10 18:11:10 +00:00
raw_ostream &VStream,
raw_ostream &CStream) const override;
Add a disassembler to the PowerPC backend The tests for the disassembler were adapted from the encoder tests, and for the most part, the output from the disassembler matches that encoder-test inputs. There are some places where more-informative mnemonics could be produced (notably for the branch instructions), and those cases are noted in the tests with FIXMEs. Future work includes: - Generating more-informative mnemonics when possible (this may also be done in the printer). - Remove the dependence on positional "numbered" operand-to-variable mapping (for both encoding and decoding). - Internally using 64-bit instruction variants in 64-bit mode (if this turns out to matter). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197693 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-19 16:13:01 +00:00
};
} // end anonymous namespace
static MCDisassembler *createPPCDisassembler(const Target &T,
[MC] Require an MCContext when constructing an MCDisassembler. This patch re-introduces the MCContext member that was removed from MCDisassembler in r206063, and requires that an MCContext be passed in at MCDisassembler construction time. (Previously the MCContext member had been initialized in an ad-hoc fashion after construction). The MCCContext member can be used by MCDisassembler sub-classes to construct constant or target-specific MCExprs. This patch updates disassemblers for in-tree targets, and provides the MCRegisterInfo instance that some disassemblers were using through the MCContext (previously those backends were constructing their own MCRegisterInfo instances). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206241 91177308-0d34-0410-b5e6-96231b3b80d8
2014-04-15 04:40:56 +00:00
const MCSubtargetInfo &STI,
MCContext &Ctx) {
return new PPCDisassembler(STI, Ctx);
Add a disassembler to the PowerPC backend The tests for the disassembler were adapted from the encoder tests, and for the most part, the output from the disassembler matches that encoder-test inputs. There are some places where more-informative mnemonics could be produced (notably for the branch instructions), and those cases are noted in the tests with FIXMEs. Future work includes: - Generating more-informative mnemonics when possible (this may also be done in the printer). - Remove the dependence on positional "numbered" operand-to-variable mapping (for both encoding and decoding). - Internally using 64-bit instruction variants in 64-bit mode (if this turns out to matter). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197693 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-19 16:13:01 +00:00
}
extern "C" void LLVMInitializePowerPCDisassembler() {
// Register the disassembler for each target.
TargetRegistry::RegisterMCDisassembler(ThePPC32Target,
createPPCDisassembler);
TargetRegistry::RegisterMCDisassembler(ThePPC64Target,
createPPCDisassembler);
TargetRegistry::RegisterMCDisassembler(ThePPC64LETarget,
createPPCDisassembler);
}
// FIXME: These can be generated by TableGen from the existing register
// encoding values!
static const unsigned CRRegs[] = {
PPC::CR0, PPC::CR1, PPC::CR2, PPC::CR3,
PPC::CR4, PPC::CR5, PPC::CR6, PPC::CR7
};
static const unsigned CRBITRegs[] = {
PPC::CR0LT, PPC::CR0GT, PPC::CR0EQ, PPC::CR0UN,
PPC::CR1LT, PPC::CR1GT, PPC::CR1EQ, PPC::CR1UN,
PPC::CR2LT, PPC::CR2GT, PPC::CR2EQ, PPC::CR2UN,
PPC::CR3LT, PPC::CR3GT, PPC::CR3EQ, PPC::CR3UN,
PPC::CR4LT, PPC::CR4GT, PPC::CR4EQ, PPC::CR4UN,
PPC::CR5LT, PPC::CR5GT, PPC::CR5EQ, PPC::CR5UN,
PPC::CR6LT, PPC::CR6GT, PPC::CR6EQ, PPC::CR6UN,
PPC::CR7LT, PPC::CR7GT, PPC::CR7EQ, PPC::CR7UN
};
static const unsigned FRegs[] = {
PPC::F0, PPC::F1, PPC::F2, PPC::F3,
PPC::F4, PPC::F5, PPC::F6, PPC::F7,
PPC::F8, PPC::F9, PPC::F10, PPC::F11,
PPC::F12, PPC::F13, PPC::F14, PPC::F15,
PPC::F16, PPC::F17, PPC::F18, PPC::F19,
PPC::F20, PPC::F21, PPC::F22, PPC::F23,
PPC::F24, PPC::F25, PPC::F26, PPC::F27,
PPC::F28, PPC::F29, PPC::F30, PPC::F31
};
static const unsigned VRegs[] = {
PPC::V0, PPC::V1, PPC::V2, PPC::V3,
PPC::V4, PPC::V5, PPC::V6, PPC::V7,
PPC::V8, PPC::V9, PPC::V10, PPC::V11,
PPC::V12, PPC::V13, PPC::V14, PPC::V15,
PPC::V16, PPC::V17, PPC::V18, PPC::V19,
PPC::V20, PPC::V21, PPC::V22, PPC::V23,
PPC::V24, PPC::V25, PPC::V26, PPC::V27,
PPC::V28, PPC::V29, PPC::V30, PPC::V31
};
[PowerPC] Initial support for the VSX instruction set VSX is an ISA extension supported on the POWER7 and later cores that enhances floating-point vector and scalar capabilities. Among other things, this adds <2 x double> support and generally helps to reduce register pressure. The interesting part of this ISA feature is the register configuration: there are 64 new 128-bit vector registers, the 32 of which are super-registers of the existing 32 scalar floating-point registers, and the second 32 of which overlap with the 32 Altivec vector registers. This makes things like vector insertion and extraction tricky: this can be free but only if we force a restriction to the right register subclass when needed. A new "minipass" PPCVSXCopy takes care of this (although it could do a more-optimal job of it; see the comment about unnecessary copies below). Please note that, currently, VSX is not enabled by default when targeting anything because it is not yet ready for that. The assembler and disassembler are fully implemented and tested. However: - CodeGen support causes miscompiles; test-suite runtime failures: MultiSource/Benchmarks/FreeBench/distray/distray MultiSource/Benchmarks/McCat/08-main/main MultiSource/Benchmarks/Olden/voronoi/voronoi MultiSource/Benchmarks/mafft/pairlocalalign MultiSource/Benchmarks/tramp3d-v4/tramp3d-v4 SingleSource/Benchmarks/CoyoteBench/almabench SingleSource/Benchmarks/Misc/matmul_f64_4x4 - The lowering currently falls back to using Altivec instructions far more than it should. Worse, there are some things that are scalarized through the stack that shouldn't be. - A lot of unnecessary copies make it past the optimizers, and this needs to be fixed. - Many more regression tests are needed. Normally, I'd fix these things prior to committing, but there are some students and other contributors who would like to work this, and so it makes sense to move this development process upstream where it can be subject to the regular code-review procedures. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@203768 91177308-0d34-0410-b5e6-96231b3b80d8
2014-03-13 07:58:58 +00:00
static const unsigned VSRegs[] = {
PPC::VSL0, PPC::VSL1, PPC::VSL2, PPC::VSL3,
PPC::VSL4, PPC::VSL5, PPC::VSL6, PPC::VSL7,
PPC::VSL8, PPC::VSL9, PPC::VSL10, PPC::VSL11,
PPC::VSL12, PPC::VSL13, PPC::VSL14, PPC::VSL15,
PPC::VSL16, PPC::VSL17, PPC::VSL18, PPC::VSL19,
PPC::VSL20, PPC::VSL21, PPC::VSL22, PPC::VSL23,
PPC::VSL24, PPC::VSL25, PPC::VSL26, PPC::VSL27,
PPC::VSL28, PPC::VSL29, PPC::VSL30, PPC::VSL31,
PPC::VSH0, PPC::VSH1, PPC::VSH2, PPC::VSH3,
PPC::VSH4, PPC::VSH5, PPC::VSH6, PPC::VSH7,
PPC::VSH8, PPC::VSH9, PPC::VSH10, PPC::VSH11,
PPC::VSH12, PPC::VSH13, PPC::VSH14, PPC::VSH15,
PPC::VSH16, PPC::VSH17, PPC::VSH18, PPC::VSH19,
PPC::VSH20, PPC::VSH21, PPC::VSH22, PPC::VSH23,
PPC::VSH24, PPC::VSH25, PPC::VSH26, PPC::VSH27,
PPC::VSH28, PPC::VSH29, PPC::VSH30, PPC::VSH31
};
[PowerPC] Add subregister classes for f64 VSX values We had stored both f64 values and v2f64, etc. values in the VSX registers. This worked, but was suboptimal because we would always spill 16-byte values even through we almost always had scalar 8-byte values. This resulted in an increase in stack-size use, extra memory bandwidth, etc. To fix this, I've added 64-bit subregisters of the Altivec registers, and combined those with the existing scalar floating-point registers to form a class of VSX scalar floating-point registers. The ABI code has also been enhanced to use this register class and some other necessary improvements have been made. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@205075 91177308-0d34-0410-b5e6-96231b3b80d8
2014-03-29 05:29:01 +00:00
static const unsigned VSFRegs[] = {
PPC::F0, PPC::F1, PPC::F2, PPC::F3,
PPC::F4, PPC::F5, PPC::F6, PPC::F7,
PPC::F8, PPC::F9, PPC::F10, PPC::F11,
PPC::F12, PPC::F13, PPC::F14, PPC::F15,
PPC::F16, PPC::F17, PPC::F18, PPC::F19,
PPC::F20, PPC::F21, PPC::F22, PPC::F23,
PPC::F24, PPC::F25, PPC::F26, PPC::F27,
PPC::F28, PPC::F29, PPC::F30, PPC::F31,
PPC::VF0, PPC::VF1, PPC::VF2, PPC::VF3,
PPC::VF4, PPC::VF5, PPC::VF6, PPC::VF7,
PPC::VF8, PPC::VF9, PPC::VF10, PPC::VF11,
PPC::VF12, PPC::VF13, PPC::VF14, PPC::VF15,
PPC::VF16, PPC::VF17, PPC::VF18, PPC::VF19,
PPC::VF20, PPC::VF21, PPC::VF22, PPC::VF23,
PPC::VF24, PPC::VF25, PPC::VF26, PPC::VF27,
PPC::VF28, PPC::VF29, PPC::VF30, PPC::VF31
};
Add a disassembler to the PowerPC backend The tests for the disassembler were adapted from the encoder tests, and for the most part, the output from the disassembler matches that encoder-test inputs. There are some places where more-informative mnemonics could be produced (notably for the branch instructions), and those cases are noted in the tests with FIXMEs. Future work includes: - Generating more-informative mnemonics when possible (this may also be done in the printer). - Remove the dependence on positional "numbered" operand-to-variable mapping (for both encoding and decoding). - Internally using 64-bit instruction variants in 64-bit mode (if this turns out to matter). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197693 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-19 16:13:01 +00:00
static const unsigned GPRegs[] = {
PPC::R0, PPC::R1, PPC::R2, PPC::R3,
PPC::R4, PPC::R5, PPC::R6, PPC::R7,
PPC::R8, PPC::R9, PPC::R10, PPC::R11,
PPC::R12, PPC::R13, PPC::R14, PPC::R15,
PPC::R16, PPC::R17, PPC::R18, PPC::R19,
PPC::R20, PPC::R21, PPC::R22, PPC::R23,
PPC::R24, PPC::R25, PPC::R26, PPC::R27,
PPC::R28, PPC::R29, PPC::R30, PPC::R31
};
static const unsigned GP0Regs[] = {
PPC::ZERO, PPC::R1, PPC::R2, PPC::R3,
PPC::R4, PPC::R5, PPC::R6, PPC::R7,
PPC::R8, PPC::R9, PPC::R10, PPC::R11,
PPC::R12, PPC::R13, PPC::R14, PPC::R15,
PPC::R16, PPC::R17, PPC::R18, PPC::R19,
PPC::R20, PPC::R21, PPC::R22, PPC::R23,
PPC::R24, PPC::R25, PPC::R26, PPC::R27,
PPC::R28, PPC::R29, PPC::R30, PPC::R31
};
static const unsigned G8Regs[] = {
PPC::X0, PPC::X1, PPC::X2, PPC::X3,
PPC::X4, PPC::X5, PPC::X6, PPC::X7,
PPC::X8, PPC::X9, PPC::X10, PPC::X11,
PPC::X12, PPC::X13, PPC::X14, PPC::X15,
PPC::X16, PPC::X17, PPC::X18, PPC::X19,
PPC::X20, PPC::X21, PPC::X22, PPC::X23,
PPC::X24, PPC::X25, PPC::X26, PPC::X27,
PPC::X28, PPC::X29, PPC::X30, PPC::X31
};
[PowerPC] Add support for the QPX vector instruction set This adds support for the QPX vector instruction set, which is used by the enhanced A2 cores on the IBM BG/Q supercomputers. QPX vectors are 256 bytes wide, holding 4 double-precision floating-point values. Boolean values, modeled here as <4 x i1> are actually also represented as floating-point values (essentially { -1, 1 } for { false, true }). QPX shares many features with Altivec and VSX, but is distinct from both of them. One major difference is that, instead of adding completely-separate vector registers, QPX vector registers are extensions of the scalar floating-point registers (lane 0 is the corresponding scalar floating-point value). The operations supported on QPX vectors mirrors that supported on the scalar floating-point values (with some additional ones for permutations and logical/comparison operations). I've been maintaining this support out-of-tree, as part of the bgclang project, for several years. This is not the entire bgclang patch set, but is most of the subset that can be cleanly integrated into LLVM proper at this time. Adding this to the LLVM backend is part of my efforts to rebase bgclang to the current LLVM trunk, but is independently useful (especially for codes that use LLVM as a JIT in library form). The assembler/disassembler test coverage is complete. The CodeGen test coverage is not, but I've included some tests, and more will be added as follow-up work. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@230413 91177308-0d34-0410-b5e6-96231b3b80d8
2015-02-25 01:06:45 +00:00
static const unsigned QFRegs[] = {
PPC::QF0, PPC::QF1, PPC::QF2, PPC::QF3,
PPC::QF4, PPC::QF5, PPC::QF6, PPC::QF7,
PPC::QF8, PPC::QF9, PPC::QF10, PPC::QF11,
PPC::QF12, PPC::QF13, PPC::QF14, PPC::QF15,
PPC::QF16, PPC::QF17, PPC::QF18, PPC::QF19,
PPC::QF20, PPC::QF21, PPC::QF22, PPC::QF23,
PPC::QF24, PPC::QF25, PPC::QF26, PPC::QF27,
PPC::QF28, PPC::QF29, PPC::QF30, PPC::QF31
};
Add a disassembler to the PowerPC backend The tests for the disassembler were adapted from the encoder tests, and for the most part, the output from the disassembler matches that encoder-test inputs. There are some places where more-informative mnemonics could be produced (notably for the branch instructions), and those cases are noted in the tests with FIXMEs. Future work includes: - Generating more-informative mnemonics when possible (this may also be done in the printer). - Remove the dependence on positional "numbered" operand-to-variable mapping (for both encoding and decoding). - Internally using 64-bit instruction variants in 64-bit mode (if this turns out to matter). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197693 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-19 16:13:01 +00:00
template <std::size_t N>
static DecodeStatus decodeRegisterClass(MCInst &Inst, uint64_t RegNo,
const unsigned (&Regs)[N]) {
assert(RegNo < N && "Invalid register number");
Inst.addOperand(MCOperand::CreateReg(Regs[RegNo]));
return MCDisassembler::Success;
}
static DecodeStatus DecodeCRRCRegisterClass(MCInst &Inst, uint64_t RegNo,
uint64_t Address,
const void *Decoder) {
return decodeRegisterClass(Inst, RegNo, CRRegs);
}
static DecodeStatus DecodeCRBITRCRegisterClass(MCInst &Inst, uint64_t RegNo,
uint64_t Address,
const void *Decoder) {
return decodeRegisterClass(Inst, RegNo, CRBITRegs);
}
static DecodeStatus DecodeF4RCRegisterClass(MCInst &Inst, uint64_t RegNo,
uint64_t Address,
const void *Decoder) {
return decodeRegisterClass(Inst, RegNo, FRegs);
}
static DecodeStatus DecodeF8RCRegisterClass(MCInst &Inst, uint64_t RegNo,
uint64_t Address,
const void *Decoder) {
return decodeRegisterClass(Inst, RegNo, FRegs);
}
static DecodeStatus DecodeVRRCRegisterClass(MCInst &Inst, uint64_t RegNo,
uint64_t Address,
const void *Decoder) {
return decodeRegisterClass(Inst, RegNo, VRegs);
}
[PowerPC] Initial support for the VSX instruction set VSX is an ISA extension supported on the POWER7 and later cores that enhances floating-point vector and scalar capabilities. Among other things, this adds <2 x double> support and generally helps to reduce register pressure. The interesting part of this ISA feature is the register configuration: there are 64 new 128-bit vector registers, the 32 of which are super-registers of the existing 32 scalar floating-point registers, and the second 32 of which overlap with the 32 Altivec vector registers. This makes things like vector insertion and extraction tricky: this can be free but only if we force a restriction to the right register subclass when needed. A new "minipass" PPCVSXCopy takes care of this (although it could do a more-optimal job of it; see the comment about unnecessary copies below). Please note that, currently, VSX is not enabled by default when targeting anything because it is not yet ready for that. The assembler and disassembler are fully implemented and tested. However: - CodeGen support causes miscompiles; test-suite runtime failures: MultiSource/Benchmarks/FreeBench/distray/distray MultiSource/Benchmarks/McCat/08-main/main MultiSource/Benchmarks/Olden/voronoi/voronoi MultiSource/Benchmarks/mafft/pairlocalalign MultiSource/Benchmarks/tramp3d-v4/tramp3d-v4 SingleSource/Benchmarks/CoyoteBench/almabench SingleSource/Benchmarks/Misc/matmul_f64_4x4 - The lowering currently falls back to using Altivec instructions far more than it should. Worse, there are some things that are scalarized through the stack that shouldn't be. - A lot of unnecessary copies make it past the optimizers, and this needs to be fixed. - Many more regression tests are needed. Normally, I'd fix these things prior to committing, but there are some students and other contributors who would like to work this, and so it makes sense to move this development process upstream where it can be subject to the regular code-review procedures. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@203768 91177308-0d34-0410-b5e6-96231b3b80d8
2014-03-13 07:58:58 +00:00
static DecodeStatus DecodeVSRCRegisterClass(MCInst &Inst, uint64_t RegNo,
uint64_t Address,
const void *Decoder) {
return decodeRegisterClass(Inst, RegNo, VSRegs);
}
[PowerPC] Add subregister classes for f64 VSX values We had stored both f64 values and v2f64, etc. values in the VSX registers. This worked, but was suboptimal because we would always spill 16-byte values even through we almost always had scalar 8-byte values. This resulted in an increase in stack-size use, extra memory bandwidth, etc. To fix this, I've added 64-bit subregisters of the Altivec registers, and combined those with the existing scalar floating-point registers to form a class of VSX scalar floating-point registers. The ABI code has also been enhanced to use this register class and some other necessary improvements have been made. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@205075 91177308-0d34-0410-b5e6-96231b3b80d8
2014-03-29 05:29:01 +00:00
static DecodeStatus DecodeVSFRCRegisterClass(MCInst &Inst, uint64_t RegNo,
uint64_t Address,
const void *Decoder) {
return decodeRegisterClass(Inst, RegNo, VSFRegs);
}
Add a disassembler to the PowerPC backend The tests for the disassembler were adapted from the encoder tests, and for the most part, the output from the disassembler matches that encoder-test inputs. There are some places where more-informative mnemonics could be produced (notably for the branch instructions), and those cases are noted in the tests with FIXMEs. Future work includes: - Generating more-informative mnemonics when possible (this may also be done in the printer). - Remove the dependence on positional "numbered" operand-to-variable mapping (for both encoding and decoding). - Internally using 64-bit instruction variants in 64-bit mode (if this turns out to matter). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197693 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-19 16:13:01 +00:00
static DecodeStatus DecodeGPRCRegisterClass(MCInst &Inst, uint64_t RegNo,
uint64_t Address,
const void *Decoder) {
return decodeRegisterClass(Inst, RegNo, GPRegs);
}
static DecodeStatus DecodeGPRC_NOR0RegisterClass(MCInst &Inst, uint64_t RegNo,
uint64_t Address,
const void *Decoder) {
return decodeRegisterClass(Inst, RegNo, GP0Regs);
}
static DecodeStatus DecodeG8RCRegisterClass(MCInst &Inst, uint64_t RegNo,
uint64_t Address,
const void *Decoder) {
return decodeRegisterClass(Inst, RegNo, G8Regs);
}
#define DecodePointerLikeRegClass0 DecodeGPRCRegisterClass
#define DecodePointerLikeRegClass1 DecodeGPRC_NOR0RegisterClass
[PowerPC] Add support for the QPX vector instruction set This adds support for the QPX vector instruction set, which is used by the enhanced A2 cores on the IBM BG/Q supercomputers. QPX vectors are 256 bytes wide, holding 4 double-precision floating-point values. Boolean values, modeled here as <4 x i1> are actually also represented as floating-point values (essentially { -1, 1 } for { false, true }). QPX shares many features with Altivec and VSX, but is distinct from both of them. One major difference is that, instead of adding completely-separate vector registers, QPX vector registers are extensions of the scalar floating-point registers (lane 0 is the corresponding scalar floating-point value). The operations supported on QPX vectors mirrors that supported on the scalar floating-point values (with some additional ones for permutations and logical/comparison operations). I've been maintaining this support out-of-tree, as part of the bgclang project, for several years. This is not the entire bgclang patch set, but is most of the subset that can be cleanly integrated into LLVM proper at this time. Adding this to the LLVM backend is part of my efforts to rebase bgclang to the current LLVM trunk, but is independently useful (especially for codes that use LLVM as a JIT in library form). The assembler/disassembler test coverage is complete. The CodeGen test coverage is not, but I've included some tests, and more will be added as follow-up work. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@230413 91177308-0d34-0410-b5e6-96231b3b80d8
2015-02-25 01:06:45 +00:00
static DecodeStatus DecodeQFRCRegisterClass(MCInst &Inst, uint64_t RegNo,
uint64_t Address,
const void *Decoder) {
return decodeRegisterClass(Inst, RegNo, QFRegs);
}
#define DecodeQSRCRegisterClass DecodeQFRCRegisterClass
#define DecodeQBRCRegisterClass DecodeQFRCRegisterClass
Add a disassembler to the PowerPC backend The tests for the disassembler were adapted from the encoder tests, and for the most part, the output from the disassembler matches that encoder-test inputs. There are some places where more-informative mnemonics could be produced (notably for the branch instructions), and those cases are noted in the tests with FIXMEs. Future work includes: - Generating more-informative mnemonics when possible (this may also be done in the printer). - Remove the dependence on positional "numbered" operand-to-variable mapping (for both encoding and decoding). - Internally using 64-bit instruction variants in 64-bit mode (if this turns out to matter). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197693 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-19 16:13:01 +00:00
template<unsigned N>
static DecodeStatus decodeUImmOperand(MCInst &Inst, uint64_t Imm,
int64_t Address, const void *Decoder) {
assert(isUInt<N>(Imm) && "Invalid immediate");
Inst.addOperand(MCOperand::CreateImm(Imm));
return MCDisassembler::Success;
}
template<unsigned N>
static DecodeStatus decodeSImmOperand(MCInst &Inst, uint64_t Imm,
int64_t Address, const void *Decoder) {
assert(isUInt<N>(Imm) && "Invalid immediate");
Inst.addOperand(MCOperand::CreateImm(SignExtend64<N>(Imm)));
return MCDisassembler::Success;
}
static DecodeStatus decodeMemRIOperands(MCInst &Inst, uint64_t Imm,
int64_t Address, const void *Decoder) {
// Decode the memri field (imm, reg), which has the low 16-bits as the
// displacement and the next 5 bits as the register #.
uint64_t Base = Imm >> 16;
uint64_t Disp = Imm & 0xFFFF;
assert(Base < 32 && "Invalid base register");
switch (Inst.getOpcode()) {
default: break;
case PPC::LBZU:
case PPC::LHAU:
case PPC::LHZU:
case PPC::LWZU:
case PPC::LFSU:
case PPC::LFDU:
// Add the tied output operand.
Inst.addOperand(MCOperand::CreateReg(GP0Regs[Base]));
break;
case PPC::STBU:
case PPC::STHU:
case PPC::STWU:
case PPC::STFSU:
case PPC::STFDU:
Inst.insert(Inst.begin(), MCOperand::CreateReg(GP0Regs[Base]));
break;
}
Inst.addOperand(MCOperand::CreateImm(SignExtend64<16>(Disp)));
Inst.addOperand(MCOperand::CreateReg(GP0Regs[Base]));
return MCDisassembler::Success;
}
static DecodeStatus decodeMemRIXOperands(MCInst &Inst, uint64_t Imm,
int64_t Address, const void *Decoder) {
// Decode the memrix field (imm, reg), which has the low 14-bits as the
// displacement and the next 5 bits as the register #.
uint64_t Base = Imm >> 14;
uint64_t Disp = Imm & 0x3FFF;
assert(Base < 32 && "Invalid base register");
if (Inst.getOpcode() == PPC::LDU)
// Add the tied output operand.
Inst.addOperand(MCOperand::CreateReg(GP0Regs[Base]));
else if (Inst.getOpcode() == PPC::STDU)
Inst.insert(Inst.begin(), MCOperand::CreateReg(GP0Regs[Base]));
Inst.addOperand(MCOperand::CreateImm(SignExtend64<16>(Disp << 2)));
Inst.addOperand(MCOperand::CreateReg(GP0Regs[Base]));
return MCDisassembler::Success;
}
static DecodeStatus decodeCRBitMOperand(MCInst &Inst, uint64_t Imm,
int64_t Address, const void *Decoder) {
// The cr bit encoding is 0x80 >> cr_reg_num.
unsigned Zeros = countTrailingZeros(Imm);
assert(Zeros < 8 && "Invalid CR bit value");
Inst.addOperand(MCOperand::CreateReg(CRRegs[7 - Zeros]));
return MCDisassembler::Success;
}
#include "PPCGenDisassemblerTables.inc"
DecodeStatus PPCDisassembler::getInstruction(MCInst &MI, uint64_t &Size,
Pass an ArrayRef to MCDisassembler::getInstruction. With this patch MCDisassembler::getInstruction takes an ArrayRef<uint8_t> instead of a MemoryObject. Even on X86 there is a maximum size an instruction can have. Given that, it seems way simpler and more efficient to just pass an ArrayRef to the disassembler instead of a MemoryObject and have it do a virtual call every time it wants some extra bytes. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221751 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-12 02:04:27 +00:00
ArrayRef<uint8_t> Bytes,
Misc style fixes. NFC. This fixes a few cases of: * Wrong variable name style. * Lines longer than 80 columns. * Repeated names in comments. * clang-format of the above. This make the next patch a lot easier to read. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221615 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-10 18:11:10 +00:00
uint64_t Address, raw_ostream &OS,
raw_ostream &CS) const {
Add a disassembler to the PowerPC backend The tests for the disassembler were adapted from the encoder tests, and for the most part, the output from the disassembler matches that encoder-test inputs. There are some places where more-informative mnemonics could be produced (notably for the branch instructions), and those cases are noted in the tests with FIXMEs. Future work includes: - Generating more-informative mnemonics when possible (this may also be done in the printer). - Remove the dependence on positional "numbered" operand-to-variable mapping (for both encoding and decoding). - Internally using 64-bit instruction variants in 64-bit mode (if this turns out to matter). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197693 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-19 16:13:01 +00:00
// Get the four bytes of the instruction.
Size = 4;
Pass an ArrayRef to MCDisassembler::getInstruction. With this patch MCDisassembler::getInstruction takes an ArrayRef<uint8_t> instead of a MemoryObject. Even on X86 there is a maximum size an instruction can have. Given that, it seems way simpler and more efficient to just pass an ArrayRef to the disassembler instead of a MemoryObject and have it do a virtual call every time it wants some extra bytes. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221751 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-12 02:04:27 +00:00
if (Bytes.size() < 4) {
Add a disassembler to the PowerPC backend The tests for the disassembler were adapted from the encoder tests, and for the most part, the output from the disassembler matches that encoder-test inputs. There are some places where more-informative mnemonics could be produced (notably for the branch instructions), and those cases are noted in the tests with FIXMEs. Future work includes: - Generating more-informative mnemonics when possible (this may also be done in the printer). - Remove the dependence on positional "numbered" operand-to-variable mapping (for both encoding and decoding). - Internally using 64-bit instruction variants in 64-bit mode (if this turns out to matter). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197693 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-19 16:13:01 +00:00
Size = 0;
return MCDisassembler::Fail;
}
// The instruction is big-endian encoded.
Misc style fixes. NFC. This fixes a few cases of: * Wrong variable name style. * Lines longer than 80 columns. * Repeated names in comments. * clang-format of the above. This make the next patch a lot easier to read. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@221615 91177308-0d34-0410-b5e6-96231b3b80d8
2014-11-10 18:11:10 +00:00
uint32_t Inst =
(Bytes[0] << 24) | (Bytes[1] << 16) | (Bytes[2] << 8) | (Bytes[3] << 0);
Add a disassembler to the PowerPC backend The tests for the disassembler were adapted from the encoder tests, and for the most part, the output from the disassembler matches that encoder-test inputs. There are some places where more-informative mnemonics could be produced (notably for the branch instructions), and those cases are noted in the tests with FIXMEs. Future work includes: - Generating more-informative mnemonics when possible (this may also be done in the printer). - Remove the dependence on positional "numbered" operand-to-variable mapping (for both encoding and decoding). - Internally using 64-bit instruction variants in 64-bit mode (if this turns out to matter). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197693 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-19 16:13:01 +00:00
[PowerPC] Add support for the QPX vector instruction set This adds support for the QPX vector instruction set, which is used by the enhanced A2 cores on the IBM BG/Q supercomputers. QPX vectors are 256 bytes wide, holding 4 double-precision floating-point values. Boolean values, modeled here as <4 x i1> are actually also represented as floating-point values (essentially { -1, 1 } for { false, true }). QPX shares many features with Altivec and VSX, but is distinct from both of them. One major difference is that, instead of adding completely-separate vector registers, QPX vector registers are extensions of the scalar floating-point registers (lane 0 is the corresponding scalar floating-point value). The operations supported on QPX vectors mirrors that supported on the scalar floating-point values (with some additional ones for permutations and logical/comparison operations). I've been maintaining this support out-of-tree, as part of the bgclang project, for several years. This is not the entire bgclang patch set, but is most of the subset that can be cleanly integrated into LLVM proper at this time. Adding this to the LLVM backend is part of my efforts to rebase bgclang to the current LLVM trunk, but is independently useful (especially for codes that use LLVM as a JIT in library form). The assembler/disassembler test coverage is complete. The CodeGen test coverage is not, but I've included some tests, and more will be added as follow-up work. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@230413 91177308-0d34-0410-b5e6-96231b3b80d8
2015-02-25 01:06:45 +00:00
if ((STI.getFeatureBits() & PPC::FeatureQPX) != 0) {
DecodeStatus result =
decodeInstruction(DecoderTableQPX32, MI, Inst, Address, this, STI);
if (result != MCDisassembler::Fail)
return result;
MI.clear();
}
Add a disassembler to the PowerPC backend The tests for the disassembler were adapted from the encoder tests, and for the most part, the output from the disassembler matches that encoder-test inputs. There are some places where more-informative mnemonics could be produced (notably for the branch instructions), and those cases are noted in the tests with FIXMEs. Future work includes: - Generating more-informative mnemonics when possible (this may also be done in the printer). - Remove the dependence on positional "numbered" operand-to-variable mapping (for both encoding and decoding). - Internally using 64-bit instruction variants in 64-bit mode (if this turns out to matter). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197693 91177308-0d34-0410-b5e6-96231b3b80d8
2013-12-19 16:13:01 +00:00
return decodeInstruction(DecoderTable32, MI, Inst, Address, this, STI);
}
Reference in New Issue Copy Permalink