maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/lib/Target/Mips/AsmParser/MipsAsmParser.cpp

2691 lines
87 KiB
C++
Raw Normal View History

remove Emacs-tag form .cpp files in Mips Backend, and fix some typo. llvm-svn: 150805
2012-02-17 16:55:11 +08:00
//===-- MipsAsmParser.cpp - Parse Mips assembly to MCInst instructions ----===//
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
[mips] Implement %hi(sym1 - sym2) and %lo(sym1 - sym2) expressions Patch implements %hi(sym1 - sym2) and %lo(sym1 - sym2) expressions for MIPS by creating target expression class MipsMCExpr. Patch by Sasa Stankovic. Differential Revision: http://llvm-reviews.chandlerc.com/D2592 llvm-svn: 200783
2014-02-05 02:41:57 +08:00
#include "MCTargetDesc/MipsMCExpr.h"
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
#include "MCTargetDesc/MipsMCTargetDesc.h"
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
#include "MipsRegisterInfo.h"
Add a MCTargetStreamer interface. This patch fixes an old FIXME by creating a MCTargetStreamer interface and moving the target specific functions for ARM, Mips and PPC to it. The ARM streamer is still declared in a common place because it is used from lib/CodeGen/ARMException.cpp, but the Mips and PPC are completely hidden in the corresponding Target directories. I will send an email to llvmdev with instructions on how to use this. llvm-svn: 192181
2013-10-08 21:08:17 +08:00
#include "MipsTargetStreamer.h"
Re-sort all of the includes with ./utils/sort_includes.py so that subsequent changes are easier to review. About to fix some layering issues, and wanted to separate out the necessary churn. Also comment and sink the include of "Windows.h" in three .inc files to match the usage in Memory.inc. llvm-svn: 198685
2014-01-07 19:48:04 +08:00
#include "llvm/ADT/APInt.h"
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
#include "llvm/ADT/StringSwitch.h"
#include "llvm/MC/MCContext.h"
#include "llvm/MC/MCExpr.h"
#include "llvm/MC/MCInst.h"
[mips] Implement shorthand add / sub forms for MIPS. Summary: - If only two registers are passed to a three-register operation, then the first argument is both source and destination register. - If a non-register is passed as the last argument, generate the immediate version of the instruction. Also mark DADD commutative and add scheduling information (to the generic scheduler), and implement DSUB. Patch by David Chisnall His work was sponsored by: DARPA, AFRL CC: theraven Differential Revision: http://llvm-reviews.chandlerc.com/D3148 llvm-svn: 204605
2014-03-24 22:05:39 +08:00
#include "llvm/MC/MCInstBuilder.h"
Use the new script to sort the includes of every file under lib. Sooooo many of these had incorrect or strange main module includes. I have manually inspected all of these, and fixed the main module include to be the nearest plausible thing I could find. If you own or care about any of these source files, I encourage you to take some time and check that these edits were sensible. I can't have broken anything (I strictly added headers, and reordered them, never removed), but they may not be the headers you'd really like to identify as containing the API being implemented. Many forward declarations and missing includes were added to a header files to allow them to parse cleanly when included first. The main module rule does in fact have its merits. =] llvm-svn: 169131
2012-12-04 00:50:05 +08:00
#include "llvm/MC/MCParser/MCAsmLexer.h"
#include "llvm/MC/MCParser/MCParsedAsmOperand.h"
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
#include "llvm/MC/MCStreamer.h"
#include "llvm/MC/MCSubtargetInfo.h"
#include "llvm/MC/MCSymbol.h"
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
#include "llvm/MC/MCTargetAsmParser.h"
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
#include "llvm/Support/Debug.h"
[mips] Add checks for alignment and maximum displacements for most of the branch instructions for mips and micromips instruction sets thus avoiding the situation of generating branches to undesired locations if offsets cannot be encoded. This patch also checks if a fixup cannot be applied and returns a fatal error if that's the case. llvm-svn: 197223
2013-12-13 19:11:02 +08:00
#include "llvm/Support/MathExtras.h"
Re-sort all of the includes with ./utils/sort_includes.py so that subsequent changes are easier to review. About to fix some layering issues, and wanted to separate out the necessary churn. Also comment and sink the include of "Windows.h" in three .inc files to match the usage in Memory.inc. llvm-svn: 198685
2014-01-07 19:48:04 +08:00
#include "llvm/Support/TargetRegistry.h"
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
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. llvm-svn: 206822
2014-04-22 06:55:11 +08:00
#define DEBUG_TYPE "mips-asm-parser"
Add an instruction deprecation feature to TableGen. The 'Deprecated' class allows you to specify a SubtargetFeature that the instruction is deprecated on. The 'ComplexDeprecationPredicate' class allows you to define a custom predicate that is called to check for deprecation. For example: ComplexDeprecationPredicate<"MCR"> would mean you would have to define the following function: bool getMCRDeprecationInfo(MCInst &MI, MCSubtargetInfo &STI, std::string &Info) Which returns 'false' for not deprecated, and 'true' for deprecated and store the warning message in 'Info'. The MCTargetAsmParser constructor was chaned to take an extra argument of the MCInstrInfo class, so out-of-tree targets will need to be changed. llvm-svn: 190598
2013-09-12 18:28:05 +08:00
namespace llvm {
class MCInstrInfo;
}
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
namespace {
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
class MipsAssemblerOptions {
public:
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
MipsAssemblerOptions() : aTReg(1), reorder(true), macro(true) {}
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
unsigned getATRegNum() { return aTReg; }
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
bool setATReg(unsigned Reg);
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
bool isReorder() { return reorder; }
void setReorder() { reorder = true; }
void setNoreorder() { reorder = false; }
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
bool isMacro() { return macro; }
void setMacro() { macro = true; }
void setNomacro() { macro = false; }
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
private:
unsigned aTReg;
bool reorder;
bool macro;
};
}
namespace {
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
class MipsAsmParser : public MCTargetAsmParser {
Add a MCTargetStreamer interface. This patch fixes an old FIXME by creating a MCTargetStreamer interface and moving the target specific functions for ARM, Mips and PPC to it. The ARM streamer is still declared in a common place because it is used from lib/CodeGen/ARMException.cpp, but the Mips and PPC are completely hidden in the corresponding Target directories. I will send an email to llvmdev with instructions on how to use this. llvm-svn: 192181
2013-10-08 21:08:17 +08:00
MipsTargetStreamer &getTargetStreamer() {
Make getTargetStreamer return a possibly null pointer. This will allow it to be called from target independent parts of the main streamer that don't know if there is a registered target streamer or not. This in turn will allow targets to perform extra actions at specified points in the interface: add extra flags for some labels, extra work during finalization, etc. llvm-svn: 199174
2014-01-14 09:21:46 +08:00
MCTargetStreamer &TS = *Parser.getStreamer().getTargetStreamer();
Add a MCTargetStreamer interface. This patch fixes an old FIXME by creating a MCTargetStreamer interface and moving the target specific functions for ARM, Mips and PPC to it. The ARM streamer is still declared in a common place because it is used from lib/CodeGen/ARMException.cpp, but the Mips and PPC are completely hidden in the corresponding Target directories. I will send an email to llvmdev with instructions on how to use this. llvm-svn: 192181
2013-10-08 21:08:17 +08:00
return static_cast<MipsTargetStreamer &>(TS);
}
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
MCSubtargetInfo &STI;
MCAsmParser &Parser;
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165346
2012-10-06 07:55:28 +08:00
MipsAssemblerOptions Options;
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
#define GET_ASSEMBLER_HEADER
#include "MipsGenAsmMatcher.inc"
[mips] Implement jr.hb and jalr.hb (Jump Register and Jump and Link Register with Hazard Barrier). Summary: These instructions are available in ISAs >= mips32/mips64. For mips32r6/mips64r6, jr.hb has a new encoding format. Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D4019 llvm-svn: 210654
2014-06-11 23:05:56 +08:00
unsigned checkTargetMatchPredicate(MCInst &Inst) override;
[ms-inline asm] Remove the MatchInstruction() function. Previously, this was the interface between the front-end and the MC layer when parsing inline assembly. Unfortunately, this is too deep into the parsing stack. Specifically, we're unable to handle target-independent assembly (i.e., assembly directives, labels, etc.). Note the MatchAndEmitInstruction() isn't the correct abstraction either. I'll be exposing target-independent hooks shortly, so this is really just a cleanup. llvm-svn: 165858
2012-10-13 08:26:04 +08:00
bool MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode,
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
OperandVector &Operands, MCStreamer &Out,
unsigned &ErrorInfo,
[C++11] Add 'override' keywords and remove 'virtual'. Additionally add 'final' and leave 'virtual' on some methods that are marked virtual without overriding anything and have no obvious overrides themselves. Mips edition llvm-svn: 207506
2014-04-29 15:58:02 +08:00
bool MatchingInlineAsm) override;
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// Parse a register as used in CFI directives
[C++11] Add 'override' keywords and remove 'virtual'. Additionally add 'final' and leave 'virtual' on some methods that are marked virtual without overriding anything and have no obvious overrides themselves. Mips edition llvm-svn: 207506
2014-04-29 15:58:02 +08:00
bool ParseRegister(unsigned &RegNo, SMLoc &StartLoc, SMLoc &EndLoc) override;
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
bool ParseParenSuffix(StringRef Name, OperandVector &Operands);
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
bool ParseBracketSuffix(StringRef Name, OperandVector &Operands);
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
bool ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
SMLoc NameLoc, OperandVector &Operands) override;
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
[C++11] Add 'override' keywords and remove 'virtual'. Additionally add 'final' and leave 'virtual' on some methods that are marked virtual without overriding anything and have no obvious overrides themselves. Mips edition llvm-svn: 207506
2014-04-29 15:58:02 +08:00
bool ParseDirective(AsmToken DirectiveID) override;
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsAsmParser::OperandMatchResultTy parseMemOperand(OperandVector &Operands);
[mips][msa] Direct Object Emission support for the MSA instruction set. In more detail, this patch adds the ability to parse, encode and decode MSA registers ($w0-$w31). The format of 2RF instructions (MipsMSAInstrFormat.td) was updated so that we could attach a test case to this patch i.e., the test case parses, encodes and decodes 2 MSA instructions. Following patches will add the remainder of the instructions. Note that DecodeMSA128BRegisterClass is missing from MipsDisassembler.td because it's not yet required at this stage and having it would cause a compiler warning (unused function). Patch by Matheus Almeida llvm-svn: 191412
2013-09-26 07:50:44 +08:00
MipsAsmParser::OperandMatchResultTy
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MatchAnyRegisterNameWithoutDollar(OperandVector &Operands,
StringRef Identifier, SMLoc S);
[mips][msa] Direct Object Emission support for the MSA instruction set. In more detail, this patch adds the ability to parse, encode and decode MSA registers ($w0-$w31). The format of 2RF instructions (MipsMSAInstrFormat.td) was updated so that we could attach a test case to this patch i.e., the test case parses, encodes and decodes 2 MSA instructions. Following patches will add the remainder of the instructions. Note that DecodeMSA128BRegisterClass is missing from MipsDisassembler.td because it's not yet required at this stage and having it would cause a compiler warning (unused function). Patch by Matheus Almeida llvm-svn: 191412
2013-09-26 07:50:44 +08:00
MipsAsmParser::OperandMatchResultTy
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MatchAnyRegisterWithoutDollar(OperandVector &Operands, SMLoc S);
[mips][msa] Direct Object Emission support for the MSA instruction set. In more detail, this patch adds the ability to parse, encode and decode MSA registers ($w0-$w31). The format of 2RF instructions (MipsMSAInstrFormat.td) was updated so that we could attach a test case to this patch i.e., the test case parses, encodes and decodes 2 MSA instructions. Following patches will add the remainder of the instructions. Note that DecodeMSA128BRegisterClass is missing from MipsDisassembler.td because it's not yet required at this stage and having it would cause a compiler warning (unused function). Patch by Matheus Almeida llvm-svn: 191412
2013-09-26 07:50:44 +08:00
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsAsmParser::OperandMatchResultTy ParseAnyRegister(OperandVector &Operands);
[mips][msa] Direct Object Emission support for the MSA instruction set. In more detail, this patch adds the ability to parse, encode and decode MSA registers ($w0-$w31). The format of 2RF instructions (MipsMSAInstrFormat.td) was updated so that we could attach a test case to this patch i.e., the test case parses, encodes and decodes 2 MSA instructions. Following patches will add the remainder of the instructions. Note that DecodeMSA128BRegisterClass is missing from MipsDisassembler.td because it's not yet required at this stage and having it would cause a compiler warning (unused function). Patch by Matheus Almeida llvm-svn: 191412
2013-09-26 07:50:44 +08:00
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsAsmParser::OperandMatchResultTy ParseImm(OperandVector &Operands);
[mips][msa] Direct Object Emission support for CTCMSA and CFCMSA. These instructions are logically related as they allow read/write of MSA control registers. Currently MSA control registers are emitted by number but hopefully that will change as soon as GAS starts accepting them by name as that would make the assembly easier to read. llvm-svn: 193078
2013-10-21 20:26:50 +08:00
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsAsmParser::OperandMatchResultTy ParseJumpTarget(OperandVector &Operands);
This patch adds aliases for Mips sub instruction with immediate operands. Corresponding test cases are added. llvm-svn: 191734
2013-10-01 17:48:56 +08:00
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsAsmParser::OperandMatchResultTy parseInvNum(OperandVector &Operands);
MipsAsmParser::OperandMatchResultTy ParseLSAImm(OperandVector &Operands);
[mips][msa] Fix immediate value of LSA instruction as it was being wrongly encoded. The immediate field should be encoded as "imm - 1" as the CPU always adds one to that field. llvm-svn: 195004
2013-11-18 20:32:49 +08:00
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
bool searchSymbolAlias(OperandVector &Operands);
This patch enables the Mips .set directive to define aliases The .set directive in the Mips the assembler can be used to set the value of a symbol to an expression. This changes the symbol's value and type to conform to the expression's. Syntax: .set symbol, expression This patch implements the parsing of the above syntax and enables the parser to use defined symbols when parsing operands. Contributor: Vladimir Medic llvm-svn: 177667
2013-03-22 05:44:16 +08:00
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
bool ParseOperand(OperandVector &, StringRef Mnemonic);
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
bool needsExpansion(MCInst &Inst);
void expandInstruction(MCInst &Inst, SMLoc IDLoc,
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
SmallVectorImpl<MCInst> &Instructions);
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
void expandLoadImm(MCInst &Inst, SMLoc IDLoc,
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
SmallVectorImpl<MCInst> &Instructions);
Initial assembler implementation of Mips load address macro This patch provides initial implementation of load address macro instruction for Mips. We have implemented two kinds of expansions with their variations depending on the size of immediate operand: 1) load address with immediate value directly: * la d,j => addiu d,$zero,j (for -32768 <= j <= 65535) * la d,j => lui d,hi16(j) ori d,d,lo16(j) (for any other 32 bit value of j) 2) load load address with register offset value * la d,j(s) => addiu d,s,j (for -32768 <= j <= 65535) * la d,j(s) => lui d,hi16(j) (for any other 32 bit value of j) ori d,d,lo16(j) addu d,d,s This patch does not cover the case when the address is loaded from the value of the label or function. Contributer: Vladimir Medic llvm-svn: 165561
2012-10-10 07:29:45 +08:00
void expandLoadAddressImm(MCInst &Inst, SMLoc IDLoc,
SmallVectorImpl<MCInst> &Instructions);
void expandLoadAddressReg(MCInst &Inst, SMLoc IDLoc,
SmallVectorImpl<MCInst> &Instructions);
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
void expandMemInst(MCInst &Inst, SMLoc IDLoc,
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
SmallVectorImpl<MCInst> &Instructions, bool isLoad,
bool isImmOpnd);
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
bool reportParseError(StringRef ErrorMsg);
[mips] Add support for .cpload. Summary: This directive is used for setting up $gp in the beginning of a function. It expands to three instructions if PIC is enabled: lui $gp, %hi(_gp_disp) addui $gp, $gp, %lo(_gp_disp) addu $gp, $gp, $reg _gp_disp is a special symbol that the linker sets to the distance between the lui instruction and the context pointer (_gp). Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D3480 llvm-svn: 207637
2014-04-30 19:28:42 +08:00
bool reportParseError(SMLoc Loc, StringRef ErrorMsg);
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
bool parseMemOffset(const MCExpr *&Res, bool isParenExpr);
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
bool parseRelocOperand(const MCExpr *&Res);
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
const MCExpr *evaluateRelocExpr(const MCExpr *Expr, StringRef RelocStr);
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
bool isEvaluated(const MCExpr *Expr);
[mips] Hoist common functionality into a new function. Given that we support multiple directives that enable a particular feature (e.g. '.set mips16'), it's best to hoist that code into a new function so that we don't repeat the same pattern w.r.t parsing and handling error cases. No functional changes. llvm-svn: 204811
2014-03-26 22:26:27 +08:00
bool parseSetFeature(uint64_t Feature);
[mips] Add support for .cpload. Summary: This directive is used for setting up $gp in the beginning of a function. It expands to three instructions if PIC is enabled: lui $gp, %hi(_gp_disp) addui $gp, $gp, %lo(_gp_disp) addu $gp, $gp, $reg _gp_disp is a special symbol that the linker sets to the distance between the lui instruction and the context pointer (_gp). Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D3480 llvm-svn: 207637
2014-04-30 19:28:42 +08:00
bool parseDirectiveCPLoad(SMLoc Loc);
[mips] Add support for .cpsetup Summary: Patch by Robert N. M. Watson His work was sponsored by: DARPA, AFRL Small corrections by myself. CC: theraven, matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3199 llvm-svn: 204924
2014-03-27 21:52:53 +08:00
bool parseDirectiveCPSetup();
[mips] Add initial support for NaN2008 in the back-end. This is so that EF_MIPS_NAN2008 is set if we are using IEEE 754-2008 NaN encoding (-mnan=2008). This patch also adds support for parsing '.nan legacy' and '.nan 2008' assembly directives. The handling of these directives should match GAS' behaviour i.e., the last directive in use sets the ELF header bit (EF_MIPS_NAN2008). Differential Revision: http://reviews.llvm.org/D3346 llvm-svn: 206396
2014-04-16 23:48:55 +08:00
bool parseDirectiveNaN();
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
bool parseDirectiveSet();
[Mips] TargetStreamer Support for .abicalls and .set pic0. This patch adds .abicalls and .set pic0 support which affects the ELF ABI and its flags. In addition the patch uses a common interface for both the MipsTargetSteamer and MipsObjectStreamer that both the integrated and standalone assemblers will use for the output for these directives. llvm-svn: 198646
2014-01-07 07:27:31 +08:00
bool parseDirectiveOption();
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
bool parseSetAtDirective();
bool parseSetNoAtDirective();
bool parseSetMacroDirective();
bool parseSetNoMacroDirective();
bool parseSetReorderDirective();
bool parseSetNoReorderDirective();
[Mips] TargetStreamer Support for .set mips16. This patch updates .set mips16 support which affects the ELF ABI and its flags. In addition the patch uses a common interface for both the MipsTargetSteamer and MipsObjectStreamer that the assembler uses for both ELF and ASCII output for these directives. llvm-svn: 199851
2014-01-23 07:08:42 +08:00
bool parseSetNoMips16Directive();
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
This patch enables the Mips .set directive to define aliases The .set directive in the Mips the assembler can be used to set the value of a symbol to an expression. This changes the symbol's value and type to conform to the expression's. Syntax: .set symbol, expression This patch implements the parsing of the above syntax and enables the parser to use defined symbols when parsing operands. Contributor: Vladimir Medic llvm-svn: 177667
2013-03-22 05:44:16 +08:00
bool parseSetAssignment();
[mips] Rename function in MipsAsmParser. parseDirectiveWord is a generic function that parses an expression which means there's no need for it to have such an specific name. Renaming it to parseDataDirective so that it can also be used to handle .dword directives[1]. [1]To be added in a follow up commit. No functional changes. llvm-svn: 204818
2014-03-26 23:24:36 +08:00
bool parseDataDirective(unsigned Size, SMLoc L);
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
bool parseDirectiveGpWord();
Capitalize the D in parseDirectiveGpDWord. DWord seems to be the canonical way to camel case dword in llvm. Thanks to Daniel Sander for noticing. llvm-svn: 205191
2014-03-31 22:15:07 +08:00
bool parseDirectiveGpDWord();
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
MCSymbolRefExpr::VariantKind getVariantKind(StringRef Symbol);
The Mips standalone assembler fpu instruction support. Test cases included Contributer: Vladimir Medic llvm-svn: 163363
2012-09-07 08:23:42 +08:00
[mips] Some uses of isMips64()/hasMips64() are really tests for 64-bit GPR's Summary: No functional change since these predicates are (currently) synonymous. Extracted from a patch by David Chisnall His work was sponsored by: DARPA, AFRL Differential Revision: http://llvm-reviews.chandlerc.com/D3202 llvm-svn: 204943
2014-03-28 00:42:17 +08:00
bool isGP64() const {
return (STI.getFeatureBits() & Mips::FeatureGP64Bit) != 0;
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
}
The Mips standalone assembler fpu instruction support. Test cases included Contributer: Vladimir Medic llvm-svn: 163363
2012-09-07 08:23:42 +08:00
bool isFP64() const {
return (STI.getFeatureBits() & Mips::FeatureFP64Bit) != 0;
}
[mips] The register names depend on the ABI being N32/N64 rather than the arch being mips64 Summary: Added test cases for O32 and N32 on MIPS64. Reviewers: matheusalmeida Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3175 llvm-svn: 204796
2014-03-26 19:39:07 +08:00
bool isN32() const { return STI.getFeatureBits() & Mips::FeatureN32; }
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
bool isN64() const { return STI.getFeatureBits() & Mips::FeatureN64; }
[mips] Use ptr_rc to simplify definitions of base+index load/store instructions. Also, fix predicates. llvm-svn: 189432
2013-08-28 08:55:15 +08:00
[mips] Add checks for alignment and maximum displacements for most of the branch instructions for mips and micromips instruction sets thus avoiding the situation of generating branches to undesired locations if offsets cannot be encoded. This patch also checks if a fixup cannot be applied and returns a fatal error if that's the case. llvm-svn: 197223
2013-12-13 19:11:02 +08:00
bool isMicroMips() const {
return STI.getFeatureBits() & Mips::FeatureMicroMips;
}
[mips][mips64r6] bc1[tf] are not available on MIPS32r6/MIPS64r6 Summary: Also tightened up the acceptable condition operand for these instructions on MIPS-I to MIPS-III. Support for $fcc[1-7] was added in MIPS-IV. Prior to that only $fcc0 is acceptable. We currently don't optimize (BEQZ (NOT $a), $target) and similar. It's probably best to do this in InstCombine. Depends on D4111 Reviewers: jkolek, zoran.jovanovic, vmedic Reviewed By: vmedic Differential Revision: http://reviews.llvm.org/D4112 llvm-svn: 210787
2014-06-12 23:00:17 +08:00
bool hasMips4() const { return STI.getFeatureBits() & Mips::FeatureMips4; }
bool hasMips32() const { return STI.getFeatureBits() & Mips::FeatureMips32; }
[mips][mips64r6] ssnop is deprecated on MIPS32r6/MIPS64r6 Summary: Depends on D4120 Reviewers: jkolek, zoran.jovanovic, vmedic Reviewed By: zoran.jovanovic, vmedic Differential Revision: http://reviews.llvm.org/D4121 llvm-svn: 211021
2014-06-16 21:25:35 +08:00
bool hasMips32r6() const {
return STI.getFeatureBits() & Mips::FeatureMips32r6;
}
bool hasMips64r6() const {
return STI.getFeatureBits() & Mips::FeatureMips64r6;
}
[mips][mips64r6] bc1[tf] are not available on MIPS32r6/MIPS64r6 Summary: Also tightened up the acceptable condition operand for these instructions on MIPS-I to MIPS-III. Support for $fcc[1-7] was added in MIPS-IV. Prior to that only $fcc0 is acceptable. We currently don't optimize (BEQZ (NOT $a), $target) and similar. It's probably best to do this in InstCombine. Depends on D4111 Reviewers: jkolek, zoran.jovanovic, vmedic Reviewed By: vmedic Differential Revision: http://reviews.llvm.org/D4112 llvm-svn: 210787
2014-06-12 23:00:17 +08:00
[mips] Add support for .cpsetup Summary: Patch by Robert N. M. Watson His work was sponsored by: DARPA, AFRL Small corrections by myself. CC: theraven, matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3199 llvm-svn: 204924
2014-03-27 21:52:53 +08:00
bool eatComma(StringRef ErrorStr);
ELF symbol table field st_other support, excluding visibility bits. Mips specific standalone assembler directive "set at". This directive changes the general purpose register that the assembler will use when given the symbolic register name $at. This does not include negative testing. That will come in a future patch. A side affect of this patch recognizes the different GPR register names for temporaries between old abi and new abi so a test case for that is included. Contributer: Vladimir Medic llvm-svn: 175686
2013-02-21 07:11:17 +08:00
int matchCPURegisterName(StringRef Symbol);
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
int matchRegisterByNumber(unsigned RegNum, unsigned RegClass);
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
int matchFPURegisterName(StringRef Name);
Optimize register parsing for MipsAsmParser. Allow symbolic aliases for FPU registers. llvm-svn: 184411
2013-06-20 19:21:49 +08:00
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
int matchFCCRegisterName(StringRef Name);
The Mips standalone assembler fpu instruction support. Test cases included Contributer: Vladimir Medic llvm-svn: 163363
2012-09-07 08:23:42 +08:00
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
int matchACRegisterName(StringRef Name);
The Mips standalone assembler fpu instruction support. Test cases included Contributer: Vladimir Medic llvm-svn: 163363
2012-09-07 08:23:42 +08:00
[mips][msa] Direct Object Emission support for the MSA instruction set. In more detail, this patch adds the ability to parse, encode and decode MSA registers ($w0-$w31). The format of 2RF instructions (MipsMSAInstrFormat.td) was updated so that we could attach a test case to this patch i.e., the test case parses, encodes and decodes 2 MSA instructions. Following patches will add the remainder of the instructions. Note that DecodeMSA128BRegisterClass is missing from MipsDisassembler.td because it's not yet required at this stage and having it would cause a compiler warning (unused function). Patch by Matheus Almeida llvm-svn: 191412
2013-09-26 07:50:44 +08:00
int matchMSA128RegisterName(StringRef Name);
[mips][msa] Direct Object Emission support for CTCMSA and CFCMSA. These instructions are logically related as they allow read/write of MSA control registers. Currently MSA control registers are emitted by number but hopefully that will change as soon as GAS starts accepting them by name as that would make the assembly easier to read. llvm-svn: 193078
2013-10-21 20:26:50 +08:00
int matchMSA128CtrlRegisterName(StringRef Name);
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
unsigned getReg(int RC, int RegNo);
[ms-inline asm] Add an interface to the GetMCInstOperandNum() function in the MCTargetAsmParser class. llvm-svn: 163122
2012-09-04 02:47:45 +08:00
[mips] Add support for .cpsetup Summary: Patch by Robert N. M. Watson His work was sponsored by: DARPA, AFRL Small corrections by myself. CC: theraven, matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3199 llvm-svn: 204924
2014-03-27 21:52:53 +08:00
unsigned getGPR(int RegNo);
ELF symbol table field st_other support, excluding visibility bits. Mips specific standalone assembler directive "set at". This directive changes the general purpose register that the assembler will use when given the symbolic register name $at. This does not include negative testing. That will come in a future patch. A side affect of this patch recognizes the different GPR register names for temporaries between old abi and new abi so a test case for that is included. Contributer: Vladimir Medic llvm-svn: 175686
2013-02-21 07:11:17 +08:00
int getATReg();
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
bool processInstruction(MCInst &Inst, SMLoc IDLoc,
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
SmallVectorImpl<MCInst> &Instructions);
[mips][msa] Direct Object Emission of INSERT.{B,H,W} instruction. INSERT is the first type of MSA instruction that requires a change to the way MSA registers are parsed. This happens because MSA registers may be suffixed by an index in the form of an immediate or a general purpose register. The changes to parseMSARegs reflect that requirement. llvm-svn: 192582
2013-10-14 19:49:30 +08:00
// Helper function that checks if the value of a vector index is within the
// boundaries of accepted values for each RegisterKind
// Example: INSERT.B $w0[n], $1 => 16 > n >= 0
bool validateMSAIndex(int Val, int RegKind);
This patch implements .set mips32r2 directive and sets appropriate feature bits. It also introduces helper functions that are used to set and clear feature bits as necessary. This directive is a counterpart of -mips32r2 command line options with the exception that it does not influence elf header flags. The usage example is gives in test file. llvm-svn: 202807
2014-03-04 17:54:09 +08:00
void setFeatureBits(unsigned Feature, StringRef FeatureString) {
if (!(STI.getFeatureBits() & Feature)) {
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
setAvailableFeatures(
ComputeAvailableFeatures(STI.ToggleFeature(FeatureString)));
This patch implements .set mips32r2 directive and sets appropriate feature bits. It also introduces helper functions that are used to set and clear feature bits as necessary. This directive is a counterpart of -mips32r2 command line options with the exception that it does not influence elf header flags. The usage example is gives in test file. llvm-svn: 202807
2014-03-04 17:54:09 +08:00
}
}
void clearFeatureBits(unsigned Feature, StringRef FeatureString) {
if (STI.getFeatureBits() & Feature) {
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
setAvailableFeatures(
ComputeAvailableFeatures(STI.ToggleFeature(FeatureString)));
This patch implements .set mips32r2 directive and sets appropriate feature bits. It also introduces helper functions that are used to set and clear feature bits as necessary. This directive is a counterpart of -mips32r2 command line options with the exception that it does not influence elf header flags. The usage example is gives in test file. llvm-svn: 202807
2014-03-04 17:54:09 +08:00
}
}
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
public:
[mips] Implement jr.hb and jalr.hb (Jump Register and Jump and Link Register with Hazard Barrier). Summary: These instructions are available in ISAs >= mips32/mips64. For mips32r6/mips64r6, jr.hb has a new encoding format. Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D4019 llvm-svn: 210654
2014-06-11 23:05:56 +08:00
enum MipsMatchResultTy {
Match_RequiresDifferentSrcAndDst = FIRST_TARGET_MATCH_RESULT_TY
#define GET_OPERAND_DIAGNOSTIC_TYPES
#include "MipsGenAsmMatcher.inc"
#undef GET_OPERAND_DIAGNOSTIC_TYPES
};
Add an instruction deprecation feature to TableGen. The 'Deprecated' class allows you to specify a SubtargetFeature that the instruction is deprecated on. The 'ComplexDeprecationPredicate' class allows you to define a custom predicate that is called to check for deprecation. For example: ComplexDeprecationPredicate<"MCR"> would mean you would have to define the following function: bool getMCRDeprecationInfo(MCInst &MI, MCSubtargetInfo &STI, std::string &Info) Which returns 'false' for not deprecated, and 'true' for deprecated and store the warning message in 'Info'. The MCTargetAsmParser constructor was chaned to take an extra argument of the MCInstrInfo class, so out-of-tree targets will need to be changed. llvm-svn: 190598
2013-09-12 18:28:05 +08:00
MipsAsmParser(MCSubtargetInfo &sti, MCAsmParser &parser,
Create MCTargetOptions. For now it contains a single flag, SanitizeAddress, which enables AddressSanitizer instrumentation of inline assembly. Patch by Yuri Gorshenin. llvm-svn: 206971
2014-04-23 19:16:03 +08:00
const MCInstrInfo &MII,
const MCTargetOptions &Options)
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
: MCTargetAsmParser(), STI(sti), Parser(parser) {
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
// Initialize the set of available features.
setAvailableFeatures(ComputeAvailableFeatures(STI.getFeatureBits()));
[mips] Make it impossible to have UnknownABI in CodeGen and Integrated Assembler. Summary: This removes the need to coerce UnknownABI to the default ABI (O32 for MIPS32, N64 for MIPS64 [*]) in both MipsSubtarget and MipsAsmParser. Clang has been updated to disable both possible default ABI's before enabling the ABI it intends to use. [*] N64 being the default for MIPS64 is not actually correct. However N32 is not fully implemented/tested yet. Depends on: D2830 Reviewers: jacksprat, matheusalmeida Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D2832 Differential Revision: http://llvm-reviews.chandlerc.com/D2846 llvm-svn: 201792
2014-02-20 22:58:19 +08:00
// Assert exactly one ABI was chosen.
assert((((STI.getFeatureBits() & Mips::FeatureO32) != 0) +
((STI.getFeatureBits() & Mips::FeatureEABI) != 0) +
((STI.getFeatureBits() & Mips::FeatureN32) != 0) +
((STI.getFeatureBits() & Mips::FeatureN64) != 0)) == 1);
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
}
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
MCAsmParser &getParser() const { return Parser; }
MCAsmLexer &getLexer() const { return Parser.getLexer(); }
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
[mips][mips64r6] bc1[tf] are not available on MIPS32r6/MIPS64r6 Summary: Also tightened up the acceptable condition operand for these instructions on MIPS-I to MIPS-III. Support for $fcc[1-7] was added in MIPS-IV. Prior to that only $fcc0 is acceptable. We currently don't optimize (BEQZ (NOT $a), $target) and similar. It's probably best to do this in InstCombine. Depends on D4111 Reviewers: jkolek, zoran.jovanovic, vmedic Reviewed By: vmedic Differential Revision: http://reviews.llvm.org/D4112 llvm-svn: 210787
2014-06-12 23:00:17 +08:00
/// True if all of $fcc0 - $fcc7 exist for the current ISA.
bool hasEightFccRegisters() const { return hasMips4() || hasMips32(); }
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// Warn if RegNo is the current assembler temporary.
void WarnIfAssemblerTemporary(int RegNo, SMLoc Loc);
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
};
}
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
namespace {
/// MipsOperand - Instances of this class represent a parsed Mips machine
/// instruction.
class MipsOperand : public MCParsedAsmOperand {
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
public:
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// Broad categories of register classes
/// The exact class is finalized by the render method.
enum RegKind {
RegKind_GPR = 1, /// GPR32 and GPR64 (depending on isGP64())
RegKind_FGR = 2, /// FGR32, FGR64, AFGR64 (depending on context and
/// isFP64())
RegKind_FCC = 4, /// FCC
RegKind_MSA128 = 8, /// MSA128[BHWD] (makes no difference which)
RegKind_MSACtrl = 16, /// MSA control registers
RegKind_COP2 = 32, /// COP2
RegKind_ACC = 64, /// HI32DSP, LO32DSP, and ACC64DSP (depending on
/// context).
RegKind_CCR = 128, /// CCR
RegKind_HWRegs = 256, /// HWRegs
[mips] Implement l[wd]c3, and s[wd]c3. Summary: These instructions were added in MIPS-I, and MIPS-II but were removed in MIPS-III. Interestingly, GAS continues to accept them when assembling for MIPS-III. For the moment, these instructions will follow GAS and accept them for MIPS-III and newer but this will be tightened up when the invalid-*.s tests are added. Depends on D3647 Reviewers: vmedic Reviewed By: vmedic Differential Revision: http://reviews.llvm.org/D3648 llvm-svn: 208311
2014-05-08 21:02:11 +08:00
RegKind_COP3 = 512, /// COP3
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// Potentially any (e.g. $1)
RegKind_Numeric = RegKind_GPR | RegKind_FGR | RegKind_FCC | RegKind_MSA128 |
RegKind_MSACtrl | RegKind_COP2 | RegKind_ACC |
[mips] Implement l[wd]c3, and s[wd]c3. Summary: These instructions were added in MIPS-I, and MIPS-II but were removed in MIPS-III. Interestingly, GAS continues to accept them when assembling for MIPS-III. For the moment, these instructions will follow GAS and accept them for MIPS-III and newer but this will be tightened up when the invalid-*.s tests are added. Depends on D3647 Reviewers: vmedic Reviewed By: vmedic Differential Revision: http://reviews.llvm.org/D3648 llvm-svn: 208311
2014-05-08 21:02:11 +08:00
RegKind_CCR | RegKind_HWRegs | RegKind_COP3
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
};
private:
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
enum KindTy {
[mips] Renamed ParseAnyRegisterWithoutDollar to MatchAnyRegisterWithoutDollar This is for consistency with other functions. The Parse* functions consume tokens and the Match* functions don't. No functional change. llvm-svn: 205305
2014-04-01 20:35:23 +08:00
k_Immediate, /// An immediate (possibly involving symbol references)
k_Memory, /// Base + Offset Memory Address
k_PhysRegister, /// A physical register from the Mips namespace
k_RegisterIndex, /// A register index in one or more RegKind.
k_Token /// A simple token
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
} Kind;
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
public:
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
MipsOperand(KindTy K, MipsAsmParser &Parser)
: MCParsedAsmOperand(), Kind(K), AsmParser(Parser) {}
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
private:
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// For diagnostics, and checking the assembler temporary
MipsAsmParser &AsmParser;
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
Silence anonymous type in anonymous union warnings. llvm-svn: 177135
2013-03-15 08:42:55 +08:00
struct Token {
const char *Data;
unsigned Length;
};
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
struct PhysRegOp {
unsigned Num; /// Register Number
};
struct RegIdxOp {
unsigned Index; /// Index into the register class
RegKind Kind; /// Bitfield of the kinds it could possibly be
const MCRegisterInfo *RegInfo;
Silence anonymous type in anonymous union warnings. llvm-svn: 177135
2013-03-15 08:42:55 +08:00
};
struct ImmOp {
const MCExpr *Val;
};
struct MemOp {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
MipsOperand *Base;
Silence anonymous type in anonymous union warnings. llvm-svn: 177135
2013-03-15 08:42:55 +08:00
const MCExpr *Off;
};
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
union {
Silence anonymous type in anonymous union warnings. llvm-svn: 177135
2013-03-15 08:42:55 +08:00
struct Token Tok;
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
struct PhysRegOp PhysReg;
struct RegIdxOp RegIdx;
Silence anonymous type in anonymous union warnings. llvm-svn: 177135
2013-03-15 08:42:55 +08:00
struct ImmOp Imm;
struct MemOp Mem;
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
};
SMLoc StartLoc, EndLoc;
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// Internal constructor for register kinds
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
static std::unique_ptr<MipsOperand> CreateReg(unsigned Index, RegKind RegKind,
const MCRegisterInfo *RegInfo,
SMLoc S, SMLoc E,
MipsAsmParser &Parser) {
auto Op = make_unique<MipsOperand>(k_RegisterIndex, Parser);
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Op->RegIdx.Index = Index;
Op->RegIdx.RegInfo = RegInfo;
Op->RegIdx.Kind = RegKind;
Op->StartLoc = S;
Op->EndLoc = E;
return Op;
}
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
public:
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// Coerce the register to GPR32 and return the real register for the current
/// target.
unsigned getGPR32Reg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_GPR) && "Invalid access!");
AsmParser.WarnIfAssemblerTemporary(RegIdx.Index, StartLoc);
unsigned ClassID = Mips::GPR32RegClassID;
return RegIdx.RegInfo->getRegClass(ClassID).getRegister(RegIdx.Index);
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205229
2014-04-01 01:43:46 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// Coerce the register to GPR64 and return the real register for the current
/// target.
unsigned getGPR64Reg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_GPR) && "Invalid access!");
unsigned ClassID = Mips::GPR64RegClassID;
return RegIdx.RegInfo->getRegClass(ClassID).getRegister(RegIdx.Index);
}
private:
/// Coerce the register to AFGR64 and return the real register for the current
/// target.
unsigned getAFGR64Reg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_FGR) && "Invalid access!");
if (RegIdx.Index % 2 != 0)
AsmParser.Warning(StartLoc, "Float register should be even.");
return RegIdx.RegInfo->getRegClass(Mips::AFGR64RegClassID)
.getRegister(RegIdx.Index / 2);
}
/// Coerce the register to FGR64 and return the real register for the current
/// target.
unsigned getFGR64Reg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_FGR) && "Invalid access!");
return RegIdx.RegInfo->getRegClass(Mips::FGR64RegClassID)
.getRegister(RegIdx.Index);
}
/// Coerce the register to FGR32 and return the real register for the current
/// target.
unsigned getFGR32Reg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_FGR) && "Invalid access!");
return RegIdx.RegInfo->getRegClass(Mips::FGR32RegClassID)
.getRegister(RegIdx.Index);
}
/// Coerce the register to FGRH32 and return the real register for the current
/// target.
unsigned getFGRH32Reg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_FGR) && "Invalid access!");
return RegIdx.RegInfo->getRegClass(Mips::FGRH32RegClassID)
.getRegister(RegIdx.Index);
}
/// Coerce the register to FCC and return the real register for the current
/// target.
unsigned getFCCReg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_FCC) && "Invalid access!");
return RegIdx.RegInfo->getRegClass(Mips::FCCRegClassID)
.getRegister(RegIdx.Index);
}
/// Coerce the register to MSA128 and return the real register for the current
/// target.
unsigned getMSA128Reg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_MSA128) && "Invalid access!");
// It doesn't matter which of the MSA128[BHWD] classes we use. They are all
// identical
unsigned ClassID = Mips::MSA128BRegClassID;
return RegIdx.RegInfo->getRegClass(ClassID).getRegister(RegIdx.Index);
}
/// Coerce the register to MSACtrl and return the real register for the
/// current target.
unsigned getMSACtrlReg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_MSACtrl) && "Invalid access!");
unsigned ClassID = Mips::MSACtrlRegClassID;
return RegIdx.RegInfo->getRegClass(ClassID).getRegister(RegIdx.Index);
}
/// Coerce the register to COP2 and return the real register for the
/// current target.
unsigned getCOP2Reg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_COP2) && "Invalid access!");
unsigned ClassID = Mips::COP2RegClassID;
return RegIdx.RegInfo->getRegClass(ClassID).getRegister(RegIdx.Index);
}
[mips] Implement l[wd]c3, and s[wd]c3. Summary: These instructions were added in MIPS-I, and MIPS-II but were removed in MIPS-III. Interestingly, GAS continues to accept them when assembling for MIPS-III. For the moment, these instructions will follow GAS and accept them for MIPS-III and newer but this will be tightened up when the invalid-*.s tests are added. Depends on D3647 Reviewers: vmedic Reviewed By: vmedic Differential Revision: http://reviews.llvm.org/D3648 llvm-svn: 208311
2014-05-08 21:02:11 +08:00
/// Coerce the register to COP3 and return the real register for the
/// current target.
unsigned getCOP3Reg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_COP3) && "Invalid access!");
unsigned ClassID = Mips::COP3RegClassID;
return RegIdx.RegInfo->getRegClass(ClassID).getRegister(RegIdx.Index);
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// Coerce the register to ACC64DSP and return the real register for the
/// current target.
unsigned getACC64DSPReg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_ACC) && "Invalid access!");
unsigned ClassID = Mips::ACC64DSPRegClassID;
return RegIdx.RegInfo->getRegClass(ClassID).getRegister(RegIdx.Index);
}
/// Coerce the register to HI32DSP and return the real register for the
/// current target.
unsigned getHI32DSPReg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_ACC) && "Invalid access!");
unsigned ClassID = Mips::HI32DSPRegClassID;
return RegIdx.RegInfo->getRegClass(ClassID).getRegister(RegIdx.Index);
}
/// Coerce the register to LO32DSP and return the real register for the
/// current target.
unsigned getLO32DSPReg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_ACC) && "Invalid access!");
unsigned ClassID = Mips::LO32DSPRegClassID;
return RegIdx.RegInfo->getRegClass(ClassID).getRegister(RegIdx.Index);
}
/// Coerce the register to CCR and return the real register for the
/// current target.
unsigned getCCRReg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_CCR) && "Invalid access!");
unsigned ClassID = Mips::CCRRegClassID;
return RegIdx.RegInfo->getRegClass(ClassID).getRegister(RegIdx.Index);
}
/// Coerce the register to HWRegs and return the real register for the
/// current target.
unsigned getHWRegsReg() const {
assert(isRegIdx() && (RegIdx.Kind & RegKind_HWRegs) && "Invalid access!");
unsigned ClassID = Mips::HWRegsRegClassID;
return RegIdx.RegInfo->getRegClass(ClassID).getRegister(RegIdx.Index);
[mips] Use ptr_rc to simplify definitions of base+index load/store instructions. Also, fix predicates. llvm-svn: 189432
2013-08-28 08:55:15 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
public:
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
void addExpr(MCInst &Inst, const MCExpr *Expr) const {
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
// Add as immediate when possible. Null MCExpr = 0.
[C++] Use 'nullptr'. Target edition. llvm-svn: 207197
2014-04-25 13:30:21 +08:00
if (!Expr)
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
Inst.addOperand(MCOperand::CreateImm(0));
else if (const MCConstantExpr *CE = dyn_cast<MCConstantExpr>(Expr))
Inst.addOperand(MCOperand::CreateImm(CE->getValue()));
else
Inst.addOperand(MCOperand::CreateExpr(Expr));
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
}
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
void addRegOperands(MCInst &Inst, unsigned N) const {
llvm_unreachable("Use a custom parser instead");
}
[mips] Renamed ParseAnyRegisterWithoutDollar to MatchAnyRegisterWithoutDollar This is for consistency with other functions. The Parse* functions consume tokens and the Match* functions don't. No functional change. llvm-svn: 205305
2014-04-01 20:35:23 +08:00
/// Render the operand to an MCInst as a GPR32
/// Asserts if the wrong number of operands are requested, or the operand
/// is not a k_RegisterIndex compatible with RegKind_GPR
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
void addGPR32AsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getGPR32Reg()));
}
[mips] Renamed ParseAnyRegisterWithoutDollar to MatchAnyRegisterWithoutDollar This is for consistency with other functions. The Parse* functions consume tokens and the Match* functions don't. No functional change. llvm-svn: 205305
2014-04-01 20:35:23 +08:00
/// Render the operand to an MCInst as a GPR64
/// Asserts if the wrong number of operands are requested, or the operand
/// is not a k_RegisterIndex compatible with RegKind_GPR
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
void addGPR64AsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getGPR64Reg()));
}
void addAFGR64AsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getAFGR64Reg()));
}
void addFGR64AsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getFGR64Reg()));
}
void addFGR32AsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getFGR32Reg()));
}
void addFGRH32AsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getFGRH32Reg()));
}
void addFCCAsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getFCCReg()));
}
void addMSA128AsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getMSA128Reg()));
}
void addMSACtrlAsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getMSACtrlReg()));
}
void addCOP2AsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getCOP2Reg()));
}
[mips] Implement l[wd]c3, and s[wd]c3. Summary: These instructions were added in MIPS-I, and MIPS-II but were removed in MIPS-III. Interestingly, GAS continues to accept them when assembling for MIPS-III. For the moment, these instructions will follow GAS and accept them for MIPS-III and newer but this will be tightened up when the invalid-*.s tests are added. Depends on D3647 Reviewers: vmedic Reviewed By: vmedic Differential Revision: http://reviews.llvm.org/D3648 llvm-svn: 208311
2014-05-08 21:02:11 +08:00
void addCOP3AsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getCOP3Reg()));
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
void addACC64DSPAsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getACC64DSPReg()));
}
void addHI32DSPAsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getHI32DSPReg()));
}
void addLO32DSPAsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getLO32DSPReg()));
}
void addCCRAsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getCCRReg()));
}
void addHWRegsAsmRegOperands(MCInst &Inst, unsigned N) const {
assert(N == 1 && "Invalid number of operands!");
Inst.addOperand(MCOperand::CreateReg(getHWRegsReg()));
}
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
void addImmOperands(MCInst &Inst, unsigned N) const {
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
assert(N == 1 && "Invalid number of operands!");
const MCExpr *Expr = getImm();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
addExpr(Inst, Expr);
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
}
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
void addMemOperands(MCInst &Inst, unsigned N) const {
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
assert(N == 2 && "Invalid number of operands!");
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Inst.addOperand(MCOperand::CreateReg(getMemBase()->getGPR32Reg()));
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
const MCExpr *Expr = getMemOff();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
addExpr(Inst, Expr);
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
}
[C++11] Add 'override' keywords and remove 'virtual'. Additionally add 'final' and leave 'virtual' on some methods that are marked virtual without overriding anything and have no obvious overrides themselves. Mips edition llvm-svn: 207506
2014-04-29 15:58:02 +08:00
bool isReg() const override {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
// As a special case until we sort out the definition of div/divu, pretend
// that $0/$zero are k_PhysRegister so that MCK_ZERO works correctly.
if (isGPRAsmReg() && RegIdx.Index == 0)
return true;
return Kind == k_PhysRegister;
}
bool isRegIdx() const { return Kind == k_RegisterIndex; }
[C++11] Add 'override' keywords and remove 'virtual'. Additionally add 'final' and leave 'virtual' on some methods that are marked virtual without overriding anything and have no obvious overrides themselves. Mips edition llvm-svn: 207506
2014-04-29 15:58:02 +08:00
bool isImm() const override { return Kind == k_Immediate; }
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
bool isConstantImm() const {
return isImm() && dyn_cast<MCConstantExpr>(getImm());
}
[C++11] Add 'override' keywords and remove 'virtual'. Additionally add 'final' and leave 'virtual' on some methods that are marked virtual without overriding anything and have no obvious overrides themselves. Mips edition llvm-svn: 207506
2014-04-29 15:58:02 +08:00
bool isToken() const override {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
// Note: It's not possible to pretend that other operand kinds are tokens.
// The matcher emitter checks tokens first.
return Kind == k_Token;
}
[C++11] Add 'override' keywords and remove 'virtual'. Additionally add 'final' and leave 'virtual' on some methods that are marked virtual without overriding anything and have no obvious overrides themselves. Mips edition llvm-svn: 207506
2014-04-29 15:58:02 +08:00
bool isMem() const override { return Kind == k_Memory; }
[mips][mips64r6] [ls][wd]c2 were re-encoded with 11-bit signed immediates rather than 16-bit in MIPS32r6/MIPS64r6 Summary: The error message for the invalid.s cases isn't very helpful. It happens because there is an instruction with a wider immediate that would have matched if the NotMips32r6 predicate were true. I have some WIP to improve the message but it affects most error messages for removed/re-encoded instructions on MIPS32r6/MIPS64r6 and should therefore be a separate commit. Depens on D4115 Reviewers: zoran.jovanovic, jkolek, vmedic Reviewed By: vmedic Differential Revision: http://reviews.llvm.org/D4117 llvm-svn: 211012
2014-06-16 18:00:45 +08:00
bool isConstantMemOff() const {
return isMem() && dyn_cast<MCConstantExpr>(getMemOff());
}
template <unsigned Bits> bool isMemWithSimmOffset() const {
return isMem() && isConstantMemOff() && isInt<Bits>(getConstantMemOff());
}
This patch adds aliases for Mips sub instruction with immediate operands. Corresponding test cases are added. llvm-svn: 191734
2013-10-01 17:48:56 +08:00
bool isInvNum() const { return Kind == k_Immediate; }
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
bool isLSAImm() const {
if (!isConstantImm())
return false;
int64_t Val = getConstantImm();
return 1 <= Val && Val <= 4;
}
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
StringRef getToken() const {
assert(Kind == k_Token && "Invalid access!");
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
return StringRef(Tok.Data, Tok.Length);
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
}
[C++11] Add 'override' keywords and remove 'virtual'. Additionally add 'final' and leave 'virtual' on some methods that are marked virtual without overriding anything and have no obvious overrides themselves. Mips edition llvm-svn: 207506
2014-04-29 15:58:02 +08:00
unsigned getReg() const override {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
// As a special case until we sort out the definition of div/divu, pretend
// that $0/$zero are k_PhysRegister so that MCK_ZERO works correctly.
if (Kind == k_RegisterIndex && RegIdx.Index == 0 &&
RegIdx.Kind & RegKind_GPR)
return getGPR32Reg(); // FIXME: GPR64 too
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
assert(Kind == k_PhysRegister && "Invalid access!");
return PhysReg.Num;
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
}
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
const MCExpr *getImm() const {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
assert((Kind == k_Immediate) && "Invalid access!");
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
return Imm.Val;
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205229
2014-04-01 01:43:46 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
int64_t getConstantImm() const {
const MCExpr *Val = getImm();
return static_cast<const MCConstantExpr *>(Val)->getValue();
}
MipsOperand *getMemBase() const {
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
assert((Kind == k_Memory) && "Invalid access!");
return Mem.Base;
}
const MCExpr *getMemOff() const {
assert((Kind == k_Memory) && "Invalid access!");
return Mem.Off;
}
[mips][mips64r6] [ls][wd]c2 were re-encoded with 11-bit signed immediates rather than 16-bit in MIPS32r6/MIPS64r6 Summary: The error message for the invalid.s cases isn't very helpful. It happens because there is an instruction with a wider immediate that would have matched if the NotMips32r6 predicate were true. I have some WIP to improve the message but it affects most error messages for removed/re-encoded instructions on MIPS32r6/MIPS64r6 and should therefore be a separate commit. Depens on D4115 Reviewers: zoran.jovanovic, jkolek, vmedic Reviewed By: vmedic Differential Revision: http://reviews.llvm.org/D4117 llvm-svn: 211012
2014-06-16 18:00:45 +08:00
int64_t getConstantMemOff() const {
return static_cast<const MCConstantExpr *>(getMemOff())->getValue();
}
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
static std::unique_ptr<MipsOperand> CreateToken(StringRef Str, SMLoc S,
MipsAsmParser &Parser) {
auto Op = make_unique<MipsOperand>(k_Token, Parser);
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
Op->Tok.Data = Str.data();
Op->Tok.Length = Str.size();
Op->StartLoc = S;
Op->EndLoc = S;
return Op;
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// Create a numeric register (e.g. $1). The exact register remains
/// unresolved until an instruction successfully matches
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
static std::unique_ptr<MipsOperand>
CreateNumericReg(unsigned Index, const MCRegisterInfo *RegInfo, SMLoc S,
SMLoc E, MipsAsmParser &Parser) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
DEBUG(dbgs() << "CreateNumericReg(" << Index << ", ...)\n");
return CreateReg(Index, RegKind_Numeric, RegInfo, S, E, Parser);
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// Create a register that is definitely a GPR.
/// This is typically only used for named registers such as $gp.
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
static std::unique_ptr<MipsOperand>
CreateGPRReg(unsigned Index, const MCRegisterInfo *RegInfo, SMLoc S, SMLoc E,
MipsAsmParser &Parser) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
return CreateReg(Index, RegKind_GPR, RegInfo, S, E, Parser);
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205229
2014-04-01 01:43:46 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// Create a register that is definitely a FGR.
/// This is typically only used for named registers such as $f0.
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
static std::unique_ptr<MipsOperand>
CreateFGRReg(unsigned Index, const MCRegisterInfo *RegInfo, SMLoc S, SMLoc E,
MipsAsmParser &Parser) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
return CreateReg(Index, RegKind_FGR, RegInfo, S, E, Parser);
}
/// Create a register that is definitely an FCC.
/// This is typically only used for named registers such as $fcc0.
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
static std::unique_ptr<MipsOperand>
CreateFCCReg(unsigned Index, const MCRegisterInfo *RegInfo, SMLoc S, SMLoc E,
MipsAsmParser &Parser) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
return CreateReg(Index, RegKind_FCC, RegInfo, S, E, Parser);
}
/// Create a register that is definitely an ACC.
/// This is typically only used for named registers such as $ac0.
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
static std::unique_ptr<MipsOperand>
CreateACCReg(unsigned Index, const MCRegisterInfo *RegInfo, SMLoc S, SMLoc E,
MipsAsmParser &Parser) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
return CreateReg(Index, RegKind_ACC, RegInfo, S, E, Parser);
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205229
2014-04-01 01:43:46 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// Create a register that is definitely an MSA128.
/// This is typically only used for named registers such as $w0.
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
static std::unique_ptr<MipsOperand>
CreateMSA128Reg(unsigned Index, const MCRegisterInfo *RegInfo, SMLoc S,
SMLoc E, MipsAsmParser &Parser) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
return CreateReg(Index, RegKind_MSA128, RegInfo, S, E, Parser);
}
/// Create a register that is definitely an MSACtrl.
/// This is typically only used for named registers such as $msaaccess.
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
static std::unique_ptr<MipsOperand>
CreateMSACtrlReg(unsigned Index, const MCRegisterInfo *RegInfo, SMLoc S,
SMLoc E, MipsAsmParser &Parser) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
return CreateReg(Index, RegKind_MSACtrl, RegInfo, S, E, Parser);
}
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
static std::unique_ptr<MipsOperand>
CreateImm(const MCExpr *Val, SMLoc S, SMLoc E, MipsAsmParser &Parser) {
auto Op = make_unique<MipsOperand>(k_Immediate, Parser);
[mips][msa] Fix immediate value of LSA instruction as it was being wrongly encoded. The immediate field should be encoded as "imm - 1" as the CPU always adds one to that field. llvm-svn: 195004
2013-11-18 20:32:49 +08:00
Op->Imm.Val = Val;
Op->StartLoc = S;
Op->EndLoc = E;
return Op;
}
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
static std::unique_ptr<MipsOperand>
CreateMem(std::unique_ptr<MipsOperand> Base, const MCExpr *Off, SMLoc S,
SMLoc E, MipsAsmParser &Parser) {
auto Op = make_unique<MipsOperand>(k_Memory, Parser);
Op->Mem.Base = Base.release();
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
Op->Mem.Off = Off;
Op->StartLoc = S;
Op->EndLoc = E;
return Op;
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
bool isGPRAsmReg() const {
return isRegIdx() && RegIdx.Kind & RegKind_GPR && RegIdx.Index <= 31;
This patch implements parsing of mips FCC register operands. The example instructions have been added to test files. llvm-svn: 187410
2013-07-30 18:12:14 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
bool isFGRAsmReg() const {
// AFGR64 is $0-$15 but we handle this in getAFGR64()
return isRegIdx() && RegIdx.Kind & RegKind_FGR && RegIdx.Index <= 31;
[mips] Replace usages of register classes with register operands. Also, remove unnecessary jalr InstAliases in Mips64InstrInfo.td and add the code to print jalr InstAliases in MipsInstPrinter::printAlias. llvm-svn: 187821
2013-08-07 06:20:40 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
bool isHWRegsAsmReg() const {
return isRegIdx() && RegIdx.Kind & RegKind_HWRegs && RegIdx.Index <= 31;
[mips] Use register operands instead of register classes in DSP instruction definitions. llvm-svn: 188343
2013-08-14 09:02:20 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
bool isCCRAsmReg() const {
return isRegIdx() && RegIdx.Kind & RegKind_CCR && RegIdx.Index <= 31;
[mips] Use register operands instead of register classes in DSP instruction definitions. llvm-svn: 188343
2013-08-14 09:02:20 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
bool isFCCAsmReg() const {
[mips][mips64r6] bc1[tf] are not available on MIPS32r6/MIPS64r6 Summary: Also tightened up the acceptable condition operand for these instructions on MIPS-I to MIPS-III. Support for $fcc[1-7] was added in MIPS-IV. Prior to that only $fcc0 is acceptable. We currently don't optimize (BEQZ (NOT $a), $target) and similar. It's probably best to do this in InstCombine. Depends on D4111 Reviewers: jkolek, zoran.jovanovic, vmedic Reviewed By: vmedic Differential Revision: http://reviews.llvm.org/D4112 llvm-svn: 210787
2014-06-12 23:00:17 +08:00
if (!(isRegIdx() && RegIdx.Kind & RegKind_FCC))
return false;
if (!AsmParser.hasEightFccRegisters())
return RegIdx.Index == 0;
return RegIdx.Index <= 7;
[mips][msa] Direct Object Emission support for the MSA instruction set. In more detail, this patch adds the ability to parse, encode and decode MSA registers ($w0-$w31). The format of 2RF instructions (MipsMSAInstrFormat.td) was updated so that we could attach a test case to this patch i.e., the test case parses, encodes and decodes 2 MSA instructions. Following patches will add the remainder of the instructions. Note that DecodeMSA128BRegisterClass is missing from MipsDisassembler.td because it's not yet required at this stage and having it would cause a compiler warning (unused function). Patch by Matheus Almeida llvm-svn: 191412
2013-09-26 07:50:44 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
bool isACCAsmReg() const {
return isRegIdx() && RegIdx.Kind & RegKind_ACC && RegIdx.Index <= 3;
[mips][msa] Direct Object Emission support for the MSA instruction set. In more detail, this patch adds the ability to parse, encode and decode MSA registers ($w0-$w31). The format of 2RF instructions (MipsMSAInstrFormat.td) was updated so that we could attach a test case to this patch i.e., the test case parses, encodes and decodes 2 MSA instructions. Following patches will add the remainder of the instructions. Note that DecodeMSA128BRegisterClass is missing from MipsDisassembler.td because it's not yet required at this stage and having it would cause a compiler warning (unused function). Patch by Matheus Almeida llvm-svn: 191412
2013-09-26 07:50:44 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
bool isCOP2AsmReg() const {
return isRegIdx() && RegIdx.Kind & RegKind_COP2 && RegIdx.Index <= 31;
[mips][msa] Direct Object Emission support for the MSA instruction set. In more detail, this patch adds the ability to parse, encode and decode MSA registers ($w0-$w31). The format of 2RF instructions (MipsMSAInstrFormat.td) was updated so that we could attach a test case to this patch i.e., the test case parses, encodes and decodes 2 MSA instructions. Following patches will add the remainder of the instructions. Note that DecodeMSA128BRegisterClass is missing from MipsDisassembler.td because it's not yet required at this stage and having it would cause a compiler warning (unused function). Patch by Matheus Almeida llvm-svn: 191412
2013-09-26 07:50:44 +08:00
}
[mips] Implement l[wd]c3, and s[wd]c3. Summary: These instructions were added in MIPS-I, and MIPS-II but were removed in MIPS-III. Interestingly, GAS continues to accept them when assembling for MIPS-III. For the moment, these instructions will follow GAS and accept them for MIPS-III and newer but this will be tightened up when the invalid-*.s tests are added. Depends on D3647 Reviewers: vmedic Reviewed By: vmedic Differential Revision: http://reviews.llvm.org/D3648 llvm-svn: 208311
2014-05-08 21:02:11 +08:00
bool isCOP3AsmReg() const {
return isRegIdx() && RegIdx.Kind & RegKind_COP3 && RegIdx.Index <= 31;
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
bool isMSA128AsmReg() const {
return isRegIdx() && RegIdx.Kind & RegKind_MSA128 && RegIdx.Index <= 31;
[mips][msa] Direct Object Emission support for the MSA instruction set. In more detail, this patch adds the ability to parse, encode and decode MSA registers ($w0-$w31). The format of 2RF instructions (MipsMSAInstrFormat.td) was updated so that we could attach a test case to this patch i.e., the test case parses, encodes and decodes 2 MSA instructions. Following patches will add the remainder of the instructions. Note that DecodeMSA128BRegisterClass is missing from MipsDisassembler.td because it's not yet required at this stage and having it would cause a compiler warning (unused function). Patch by Matheus Almeida llvm-svn: 191412
2013-09-26 07:50:44 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
bool isMSACtrlAsmReg() const {
return isRegIdx() && RegIdx.Kind & RegKind_MSACtrl && RegIdx.Index <= 7;
[mips][msa] Direct Object Emission support for CTCMSA and CFCMSA. These instructions are logically related as they allow read/write of MSA control registers. Currently MSA control registers are emitted by number but hopefully that will change as soon as GAS starts accepting them by name as that would make the assembly easier to read. llvm-svn: 193078
2013-10-21 20:26:50 +08:00
}
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
/// getStartLoc - Get the location of the first token of this operand.
[C++11] Add 'override' keywords and remove 'virtual'. Additionally add 'final' and leave 'virtual' on some methods that are marked virtual without overriding anything and have no obvious overrides themselves. Mips edition llvm-svn: 207506
2014-04-29 15:58:02 +08:00
SMLoc getStartLoc() const override { return StartLoc; }
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
/// getEndLoc - Get the location of the last token of this operand.
[C++11] Add 'override' keywords and remove 'virtual'. Additionally add 'final' and leave 'virtual' on some methods that are marked virtual without overriding anything and have no obvious overrides themselves. Mips edition llvm-svn: 207506
2014-04-29 15:58:02 +08:00
SMLoc getEndLoc() const override { return EndLoc; }
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
MipsAsmParser.cpp: Fix vg_leak in MipsOperand::CreateMem(). Mem.Base is managed by k_Memory itself. llvm-svn: 206293
2014-04-15 22:13:21 +08:00
virtual ~MipsOperand() {
switch (Kind) {
case k_Immediate:
break;
case k_Memory:
delete Mem.Base;
break;
case k_PhysRegister:
case k_RegisterIndex:
case k_Token:
break;
}
}
[C++11] Add 'override' keywords and remove 'virtual'. Additionally add 'final' and leave 'virtual' on some methods that are marked virtual without overriding anything and have no obvious overrides themselves. Mips edition llvm-svn: 207506
2014-04-29 15:58:02 +08:00
void print(raw_ostream &OS) const override {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
switch (Kind) {
case k_Immediate:
OS << "Imm<";
Imm.Val->print(OS);
OS << ">";
break;
case k_Memory:
OS << "Mem<";
Mem.Base->print(OS);
OS << ", ";
Mem.Off->print(OS);
OS << ">";
break;
case k_PhysRegister:
OS << "PhysReg<" << PhysReg.Num << ">";
break;
case k_RegisterIndex:
OS << "RegIdx<" << RegIdx.Index << ":" << RegIdx.Kind << ">";
break;
case k_Token:
OS << Tok.Data;
break;
}
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
}; // class MipsOperand
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
} // namespace
Add stub methods for mips assembly matcher. Patch by Vladimir Medic. llvm-svn: 162124
2012-08-18 04:16:42 +08:00
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
namespace llvm {
extern const MCInstrDesc MipsInsts[];
}
static const MCInstrDesc &getInstDesc(unsigned Opcode) {
return MipsInsts[Opcode];
}
bool MipsAsmParser::processInstruction(MCInst &Inst, SMLoc IDLoc,
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
SmallVectorImpl<MCInst> &Instructions) {
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
const MCInstrDesc &MCID = getInstDesc(Inst.getOpcode());
[mips] Implement shorthand add / sub forms for MIPS. Summary: - If only two registers are passed to a three-register operation, then the first argument is both source and destination register. - If a non-register is passed as the last argument, generate the immediate version of the instruction. Also mark DADD commutative and add scheduling information (to the generic scheduler), and implement DSUB. Patch by David Chisnall His work was sponsored by: DARPA, AFRL CC: theraven Differential Revision: http://llvm-reviews.chandlerc.com/D3148 llvm-svn: 204605
2014-03-24 22:05:39 +08:00
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
Inst.setLoc(IDLoc);
[mips] Add checks for alignment and maximum displacements for most of the branch instructions for mips and micromips instruction sets thus avoiding the situation of generating branches to undesired locations if offsets cannot be encoded. This patch also checks if a fixup cannot be applied and returns a fatal error if that's the case. llvm-svn: 197223
2013-12-13 19:11:02 +08:00
if (MCID.isBranch() || MCID.isCall()) {
const unsigned Opcode = Inst.getOpcode();
MCOperand Offset;
switch (Opcode) {
default:
break;
case Mips::BEQ:
case Mips::BNE:
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
case Mips::BEQ_MM:
case Mips::BNE_MM:
[Mips] formatting through clang-format llvm-svn: 199853
2014-01-23 07:31:38 +08:00
assert(MCID.getNumOperands() == 3 && "unexpected number of operands");
[mips] Add checks for alignment and maximum displacements for most of the branch instructions for mips and micromips instruction sets thus avoiding the situation of generating branches to undesired locations if offsets cannot be encoded. This patch also checks if a fixup cannot be applied and returns a fatal error if that's the case. llvm-svn: 197223
2013-12-13 19:11:02 +08:00
Offset = Inst.getOperand(2);
if (!Offset.isImm())
break; // We'll deal with this situation later on when applying fixups.
if (!isIntN(isMicroMips() ? 17 : 18, Offset.getImm()))
return Error(IDLoc, "branch target out of range");
[Mips] formatting through clang-format llvm-svn: 199853
2014-01-23 07:31:38 +08:00
if (OffsetToAlignment(Offset.getImm(), 1LL << (isMicroMips() ? 1 : 2)))
[mips] Add checks for alignment and maximum displacements for most of the branch instructions for mips and micromips instruction sets thus avoiding the situation of generating branches to undesired locations if offsets cannot be encoded. This patch also checks if a fixup cannot be applied and returns a fatal error if that's the case. llvm-svn: 197223
2013-12-13 19:11:02 +08:00
return Error(IDLoc, "branch to misaligned address");
break;
case Mips::BGEZ:
case Mips::BGTZ:
case Mips::BLEZ:
case Mips::BLTZ:
case Mips::BGEZAL:
case Mips::BLTZAL:
case Mips::BC1F:
case Mips::BC1T:
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
case Mips::BGEZ_MM:
case Mips::BGTZ_MM:
case Mips::BLEZ_MM:
case Mips::BLTZ_MM:
case Mips::BGEZAL_MM:
case Mips::BLTZAL_MM:
case Mips::BC1F_MM:
case Mips::BC1T_MM:
[Mips] formatting through clang-format llvm-svn: 199853
2014-01-23 07:31:38 +08:00
assert(MCID.getNumOperands() == 2 && "unexpected number of operands");
[mips] Add checks for alignment and maximum displacements for most of the branch instructions for mips and micromips instruction sets thus avoiding the situation of generating branches to undesired locations if offsets cannot be encoded. This patch also checks if a fixup cannot be applied and returns a fatal error if that's the case. llvm-svn: 197223
2013-12-13 19:11:02 +08:00
Offset = Inst.getOperand(1);
if (!Offset.isImm())
break; // We'll deal with this situation later on when applying fixups.
if (!isIntN(isMicroMips() ? 17 : 18, Offset.getImm()))
return Error(IDLoc, "branch target out of range");
[Mips] formatting through clang-format llvm-svn: 199853
2014-01-23 07:31:38 +08:00
if (OffsetToAlignment(Offset.getImm(), 1LL << (isMicroMips() ? 1 : 2)))
[mips] Add checks for alignment and maximum displacements for most of the branch instructions for mips and micromips instruction sets thus avoiding the situation of generating branches to undesired locations if offsets cannot be encoded. This patch also checks if a fixup cannot be applied and returns a fatal error if that's the case. llvm-svn: 197223
2013-12-13 19:11:02 +08:00
return Error(IDLoc, "branch to misaligned address");
break;
}
}
[mips][mips64r6] ssnop is deprecated on MIPS32r6/MIPS64r6 Summary: Depends on D4120 Reviewers: jkolek, zoran.jovanovic, vmedic Reviewed By: zoran.jovanovic, vmedic Differential Revision: http://reviews.llvm.org/D4121 llvm-svn: 211021
2014-06-16 21:25:35 +08:00
// SSNOP is deprecated on MIPS32r6/MIPS64r6
// We still accept it but it is a normal nop.
if (hasMips32r6() && Inst.getOpcode() == Mips::SSNOP) {
std::string ISA = hasMips64r6() ? "MIPS64r6" : "MIPS32r6";
Warning(IDLoc, "ssnop is deprecated for " + ISA + " and is equivalent to a "
"nop instruction");
}
Mips assembler: .set reorder support Mips have delayslots for certain instructions like jumps and branches. These are instructions that follow the branch or jump and are executed before the jump or branch is completed. Early Mips compilers could not cope with delayslots and left them up to the assembler. The assembler would fill the delayslots with the appropriate instruction, usually just a nop to allow correct runtime behavior. The default behavior for this is set with .set reorder. To tell the assembler that you don't want it to mess with the delayslot one used .set noreorder. For backwards compatibility we need to support .set reorder and have it be the default behavior in the assembler. Our support for it is to insert a NOP directly after an instruction with a delayslot when in .set reorder mode. Contributer: Vladimir Medic llvm-svn: 180584
2013-04-26 07:31:35 +08:00
if (MCID.hasDelaySlot() && Options.isReorder()) {
// If this instruction has a delay slot and .set reorder is active,
// emit a NOP after it.
Instructions.push_back(Inst);
MCInst NopInst;
NopInst.setOpcode(Mips::SLL);
NopInst.addOperand(MCOperand::CreateReg(Mips::ZERO));
NopInst.addOperand(MCOperand::CreateReg(Mips::ZERO));
NopInst.addOperand(MCOperand::CreateImm(0));
Instructions.push_back(NopInst);
return false;
}
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
if (MCID.mayLoad() || MCID.mayStore()) {
// Check the offset of memory operand, if it is a symbol
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// reference or immediate we may have to expand instructions.
for (unsigned i = 0; i < MCID.getNumOperands(); i++) {
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
const MCOperandInfo &OpInfo = MCID.OpInfo[i];
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
if ((OpInfo.OperandType == MCOI::OPERAND_MEMORY) ||
(OpInfo.OperandType == MCOI::OPERAND_UNKNOWN)) {
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
MCOperand &Op = Inst.getOperand(i);
if (Op.isImm()) {
int MemOffset = Op.getImm();
if (MemOffset < -32768 || MemOffset > 32767) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Offset can't exceed 16bit value.
expandMemInst(Inst, IDLoc, Instructions, MCID.mayLoad(), true);
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
return false;
}
} else if (Op.isExpr()) {
const MCExpr *Expr = Op.getExpr();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
if (Expr->getKind() == MCExpr::SymbolRef) {
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
const MCSymbolRefExpr *SR =
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
static_cast<const MCSymbolRefExpr *>(Expr);
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
if (SR->getKind() == MCSymbolRefExpr::VK_None) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Expand symbol.
expandMemInst(Inst, IDLoc, Instructions, MCID.mayLoad(), false);
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
return false;
}
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
} else if (!isEvaluated(Expr)) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
expandMemInst(Inst, IDLoc, Instructions, MCID.mayLoad(), false);
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
return false;
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
}
}
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
} // for
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
} // if load/store
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
if (needsExpansion(Inst))
expandInstruction(Inst, IDLoc, Instructions);
else
Instructions.push_back(Inst);
return false;
}
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
bool MipsAsmParser::needsExpansion(MCInst &Inst) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
switch (Inst.getOpcode()) {
case Mips::LoadImm32Reg:
case Mips::LoadAddr32Imm:
case Mips::LoadAddr32Reg:
return true;
default:
return false;
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
}
}
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
void MipsAsmParser::expandInstruction(MCInst &Inst, SMLoc IDLoc,
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
SmallVectorImpl<MCInst> &Instructions) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
switch (Inst.getOpcode()) {
case Mips::LoadImm32Reg:
return expandLoadImm(Inst, IDLoc, Instructions);
case Mips::LoadAddr32Imm:
return expandLoadAddressImm(Inst, IDLoc, Instructions);
case Mips::LoadAddr32Reg:
return expandLoadAddressReg(Inst, IDLoc, Instructions);
}
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
}
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
void MipsAsmParser::expandLoadImm(MCInst &Inst, SMLoc IDLoc,
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
SmallVectorImpl<MCInst> &Instructions) {
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
MCInst tmpInst;
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
const MCOperand &ImmOp = Inst.getOperand(1);
Initial assembler implementation of Mips load address macro This patch provides initial implementation of load address macro instruction for Mips. We have implemented two kinds of expansions with their variations depending on the size of immediate operand: 1) load address with immediate value directly: * la d,j => addiu d,$zero,j (for -32768 <= j <= 65535) * la d,j => lui d,hi16(j) ori d,d,lo16(j) (for any other 32 bit value of j) 2) load load address with register offset value * la d,j(s) => addiu d,s,j (for -32768 <= j <= 65535) * la d,j(s) => lui d,hi16(j) (for any other 32 bit value of j) ori d,d,lo16(j) addu d,d,s This patch does not cover the case when the address is loaded from the value of the label or function. Contributer: Vladimir Medic llvm-svn: 165561
2012-10-10 07:29:45 +08:00
assert(ImmOp.isImm() && "expected immediate operand kind");
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
const MCOperand &RegOp = Inst.getOperand(0);
assert(RegOp.isReg() && "expected register operand kind");
int ImmValue = ImmOp.getImm();
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
tmpInst.setLoc(IDLoc);
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
if (0 <= ImmValue && ImmValue <= 65535) {
// For 0 <= j <= 65535.
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
// li d,j => ori d,$zero,j
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
tmpInst.setOpcode(Mips::ORi);
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
tmpInst.addOperand(MCOperand::CreateReg(RegOp.getReg()));
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
tmpInst.addOperand(MCOperand::CreateReg(Mips::ZERO));
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
tmpInst.addOperand(MCOperand::CreateImm(ImmValue));
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
Instructions.push_back(tmpInst);
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
} else if (ImmValue < 0 && ImmValue >= -32768) {
// For -32768 <= j < 0.
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
// li d,j => addiu d,$zero,j
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
tmpInst.setOpcode(Mips::ADDiu);
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
tmpInst.addOperand(MCOperand::CreateReg(RegOp.getReg()));
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
tmpInst.addOperand(MCOperand::CreateReg(Mips::ZERO));
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
tmpInst.addOperand(MCOperand::CreateImm(ImmValue));
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
Instructions.push_back(tmpInst);
} else {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// For any other value of j that is representable as a 32-bit integer.
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
// li d,j => lui d,hi16(j)
Initial assembler implementation of Mips load address macro This patch provides initial implementation of load address macro instruction for Mips. We have implemented two kinds of expansions with their variations depending on the size of immediate operand: 1) load address with immediate value directly: * la d,j => addiu d,$zero,j (for -32768 <= j <= 65535) * la d,j => lui d,hi16(j) ori d,d,lo16(j) (for any other 32 bit value of j) 2) load load address with register offset value * la d,j(s) => addiu d,s,j (for -32768 <= j <= 65535) * la d,j(s) => lui d,hi16(j) (for any other 32 bit value of j) ori d,d,lo16(j) addu d,d,s This patch does not cover the case when the address is loaded from the value of the label or function. Contributer: Vladimir Medic llvm-svn: 165561
2012-10-10 07:29:45 +08:00
// ori d,d,lo16(j)
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
tmpInst.setOpcode(Mips::LUi);
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
tmpInst.addOperand(MCOperand::CreateReg(RegOp.getReg()));
tmpInst.addOperand(MCOperand::CreateImm((ImmValue & 0xffff0000) >> 16));
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
Instructions.push_back(tmpInst);
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
tmpInst.clear();
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
tmpInst.setOpcode(Mips::ORi);
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
tmpInst.addOperand(MCOperand::CreateReg(RegOp.getReg()));
tmpInst.addOperand(MCOperand::CreateReg(RegOp.getReg()));
tmpInst.addOperand(MCOperand::CreateImm(ImmValue & 0xffff));
tmpInst.setLoc(IDLoc);
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
Instructions.push_back(tmpInst);
}
}
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165350
2012-10-06 08:53:28 +08:00
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
void
MipsAsmParser::expandLoadAddressReg(MCInst &Inst, SMLoc IDLoc,
SmallVectorImpl<MCInst> &Instructions) {
Initial assembler implementation of Mips load address macro This patch provides initial implementation of load address macro instruction for Mips. We have implemented two kinds of expansions with their variations depending on the size of immediate operand: 1) load address with immediate value directly: * la d,j => addiu d,$zero,j (for -32768 <= j <= 65535) * la d,j => lui d,hi16(j) ori d,d,lo16(j) (for any other 32 bit value of j) 2) load load address with register offset value * la d,j(s) => addiu d,s,j (for -32768 <= j <= 65535) * la d,j(s) => lui d,hi16(j) (for any other 32 bit value of j) ori d,d,lo16(j) addu d,d,s This patch does not cover the case when the address is loaded from the value of the label or function. Contributer: Vladimir Medic llvm-svn: 165561
2012-10-10 07:29:45 +08:00
MCInst tmpInst;
const MCOperand &ImmOp = Inst.getOperand(2);
assert(ImmOp.isImm() && "expected immediate operand kind");
const MCOperand &SrcRegOp = Inst.getOperand(1);
assert(SrcRegOp.isReg() && "expected register operand kind");
const MCOperand &DstRegOp = Inst.getOperand(0);
assert(DstRegOp.isReg() && "expected register operand kind");
int ImmValue = ImmOp.getImm();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
if (-32768 <= ImmValue && ImmValue <= 65535) {
// For -32768 <= j <= 65535.
// la d,j(s) => addiu d,s,j
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
tmpInst.setOpcode(Mips::ADDiu);
Initial assembler implementation of Mips load address macro This patch provides initial implementation of load address macro instruction for Mips. We have implemented two kinds of expansions with their variations depending on the size of immediate operand: 1) load address with immediate value directly: * la d,j => addiu d,$zero,j (for -32768 <= j <= 65535) * la d,j => lui d,hi16(j) ori d,d,lo16(j) (for any other 32 bit value of j) 2) load load address with register offset value * la d,j(s) => addiu d,s,j (for -32768 <= j <= 65535) * la d,j(s) => lui d,hi16(j) (for any other 32 bit value of j) ori d,d,lo16(j) addu d,d,s This patch does not cover the case when the address is loaded from the value of the label or function. Contributer: Vladimir Medic llvm-svn: 165561
2012-10-10 07:29:45 +08:00
tmpInst.addOperand(MCOperand::CreateReg(DstRegOp.getReg()));
tmpInst.addOperand(MCOperand::CreateReg(SrcRegOp.getReg()));
tmpInst.addOperand(MCOperand::CreateImm(ImmValue));
Instructions.push_back(tmpInst);
} else {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// For any other value of j that is representable as a 32-bit integer.
// la d,j(s) => lui d,hi16(j)
// ori d,d,lo16(j)
// addu d,d,s
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
tmpInst.setOpcode(Mips::LUi);
Initial assembler implementation of Mips load address macro This patch provides initial implementation of load address macro instruction for Mips. We have implemented two kinds of expansions with their variations depending on the size of immediate operand: 1) load address with immediate value directly: * la d,j => addiu d,$zero,j (for -32768 <= j <= 65535) * la d,j => lui d,hi16(j) ori d,d,lo16(j) (for any other 32 bit value of j) 2) load load address with register offset value * la d,j(s) => addiu d,s,j (for -32768 <= j <= 65535) * la d,j(s) => lui d,hi16(j) (for any other 32 bit value of j) ori d,d,lo16(j) addu d,d,s This patch does not cover the case when the address is loaded from the value of the label or function. Contributer: Vladimir Medic llvm-svn: 165561
2012-10-10 07:29:45 +08:00
tmpInst.addOperand(MCOperand::CreateReg(DstRegOp.getReg()));
tmpInst.addOperand(MCOperand::CreateImm((ImmValue & 0xffff0000) >> 16));
Instructions.push_back(tmpInst);
tmpInst.clear();
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
tmpInst.setOpcode(Mips::ORi);
Initial assembler implementation of Mips load address macro This patch provides initial implementation of load address macro instruction for Mips. We have implemented two kinds of expansions with their variations depending on the size of immediate operand: 1) load address with immediate value directly: * la d,j => addiu d,$zero,j (for -32768 <= j <= 65535) * la d,j => lui d,hi16(j) ori d,d,lo16(j) (for any other 32 bit value of j) 2) load load address with register offset value * la d,j(s) => addiu d,s,j (for -32768 <= j <= 65535) * la d,j(s) => lui d,hi16(j) (for any other 32 bit value of j) ori d,d,lo16(j) addu d,d,s This patch does not cover the case when the address is loaded from the value of the label or function. Contributer: Vladimir Medic llvm-svn: 165561
2012-10-10 07:29:45 +08:00
tmpInst.addOperand(MCOperand::CreateReg(DstRegOp.getReg()));
tmpInst.addOperand(MCOperand::CreateReg(DstRegOp.getReg()));
tmpInst.addOperand(MCOperand::CreateImm(ImmValue & 0xffff));
Instructions.push_back(tmpInst);
tmpInst.clear();
tmpInst.setOpcode(Mips::ADDu);
tmpInst.addOperand(MCOperand::CreateReg(DstRegOp.getReg()));
tmpInst.addOperand(MCOperand::CreateReg(DstRegOp.getReg()));
tmpInst.addOperand(MCOperand::CreateReg(SrcRegOp.getReg()));
Instructions.push_back(tmpInst);
}
}
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
void
MipsAsmParser::expandLoadAddressImm(MCInst &Inst, SMLoc IDLoc,
SmallVectorImpl<MCInst> &Instructions) {
Initial assembler implementation of Mips load address macro This patch provides initial implementation of load address macro instruction for Mips. We have implemented two kinds of expansions with their variations depending on the size of immediate operand: 1) load address with immediate value directly: * la d,j => addiu d,$zero,j (for -32768 <= j <= 65535) * la d,j => lui d,hi16(j) ori d,d,lo16(j) (for any other 32 bit value of j) 2) load load address with register offset value * la d,j(s) => addiu d,s,j (for -32768 <= j <= 65535) * la d,j(s) => lui d,hi16(j) (for any other 32 bit value of j) ori d,d,lo16(j) addu d,d,s This patch does not cover the case when the address is loaded from the value of the label or function. Contributer: Vladimir Medic llvm-svn: 165561
2012-10-10 07:29:45 +08:00
MCInst tmpInst;
const MCOperand &ImmOp = Inst.getOperand(1);
assert(ImmOp.isImm() && "expected immediate operand kind");
const MCOperand &RegOp = Inst.getOperand(0);
assert(RegOp.isReg() && "expected register operand kind");
int ImmValue = ImmOp.getImm();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
if (-32768 <= ImmValue && ImmValue <= 65535) {
// For -32768 <= j <= 65535.
// la d,j => addiu d,$zero,j
Initial assembler implementation of Mips load address macro This patch provides initial implementation of load address macro instruction for Mips. We have implemented two kinds of expansions with their variations depending on the size of immediate operand: 1) load address with immediate value directly: * la d,j => addiu d,$zero,j (for -32768 <= j <= 65535) * la d,j => lui d,hi16(j) ori d,d,lo16(j) (for any other 32 bit value of j) 2) load load address with register offset value * la d,j(s) => addiu d,s,j (for -32768 <= j <= 65535) * la d,j(s) => lui d,hi16(j) (for any other 32 bit value of j) ori d,d,lo16(j) addu d,d,s This patch does not cover the case when the address is loaded from the value of the label or function. Contributer: Vladimir Medic llvm-svn: 165561
2012-10-10 07:29:45 +08:00
tmpInst.setOpcode(Mips::ADDiu);
tmpInst.addOperand(MCOperand::CreateReg(RegOp.getReg()));
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
tmpInst.addOperand(MCOperand::CreateReg(Mips::ZERO));
Initial assembler implementation of Mips load address macro This patch provides initial implementation of load address macro instruction for Mips. We have implemented two kinds of expansions with their variations depending on the size of immediate operand: 1) load address with immediate value directly: * la d,j => addiu d,$zero,j (for -32768 <= j <= 65535) * la d,j => lui d,hi16(j) ori d,d,lo16(j) (for any other 32 bit value of j) 2) load load address with register offset value * la d,j(s) => addiu d,s,j (for -32768 <= j <= 65535) * la d,j(s) => lui d,hi16(j) (for any other 32 bit value of j) ori d,d,lo16(j) addu d,d,s This patch does not cover the case when the address is loaded from the value of the label or function. Contributer: Vladimir Medic llvm-svn: 165561
2012-10-10 07:29:45 +08:00
tmpInst.addOperand(MCOperand::CreateImm(ImmValue));
Instructions.push_back(tmpInst);
} else {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// For any other value of j that is representable as a 32-bit integer.
// la d,j => lui d,hi16(j)
// ori d,d,lo16(j)
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
tmpInst.setOpcode(Mips::LUi);
Initial assembler implementation of Mips load address macro This patch provides initial implementation of load address macro instruction for Mips. We have implemented two kinds of expansions with their variations depending on the size of immediate operand: 1) load address with immediate value directly: * la d,j => addiu d,$zero,j (for -32768 <= j <= 65535) * la d,j => lui d,hi16(j) ori d,d,lo16(j) (for any other 32 bit value of j) 2) load load address with register offset value * la d,j(s) => addiu d,s,j (for -32768 <= j <= 65535) * la d,j(s) => lui d,hi16(j) (for any other 32 bit value of j) ori d,d,lo16(j) addu d,d,s This patch does not cover the case when the address is loaded from the value of the label or function. Contributer: Vladimir Medic llvm-svn: 165561
2012-10-10 07:29:45 +08:00
tmpInst.addOperand(MCOperand::CreateReg(RegOp.getReg()));
tmpInst.addOperand(MCOperand::CreateImm((ImmValue & 0xffff0000) >> 16));
Instructions.push_back(tmpInst);
tmpInst.clear();
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
tmpInst.setOpcode(Mips::ORi);
Initial assembler implementation of Mips load address macro This patch provides initial implementation of load address macro instruction for Mips. We have implemented two kinds of expansions with their variations depending on the size of immediate operand: 1) load address with immediate value directly: * la d,j => addiu d,$zero,j (for -32768 <= j <= 65535) * la d,j => lui d,hi16(j) ori d,d,lo16(j) (for any other 32 bit value of j) 2) load load address with register offset value * la d,j(s) => addiu d,s,j (for -32768 <= j <= 65535) * la d,j(s) => lui d,hi16(j) (for any other 32 bit value of j) ori d,d,lo16(j) addu d,d,s This patch does not cover the case when the address is loaded from the value of the label or function. Contributer: Vladimir Medic llvm-svn: 165561
2012-10-10 07:29:45 +08:00
tmpInst.addOperand(MCOperand::CreateReg(RegOp.getReg()));
tmpInst.addOperand(MCOperand::CreateReg(RegOp.getReg()));
tmpInst.addOperand(MCOperand::CreateImm(ImmValue & 0xffff));
Instructions.push_back(tmpInst);
}
}
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
void MipsAsmParser::expandMemInst(MCInst &Inst, SMLoc IDLoc,
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
SmallVectorImpl<MCInst> &Instructions,
bool isLoad, bool isImmOpnd) {
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
const MCSymbolRefExpr *SR;
MCInst TempInst;
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
unsigned ImmOffset, HiOffset, LoOffset;
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
const MCExpr *ExprOffset;
unsigned TmpRegNum;
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
unsigned AtRegNum = getReg(
[mips] Some uses of isMips64()/hasMips64() are really tests for 64-bit GPR's Summary: No functional change since these predicates are (currently) synonymous. Extracted from a patch by David Chisnall His work was sponsored by: DARPA, AFRL Differential Revision: http://llvm-reviews.chandlerc.com/D3202 llvm-svn: 204943
2014-03-28 00:42:17 +08:00
(isGP64()) ? Mips::GPR64RegClassID : Mips::GPR32RegClassID, getATReg());
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// 1st operand is either the source or destination register.
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
assert(Inst.getOperand(0).isReg() && "expected register operand kind");
unsigned RegOpNum = Inst.getOperand(0).getReg();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// 2nd operand is the base register.
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
assert(Inst.getOperand(1).isReg() && "expected register operand kind");
unsigned BaseRegNum = Inst.getOperand(1).getReg();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// 3rd operand is either an immediate or expression.
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
if (isImmOpnd) {
assert(Inst.getOperand(2).isImm() && "expected immediate operand kind");
ImmOffset = Inst.getOperand(2).getImm();
LoOffset = ImmOffset & 0x0000ffff;
HiOffset = (ImmOffset & 0xffff0000) >> 16;
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// If msb of LoOffset is 1(negative number) we must increment HiOffset.
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
if (LoOffset & 0x8000)
HiOffset++;
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
} else
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
ExprOffset = Inst.getOperand(2).getExpr();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// All instructions will have the same location.
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
TempInst.setLoc(IDLoc);
// 1st instruction in expansion is LUi. For load instruction we can use
// the dst register as a temporary if base and dst are different,
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// but for stores we must use $at.
TmpRegNum = (isLoad && (BaseRegNum != RegOpNum)) ? RegOpNum : AtRegNum;
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
TempInst.setOpcode(Mips::LUi);
TempInst.addOperand(MCOperand::CreateReg(TmpRegNum));
if (isImmOpnd)
TempInst.addOperand(MCOperand::CreateImm(HiOffset));
else {
if (ExprOffset->getKind() == MCExpr::SymbolRef) {
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
SR = static_cast<const MCSymbolRefExpr *>(ExprOffset);
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
const MCSymbolRefExpr *HiExpr = MCSymbolRefExpr::Create(
SR->getSymbol().getName(), MCSymbolRefExpr::VK_Mips_ABS_HI,
getContext());
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
TempInst.addOperand(MCOperand::CreateExpr(HiExpr));
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
} else {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
const MCExpr *HiExpr = evaluateRelocExpr(ExprOffset, "hi");
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
TempInst.addOperand(MCOperand::CreateExpr(HiExpr));
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
}
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Add the instruction to the list.
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
Instructions.push_back(TempInst);
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Prepare TempInst for next instruction.
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
TempInst.clear();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Add temp register to base.
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
TempInst.setOpcode(Mips::ADDu);
TempInst.addOperand(MCOperand::CreateReg(TmpRegNum));
TempInst.addOperand(MCOperand::CreateReg(TmpRegNum));
TempInst.addOperand(MCOperand::CreateReg(BaseRegNum));
Instructions.push_back(TempInst);
TempInst.clear();
Fix known typos Sweep the codebase for common typos. Includes some changes to visible function names that were misspelt. llvm-svn: 200018
2014-01-25 01:20:08 +08:00
// And finally, create original instruction with low part
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// of offset and new base.
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
TempInst.setOpcode(Inst.getOpcode());
TempInst.addOperand(MCOperand::CreateReg(RegOpNum));
TempInst.addOperand(MCOperand::CreateReg(TmpRegNum));
if (isImmOpnd)
TempInst.addOperand(MCOperand::CreateImm(LoOffset));
else {
if (ExprOffset->getKind() == MCExpr::SymbolRef) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
const MCSymbolRefExpr *LoExpr = MCSymbolRefExpr::Create(
SR->getSymbol().getName(), MCSymbolRefExpr::VK_Mips_ABS_LO,
getContext());
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
TempInst.addOperand(MCOperand::CreateExpr(LoExpr));
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
} else {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
const MCExpr *LoExpr = evaluateRelocExpr(ExprOffset, "lo");
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
TempInst.addOperand(MCOperand::CreateExpr(LoExpr));
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
}
}
Instructions.push_back(TempInst);
TempInst.clear();
}
[mips] Implement jr.hb and jalr.hb (Jump Register and Jump and Link Register with Hazard Barrier). Summary: These instructions are available in ISAs >= mips32/mips64. For mips32r6/mips64r6, jr.hb has a new encoding format. Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D4019 llvm-svn: 210654
2014-06-11 23:05:56 +08:00
unsigned MipsAsmParser::checkTargetMatchPredicate(MCInst &Inst) {
// As described by the Mips32r2 spec, the registers Rd and Rs for
// jalr.hb must be different.
unsigned Opcode = Inst.getOpcode();
if (Opcode == Mips::JALR_HB &&
(Inst.getOperand(0).getReg() == Inst.getOperand(1).getReg()))
return Match_RequiresDifferentSrcAndDst;
return Match_Success;
}
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
bool MipsAsmParser::MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode,
OperandVector &Operands,
MCStreamer &Out,
unsigned &ErrorInfo,
bool MatchingInlineAsm) {
[mips] Implement jr.hb and jalr.hb (Jump Register and Jump and Link Register with Hazard Barrier). Summary: These instructions are available in ISAs >= mips32/mips64. For mips32r6/mips64r6, jr.hb has a new encoding format. Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D4019 llvm-svn: 210654
2014-06-11 23:05:56 +08:00
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
MCInst Inst;
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
SmallVector<MCInst, 8> Instructions;
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
unsigned MatchResult =
MatchInstructionImpl(Operands, Inst, ErrorInfo, MatchingInlineAsm);
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
switch (MatchResult) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
default:
break;
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
case Match_Success: {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
if (processInstruction(Inst, IDLoc, Instructions))
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
return true;
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
for (unsigned i = 0; i < Instructions.size(); i++)
Change MCStreamer EmitInstruction interface to take subtarget info llvm-svn: 200345
2014-01-29 07:12:42 +08:00
Out.EmitInstruction(Instructions[i], STI);
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
return false;
}
case Match_MissingFeature:
Error(IDLoc, "instruction requires a CPU feature not currently enabled");
return true;
case Match_InvalidOperand: {
SMLoc ErrorLoc = IDLoc;
if (ErrorInfo != ~0U) {
if (ErrorInfo >= Operands.size())
return Error(IDLoc, "too few operands for instruction");
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
ErrorLoc = ((MipsOperand &)*Operands[ErrorInfo]).getStartLoc();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
if (ErrorLoc == SMLoc())
ErrorLoc = IDLoc;
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
}
return Error(ErrorLoc, "invalid operand for instruction");
}
case Match_MnemonicFail:
return Error(IDLoc, "invalid instruction");
[mips] Implement jr.hb and jalr.hb (Jump Register and Jump and Link Register with Hazard Barrier). Summary: These instructions are available in ISAs >= mips32/mips64. For mips32r6/mips64r6, jr.hb has a new encoding format. Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D4019 llvm-svn: 210654
2014-06-11 23:05:56 +08:00
case Match_RequiresDifferentSrcAndDst:
return Error(IDLoc, "source and destination must be different");
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
}
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
return true;
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
void MipsAsmParser::WarnIfAssemblerTemporary(int RegIndex, SMLoc Loc) {
if ((RegIndex != 0) && ((int)Options.getATRegNum() == RegIndex)) {
if (RegIndex == 1)
Warning(Loc, "Used $at without \".set noat\"");
[mips] Correct testcase for .set at=$reg and emit the new warnings for numeric registers too. Summary: Remove the XFAIL added in my previous commit and correct the test such that it correctly tests the expansion of the assembler temporary. Also added a test to check that $at is always $1 when written by the user. Corrected the new assembler temporary warnings so that they are emitted for numeric registers too. Differential Revision: http://llvm-reviews.chandlerc.com/D3169 llvm-svn: 204711
2014-03-25 19:16:03 +08:00
else
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Warning(Loc, Twine("Used $") + Twine(RegIndex) + " with \".set at=$" +
Twine(RegIndex) + "\"");
[mips] Fix assembler temporary expansion and add associated warnings about the use of $at. Summary: The assembler temporary is normally $at ($1) but can be reassigned using '.set at=$reg'. Regardless of which register is nominated as the assembler temporary, $at remains $1 when written by the user. Adds warnings under the following conditions: * The register nominated as the assembler temporary is used by the user. * '.set noat' is in effect and $at is used by the user. Both of these only work for named registers. I have a follow up commit that makes it work for numeric registers as well. XFAIL set-at-directive.s since it incorrectly tests that $at is redefined by '.set at=$reg'. Testcases will follow in a separate commit. Patch by David Chisnall His work was sponsored by: DARPA, AFRL Differential Revision: http://llvm-reviews.chandlerc.com/D3167 llvm-svn: 204710
2014-03-25 18:57:07 +08:00
}
[mips] Correct testcase for .set at=$reg and emit the new warnings for numeric registers too. Summary: Remove the XFAIL added in my previous commit and correct the test such that it correctly tests the expansion of the assembler temporary. Also added a test to check that $at is always $1 when written by the user. Corrected the new assembler temporary warnings so that they are emitted for numeric registers too. Differential Revision: http://llvm-reviews.chandlerc.com/D3169 llvm-svn: 204711
2014-03-25 19:16:03 +08:00
}
int MipsAsmParser::matchCPURegisterName(StringRef Name) {
int CC;
ELF symbol table field st_other support, excluding visibility bits. Mips specific standalone assembler directive "set at". This directive changes the general purpose register that the assembler will use when given the symbolic register name $at. This does not include negative testing. That will come in a future patch. A side affect of this patch recognizes the different GPR register names for temporaries between old abi and new abi so a test case for that is included. Contributer: Vladimir Medic llvm-svn: 175686
2013-02-21 07:11:17 +08:00
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
CC = StringSwitch<unsigned>(Name)
.Case("zero", 0)
[mips] Correct testcase for .set at=$reg and emit the new warnings for numeric registers too. Summary: Remove the XFAIL added in my previous commit and correct the test such that it correctly tests the expansion of the assembler temporary. Also added a test to check that $at is always $1 when written by the user. Corrected the new assembler temporary warnings so that they are emitted for numeric registers too. Differential Revision: http://llvm-reviews.chandlerc.com/D3169 llvm-svn: 204711
2014-03-25 19:16:03 +08:00
.Case("at", 1)
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
.Case("a0", 4)
.Case("a1", 5)
.Case("a2", 6)
.Case("a3", 7)
.Case("v0", 2)
.Case("v1", 3)
.Case("s0", 16)
.Case("s1", 17)
.Case("s2", 18)
.Case("s3", 19)
.Case("s4", 20)
.Case("s5", 21)
.Case("s6", 22)
.Case("s7", 23)
.Case("k0", 26)
.Case("k1", 27)
[mips] $s8 is an alias for $fp in all ABI's, not just N32/N64. llvm-svn: 204793
2014-03-26 19:05:24 +08:00
.Case("gp", 28)
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
.Case("sp", 29)
.Case("fp", 30)
[mips] $s8 is an alias for $fp in all ABI's, not just N32/N64. llvm-svn: 204793
2014-03-26 19:05:24 +08:00
.Case("s8", 30)
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
.Case("ra", 31)
.Case("t0", 8)
.Case("t1", 9)
.Case("t2", 10)
.Case("t3", 11)
.Case("t4", 12)
.Case("t5", 13)
.Case("t6", 14)
.Case("t7", 15)
.Case("t8", 24)
.Case("t9", 25)
.Default(-1);
ELF symbol table field st_other support, excluding visibility bits. Mips specific standalone assembler directive "set at". This directive changes the general purpose register that the assembler will use when given the symbolic register name $at. This does not include negative testing. That will come in a future patch. A side affect of this patch recognizes the different GPR register names for temporaries between old abi and new abi so a test case for that is included. Contributer: Vladimir Medic llvm-svn: 175686
2013-02-21 07:11:17 +08:00
[mips] The register names depend on the ABI being N32/N64 rather than the arch being mips64 Summary: Added test cases for O32 and N32 on MIPS64. Reviewers: matheusalmeida Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3175 llvm-svn: 204796
2014-03-26 19:39:07 +08:00
if (isN32() || isN64()) {
// Although SGI documentation just cuts out t0-t3 for n32/n64,
// GNU pushes the values of t0-t3 to override the o32/o64 values for t4-t7
// We are supporting both cases, so for t0-t3 we'll just push them to t4-t7.
if (8 <= CC && CC <= 11)
CC += 4;
if (CC == -1)
CC = StringSwitch<unsigned>(Name)
.Case("a4", 8)
.Case("a5", 9)
.Case("a6", 10)
.Case("a7", 11)
.Case("kt0", 26)
.Case("kt1", 27)
.Default(-1);
}
Improvements to MIPS64 assembler: - Teach it about dadd[i] instructions and move pseudo-instruction - Make it parse the register names correctly (for N32 / N64) llvm-svn: 165506
2012-10-10 00:27:43 +08:00
ELF symbol table field st_other support, excluding visibility bits. Mips specific standalone assembler directive "set at". This directive changes the general purpose register that the assembler will use when given the symbolic register name $at. This does not include negative testing. That will come in a future patch. A side affect of this patch recognizes the different GPR register names for temporaries between old abi and new abi so a test case for that is included. Contributer: Vladimir Medic llvm-svn: 175686
2013-02-21 07:11:17 +08:00
return CC;
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
int MipsAsmParser::matchFPURegisterName(StringRef Name) {
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
The Mips standalone assembler fpu instruction support. Test cases included Contributer: Vladimir Medic llvm-svn: 163363
2012-09-07 08:23:42 +08:00
if (Name[0] == 'f') {
StringRef NumString = Name.substr(1);
unsigned IntVal;
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
if (NumString.getAsInteger(10, IntVal))
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
return -1; // This is not an integer.
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
if (IntVal > 31) // Maximum index for fpu register.
The Mips standalone assembler fpu instruction support. Test cases included Contributer: Vladimir Medic llvm-svn: 163363
2012-09-07 08:23:42 +08:00
return -1;
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
return IntVal;
}
return -1;
}
The Mips standalone assembler fpu instruction support. Test cases included Contributer: Vladimir Medic llvm-svn: 163363
2012-09-07 08:23:42 +08:00
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
int MipsAsmParser::matchFCCRegisterName(StringRef Name) {
if (Name.startswith("fcc")) {
StringRef NumString = Name.substr(3);
unsigned IntVal;
if (NumString.getAsInteger(10, IntVal))
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
return -1; // This is not an integer.
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
if (IntVal > 7) // There are only 8 fcc registers.
return -1;
return IntVal;
The Mips standalone assembler fpu instruction support. Test cases included Contributer: Vladimir Medic llvm-svn: 163363
2012-09-07 08:23:42 +08:00
}
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
return -1;
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
int MipsAsmParser::matchACRegisterName(StringRef Name) {
[mips] Fix bug in parsing accumulator registers. llvm-svn: 188344
2013-08-14 09:15:52 +08:00
if (Name.startswith("ac")) {
StringRef NumString = Name.substr(2);
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
unsigned IntVal;
if (NumString.getAsInteger(10, IntVal))
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
return -1; // This is not an integer.
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
if (IntVal > 3) // There are only 3 acc registers.
return -1;
return IntVal;
}
return -1;
}
Optimize register parsing for MipsAsmParser. Allow symbolic aliases for FPU registers. llvm-svn: 184411
2013-06-20 19:21:49 +08:00
[mips][msa] Direct Object Emission support for the MSA instruction set. In more detail, this patch adds the ability to parse, encode and decode MSA registers ($w0-$w31). The format of 2RF instructions (MipsMSAInstrFormat.td) was updated so that we could attach a test case to this patch i.e., the test case parses, encodes and decodes 2 MSA instructions. Following patches will add the remainder of the instructions. Note that DecodeMSA128BRegisterClass is missing from MipsDisassembler.td because it's not yet required at this stage and having it would cause a compiler warning (unused function). Patch by Matheus Almeida llvm-svn: 191412
2013-09-26 07:50:44 +08:00
int MipsAsmParser::matchMSA128RegisterName(StringRef Name) {
unsigned IntVal;
if (Name.front() != 'w' || Name.drop_front(1).getAsInteger(10, IntVal))
return -1;
if (IntVal > 31)
return -1;
return IntVal;
}
[mips][msa] Direct Object Emission support for CTCMSA and CFCMSA. These instructions are logically related as they allow read/write of MSA control registers. Currently MSA control registers are emitted by number but hopefully that will change as soon as GAS starts accepting them by name as that would make the assembly easier to read. llvm-svn: 193078
2013-10-21 20:26:50 +08:00
int MipsAsmParser::matchMSA128CtrlRegisterName(StringRef Name) {
int CC;
CC = StringSwitch<unsigned>(Name)
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
.Case("msair", 0)
.Case("msacsr", 1)
.Case("msaaccess", 2)
.Case("msasave", 3)
.Case("msamodify", 4)
.Case("msarequest", 5)
.Case("msamap", 6)
.Case("msaunmap", 7)
.Default(-1);
[mips][msa] Direct Object Emission support for CTCMSA and CFCMSA. These instructions are logically related as they allow read/write of MSA control registers. Currently MSA control registers are emitted by number but hopefully that will change as soon as GAS starts accepting them by name as that would make the assembly easier to read. llvm-svn: 193078
2013-10-21 20:26:50 +08:00
return CC;
}
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
bool MipsAssemblerOptions::setATReg(unsigned Reg) {
if (Reg > 31)
return false;
aTReg = Reg;
return true;
}
[mips] Add error message when trying to use $at in '.set noat' mode. Summary: Patch by David Chisnall His work was sponsored by: DARPA, AFRL Differential Revision: http://llvm-reviews.chandlerc.com/D3158 llvm-svn: 204621
2014-03-25 00:48:01 +08:00
int MipsAsmParser::getATReg() {
int AT = Options.getATRegNum();
if (AT == 0)
TokError("Pseudo instruction requires $at, which is not available");
return AT;
}
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
unsigned MipsAsmParser::getReg(int RC, int RegNo) {
Use pointers to the MCAsmInfo and MCRegInfo. Someone may want to do something crazy, like replace these objects if they change or something. No functionality change intended. llvm-svn: 184175
2013-06-18 15:20:20 +08:00
return *(getContext().getRegisterInfo()->getRegClass(RC).begin() + RegNo);
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
}
[mips] Add support for .cpsetup Summary: Patch by Robert N. M. Watson His work was sponsored by: DARPA, AFRL Small corrections by myself. CC: theraven, matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3199 llvm-svn: 204924
2014-03-27 21:52:53 +08:00
unsigned MipsAsmParser::getGPR(int RegNo) {
[mips] Some uses of isMips64()/hasMips64() are really tests for 64-bit GPR's Summary: No functional change since these predicates are (currently) synonymous. Extracted from a patch by David Chisnall His work was sponsored by: DARPA, AFRL Differential Revision: http://llvm-reviews.chandlerc.com/D3202 llvm-svn: 204943
2014-03-28 00:42:17 +08:00
return getReg(isGP64() ? Mips::GPR64RegClassID : Mips::GPR32RegClassID,
RegNo);
[mips] Add support for .cpsetup Summary: Patch by Robert N. M. Watson His work was sponsored by: DARPA, AFRL Small corrections by myself. CC: theraven, matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3199 llvm-svn: 204924
2014-03-27 21:52:53 +08:00
}
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
int MipsAsmParser::matchRegisterByNumber(unsigned RegNum, unsigned RegClass) {
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
if (RegNum >
[mips] Attempting to use register $32 should be an error instead of an assertion. Reviewers: matheusalmeida Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3201 llvm-svn: 204932
2014-03-27 23:00:44 +08:00
getContext().getRegisterInfo()->getRegClass(RegClass).getNumRegs() - 1)
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
return -1;
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
return getReg(RegClass, RegNum);
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
}
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
bool MipsAsmParser::ParseOperand(OperandVector &Operands, StringRef Mnemonic) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
DEBUG(dbgs() << "ParseOperand\n");
Implement methods that enable expansion of load immediate macro instruction (li) in the assembler. We have identified three possible expansions depending on the size of immediate operand: 1) for 0 ≤ j ≤ 65535. li d,j => ori d,$zero,j 2) for −32768 ≤ j < 0. li d,j => addiu d,$zero,j 3) for any other value of j that is representable as a 32-bit integer. li d,j => lui d,hi16(j) ori d,d,lo16(j) All of the above have been implemented in ths patch. Contributer: Vladimir Medic llvm-svn: 165199
2012-10-04 12:03:53 +08:00
// Check if the current operand has a custom associated parser, if so, try to
// custom parse the operand, or fallback to the general approach.
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
OperandMatchResultTy ResTy = MatchOperandParserImpl(Operands, Mnemonic);
if (ResTy == MatchOperand_Success)
return false;
// If there wasn't a custom match, try the generic matcher below. Otherwise,
// there was a match, but an error occurred, in which case, just return that
// the operand parsing failed.
if (ResTy == MatchOperand_ParseFail)
return true;
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
DEBUG(dbgs() << ".. Generic Parser\n");
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
switch (getLexer().getKind()) {
default:
Error(Parser.getTok().getLoc(), "unexpected token in operand");
return true;
case AsmToken::Dollar: {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Parse the register.
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
SMLoc S = Parser.getTok().getLoc();
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205229
2014-04-01 01:43:46 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
// Almost all registers have been parsed by custom parsers. There is only
// one exception to this. $zero (and it's alias $0) will reach this point
// for div, divu, and similar instructions because it is not an operand
// to the instruction definition but an explicit register. Special case
// this situation for now.
if (ParseAnyRegister(Operands) != MatchOperand_NoMatch)
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
return false;
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Maybe it is a symbol reference.
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
StringRef Identifier;
MCParser: Update method names per coding guidelines. s/AddDirectiveHandler/addDirectiveHandler/ s/ParseMSInlineAsm/parseMSInlineAsm/ s/ParseIdentifier/parseIdentifier/ s/ParseStringToEndOfStatement/parseStringToEndOfStatement/ s/ParseEscapedString/parseEscapedString/ s/EatToEndOfStatement/eatToEndOfStatement/ s/ParseExpression/parseExpression/ s/ParseParenExpression/parseParenExpression/ s/ParseAbsoluteExpression/parseAbsoluteExpression/ s/CheckForValidSection/checkForValidSection/ http://llvm.org/docs/CodingStandards.html#name-types-functions-variables-and-enumerators-properly No functional change intended. llvm-svn: 175675
2013-02-21 06:21:35 +08:00
if (Parser.parseIdentifier(Identifier))
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
return true;
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
SMLoc E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
MipsAsmParser: Fix a couple of string use-after-frees and misuses of classof. llvm-svn: 163383
2012-09-07 17:47:42 +08:00
MCSymbol *Sym = getContext().GetOrCreateSymbol("$" + Identifier);
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Otherwise create a symbol reference.
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
const MCExpr *Res =
MCSymbolRefExpr::Create(Sym, MCSymbolRefExpr::VK_None, getContext());
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Operands.push_back(MipsOperand::CreateImm(Res, S, E, *this));
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
return false;
}
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
// Else drop to expression parsing.
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
case AsmToken::LParen:
case AsmToken::Minus:
case AsmToken::Plus:
case AsmToken::Integer:
[mips] Update MipsAsmParser so that it's possible to handle immediates that start with the binary operator NOT (~). Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D4158 llvm-svn: 211163
2014-06-18 21:55:18 +08:00
case AsmToken::Tilde:
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
case AsmToken::String: {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
DEBUG(dbgs() << ".. generic integer\n");
OperandMatchResultTy ResTy = ParseImm(Operands);
return ResTy != MatchOperand_Success;
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
}
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
case AsmToken::Percent: {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// It is a symbol reference or constant expression.
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
const MCExpr *IdVal;
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
SMLoc S = Parser.getTok().getLoc(); // Start location of the operand.
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
if (parseRelocOperand(IdVal))
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
return true;
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
SMLoc E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Operands.push_back(MipsOperand::CreateImm(IdVal, S, E, *this));
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
return false;
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
} // case AsmToken::Percent
} // switch(getLexer().getKind())
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
return true;
}
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
const MCExpr *MipsAsmParser::evaluateRelocExpr(const MCExpr *Expr,
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
StringRef RelocStr) {
const MCExpr *Res;
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Check the type of the expression.
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Expr)) {
[mips] Extend MipsMCExpr class to handle %higher(sym1 - sym2 + const) and %highest(sym1 - sym2 + const) relocations. Remove "ABS_" from VK_Mips_HI and VK_Mips_LO enums in MipsMCExpr, to be consistent with VK_Mips_HIGHER and VK_Mips_HIGHEST. This change also deletes test file test/MC/Mips/higher_highest.ll and moves its CHECK's to the new test file test/MC/Mips/higher-highest-addressing.s. The deleted file tests that R_MIPS_HIGHER and R_MIPS_HIGHEST relocations are emitted in the .o file. Since it uses -force-mips-long-branch option, it was created when MipsLongBranch's implementation was emitting R_MIPS_HIGHER and R_MIPS_HIGHEST relocations in the .o file. It was disabled when MipsLongBranch started to directly calculate offsets. Differential Revision: http://llvm-reviews.chandlerc.com/D3230 llvm-svn: 205522
2014-04-03 18:37:45 +08:00
// It's a constant, evaluate reloc value.
int16_t Val;
switch (getVariantKind(RelocStr)) {
case MCSymbolRefExpr::VK_Mips_ABS_LO:
// Get the 1st 16-bits.
Val = MCE->getValue() & 0xffff;
break;
case MCSymbolRefExpr::VK_Mips_ABS_HI:
// Get the 2nd 16-bits. Also add 1 if bit 15 is 1, to compensate for low
// 16 bits being negative.
Val = ((MCE->getValue() + 0x8000) >> 16) & 0xffff;
break;
case MCSymbolRefExpr::VK_Mips_HIGHER:
// Get the 3rd 16-bits.
Val = ((MCE->getValue() + 0x80008000LL) >> 32) & 0xffff;
break;
case MCSymbolRefExpr::VK_Mips_HIGHEST:
// Get the 4th 16-bits.
Val = ((MCE->getValue() + 0x800080008000LL) >> 48) & 0xffff;
break;
default:
report_fatal_error("Unsupported reloc value!");
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
}
[mips] Extend MipsMCExpr class to handle %higher(sym1 - sym2 + const) and %highest(sym1 - sym2 + const) relocations. Remove "ABS_" from VK_Mips_HI and VK_Mips_LO enums in MipsMCExpr, to be consistent with VK_Mips_HIGHER and VK_Mips_HIGHEST. This change also deletes test file test/MC/Mips/higher_highest.ll and moves its CHECK's to the new test file test/MC/Mips/higher-highest-addressing.s. The deleted file tests that R_MIPS_HIGHER and R_MIPS_HIGHEST relocations are emitted in the .o file. Since it uses -force-mips-long-branch option, it was created when MipsLongBranch's implementation was emitting R_MIPS_HIGHER and R_MIPS_HIGHEST relocations in the .o file. It was disabled when MipsLongBranch started to directly calculate offsets. Differential Revision: http://llvm-reviews.chandlerc.com/D3230 llvm-svn: 205522
2014-04-03 18:37:45 +08:00
return MCConstantExpr::Create(Val, getContext());
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
}
if (const MCSymbolRefExpr *MSRE = dyn_cast<MCSymbolRefExpr>(Expr)) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// It's a symbol, create a symbolic expression from the symbol.
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
StringRef Symbol = MSRE->getSymbol().getName();
MCSymbolRefExpr::VariantKind VK = getVariantKind(RelocStr);
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Res = MCSymbolRefExpr::Create(Symbol, VK, getContext());
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
return Res;
}
if (const MCBinaryExpr *BE = dyn_cast<MCBinaryExpr>(Expr)) {
[mips] Implement %hi(sym1 - sym2) and %lo(sym1 - sym2) expressions Patch implements %hi(sym1 - sym2) and %lo(sym1 - sym2) expressions for MIPS by creating target expression class MipsMCExpr. Patch by Sasa Stankovic. Differential Revision: http://llvm-reviews.chandlerc.com/D2592 llvm-svn: 200783
2014-02-05 02:41:57 +08:00
MCSymbolRefExpr::VariantKind VK = getVariantKind(RelocStr);
[mips] Extend MipsMCExpr class to handle %higher(sym1 - sym2 + const) and %highest(sym1 - sym2 + const) relocations. Remove "ABS_" from VK_Mips_HI and VK_Mips_LO enums in MipsMCExpr, to be consistent with VK_Mips_HIGHER and VK_Mips_HIGHEST. This change also deletes test file test/MC/Mips/higher_highest.ll and moves its CHECK's to the new test file test/MC/Mips/higher-highest-addressing.s. The deleted file tests that R_MIPS_HIGHER and R_MIPS_HIGHEST relocations are emitted in the .o file. Since it uses -force-mips-long-branch option, it was created when MipsLongBranch's implementation was emitting R_MIPS_HIGHER and R_MIPS_HIGHEST relocations in the .o file. It was disabled when MipsLongBranch started to directly calculate offsets. Differential Revision: http://llvm-reviews.chandlerc.com/D3230 llvm-svn: 205522
2014-04-03 18:37:45 +08:00
// Try to create target expression.
if (MipsMCExpr::isSupportedBinaryExpr(VK, BE))
return MipsMCExpr::Create(VK, Expr, getContext());
[mips] Implement %hi(sym1 - sym2) and %lo(sym1 - sym2) expressions Patch implements %hi(sym1 - sym2) and %lo(sym1 - sym2) expressions for MIPS by creating target expression class MipsMCExpr. Patch by Sasa Stankovic. Differential Revision: http://llvm-reviews.chandlerc.com/D2592 llvm-svn: 200783
2014-02-05 02:41:57 +08:00
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
const MCExpr *LExp = evaluateRelocExpr(BE->getLHS(), RelocStr);
const MCExpr *RExp = evaluateRelocExpr(BE->getRHS(), RelocStr);
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
Res = MCBinaryExpr::Create(BE->getOpcode(), LExp, RExp, getContext());
return Res;
}
if (const MCUnaryExpr *UN = dyn_cast<MCUnaryExpr>(Expr)) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
const MCExpr *UnExp = evaluateRelocExpr(UN->getSubExpr(), RelocStr);
Res = MCUnaryExpr::Create(UN->getOpcode(), UnExp, getContext());
return Res;
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Just return the original expression.
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
return Expr;
}
bool MipsAsmParser::isEvaluated(const MCExpr *Expr) {
switch (Expr->getKind()) {
case MCExpr::Constant:
return true;
case MCExpr::SymbolRef:
return (cast<MCSymbolRefExpr>(Expr)->getKind() != MCSymbolRefExpr::VK_None);
case MCExpr::Binary:
if (const MCBinaryExpr *BE = dyn_cast<MCBinaryExpr>(Expr)) {
if (!isEvaluated(BE->getLHS()))
return false;
return isEvaluated(BE->getRHS());
}
case MCExpr::Unary:
return isEvaluated(cast<MCUnaryExpr>(Expr)->getSubExpr());
[mips] Implement %hi(sym1 - sym2) and %lo(sym1 - sym2) expressions Patch implements %hi(sym1 - sym2) and %lo(sym1 - sym2) expressions for MIPS by creating target expression class MipsMCExpr. Patch by Sasa Stankovic. Differential Revision: http://llvm-reviews.chandlerc.com/D2592 llvm-svn: 200783
2014-02-05 02:41:57 +08:00
case MCExpr::Target:
return true;
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
}
return false;
}
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
bool MipsAsmParser::parseRelocOperand(const MCExpr *&Res) {
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
Parser.Lex(); // Eat the % token.
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
const AsmToken &Tok = Parser.getTok(); // Get next token, operation.
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
if (Tok.isNot(AsmToken::Identifier))
return true;
MipsAsmParser: Fix a couple of string use-after-frees and misuses of classof. llvm-svn: 163383
2012-09-07 17:47:42 +08:00
std::string Str = Tok.getIdentifier().str();
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Eat the identifier.
// Now make an expression from the rest of the operand.
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
const MCExpr *IdVal;
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
SMLoc EndLoc;
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
if (getLexer().getKind() == AsmToken::LParen) {
while (1) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Eat the '(' token.
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
if (getLexer().getKind() == AsmToken::Percent) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Eat the % token.
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
const AsmToken &nextTok = Parser.getTok();
if (nextTok.isNot(AsmToken::Identifier))
return true;
MipsAsmParser: Fix a couple of string use-after-frees and misuses of classof. llvm-svn: 163383
2012-09-07 17:47:42 +08:00
Str += "(%";
Str += nextTok.getIdentifier();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Eat the identifier.
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
if (getLexer().getKind() != AsmToken::LParen)
return true;
} else
break;
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
if (getParser().parseParenExpression(IdVal, EndLoc))
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
return true;
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
while (getLexer().getKind() == AsmToken::RParen)
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Eat the ')' token.
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
} else
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
return true; // Parenthesis must follow the relocation operand.
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Res = evaluateRelocExpr(IdVal, Str);
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
return false;
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
}
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
bool MipsAsmParser::ParseRegister(unsigned &RegNo, SMLoc &StartLoc,
SMLoc &EndLoc) {
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
SmallVector<std::unique_ptr<MCParsedAsmOperand>, 1> Operands;
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
OperandMatchResultTy ResTy = ParseAnyRegister(Operands);
if (ResTy == MatchOperand_Success) {
assert(Operands.size() == 1);
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsOperand &Operand = static_cast<MipsOperand &>(*Operands.front());
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
StartLoc = Operand.getStartLoc();
EndLoc = Operand.getEndLoc();
// AFAIK, we only support numeric registers and named GPR's in CFI
// directives.
// Don't worry about eating tokens before failing. Using an unrecognised
// register is a parse error.
if (Operand.isGPRAsmReg()) {
// Resolve to GPR32 or GPR64 appropriately.
RegNo = isGP64() ? Operand.getGPR64Reg() : Operand.getGPR32Reg();
}
return (RegNo == (unsigned)-1);
}
assert(Operands.size() == 0);
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
return (RegNo == (unsigned)-1);
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
}
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
bool MipsAsmParser::parseMemOffset(const MCExpr *&Res, bool isParenExpr) {
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
SMLoc S;
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
bool Result = true;
while (getLexer().getKind() == AsmToken::LParen)
Parser.Lex();
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
switch (getLexer().getKind()) {
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
default:
return true;
This patch that enables the Mips assembler to use symbols for offset for instructions This patch uses the generated instruction info tables to identify memory/load store instructions. After successful matching and based on the operand type and size, it generates additional instructions to the output. Contributor: Vladimir Medic llvm-svn: 177685
2013-03-22 08:05:30 +08:00
case AsmToken::Identifier:
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
case AsmToken::LParen:
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
case AsmToken::Integer:
case AsmToken::Minus:
case AsmToken::Plus:
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
if (isParenExpr)
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Result = getParser().parseParenExpression(Res, S);
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
else
Result = (getParser().parseExpression(Res));
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
while (getLexer().getKind() == AsmToken::RParen)
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
Parser.Lex();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
break;
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
case AsmToken::Percent:
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
Result = parseRelocOperand(Res);
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
}
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
return Result;
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
}
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsAsmParser::OperandMatchResultTy
MipsAsmParser::parseMemOperand(OperandVector &Operands) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
DEBUG(dbgs() << "parseMemOperand\n");
[C++] Use 'nullptr'. Target edition. llvm-svn: 207197
2014-04-25 13:30:21 +08:00
const MCExpr *IdVal = nullptr;
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
SMLoc S;
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
bool isParenExpr = false;
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
MipsAsmParser::OperandMatchResultTy Res = MatchOperand_NoMatch;
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// First operand is the offset.
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
S = Parser.getTok().getLoc();
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
if (getLexer().getKind() == AsmToken::LParen) {
Parser.Lex();
isParenExpr = true;
}
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
if (getLexer().getKind() != AsmToken::Dollar) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
if (parseMemOffset(IdVal, isParenExpr))
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
return MatchOperand_ParseFail;
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
const AsmToken &Tok = Parser.getTok(); // Get the next token.
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
if (Tok.isNot(AsmToken::LParen)) {
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsOperand &Mnemonic = static_cast<MipsOperand &>(*Operands[0]);
if (Mnemonic.getToken() == "la") {
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
SMLoc E =
SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Operands.push_back(MipsOperand::CreateImm(IdVal, S, E, *this));
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
return MatchOperand_Success;
}
if (Tok.is(AsmToken::EndOfStatement)) {
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
SMLoc E =
SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Zero register assumed, add a memory operand with ZERO as its base.
MipsAsmParser.cpp: Fix vg_leak in MipsOperand::CreateMem(). Mem.Base is managed by k_Memory itself. llvm-svn: 206293
2014-04-15 22:13:21 +08:00
// "Base" will be managed by k_Memory.
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
auto Base = MipsOperand::CreateGPRReg(0, getContext().getRegisterInfo(),
S, E, *this);
Operands.push_back(
MipsOperand::CreateMem(std::move(Base), IdVal, S, E, *this));
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
return MatchOperand_Success;
}
Error(Parser.getTok().getLoc(), "'(' expected");
return MatchOperand_ParseFail;
Initial assembler implementation of Mips load address macro This patch provides initial implementation of load address macro instruction for Mips. We have implemented two kinds of expansions with their variations depending on the size of immediate operand: 1) load address with immediate value directly: * la d,j => addiu d,$zero,j (for -32768 <= j <= 65535) * la d,j => lui d,hi16(j) ori d,d,lo16(j) (for any other 32 bit value of j) 2) load load address with register offset value * la d,j(s) => addiu d,s,j (for -32768 <= j <= 65535) * la d,j(s) => lui d,hi16(j) (for any other 32 bit value of j) ori d,d,lo16(j) addu d,d,s This patch does not cover the case when the address is loaded from the value of the label or function. Contributer: Vladimir Medic llvm-svn: 165561
2012-10-10 07:29:45 +08:00
}
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Eat the '(' token.
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
}
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Res = ParseAnyRegister(Operands);
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
if (Res != MatchOperand_Success)
return Res;
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
This patch introduces changes to MipsAsmParser register parsing routines. The code now follows more deterministic path and makes the code more efficient and easier to maintain. llvm-svn: 188264
2013-08-13 21:07:09 +08:00
if (Parser.getTok().isNot(AsmToken::RParen)) {
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
Error(Parser.getTok().getLoc(), "')' expected");
return MatchOperand_ParseFail;
}
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
SMLoc E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Eat the ')' token.
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
[C++] Use 'nullptr'. Target edition. llvm-svn: 207197
2014-04-25 13:30:21 +08:00
if (!IdVal)
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
IdVal = MCConstantExpr::Create(0, getContext());
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Replace the register operand with the memory operand.
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
std::unique_ptr<MipsOperand> op(
static_cast<MipsOperand *>(Operands.back().release()));
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Remove the register from the operands.
MipsAsmParser.cpp: Fix vg_leak in MipsOperand::CreateMem(). Mem.Base is managed by k_Memory itself. llvm-svn: 206293
2014-04-15 22:13:21 +08:00
// "op" will be managed by k_Memory.
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
Operands.pop_back();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Add the memory operand.
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
if (const MCBinaryExpr *BE = dyn_cast<MCBinaryExpr>(IdVal)) {
int64_t Imm;
if (IdVal->EvaluateAsAbsolute(Imm))
IdVal = MCConstantExpr::Create(Imm, getContext());
else if (BE->getLHS()->getKind() != MCExpr::SymbolRef)
IdVal = MCBinaryExpr::Create(BE->getOpcode(), BE->getRHS(), BE->getLHS(),
getContext());
}
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
Operands.push_back(MipsOperand::CreateMem(std::move(op), IdVal, S, E, *this));
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
return MatchOperand_Success;
}
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
bool MipsAsmParser::searchSymbolAlias(OperandVector &Operands) {
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
MCSymbol *Sym = getContext().LookupSymbol(Parser.getTok().getIdentifier());
if (Sym) {
SMLoc S = Parser.getTok().getLoc();
const MCExpr *Expr;
if (Sym->isVariable())
Expr = Sym->getVariableValue();
else
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
return false;
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
if (Expr->getKind() == MCExpr::SymbolRef) {
const MCSymbolRefExpr *Ref = static_cast<const MCSymbolRefExpr *>(Expr);
const StringRef DefSymbol = Ref->getSymbol().getName();
if (DefSymbol.startswith("$")) {
OperandMatchResultTy ResTy =
MatchAnyRegisterNameWithoutDollar(Operands, DefSymbol.substr(1), S);
[mips] Hoist Parser.Lex() calls out of MatchAnyRegisterNameWithoutDollar() Summary: No functional change Depends on D3222 Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3232 llvm-svn: 205295
2014-04-01 18:37:46 +08:00
if (ResTy == MatchOperand_Success) {
Parser.Lex();
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
return true;
[mips] Hoist Parser.Lex() calls out of MatchAnyRegisterNameWithoutDollar() Summary: No functional change Depends on D3222 Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3232 llvm-svn: 205295
2014-04-01 18:37:46 +08:00
} else if (ResTy == MatchOperand_ParseFail)
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
llvm_unreachable("Should never ParseFail");
return false;
}
} else if (Expr->getKind() == MCExpr::Constant) {
Parser.Lex();
const MCConstantExpr *Const = static_cast<const MCConstantExpr *>(Expr);
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
Operands.push_back(
MipsOperand::CreateImm(Const, S, Parser.getTok().getLoc(), *this));
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
return true;
}
[mips] Use ptr_rc to simplify definitions of base+index load/store instructions. Also, fix predicates. llvm-svn: 189432
2013-08-28 08:55:15 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
return false;
[mips] Use ptr_rc to simplify definitions of base+index load/store instructions. Also, fix predicates. llvm-svn: 189432
2013-08-28 08:55:15 +08:00
}
MipsAsmParser::OperandMatchResultTy
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsAsmParser::MatchAnyRegisterNameWithoutDollar(OperandVector &Operands,
StringRef Identifier,
SMLoc S) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
int Index = matchCPURegisterName(Identifier);
if (Index != -1) {
Operands.push_back(MipsOperand::CreateGPRReg(
Index, getContext().getRegisterInfo(), S, getLexer().getLoc(), *this));
return MatchOperand_Success;
}
[mips] Use ptr_rc to simplify definitions of base+index load/store instructions. Also, fix predicates. llvm-svn: 189432
2013-08-28 08:55:15 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Index = matchFPURegisterName(Identifier);
if (Index != -1) {
Operands.push_back(MipsOperand::CreateFGRReg(
Index, getContext().getRegisterInfo(), S, getLexer().getLoc(), *this));
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
return MatchOperand_Success;
}
This patch represents Mips utilization of r186388 code that alows asm matcher to emit mnemonics contain '.' characters. This makes asm parser code simpler and more efficient. llvm-svn: 186397
2013-07-16 18:07:14 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Index = matchFCCRegisterName(Identifier);
if (Index != -1) {
Operands.push_back(MipsOperand::CreateFCCReg(
Index, getContext().getRegisterInfo(), S, getLexer().getLoc(), *this));
return MatchOperand_Success;
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205229
2014-04-01 01:43:46 +08:00
}
[mips][msa] Direct Object Emission of INSERT.{B,H,W} instruction. INSERT is the first type of MSA instruction that requires a change to the way MSA registers are parsed. This happens because MSA registers may be suffixed by an index in the form of an immediate or a general purpose register. The changes to parseMSARegs reflect that requirement. llvm-svn: 192582
2013-10-14 19:49:30 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Index = matchACRegisterName(Identifier);
if (Index != -1) {
Operands.push_back(MipsOperand::CreateACCReg(
Index, getContext().getRegisterInfo(), S, getLexer().getLoc(), *this));
[mips][msa] Direct Object Emission of INSERT.{B,H,W} instruction. INSERT is the first type of MSA instruction that requires a change to the way MSA registers are parsed. This happens because MSA registers may be suffixed by an index in the form of an immediate or a general purpose register. The changes to parseMSARegs reflect that requirement. llvm-svn: 192582
2013-10-14 19:49:30 +08:00
return MatchOperand_Success;
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
}
[mips][msa] Direct Object Emission of INSERT.{B,H,W} instruction. INSERT is the first type of MSA instruction that requires a change to the way MSA registers are parsed. This happens because MSA registers may be suffixed by an index in the form of an immediate or a general purpose register. The changes to parseMSARegs reflect that requirement. llvm-svn: 192582
2013-10-14 19:49:30 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Index = matchMSA128RegisterName(Identifier);
if (Index != -1) {
Operands.push_back(MipsOperand::CreateMSA128Reg(
Index, getContext().getRegisterInfo(), S, getLexer().getLoc(), *this));
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205229
2014-04-01 01:43:46 +08:00
return MatchOperand_Success;
[mips][msa] Direct Object Emission of INSERT.{B,H,W} instruction. INSERT is the first type of MSA instruction that requires a change to the way MSA registers are parsed. This happens because MSA registers may be suffixed by an index in the form of an immediate or a general purpose register. The changes to parseMSARegs reflect that requirement. llvm-svn: 192582
2013-10-14 19:49:30 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Index = matchMSA128CtrlRegisterName(Identifier);
if (Index != -1) {
Operands.push_back(MipsOperand::CreateMSACtrlReg(
Index, getContext().getRegisterInfo(), S, getLexer().getLoc(), *this));
return MatchOperand_Success;
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
return MatchOperand_NoMatch;
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
}
MipsAsmParser::OperandMatchResultTy
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsAsmParser::MatchAnyRegisterWithoutDollar(OperandVector &Operands, SMLoc S) {
[mips] Use AsmLexer::peekTok() to resolve the conflict between $reg and $sym Summary: Parsing registers no longer consume the $ token before it's confirmed whether it really has a register or not, therefore it's no longer impossible to match symbols if registers were tried first. Depends on D3232 Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3233 llvm-svn: 205297
2014-04-01 18:40:14 +08:00
auto Token = Parser.getLexer().peekTok(false);
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
if (Token.is(AsmToken::Identifier)) {
DEBUG(dbgs() << ".. identifier\n");
StringRef Identifier = Token.getIdentifier();
[mips] Hoist Parser.Lex() calls out of MatchAnyRegisterNameWithoutDollar() Summary: No functional change Depends on D3222 Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3232 llvm-svn: 205295
2014-04-01 18:37:46 +08:00
OperandMatchResultTy ResTy =
MatchAnyRegisterNameWithoutDollar(Operands, Identifier, S);
return ResTy;
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
} else if (Token.is(AsmToken::Integer)) {
DEBUG(dbgs() << ".. integer\n");
Operands.push_back(MipsOperand::CreateNumericReg(
Token.getIntVal(), getContext().getRegisterInfo(), S, Token.getLoc(),
*this));
return MatchOperand_Success;
}
[mips] Use register operands instead of register classes in DSP instruction definitions. llvm-svn: 188343
2013-08-14 09:02:20 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
DEBUG(dbgs() << Parser.getTok().getKind() << "\n");
[mips] Use register operands instead of register classes in DSP instruction definitions. llvm-svn: 188343
2013-08-14 09:02:20 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
return MatchOperand_NoMatch;
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
}
[mips] Use register operands instead of register classes in DSP instruction definitions. llvm-svn: 188343
2013-08-14 09:02:20 +08:00
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsAsmParser::OperandMatchResultTy
MipsAsmParser::ParseAnyRegister(OperandVector &Operands) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
DEBUG(dbgs() << "ParseAnyRegister\n");
[mips] Use register operands instead of register classes in DSP instruction definitions. llvm-svn: 188343
2013-08-14 09:02:20 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
auto Token = Parser.getTok();
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
SMLoc S = Token.getLoc();
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
if (Token.isNot(AsmToken::Dollar)) {
DEBUG(dbgs() << ".. !$ -> try sym aliasing\n");
if (Token.is(AsmToken::Identifier)) {
if (searchSymbolAlias(Operands))
return MatchOperand_Success;
}
DEBUG(dbgs() << ".. !symalias -> NoMatch\n");
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
return MatchOperand_NoMatch;
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
}
DEBUG(dbgs() << ".. $\n");
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
[mips] Renamed ParseAnyRegisterWithoutDollar to MatchAnyRegisterWithoutDollar This is for consistency with other functions. The Parse* functions consume tokens and the Match* functions don't. No functional change. llvm-svn: 205305
2014-04-01 20:35:23 +08:00
OperandMatchResultTy ResTy = MatchAnyRegisterWithoutDollar(Operands, S);
[mips] Use AsmLexer::peekTok() to resolve the conflict between $reg and $sym Summary: Parsing registers no longer consume the $ token before it's confirmed whether it really has a register or not, therefore it's no longer impossible to match symbols if registers were tried first. Depends on D3232 Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3233 llvm-svn: 205297
2014-04-01 18:40:14 +08:00
if (ResTy == MatchOperand_Success) {
Parser.Lex(); // $
Parser.Lex(); // identifier
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
return ResTy;
[mips] Use register operands instead of register classes in DSP instruction definitions. llvm-svn: 188343
2013-08-14 09:02:20 +08:00
}
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
MipsAsmParser::OperandMatchResultTy
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsAsmParser::ParseImm(OperandVector &Operands) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
switch (getLexer().getKind()) {
default:
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
return MatchOperand_NoMatch;
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
case AsmToken::LParen:
case AsmToken::Minus:
case AsmToken::Plus:
case AsmToken::Integer:
[mips] Update MipsAsmParser so that it's possible to handle immediates that start with the binary operator NOT (~). Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D4158 llvm-svn: 211163
2014-06-18 21:55:18 +08:00
case AsmToken::Tilde:
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
case AsmToken::String:
break;
}
This patch implements Mips load/store instructions from/to coprocessor 2. Test cases are added. llvm-svn: 190780
2013-09-16 18:29:42 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
const MCExpr *IdVal;
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
SMLoc S = Parser.getTok().getLoc();
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
if (getParser().parseExpression(IdVal))
return MatchOperand_ParseFail;
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
SMLoc E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
Operands.push_back(MipsOperand::CreateImm(IdVal, S, E, *this));
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205229
2014-04-01 01:43:46 +08:00
return MatchOperand_Success;
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
}
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsAsmParser::OperandMatchResultTy
MipsAsmParser::ParseJumpTarget(OperandVector &Operands) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
DEBUG(dbgs() << "ParseJumpTarget\n");
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
SMLoc S = getLexer().getLoc();
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
// Integers and expressions are acceptable
OperandMatchResultTy ResTy = ParseImm(Operands);
if (ResTy != MatchOperand_NoMatch)
return ResTy;
Revert: [mips] Rewrite MipsAsmParser and MipsOperand.' due to buildbot errors in lld tests. It's currently unable to parse 'sym + imm' without surrounding parenthesis. llvm-svn: 205237
2014-04-01 02:51:43 +08:00
[mips] Use AsmLexer::peekTok() to resolve the conflict between $reg and $sym Summary: Parsing registers no longer consume the $ token before it's confirmed whether it really has a register or not, therefore it's no longer impossible to match symbols if registers were tried first. Depends on D3232 Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3233 llvm-svn: 205297
2014-04-01 18:40:14 +08:00
// Registers are a valid target and have priority over symbols.
ResTy = ParseAnyRegister(Operands);
if (ResTy != MatchOperand_NoMatch)
return ResTy;
[mips] Extend ParseJumpTarget to support the full symbol expression syntax. Summary: This should fix the issues the D3222 caused in lld. Testcase is based on the one that failed in the buildbot. Depends on D3233 Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3234 llvm-svn: 205298
2014-04-01 18:41:48 +08:00
const MCExpr *Expr = nullptr;
if (Parser.parseExpression(Expr)) {
// We have no way of knowing if a symbol was consumed so we must ParseFail
return MatchOperand_ParseFail;
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
}
[mips] Extend ParseJumpTarget to support the full symbol expression syntax. Summary: This should fix the issues the D3222 caused in lld. Testcase is based on the one that failed in the buildbot. Depends on D3233 Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3234 llvm-svn: 205298
2014-04-01 18:41:48 +08:00
Operands.push_back(
MipsOperand::CreateImm(Expr, S, getLexer().getLoc(), *this));
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
return MatchOperand_Success;
This patch tackles the problem of parsing Mips register names in the standalone assembler llvm-mc. Registers such as $A1 can represent either a 32 or 64 bit register based on the instruction using it. In addition, based on the abi, $T0 can represent different 32 bit registers. The problem is resolved by the Mips specific AsmParser td definitions changing to work together. Many cases of RegisterClass parameters are now RegisterOperand. Contributer: Vladimir Medic llvm-svn: 172284
2013-01-12 09:03:14 +08:00
}
This patch adds aliases for Mips sub instruction with immediate operands. Corresponding test cases are added. llvm-svn: 191734
2013-10-01 17:48:56 +08:00
MipsAsmParser::OperandMatchResultTy
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsAsmParser::parseInvNum(OperandVector &Operands) {
This patch adds aliases for Mips sub instruction with immediate operands. Corresponding test cases are added. llvm-svn: 191734
2013-10-01 17:48:56 +08:00
const MCExpr *IdVal;
// If the first token is '$' we may have register operand.
if (Parser.getTok().is(AsmToken::Dollar))
return MatchOperand_NoMatch;
SMLoc S = Parser.getTok().getLoc();
if (getParser().parseExpression(IdVal))
return MatchOperand_ParseFail;
const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(IdVal);
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
assert(MCE && "Unexpected MCExpr type.");
This patch adds aliases for Mips sub instruction with immediate operands. Corresponding test cases are added. llvm-svn: 191734
2013-10-01 17:48:56 +08:00
int64_t Val = MCE->getValue();
SMLoc E = SMLoc::getFromPointer(Parser.getTok().getLoc().getPointer() - 1);
Operands.push_back(MipsOperand::CreateImm(
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
MCConstantExpr::Create(0 - Val, getContext()), S, E, *this));
This patch adds aliases for Mips sub instruction with immediate operands. Corresponding test cases are added. llvm-svn: 191734
2013-10-01 17:48:56 +08:00
return MatchOperand_Success;
}
[mips][msa] Fix immediate value of LSA instruction as it was being wrongly encoded. The immediate field should be encoded as "imm - 1" as the CPU always adds one to that field. llvm-svn: 195004
2013-11-18 20:32:49 +08:00
MipsAsmParser::OperandMatchResultTy
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsAsmParser::ParseLSAImm(OperandVector &Operands) {
[mips][msa] Fix immediate value of LSA instruction as it was being wrongly encoded. The immediate field should be encoded as "imm - 1" as the CPU always adds one to that field. llvm-svn: 195004
2013-11-18 20:32:49 +08:00
switch (getLexer().getKind()) {
default:
return MatchOperand_NoMatch;
case AsmToken::LParen:
case AsmToken::Plus:
case AsmToken::Minus:
case AsmToken::Integer:
break;
}
const MCExpr *Expr;
SMLoc S = Parser.getTok().getLoc();
if (getParser().parseExpression(Expr))
return MatchOperand_ParseFail;
int64_t Val;
if (!Expr->EvaluateAsAbsolute(Val)) {
Error(S, "expected immediate value");
return MatchOperand_ParseFail;
}
// The LSA instruction allows a 2-bit unsigned immediate. For this reason
// and because the CPU always adds one to the immediate field, the allowed
// range becomes 1..4. We'll only check the range here and will deal
// with the addition/subtraction when actually decoding/encoding
// the instruction.
if (Val < 1 || Val > 4) {
Error(S, "immediate not in range (1..4)");
return MatchOperand_ParseFail;
}
[Mips] formatting through clang-format llvm-svn: 199853
2014-01-23 07:31:38 +08:00
Operands.push_back(
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
MipsOperand::CreateImm(Expr, S, Parser.getTok().getLoc(), *this));
[mips][msa] Fix immediate value of LSA instruction as it was being wrongly encoded. The immediate field should be encoded as "imm - 1" as the CPU always adds one to that field. llvm-svn: 195004
2013-11-18 20:32:49 +08:00
return MatchOperand_Success;
}
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
MCSymbolRefExpr::VariantKind MipsAsmParser::getVariantKind(StringRef Symbol) {
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
MCSymbolRefExpr::VariantKind VK =
StringSwitch<MCSymbolRefExpr::VariantKind>(Symbol)
.Case("hi", MCSymbolRefExpr::VK_Mips_ABS_HI)
.Case("lo", MCSymbolRefExpr::VK_Mips_ABS_LO)
.Case("gp_rel", MCSymbolRefExpr::VK_Mips_GPREL)
.Case("call16", MCSymbolRefExpr::VK_Mips_GOT_CALL)
.Case("got", MCSymbolRefExpr::VK_Mips_GOT)
.Case("tlsgd", MCSymbolRefExpr::VK_Mips_TLSGD)
.Case("tlsldm", MCSymbolRefExpr::VK_Mips_TLSLDM)
.Case("dtprel_hi", MCSymbolRefExpr::VK_Mips_DTPREL_HI)
.Case("dtprel_lo", MCSymbolRefExpr::VK_Mips_DTPREL_LO)
.Case("gottprel", MCSymbolRefExpr::VK_Mips_GOTTPREL)
.Case("tprel_hi", MCSymbolRefExpr::VK_Mips_TPREL_HI)
.Case("tprel_lo", MCSymbolRefExpr::VK_Mips_TPREL_LO)
.Case("got_disp", MCSymbolRefExpr::VK_Mips_GOT_DISP)
.Case("got_page", MCSymbolRefExpr::VK_Mips_GOT_PAGE)
.Case("got_ofst", MCSymbolRefExpr::VK_Mips_GOT_OFST)
.Case("hi(%neg(%gp_rel", MCSymbolRefExpr::VK_Mips_GPOFF_HI)
.Case("lo(%neg(%gp_rel", MCSymbolRefExpr::VK_Mips_GPOFF_LO)
[mips] Implement missing relocations in the integrated assembler. %got_hi, %got_lo, %call_hi, %call_lo, %higher, and %highest are now recognised by MipsAsmParser::getVariantKind(). To prevent future issues with missing entries in this StringSwitch, I've added an assertion to the default case. llvm-svn: 205200
2014-03-31 23:15:02 +08:00
.Case("got_hi", MCSymbolRefExpr::VK_Mips_GOT_HI16)
.Case("got_lo", MCSymbolRefExpr::VK_Mips_GOT_LO16)
.Case("call_hi", MCSymbolRefExpr::VK_Mips_CALL_HI16)
.Case("call_lo", MCSymbolRefExpr::VK_Mips_CALL_LO16)
.Case("higher", MCSymbolRefExpr::VK_Mips_HIGHER)
.Case("highest", MCSymbolRefExpr::VK_Mips_HIGHEST)
[mips][mips64r6] Add Relocations R_MIPS_PCHI16, R_MIPS_PCLO16 Differential Revision: http://reviews.llvm.org/D3860 llvm-svn: 209659
2014-05-27 22:58:51 +08:00
.Case("pcrel_hi", MCSymbolRefExpr::VK_Mips_PCREL_HI16)
.Case("pcrel_lo", MCSymbolRefExpr::VK_Mips_PCREL_LO16)
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
.Default(MCSymbolRefExpr::VK_None);
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
assert(VK != MCSymbolRefExpr::VK_None);
[mips] Implement missing relocations in the integrated assembler. %got_hi, %got_lo, %call_hi, %call_lo, %higher, and %highest are now recognised by MipsAsmParser::getVariantKind(). To prevent future issues with missing entries in this StringSwitch, I've added an assertion to the default case. llvm-svn: 205200
2014-03-31 23:15:02 +08:00
The Mips standalone assembler memory instruction support. This includes sb,sc,sh,sw,lb,lw,lbu,lh,lhu,ll,lw Test case included Contributer: Vladimir Medic llvm-svn: 163346
2012-09-07 04:00:02 +08:00
return VK;
}
The Mips standalone assembler fpu instruction support. Test cases included Contributer: Vladimir Medic llvm-svn: 163363
2012-09-07 08:23:42 +08:00
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
/// Sometimes (i.e. load/stores) the operand may be followed immediately by
/// either this.
/// ::= '(', register, ')'
/// handle it before we iterate so we don't get tripped up by the lack of
/// a comma.
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
bool MipsAsmParser::ParseParenSuffix(StringRef Name, OperandVector &Operands) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
if (getLexer().is(AsmToken::LParen)) {
Operands.push_back(
MipsOperand::CreateToken("(", getLexer().getLoc(), *this));
Parser.Lex();
if (ParseOperand(Operands, Name)) {
SMLoc Loc = getLexer().getLoc();
Parser.eatToEndOfStatement();
return Error(Loc, "unexpected token in argument list");
}
if (Parser.getTok().isNot(AsmToken::RParen)) {
SMLoc Loc = getLexer().getLoc();
Parser.eatToEndOfStatement();
return Error(Loc, "unexpected token, expected ')'");
}
Operands.push_back(
MipsOperand::CreateToken(")", getLexer().getLoc(), *this));
Parser.Lex();
}
return false;
}
/// Sometimes (i.e. in MSA) the operand may be followed immediately by
/// either one of these.
/// ::= '[', register, ']'
/// ::= '[', integer, ']'
/// handle it before we iterate so we don't get tripped up by the lack of
/// a comma.
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
bool MipsAsmParser::ParseBracketSuffix(StringRef Name,
OperandVector &Operands) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
if (getLexer().is(AsmToken::LBrac)) {
Operands.push_back(
MipsOperand::CreateToken("[", getLexer().getLoc(), *this));
Parser.Lex();
if (ParseOperand(Operands, Name)) {
SMLoc Loc = getLexer().getLoc();
Parser.eatToEndOfStatement();
return Error(Loc, "unexpected token in argument list");
}
if (Parser.getTok().isNot(AsmToken::RBrac)) {
SMLoc Loc = getLexer().getLoc();
Parser.eatToEndOfStatement();
return Error(Loc, "unexpected token, expected ']'");
}
Operands.push_back(
MipsOperand::CreateToken("]", getLexer().getLoc(), *this));
Parser.Lex();
}
return false;
}
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
bool MipsAsmParser::ParseInstruction(ParseInstructionInfo &Info, StringRef Name,
SMLoc NameLoc, OperandVector &Operands) {
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
DEBUG(dbgs() << "ParseInstruction\n");
This patch checks for valid mnemonics at the beginning of parseInstruction method, thus giving the user the right error message for non-existing instructions. llvm-svn: 186512
2013-07-17 23:00:42 +08:00
// Check if we have valid mnemonic
Split generated asm mnemonic matching table into a separate table for each asm variant. This removes the need to store the asm variant in each row of the single table that existed before. Shaves ~16K off the size of X86AsmParser.o. llvm-svn: 187026
2013-07-24 15:33:14 +08:00
if (!mnemonicIsValid(Name, 0)) {
This patch checks for valid mnemonics at the beginning of parseInstruction method, thus giving the user the right error message for non-existing instructions. llvm-svn: 186512
2013-07-17 23:00:42 +08:00
Parser.eatToEndOfStatement();
return Error(NameLoc, "Unknown instruction");
}
This patch represents Mips utilization of r186388 code that alows asm matcher to emit mnemonics contain '.' characters. This makes asm parser code simpler and more efficient. llvm-svn: 186397
2013-07-16 18:07:14 +08:00
// First operand in MCInst is instruction mnemonic.
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
Operands.push_back(MipsOperand::CreateToken(Name, NameLoc, *this));
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
// Read the remaining operands.
if (getLexer().isNot(AsmToken::EndOfStatement)) {
// Read the first operand.
This patch represents Mips utilization of r186388 code that alows asm matcher to emit mnemonics contain '.' characters. This makes asm parser code simpler and more efficient. llvm-svn: 186397
2013-07-16 18:07:14 +08:00
if (ParseOperand(Operands, Name)) {
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
SMLoc Loc = getLexer().getLoc();
MCParser: Update method names per coding guidelines. s/AddDirectiveHandler/addDirectiveHandler/ s/ParseMSInlineAsm/parseMSInlineAsm/ s/ParseIdentifier/parseIdentifier/ s/ParseStringToEndOfStatement/parseStringToEndOfStatement/ s/ParseEscapedString/parseEscapedString/ s/EatToEndOfStatement/eatToEndOfStatement/ s/ParseExpression/parseExpression/ s/ParseParenExpression/parseParenExpression/ s/ParseAbsoluteExpression/parseAbsoluteExpression/ s/CheckForValidSection/checkForValidSection/ http://llvm.org/docs/CodingStandards.html#name-types-functions-variables-and-enumerators-properly No functional change intended. llvm-svn: 175675
2013-02-21 06:21:35 +08:00
Parser.eatToEndOfStatement();
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
return Error(Loc, "unexpected token in argument list");
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
if (getLexer().is(AsmToken::LBrac) && ParseBracketSuffix(Name, Operands))
return true;
// AFAIK, parenthesis suffixes are never on the first operand
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
while (getLexer().is(AsmToken::Comma)) {
Parser.Lex(); // Eat the comma.
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
// Parse and remember the operand.
if (ParseOperand(Operands, Name)) {
SMLoc Loc = getLexer().getLoc();
MCParser: Update method names per coding guidelines. s/AddDirectiveHandler/addDirectiveHandler/ s/ParseMSInlineAsm/parseMSInlineAsm/ s/ParseIdentifier/parseIdentifier/ s/ParseStringToEndOfStatement/parseStringToEndOfStatement/ s/ParseEscapedString/parseEscapedString/ s/EatToEndOfStatement/eatToEndOfStatement/ s/ParseExpression/parseExpression/ s/ParseParenExpression/parseParenExpression/ s/ParseAbsoluteExpression/parseAbsoluteExpression/ s/CheckForValidSection/checkForValidSection/ http://llvm.org/docs/CodingStandards.html#name-types-functions-variables-and-enumerators-properly No functional change intended. llvm-svn: 175675
2013-02-21 06:21:35 +08:00
Parser.eatToEndOfStatement();
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
return Error(Loc, "unexpected token in argument list");
}
[mips] Rewrite MipsAsmParser and MipsOperand. Summary: Highlights: - Registers are resolved much later (by the render method). Prior to that point, GPR32's/GPR64's are GPR's regardless of register size. Similarly FGR32's/FGR64's/AFGR64's are FGR's regardless of register size or FR mode. Numeric registers can be anything. - All registers are parsed the same way everywhere (even when handling symbol aliasing) - One consequence is that all registers can be specified numerically almost anywhere (e.g. $fccX, $wX). The exception is symbol aliasing but that can be easily resolved. - Removes the need for the hasConsumedDollar hack - Parenthesis and Bracket suffixes are handled generically - Micromips instructions are parsed directly instead of going through the standard encodings first. - rdhwr accepts all 32 registers, and the following instructions that previously xfailed now work: ddiv, ddivu, div, divu, cvt.l.[ds], se[bh], wsbh, floor.w.[ds], c.ngl.d, c.sf.s, dsbh, dshd, madd.s, msub.s, nmadd.s, nmsub.s, swxc1 - Diagnostics involving registers point at the correct character (the $) - There's only one kind of immediate in MipsOperand. LSA immediates are handled by the predicate and renderer. Lowlights: - Hardcoded '$zero' in the div patterns is handled with a hack. MipsOperand::isReg() will return true for a k_RegisterIndex token with Index == 0 and getReg() will return ZERO for this case. Note that it doesn't return ZERO_64 on isGP64() targets. - I haven't cleaned up all of the now-unused functions. Some more of the generic parser could be removed too (integers and relocs for example). - insve.df needed a custom decoder to handle the implicit fourth operand that was needed to make it parse correctly. The difficulty was that the matcher expected a Token<'0'> but gets an Imm<0>. Adding an implicit zero solved this. Reviewers: matheusalmeida, vmedic Reviewed By: matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3222 llvm-svn: 205292
2014-04-01 18:35:28 +08:00
// Parse bracket and parenthesis suffixes before we iterate
if (getLexer().is(AsmToken::LBrac)) {
if (ParseBracketSuffix(Name, Operands))
return true;
} else if (getLexer().is(AsmToken::LParen) &&
ParseParenSuffix(Name, Operands))
return true;
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
}
}
if (getLexer().isNot(AsmToken::EndOfStatement)) {
SMLoc Loc = getLexer().getLoc();
MCParser: Update method names per coding guidelines. s/AddDirectiveHandler/addDirectiveHandler/ s/ParseMSInlineAsm/parseMSInlineAsm/ s/ParseIdentifier/parseIdentifier/ s/ParseStringToEndOfStatement/parseStringToEndOfStatement/ s/ParseEscapedString/parseEscapedString/ s/EatToEndOfStatement/eatToEndOfStatement/ s/ParseExpression/parseExpression/ s/ParseParenExpression/parseParenExpression/ s/ParseAbsoluteExpression/parseAbsoluteExpression/ s/CheckForValidSection/checkForValidSection/ http://llvm.org/docs/CodingStandards.html#name-types-functions-variables-and-enumerators-properly No functional change intended. llvm-svn: 175675
2013-02-21 06:21:35 +08:00
Parser.eatToEndOfStatement();
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
return Error(Loc, "unexpected token in argument list");
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Consume the EndOfStatement.
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
return false;
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
}
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
bool MipsAsmParser::reportParseError(StringRef ErrorMsg) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
SMLoc Loc = getLexer().getLoc();
Parser.eatToEndOfStatement();
return Error(Loc, ErrorMsg);
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
}
[mips] Add support for .cpload. Summary: This directive is used for setting up $gp in the beginning of a function. It expands to three instructions if PIC is enabled: lui $gp, %hi(_gp_disp) addui $gp, $gp, %lo(_gp_disp) addu $gp, $gp, $reg _gp_disp is a special symbol that the linker sets to the distance between the lui instruction and the context pointer (_gp). Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D3480 llvm-svn: 207637
2014-04-30 19:28:42 +08:00
bool MipsAsmParser::reportParseError(SMLoc Loc, StringRef ErrorMsg) {
return Error(Loc, ErrorMsg);
}
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
bool MipsAsmParser::parseSetNoAtDirective() {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Line should look like: ".set noat".
// set at reg to 0.
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165346
2012-10-06 07:55:28 +08:00
Options.setATReg(0);
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
// eat noat
Parser.Lex();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// If this is not the end of the statement, report an error.
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
if (getLexer().isNot(AsmToken::EndOfStatement)) {
reportParseError("unexpected token in statement");
return false;
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Consume the EndOfStatement.
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
return false;
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
bool MipsAsmParser::parseSetAtDirective() {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Line can be .set at - defaults to $1
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
// or .set at=$reg
ELF symbol table field st_other support, excluding visibility bits. Mips specific standalone assembler directive "set at". This directive changes the general purpose register that the assembler will use when given the symbolic register name $at. This does not include negative testing. That will come in a future patch. A side affect of this patch recognizes the different GPR register names for temporaries between old abi and new abi so a test case for that is included. Contributer: Vladimir Medic llvm-svn: 175686
2013-02-21 07:11:17 +08:00
int AtRegNo;
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
getParser().Lex();
if (getLexer().is(AsmToken::EndOfStatement)) {
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165346
2012-10-06 07:55:28 +08:00
Options.setATReg(1);
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Consume the EndOfStatement.
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
return false;
} else if (getLexer().is(AsmToken::Equal)) {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
getParser().Lex(); // Eat the '='.
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
if (getLexer().isNot(AsmToken::Dollar)) {
reportParseError("unexpected token in statement");
return false;
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Eat the '$'.
ELF symbol table field st_other support, excluding visibility bits. Mips specific standalone assembler directive "set at". This directive changes the general purpose register that the assembler will use when given the symbolic register name $at. This does not include negative testing. That will come in a future patch. A side affect of this patch recognizes the different GPR register names for temporaries between old abi and new abi so a test case for that is included. Contributer: Vladimir Medic llvm-svn: 175686
2013-02-21 07:11:17 +08:00
const AsmToken &Reg = Parser.getTok();
if (Reg.is(AsmToken::Identifier)) {
AtRegNo = matchCPURegisterName(Reg.getIdentifier());
} else if (Reg.is(AsmToken::Integer)) {
AtRegNo = Reg.getIntVal();
} else {
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
reportParseError("unexpected token in statement");
return false;
}
ELF symbol table field st_other support, excluding visibility bits. Mips specific standalone assembler directive "set at". This directive changes the general purpose register that the assembler will use when given the symbolic register name $at. This does not include negative testing. That will come in a future patch. A side affect of this patch recognizes the different GPR register names for temporaries between old abi and new abi so a test case for that is included. Contributer: Vladimir Medic llvm-svn: 175686
2013-02-21 07:11:17 +08:00
[mips] '.set at=$0' should be equivalent to '.set noat' Differential Revision: http://llvm-reviews.chandlerc.com/D3171 llvm-svn: 204714
2014-03-25 21:01:06 +08:00
if (AtRegNo < 0 || AtRegNo > 31) {
ELF symbol table field st_other support, excluding visibility bits. Mips specific standalone assembler directive "set at". This directive changes the general purpose register that the assembler will use when given the symbolic register name $at. This does not include negative testing. That will come in a future patch. A side affect of this patch recognizes the different GPR register names for temporaries between old abi and new abi so a test case for that is included. Contributer: Vladimir Medic llvm-svn: 175686
2013-02-21 07:11:17 +08:00
reportParseError("unexpected token in statement");
return false;
}
if (!Options.setATReg(AtRegNo)) {
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
reportParseError("unexpected token in statement");
return false;
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
getParser().Lex(); // Eat the register.
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
if (getLexer().isNot(AsmToken::EndOfStatement)) {
reportParseError("unexpected token in statement");
return false;
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
}
Parser.Lex(); // Consume the EndOfStatement.
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
return false;
} else {
reportParseError("unexpected token in statement");
return false;
}
}
bool MipsAsmParser::parseSetReorderDirective() {
Parser.Lex();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// If this is not the end of the statement, report an error.
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
if (getLexer().isNot(AsmToken::EndOfStatement)) {
reportParseError("unexpected token in statement");
return false;
}
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165346
2012-10-06 07:55:28 +08:00
Options.setReorder();
[mips] Assembly parser must invoke the target streamer to handle .set reorder macro. llvm-svn: 203459
2014-03-10 21:21:10 +08:00
getTargetStreamer().emitDirectiveSetReorder();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Consume the EndOfStatement.
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
return false;
}
bool MipsAsmParser::parseSetNoReorderDirective() {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex();
// If this is not the end of the statement, report an error.
if (getLexer().isNot(AsmToken::EndOfStatement)) {
reportParseError("unexpected token in statement");
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
return false;
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
}
Options.setNoreorder();
Implement the missing bits corresponding to .mips_hack_elf_flags. These were: * noreorder handling on the target object streamer and asm parser. * setting the initial flag bits based on the enabled features. * setting the elf header flag for micromips It is *really* depressing I am the one doing this instead of someone at mips actually taking the time to understand the infrastructure. llvm-svn: 200138
2014-01-26 14:57:13 +08:00
getTargetStreamer().emitDirectiveSetNoReorder();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Consume the EndOfStatement.
return false;
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
}
bool MipsAsmParser::parseSetMacroDirective() {
Parser.Lex();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// If this is not the end of the statement, report an error.
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
if (getLexer().isNot(AsmToken::EndOfStatement)) {
reportParseError("unexpected token in statement");
return false;
}
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165346
2012-10-06 07:55:28 +08:00
Options.setMacro();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Consume the EndOfStatement.
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
return false;
}
bool MipsAsmParser::parseSetNoMacroDirective() {
Parser.Lex();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// If this is not the end of the statement, report an error.
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
if (getLexer().isNot(AsmToken::EndOfStatement)) {
reportParseError("`noreorder' must be set before `nomacro'");
return false;
}
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165346
2012-10-06 07:55:28 +08:00
if (Options.isReorder()) {
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
reportParseError("`noreorder' must be set before `nomacro'");
return false;
}
Minor changes based on post commit review: Contributer: Vladimir Medic llvm-svn: 165346
2012-10-06 07:55:28 +08:00
Options.setNomacro();
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
Parser.Lex(); // Consume the EndOfStatement.
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
return false;
}
This patch enables the Mips .set directive to define aliases The .set directive in the Mips the assembler can be used to set the value of a symbol to an expression. This changes the symbol's value and type to conform to the expression's. Syntax: .set symbol, expression This patch implements the parsing of the above syntax and enables the parser to use defined symbols when parsing operands. Contributor: Vladimir Medic llvm-svn: 177667
2013-03-22 05:44:16 +08:00
[Mips] TargetStreamer Support for .set mips16. This patch updates .set mips16 support which affects the ELF ABI and its flags. In addition the patch uses a common interface for both the MipsTargetSteamer and MipsObjectStreamer that the assembler uses for both ELF and ASCII output for these directives. llvm-svn: 199851
2014-01-23 07:08:42 +08:00
bool MipsAsmParser::parseSetNoMips16Directive() {
Parser.Lex();
// If this is not the end of the statement, report an error.
if (getLexer().isNot(AsmToken::EndOfStatement)) {
reportParseError("unexpected token in statement");
return false;
}
// For now do nothing.
Parser.Lex(); // Consume the EndOfStatement.
return false;
}
This patch enables the Mips .set directive to define aliases The .set directive in the Mips the assembler can be used to set the value of a symbol to an expression. This changes the symbol's value and type to conform to the expression's. Syntax: .set symbol, expression This patch implements the parsing of the above syntax and enables the parser to use defined symbols when parsing operands. Contributor: Vladimir Medic llvm-svn: 177667
2013-03-22 05:44:16 +08:00
bool MipsAsmParser::parseSetAssignment() {
StringRef Name;
const MCExpr *Value;
if (Parser.parseIdentifier(Name))
reportParseError("expected identifier after .set");
if (getLexer().isNot(AsmToken::Comma))
return reportParseError("unexpected token in .set directive");
Mips assembler: Enable handling of nested expressions This patch allows the Mips assembler to parse and emit nested expressions as instruction operands. It also extends the expansion of memory instructions when an offset is given as an expression. Contributer: Vladimir Medic llvm-svn: 179657
2013-04-17 08:18:04 +08:00
Lex(); // Eat comma
This patch enables the Mips .set directive to define aliases The .set directive in the Mips the assembler can be used to set the value of a symbol to an expression. This changes the symbol's value and type to conform to the expression's. Syntax: .set symbol, expression This patch implements the parsing of the above syntax and enables the parser to use defined symbols when parsing operands. Contributor: Vladimir Medic llvm-svn: 177667
2013-03-22 05:44:16 +08:00
[Mips] formatting through clang-format llvm-svn: 199853
2014-01-23 07:31:38 +08:00
if (Parser.parseExpression(Value))
Mips assembler: Improve set register alias handling This patch solves the problem of numeric register values not being accepted: ../set_alias.s:1:11: error: expected valid expression after comma .set r4,$4 ^ The parsing of .set directive is changed and handling of symbols in code as well to enable this feature. The test example is added. Patch by Vladimir Medic llvm-svn: 182807
2013-05-29 06:21:05 +08:00
return reportParseError("expected valid expression after comma");
This patch enables the Mips .set directive to define aliases The .set directive in the Mips the assembler can be used to set the value of a symbol to an expression. This changes the symbol's value and type to conform to the expression's. Syntax: .set symbol, expression This patch implements the parsing of the above syntax and enables the parser to use defined symbols when parsing operands. Contributor: Vladimir Medic llvm-svn: 177667
2013-03-22 05:44:16 +08:00
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Check if the Name already exists as a symbol.
This patch enables the Mips .set directive to define aliases The .set directive in the Mips the assembler can be used to set the value of a symbol to an expression. This changes the symbol's value and type to conform to the expression's. Syntax: .set symbol, expression This patch implements the parsing of the above syntax and enables the parser to use defined symbols when parsing operands. Contributor: Vladimir Medic llvm-svn: 177667
2013-03-22 05:44:16 +08:00
MCSymbol *Sym = getContext().LookupSymbol(Name);
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
if (Sym)
This patch enables the Mips .set directive to define aliases The .set directive in the Mips the assembler can be used to set the value of a symbol to an expression. This changes the symbol's value and type to conform to the expression's. Syntax: .set symbol, expression This patch implements the parsing of the above syntax and enables the parser to use defined symbols when parsing operands. Contributor: Vladimir Medic llvm-svn: 177667
2013-03-22 05:44:16 +08:00
return reportParseError("symbol already defined");
Sym = getContext().GetOrCreateSymbol(Name);
Sym->setVariableValue(Value);
return false;
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
[mips] Hoist common functionality into a new function. Given that we support multiple directives that enable a particular feature (e.g. '.set mips16'), it's best to hoist that code into a new function so that we don't repeat the same pattern w.r.t parsing and handling error cases. No functional changes. llvm-svn: 204811
2014-03-26 22:26:27 +08:00
bool MipsAsmParser::parseSetFeature(uint64_t Feature) {
Parser.Lex();
if (getLexer().isNot(AsmToken::EndOfStatement))
return reportParseError("unexpected token in .set directive");
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
switch (Feature) {
default:
llvm_unreachable("Unimplemented feature");
case Mips::FeatureDSP:
setFeatureBits(Mips::FeatureDSP, "dsp");
getTargetStreamer().emitDirectiveSetDsp();
[mips] Hoist common functionality into a new function. Given that we support multiple directives that enable a particular feature (e.g. '.set mips16'), it's best to hoist that code into a new function so that we don't repeat the same pattern w.r.t parsing and handling error cases. No functional changes. llvm-svn: 204811
2014-03-26 22:26:27 +08:00
break;
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
case Mips::FeatureMicroMips:
getTargetStreamer().emitDirectiveSetMicroMips();
[mips] Hoist common functionality into a new function. Given that we support multiple directives that enable a particular feature (e.g. '.set mips16'), it's best to hoist that code into a new function so that we don't repeat the same pattern w.r.t parsing and handling error cases. No functional changes. llvm-svn: 204811
2014-03-26 22:26:27 +08:00
break;
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
case Mips::FeatureMips16:
getTargetStreamer().emitDirectiveSetMips16();
[mips] Hoist common functionality into a new function. Given that we support multiple directives that enable a particular feature (e.g. '.set mips16'), it's best to hoist that code into a new function so that we don't repeat the same pattern w.r.t parsing and handling error cases. No functional changes. llvm-svn: 204811
2014-03-26 22:26:27 +08:00
break;
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
case Mips::FeatureMips32r2:
setFeatureBits(Mips::FeatureMips32r2, "mips32r2");
getTargetStreamer().emitDirectiveSetMips32R2();
[mips] Hoist common functionality into a new function. Given that we support multiple directives that enable a particular feature (e.g. '.set mips16'), it's best to hoist that code into a new function so that we don't repeat the same pattern w.r.t parsing and handling error cases. No functional changes. llvm-svn: 204811
2014-03-26 22:26:27 +08:00
break;
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
case Mips::FeatureMips64:
setFeatureBits(Mips::FeatureMips64, "mips64");
getTargetStreamer().emitDirectiveSetMips64();
[mips] Add support to '.set mips64'. The '.set mips64' directive enables the feature Mips:FeatureMips64 from assembly. Note that it doesn't modify the ELF header as opposed to the use of -mips64 from the command-line. The reason for this is that we want to be as compatible as possible with existing assemblers like GAS. llvm-svn: 204817
2014-03-26 23:14:32 +08:00
break;
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
case Mips::FeatureMips64r2:
setFeatureBits(Mips::FeatureMips64r2, "mips64r2");
getTargetStreamer().emitDirectiveSetMips64R2();
[mips] Add support to '.set mips64r2'. The '.set mips64r2' directive enables the feature Mips:FeatureMips64r2 from assembly. Note that it doesn't modify the ELF header as opposed to the use of -mips64r2 from the command-line. The reason for this is that we want to be as compatible as possible with existing assemblers like GAS. llvm-svn: 204815
2014-03-26 22:52:22 +08:00
break;
[mips] Hoist common functionality into a new function. Given that we support multiple directives that enable a particular feature (e.g. '.set mips16'), it's best to hoist that code into a new function so that we don't repeat the same pattern w.r.t parsing and handling error cases. No functional changes. llvm-svn: 204811
2014-03-26 22:26:27 +08:00
}
return false;
}
[mips] Add support for .cpsetup Summary: Patch by Robert N. M. Watson His work was sponsored by: DARPA, AFRL Small corrections by myself. CC: theraven, matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3199 llvm-svn: 204924
2014-03-27 21:52:53 +08:00
bool MipsAsmParser::eatComma(StringRef ErrorStr) {
if (getLexer().isNot(AsmToken::Comma)) {
SMLoc Loc = getLexer().getLoc();
Parser.eatToEndOfStatement();
return Error(Loc, ErrorStr);
}
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
Parser.Lex(); // Eat the comma.
[mips] Add support for .cpsetup Summary: Patch by Robert N. M. Watson His work was sponsored by: DARPA, AFRL Small corrections by myself. CC: theraven, matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3199 llvm-svn: 204924
2014-03-27 21:52:53 +08:00
return true;
}
[mips] Add support for .cpload. Summary: This directive is used for setting up $gp in the beginning of a function. It expands to three instructions if PIC is enabled: lui $gp, %hi(_gp_disp) addui $gp, $gp, %lo(_gp_disp) addu $gp, $gp, $reg _gp_disp is a special symbol that the linker sets to the distance between the lui instruction and the context pointer (_gp). Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D3480 llvm-svn: 207637
2014-04-30 19:28:42 +08:00
bool MipsAsmParser::parseDirectiveCPLoad(SMLoc Loc) {
if (Options.isReorder())
Warning(Loc, ".cpload in reorder section");
// FIXME: Warn if cpload is used in Mips16 mode.
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
SmallVector<std::unique_ptr<MCParsedAsmOperand>, 1> Reg;
[mips] Add support for .cpload. Summary: This directive is used for setting up $gp in the beginning of a function. It expands to three instructions if PIC is enabled: lui $gp, %hi(_gp_disp) addui $gp, $gp, %lo(_gp_disp) addu $gp, $gp, $reg _gp_disp is a special symbol that the linker sets to the distance between the lui instruction and the context pointer (_gp). Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D3480 llvm-svn: 207637
2014-04-30 19:28:42 +08:00
OperandMatchResultTy ResTy = ParseAnyRegister(Reg);
if (ResTy == MatchOperand_NoMatch || ResTy == MatchOperand_ParseFail) {
reportParseError("expected register containing function address");
return false;
}
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
MipsOperand &RegOpnd = static_cast<MipsOperand &>(*Reg[0]);
if (!RegOpnd.isGPRAsmReg()) {
reportParseError(RegOpnd.getStartLoc(), "invalid register");
[mips] Add support for .cpload. Summary: This directive is used for setting up $gp in the beginning of a function. It expands to three instructions if PIC is enabled: lui $gp, %hi(_gp_disp) addui $gp, $gp, %lo(_gp_disp) addu $gp, $gp, $reg _gp_disp is a special symbol that the linker sets to the distance between the lui instruction and the context pointer (_gp). Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D3480 llvm-svn: 207637
2014-04-30 19:28:42 +08:00
return false;
}
AsmMatchers: Use unique_ptr to manage ownership of MCParsedAsmOperand I saw at least a memory leak or two from inspection (on probably untested error paths) and r206991, which was the original inspiration for this change. I ran this idea by Jim Grosbach a few weeks ago & he was OK with it. Since it's a basically mechanical patch that seemed sufficient - usual post-commit review, revert, etc, as needed. llvm-svn: 210427
2014-06-09 00:18:35 +08:00
getTargetStreamer().emitDirectiveCpload(RegOpnd.getGPR32Reg());
[mips] Add support for .cpload. Summary: This directive is used for setting up $gp in the beginning of a function. It expands to three instructions if PIC is enabled: lui $gp, %hi(_gp_disp) addui $gp, $gp, %lo(_gp_disp) addu $gp, $gp, $reg _gp_disp is a special symbol that the linker sets to the distance between the lui instruction and the context pointer (_gp). Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D3480 llvm-svn: 207637
2014-04-30 19:28:42 +08:00
return false;
}
[mips] Add support for .cpsetup Summary: Patch by Robert N. M. Watson His work was sponsored by: DARPA, AFRL Small corrections by myself. CC: theraven, matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3199 llvm-svn: 204924
2014-03-27 21:52:53 +08:00
bool MipsAsmParser::parseDirectiveCPSetup() {
unsigned FuncReg;
unsigned Save;
bool SaveIsReg = true;
[mips] Remove the last usage of parseRegister from MipsAsmParser. Summary: Added negative test case so that we can be sure we handle erroneous situations while parsing the .cpsetup directive. Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D3681 llvm-svn: 211160
2014-06-18 21:08:59 +08:00
SmallVector<std::unique_ptr<MCParsedAsmOperand>, 1> TmpReg;
OperandMatchResultTy ResTy = ParseAnyRegister(TmpReg);
if (ResTy == MatchOperand_NoMatch) {
reportParseError("expected register containing function address");
Parser.eatToEndOfStatement();
return false;
}
MipsOperand &FuncRegOpnd = static_cast<MipsOperand &>(*TmpReg[0]);
if (!FuncRegOpnd.isGPRAsmReg()) {
reportParseError(FuncRegOpnd.getStartLoc(), "invalid register");
Parser.eatToEndOfStatement();
return false;
}
FuncReg = FuncRegOpnd.getGPR32Reg();
TmpReg.clear();
[mips] Add support for .cpsetup Summary: Patch by Robert N. M. Watson His work was sponsored by: DARPA, AFRL Small corrections by myself. CC: theraven, matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3199 llvm-svn: 204924
2014-03-27 21:52:53 +08:00
if (!eatComma("expected comma parsing directive"))
return true;
[mips] Remove the last usage of parseRegister from MipsAsmParser. Summary: Added negative test case so that we can be sure we handle erroneous situations while parsing the .cpsetup directive. Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D3681 llvm-svn: 211160
2014-06-18 21:08:59 +08:00
ResTy = ParseAnyRegister(TmpReg);
if (ResTy == MatchOperand_NoMatch) {
[mips] Add support for .cpsetup Summary: Patch by Robert N. M. Watson His work was sponsored by: DARPA, AFRL Small corrections by myself. CC: theraven, matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3199 llvm-svn: 204924
2014-03-27 21:52:53 +08:00
const AsmToken &Tok = Parser.getTok();
if (Tok.is(AsmToken::Integer)) {
Save = Tok.getIntVal();
SaveIsReg = false;
Parser.Lex();
[mips] Remove the last usage of parseRegister from MipsAsmParser. Summary: Added negative test case so that we can be sure we handle erroneous situations while parsing the .cpsetup directive. Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D3681 llvm-svn: 211160
2014-06-18 21:08:59 +08:00
} else {
reportParseError("expected save register or stack offset");
Parser.eatToEndOfStatement();
return false;
}
} else {
MipsOperand &SaveOpnd = static_cast<MipsOperand &>(*TmpReg[0]);
if (!SaveOpnd.isGPRAsmReg()) {
reportParseError(SaveOpnd.getStartLoc(), "invalid register");
Parser.eatToEndOfStatement();
return false;
}
Save = SaveOpnd.getGPR32Reg();
}
[mips] Add support for .cpsetup Summary: Patch by Robert N. M. Watson His work was sponsored by: DARPA, AFRL Small corrections by myself. CC: theraven, matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3199 llvm-svn: 204924
2014-03-27 21:52:53 +08:00
if (!eatComma("expected comma parsing directive"))
return true;
StringRef Name;
if (Parser.parseIdentifier(Name))
reportParseError("expected identifier");
MCSymbol *Sym = getContext().GetOrCreateSymbol(Name);
[mips] Move expansion of .cpsetup to target streamer. Summary: There are two functional changes: 1) The directive is not expanded for the ASM->ASM code path. 2) If PIC is not set, there's no expansion for the ASM->OBJ code path (same behaviour as GAS). Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D3482 llvm-svn: 207741
2014-05-01 18:24:46 +08:00
getTargetStreamer().emitDirectiveCpsetup(FuncReg, Save, *Sym, SaveIsReg);
[mips] Add support for .cpsetup Summary: Patch by Robert N. M. Watson His work was sponsored by: DARPA, AFRL Small corrections by myself. CC: theraven, matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3199 llvm-svn: 204924
2014-03-27 21:52:53 +08:00
return false;
}
[mips] Add initial support for NaN2008 in the back-end. This is so that EF_MIPS_NAN2008 is set if we are using IEEE 754-2008 NaN encoding (-mnan=2008). This patch also adds support for parsing '.nan legacy' and '.nan 2008' assembly directives. The handling of these directives should match GAS' behaviour i.e., the last directive in use sets the ELF header bit (EF_MIPS_NAN2008). Differential Revision: http://reviews.llvm.org/D3346 llvm-svn: 206396
2014-04-16 23:48:55 +08:00
bool MipsAsmParser::parseDirectiveNaN() {
if (getLexer().isNot(AsmToken::EndOfStatement)) {
const AsmToken &Tok = Parser.getTok();
if (Tok.getString() == "2008") {
Parser.Lex();
getTargetStreamer().emitDirectiveNaN2008();
return false;
} else if (Tok.getString() == "legacy") {
Parser.Lex();
getTargetStreamer().emitDirectiveNaNLegacy();
return false;
}
}
// If we don't recognize the option passed to the .nan
// directive (e.g. no option or unknown option), emit an error.
reportParseError("invalid option in .nan directive");
return false;
}
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
bool MipsAsmParser::parseDirectiveSet() {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Get the next token.
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
const AsmToken &Tok = Parser.getTok();
if (Tok.getString() == "noat") {
return parseSetNoAtDirective();
} else if (Tok.getString() == "at") {
return parseSetAtDirective();
} else if (Tok.getString() == "reorder") {
return parseSetReorderDirective();
} else if (Tok.getString() == "noreorder") {
return parseSetNoReorderDirective();
} else if (Tok.getString() == "macro") {
return parseSetMacroDirective();
} else if (Tok.getString() == "nomacro") {
return parseSetNoMacroDirective();
[Mips] TargetStreamer Support for .set mips16. This patch updates .set mips16 support which affects the ELF ABI and its flags. In addition the patch uses a common interface for both the MipsTargetSteamer and MipsObjectStreamer that the assembler uses for both ELF and ASCII output for these directives. llvm-svn: 199851
2014-01-23 07:08:42 +08:00
} else if (Tok.getString() == "mips16") {
[mips] Hoist common functionality into a new function. Given that we support multiple directives that enable a particular feature (e.g. '.set mips16'), it's best to hoist that code into a new function so that we don't repeat the same pattern w.r.t parsing and handling error cases. No functional changes. llvm-svn: 204811
2014-03-26 22:26:27 +08:00
return parseSetFeature(Mips::FeatureMips16);
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
} else if (Tok.getString() == "nomips16") {
[Mips] TargetStreamer Support for .set mips16. This patch updates .set mips16 support which affects the ELF ABI and its flags. In addition the patch uses a common interface for both the MipsTargetSteamer and MipsObjectStreamer that the assembler uses for both ELF and ASCII output for these directives. llvm-svn: 199851
2014-01-23 07:08:42 +08:00
return parseSetNoMips16Directive();
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
} else if (Tok.getString() == "nomicromips") {
Replace .mips_hack_stocg with ".set micromips" and ".set nomicromips". This matches what gnu as does and implementing this is easier than arguing about it. llvm-svn: 199181
2014-01-14 12:25:13 +08:00
getTargetStreamer().emitDirectiveSetNoMicroMips();
Parser.eatToEndOfStatement();
return false;
} else if (Tok.getString() == "micromips") {
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
return parseSetFeature(Mips::FeatureMicroMips);
This patch implements .set mips32r2 directive and sets appropriate feature bits. It also introduces helper functions that are used to set and clear feature bits as necessary. This directive is a counterpart of -mips32r2 command line options with the exception that it does not influence elf header flags. The usage example is gives in test file. llvm-svn: 202807
2014-03-04 17:54:09 +08:00
} else if (Tok.getString() == "mips32r2") {
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
return parseSetFeature(Mips::FeatureMips32r2);
[mips] Add support to '.set mips64'. The '.set mips64' directive enables the feature Mips:FeatureMips64 from assembly. Note that it doesn't modify the ELF header as opposed to the use of -mips64 from the command-line. The reason for this is that we want to be as compatible as possible with existing assemblers like GAS. llvm-svn: 204817
2014-03-26 23:14:32 +08:00
} else if (Tok.getString() == "mips64") {
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
return parseSetFeature(Mips::FeatureMips64);
[mips] Add support to '.set mips64r2'. The '.set mips64r2' directive enables the feature Mips:FeatureMips64r2 from assembly. Note that it doesn't modify the ELF header as opposed to the use of -mips64r2 from the command-line. The reason for this is that we want to be as compatible as possible with existing assemblers like GAS. llvm-svn: 204815
2014-03-26 22:52:22 +08:00
} else if (Tok.getString() == "mips64r2") {
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
return parseSetFeature(Mips::FeatureMips64r2);
This patch implements .set dsp directive and sets appropriate feature bits.This directive is a counterpart of -mattr=dsp command line option with the exception that it does not influence elf header flags. The usage example is gives in test file. llvm-svn: 202966
2014-03-05 19:05:09 +08:00
} else if (Tok.getString() == "dsp") {
[mips] Clang-format MipsAsmParser. No functional changes. llvm-svn: 206878
2014-04-22 18:15:54 +08:00
return parseSetFeature(Mips::FeatureDSP);
This patch enables the Mips .set directive to define aliases The .set directive in the Mips the assembler can be used to set the value of a symbol to an expression. This changes the symbol's value and type to conform to the expression's. Syntax: .set symbol, expression This patch implements the parsing of the above syntax and enables the parser to use defined symbols when parsing operands. Contributor: Vladimir Medic llvm-svn: 177667
2013-03-22 05:44:16 +08:00
} else {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// It is just an identifier, look for an assignment.
This patch enables the Mips .set directive to define aliases The .set directive in the Mips the assembler can be used to set the value of a symbol to an expression. This changes the symbol's value and type to conform to the expression's. Syntax: .set symbol, expression This patch implements the parsing of the above syntax and enables the parser to use defined symbols when parsing operands. Contributor: Vladimir Medic llvm-svn: 177667
2013-03-22 05:44:16 +08:00
parseSetAssignment();
return false;
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
}
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
return true;
}
[mips] Rename function in MipsAsmParser. parseDirectiveWord is a generic function that parses an expression which means there's no need for it to have such an specific name. Renaming it to parseDataDirective so that it can also be used to handle .dword directives[1]. [1]To be added in a follow up commit. No functional changes. llvm-svn: 204818
2014-03-26 23:24:36 +08:00
/// parseDataDirective
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
/// ::= .word [ expression (, expression)* ]
[mips] Rename function in MipsAsmParser. parseDirectiveWord is a generic function that parses an expression which means there's no need for it to have such an specific name. Renaming it to parseDataDirective so that it can also be used to handle .dword directives[1]. [1]To be added in a follow up commit. No functional changes. llvm-svn: 204818
2014-03-26 23:24:36 +08:00
bool MipsAsmParser::parseDataDirective(unsigned Size, SMLoc L) {
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
if (getLexer().isNot(AsmToken::EndOfStatement)) {
for (;;) {
const MCExpr *Value;
MCParser: Update method names per coding guidelines. s/AddDirectiveHandler/addDirectiveHandler/ s/ParseMSInlineAsm/parseMSInlineAsm/ s/ParseIdentifier/parseIdentifier/ s/ParseStringToEndOfStatement/parseStringToEndOfStatement/ s/ParseEscapedString/parseEscapedString/ s/EatToEndOfStatement/eatToEndOfStatement/ s/ParseExpression/parseExpression/ s/ParseParenExpression/parseParenExpression/ s/ParseAbsoluteExpression/parseAbsoluteExpression/ s/CheckForValidSection/checkForValidSection/ http://llvm.org/docs/CodingStandards.html#name-types-functions-variables-and-enumerators-properly No functional change intended. llvm-svn: 175675
2013-02-21 06:21:35 +08:00
if (getParser().parseExpression(Value))
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
return true;
getParser().getStreamer().EmitValue(Value, Size);
if (getLexer().is(AsmToken::EndOfStatement))
break;
// FIXME: Improve diagnostic.
if (getLexer().isNot(AsmToken::Comma))
return Error(L, "unexpected token in directive");
Parser.Lex();
}
}
Parser.Lex();
return false;
}
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
/// parseDirectiveGpWord
/// ::= .gpword local_sym
bool MipsAsmParser::parseDirectiveGpWord() {
const MCExpr *Value;
// EmitGPRel32Value requires an expression, so we are using base class
// method to evaluate the expression.
if (getParser().parseExpression(Value))
return true;
getParser().getStreamer().EmitGPRel32Value(Value);
Fix bug in .gpword directive parsing. llvm-svn: 194570
2013-11-13 21:18:04 +08:00
if (getLexer().isNot(AsmToken::EndOfStatement))
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
return Error(getLexer().getLoc(), "unexpected token in directive");
Fix bug in .gpword directive parsing. llvm-svn: 194570
2013-11-13 21:18:04 +08:00
Parser.Lex(); // Eat EndOfStatement token.
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
return false;
}
Capitalize the D in parseDirectiveGpDWord. DWord seems to be the canonical way to camel case dword in llvm. Thanks to Daniel Sander for noticing. llvm-svn: 205191
2014-03-31 22:15:07 +08:00
/// parseDirectiveGpDWord
Parse .gpdword and convert another llc -filetype=obj test. llvm-svn: 205028
2014-03-29 02:50:26 +08:00
/// ::= .gpdword local_sym
Capitalize the D in parseDirectiveGpDWord. DWord seems to be the canonical way to camel case dword in llvm. Thanks to Daniel Sander for noticing. llvm-svn: 205191
2014-03-31 22:15:07 +08:00
bool MipsAsmParser::parseDirectiveGpDWord() {
Parse .gpdword and convert another llc -filetype=obj test. llvm-svn: 205028
2014-03-29 02:50:26 +08:00
const MCExpr *Value;
// EmitGPRel64Value requires an expression, so we are using base class
// method to evaluate the expression.
if (getParser().parseExpression(Value))
return true;
getParser().getStreamer().EmitGPRel64Value(Value);
if (getLexer().isNot(AsmToken::EndOfStatement))
return Error(getLexer().getLoc(), "unexpected token in directive");
Parser.Lex(); // Eat EndOfStatement token.
return false;
}
[Mips] TargetStreamer Support for .abicalls and .set pic0. This patch adds .abicalls and .set pic0 support which affects the ELF ABI and its flags. In addition the patch uses a common interface for both the MipsTargetSteamer and MipsObjectStreamer that both the integrated and standalone assemblers will use for the output for these directives. llvm-svn: 198646
2014-01-07 07:27:31 +08:00
bool MipsAsmParser::parseDirectiveOption() {
// Get the option token.
AsmToken Tok = Parser.getTok();
// At the moment only identifiers are supported.
if (Tok.isNot(AsmToken::Identifier)) {
Error(Parser.getTok().getLoc(), "unexpected token in .option directive");
Parser.eatToEndOfStatement();
return false;
}
StringRef Option = Tok.getIdentifier();
if (Option == "pic0") {
getTargetStreamer().emitDirectiveOptionPic0();
Parser.Lex();
if (Parser.getTok().isNot(AsmToken::EndOfStatement)) {
Error(Parser.getTok().getLoc(),
"unexpected token in .option pic0 directive");
Parser.eatToEndOfStatement();
}
return false;
}
[mips] Add support for '.option pic2'. The directive '.option pic2' enables PIC from assembly source. At the moment none of the macros/directives check the PIC bit but that's going to be fixed relatively soon. For example, the expansion of macros like 'la' depend on the relocation model. llvm-svn: 204803
2014-03-26 21:40:29 +08:00
if (Option == "pic2") {
getTargetStreamer().emitDirectiveOptionPic2();
Parser.Lex();
if (Parser.getTok().isNot(AsmToken::EndOfStatement)) {
Error(Parser.getTok().getLoc(),
"unexpected token in .option pic2 directive");
Parser.eatToEndOfStatement();
}
return false;
}
[Mips] TargetStreamer Support for .abicalls and .set pic0. This patch adds .abicalls and .set pic0 support which affects the ELF ABI and its flags. In addition the patch uses a common interface for both the MipsTargetSteamer and MipsObjectStreamer that both the integrated and standalone assemblers will use for the output for these directives. llvm-svn: 198646
2014-01-07 07:27:31 +08:00
// Unknown option.
Warning(Parser.getTok().getLoc(), "unknown option in .option directive");
Parser.eatToEndOfStatement();
return false;
}
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
bool MipsAsmParser::ParseDirective(AsmToken DirectiveID) {
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
StringRef IDVal = DirectiveID.getString();
[mips] Add support for .cpload. Summary: This directive is used for setting up $gp in the beginning of a function. It expands to three instructions if PIC is enabled: lui $gp, %hi(_gp_disp) addui $gp, $gp, %lo(_gp_disp) addu $gp, $gp, $reg _gp_disp is a special symbol that the linker sets to the distance between the lui instruction and the context pointer (_gp). Reviewers: dsanders Reviewed By: dsanders Differential Revision: http://reviews.llvm.org/D3480 llvm-svn: 207637
2014-04-30 19:28:42 +08:00
if (IDVal == ".cpload")
return parseDirectiveCPLoad(DirectiveID.getLoc());
[mips] Add support to the '.dword' directive. The '.dword' directive accepts a list of expressions and emits them in 8-byte chunks in successive locations. llvm-svn: 204822
2014-03-26 23:44:18 +08:00
if (IDVal == ".dword") {
parseDataDirective(8, DirectiveID.getLoc());
return false;
}
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
if (IDVal == ".ent") {
// Ignore this directive for now.
The Mips standalone assembler intial directive support. Actually these are just stubs for parsing the directives. Semantic support will come later. Test cases included Contributer: Vladimir Medic llvm-svn: 163364
2012-09-07 08:48:02 +08:00
Parser.Lex();
return false;
}
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
if (IDVal == ".end") {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Ignore this directive for now.
The Mips standalone assembler intial directive support. Actually these are just stubs for parsing the directives. Semantic support will come later. Test cases included Contributer: Vladimir Medic llvm-svn: 163364
2012-09-07 08:48:02 +08:00
Parser.Lex();
return false;
}
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
if (IDVal == ".frame") {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Ignore this directive for now.
MCParser: Update method names per coding guidelines. s/AddDirectiveHandler/addDirectiveHandler/ s/ParseMSInlineAsm/parseMSInlineAsm/ s/ParseIdentifier/parseIdentifier/ s/ParseStringToEndOfStatement/parseStringToEndOfStatement/ s/ParseEscapedString/parseEscapedString/ s/EatToEndOfStatement/eatToEndOfStatement/ s/ParseExpression/parseExpression/ s/ParseParenExpression/parseParenExpression/ s/ParseAbsoluteExpression/parseAbsoluteExpression/ s/CheckForValidSection/checkForValidSection/ http://llvm.org/docs/CodingStandards.html#name-types-functions-variables-and-enumerators-properly No functional change intended. llvm-svn: 175675
2013-02-21 06:21:35 +08:00
Parser.eatToEndOfStatement();
The Mips standalone assembler intial directive support. Actually these are just stubs for parsing the directives. Semantic support will come later. Test cases included Contributer: Vladimir Medic llvm-svn: 163364
2012-09-07 08:48:02 +08:00
return false;
}
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
if (IDVal == ".set") {
This patch is a partial implementation of mips .set assembler directive. Directive is defined as follows: .set option The patch implements following options at - lets the assembler use the $at register for macros, but generates warnings if the source program uses $at noat - let source programs use $at without issuingwarnings. noreorder - prevents the assembler from reordering machine language instructions. nomacro - causes the assembler to print a warning whenever an assembler operation generates more than one machine language instruction. macro - lets the assembler generate multiple machine instructions from a single assembler instruction reorder - lets the assembler reorder machine language instructions to improve performance The above variants are parsed and their boolean values set or unset. The code to actually use them will come later. Following options are not implemented yet: nomips16 nomicromips move nomove Contributer: Vladimir Medic llvm-svn: 165194
2012-10-04 10:29:46 +08:00
return parseDirectiveSet();
The Mips standalone assembler intial directive support. Actually these are just stubs for parsing the directives. Semantic support will come later. Test cases included Contributer: Vladimir Medic llvm-svn: 163364
2012-09-07 08:48:02 +08:00
}
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
if (IDVal == ".fmask") {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Ignore this directive for now.
MCParser: Update method names per coding guidelines. s/AddDirectiveHandler/addDirectiveHandler/ s/ParseMSInlineAsm/parseMSInlineAsm/ s/ParseIdentifier/parseIdentifier/ s/ParseStringToEndOfStatement/parseStringToEndOfStatement/ s/ParseEscapedString/parseEscapedString/ s/EatToEndOfStatement/eatToEndOfStatement/ s/ParseExpression/parseExpression/ s/ParseParenExpression/parseParenExpression/ s/ParseAbsoluteExpression/parseAbsoluteExpression/ s/CheckForValidSection/checkForValidSection/ http://llvm.org/docs/CodingStandards.html#name-types-functions-variables-and-enumerators-properly No functional change intended. llvm-svn: 175675
2013-02-21 06:21:35 +08:00
Parser.eatToEndOfStatement();
The Mips standalone assembler intial directive support. Actually these are just stubs for parsing the directives. Semantic support will come later. Test cases included Contributer: Vladimir Medic llvm-svn: 163364
2012-09-07 08:48:02 +08:00
return false;
}
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
if (IDVal == ".mask") {
Mips assembler: formatting and comment changes. This patch should not have any functional changes. llvm-svn: 179737
2013-04-18 08:41:53 +08:00
// Ignore this directive for now.
MCParser: Update method names per coding guidelines. s/AddDirectiveHandler/addDirectiveHandler/ s/ParseMSInlineAsm/parseMSInlineAsm/ s/ParseIdentifier/parseIdentifier/ s/ParseStringToEndOfStatement/parseStringToEndOfStatement/ s/ParseEscapedString/parseEscapedString/ s/EatToEndOfStatement/eatToEndOfStatement/ s/ParseExpression/parseExpression/ s/ParseParenExpression/parseParenExpression/ s/ParseAbsoluteExpression/parseAbsoluteExpression/ s/CheckForValidSection/checkForValidSection/ http://llvm.org/docs/CodingStandards.html#name-types-functions-variables-and-enumerators-properly No functional change intended. llvm-svn: 175675
2013-02-21 06:21:35 +08:00
Parser.eatToEndOfStatement();
The Mips standalone assembler intial directive support. Actually these are just stubs for parsing the directives. Semantic support will come later. Test cases included Contributer: Vladimir Medic llvm-svn: 163364
2012-09-07 08:48:02 +08:00
return false;
}
[mips] Add initial support for NaN2008 in the back-end. This is so that EF_MIPS_NAN2008 is set if we are using IEEE 754-2008 NaN encoding (-mnan=2008). This patch also adds support for parsing '.nan legacy' and '.nan 2008' assembly directives. The handling of these directives should match GAS' behaviour i.e., the last directive in use sets the ELF header bit (EF_MIPS_NAN2008). Differential Revision: http://reviews.llvm.org/D3346 llvm-svn: 206396
2014-04-16 23:48:55 +08:00
if (IDVal == ".nan")
return parseDirectiveNaN();
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
if (IDVal == ".gpword") {
Implement gpword directive for mips, test case added. Stype changes using clang-format are also included. llvm-svn: 194145
2013-11-06 19:27:05 +08:00
parseDirectiveGpWord();
The Mips standalone assembler intial directive support. Actually these are just stubs for parsing the directives. Semantic support will come later. Test cases included Contributer: Vladimir Medic llvm-svn: 163364
2012-09-07 08:48:02 +08:00
return false;
}
Parse .gpdword and convert another llc -filetype=obj test. llvm-svn: 205028
2014-03-29 02:50:26 +08:00
if (IDVal == ".gpdword") {
Capitalize the D in parseDirectiveGpDWord. DWord seems to be the canonical way to camel case dword in llvm. Thanks to Daniel Sander for noticing. llvm-svn: 205191
2014-03-31 22:15:07 +08:00
parseDirectiveGpDWord();
Parse .gpdword and convert another llc -filetype=obj test. llvm-svn: 205028
2014-03-29 02:50:26 +08:00
return false;
}
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
if (IDVal == ".word") {
[mips] Rename function in MipsAsmParser. parseDirectiveWord is a generic function that parses an expression which means there's no need for it to have such an specific name. Renaming it to parseDataDirective so that it can also be used to handle .dword directives[1]. [1]To be added in a follow up commit. No functional changes. llvm-svn: 204818
2014-03-26 23:24:36 +08:00
parseDataDirective(4, DirectiveID.getLoc());
This patch implements parsing the .word directive for the Mips assembler. Contributer: Vladimir Medic llvm-svn: 173407
2013-01-25 09:31:34 +08:00
return false;
}
[Mips] TargetStreamer Support for .abicalls and .set pic0. This patch adds .abicalls and .set pic0 support which affects the ELF ABI and its flags. In addition the patch uses a common interface for both the MipsTargetSteamer and MipsObjectStreamer that both the integrated and standalone assemblers will use for the output for these directives. llvm-svn: 198646
2014-01-07 07:27:31 +08:00
if (IDVal == ".option")
return parseDirectiveOption();
if (IDVal == ".abicalls") {
getTargetStreamer().emitDirectiveAbiCalls();
if (Parser.getTok().isNot(AsmToken::EndOfStatement)) {
Error(Parser.getTok().getLoc(), "unexpected token in directive");
// Clear line
Parser.eatToEndOfStatement();
}
return false;
}
[mips] Add support for .cpsetup Summary: Patch by Robert N. M. Watson His work was sponsored by: DARPA, AFRL Small corrections by myself. CC: theraven, matheusalmeida Differential Revision: http://llvm-reviews.chandlerc.com/D3199 llvm-svn: 204924
2014-03-27 21:52:53 +08:00
if (IDVal == ".cpsetup")
return parseDirectiveCPSetup();
Add the skeleton of an asm parser for mips. llvm-svn: 147923
2012-01-11 11:56:41 +08:00
return true;
}
extern "C" void LLVMInitializeMipsAsmParser() {
RegisterMCAsmParser<MipsAsmParser> X(TheMipsTarget);
RegisterMCAsmParser<MipsAsmParser> Y(TheMipselTarget);
RegisterMCAsmParser<MipsAsmParser> A(TheMips64Target);
RegisterMCAsmParser<MipsAsmParser> B(TheMips64elTarget);
}
Mips specific llvm assembler support for ALU instructions. This includes register support. Test case included. Contributer: Vladimir Medic llvm-svn: 163268
2012-09-06 07:34:03 +08:00
#define GET_REGISTER_MATCHER
#define GET_MATCHER_IMPLEMENTATION
#include "MipsGenAsmMatcher.inc"