LLVM 中物理寄存器的表示

LLVM 物理寄存器的表示

LLVM中是如何定义，描述物理寄存器的？在回答这个问题之前，我们先看看与物理寄存器有关数据结构的定义，由于这些定义主要写在td文件中，我们先用table-gen输出完整定义。

table-gen 生成register的信息

在下面的介绍中，我们会用到table-gen的数据，这里先列出相关数据生成命令。

$llvm-tblgen --gen-register-info -I ./llvm/lib/Target/X86 -I ./llvm/include  -
I ./llvm/lib/Target ./llvm/lib/Target/X86/X86.td -o outreg.inf

$llvm-tblgen --gen-register-bank -I ./llvm/lib/Target/X86 -I ./llvm/include  -
I ./llvm/lib/Target ./llvm/lib/Target/X86/X86.td -o outreg.bank

$llvm-tblgen  -I ./llvm/lib/Target/X86 -I ./llvm/include  -
I ./llvm/lib/Target ./llvm/lib/Target/X86/X86.td -o out.info

认识MCRegisterInfo

LLVM中主要通过MCRegisterInfo对物理寄存器相关信息进行访问和遍历。

class MCRegisterInfo {
   
  const MCRegisterDesc *Desc;                 // Pointer to the descriptor array
  unsigned NumRegs;                           // Number of entries in the array
  MCRegister RAReg;                           // Return Address register
  MCRegister PCReg;                           // Program Counter register
  const MCRegisterClass *Classes;             // Pointer to the regclass array
  unsigned NumClasses;                        // Number of entries in the array
  unsigned NumRegUnits;                       // Number of regunits.
  const MCPhysReg (*RegUnitRoots)[2];         // Pointer to regunit root table.
  const MCPhysReg *DiffLists;                 // Pointer to the difflists array
  const LaneBitmask *RegUnitMaskSequences;    // Pointer to lane mask sequences
                                              // for register units.
  const char *RegStrings;                     // Pointer to the string table.
  const char *RegClassStrings;                // Pointer to the class strings.
  const uint16_t *SubRegIndices;              // Pointer to the subreg lookup
                                              // array.
  // struct SubRegCoveredBits { uint16_t Offset; uint16_t Size; };
  const SubRegCoveredBits *SubRegIdxRanges;   // Pointer to the subreg covered
                                              // bit ranges array.
  unsigned NumSubRegIndices;                  // Number of subreg indices.
  const uint16_t *RegEncodingTable;           // Pointer to array of register
                                              // encodings.
...

public:
  /// DiffListIterator - Base iterator class that can traverse the
  /// differentially encoded register and regunit lists in DiffLists.
  /// Don't use this class directly, use one of the specialized sub-classes
  /// defined below.
  class DiffListIterator {
   
    uint16_t Val = 0;
    const MCPhysReg *List = nullptr;
  }
  ...
}

如何描述寄存器

每个寄存器的描述被放在一张Desc表里，一共有NumRegs个寄存器。

class MCRegisterInfo {
   
  const MCRegisterDesc *Desc;                 // Pointer to the descriptor array
  unsigned NumRegs;                           // Number of entries in the array
...}

单个寄存器描述信息如下：

struct MCRegisterDesc {
   
                      //   extern const char X86RegStrings[] = {
   
                      //     /* 0 */ "XMM10\0"
                      //     /* 6 */ "YMM10\0"
                      //    /* 12 */ "ZMM10\0"  ... }
  uint32_t Name;      // 在X86RegStrings字符串表中的偏移，比如YMM10为6
  uint32_t SubRegs;   // Sub-register set (在DiffLists中的偏移)
  uint32_t SuperRegs; // Super-register set (在DiffLists中的偏移)

  // Offset into MCRI::SubRegIndices of a list of sub-register indices for each
  // sub-register in SubRegs.
  uint32_t SubRegIndices; // 该Reg的SubRegIndices在 MCRI::SubRegIndices 中的偏移

  // RegUnits - Points to the list of register units. 
  // 低 4 bits 存储 Scale， 其它 bits 表示在 DiffLists 中的偏移. See MCRegUnitIterator.
  uint32_t RegUnits;

  /// Index into list with lane mask sequences. The sequence contains a lanemask
  /// for every register unit.
  uint16_t RegUnitLaneMasks;
};

X86后端如何初始化MCRegisterInfo

在outreg.inf中我们发现有一段初始化MCRegisterInfo 的代码，以此我们很容易将table-gen中生成的一些表项与MCRegisterInfo中的成员对应起来：

 static inline void InitX86MCRegisterInfo(MCRegisterInfo *RI, unsigned RA,
                                          unsigned DwarfFlavour = 0,
                                          unsigned EHFlavour = 0,
                                          unsigned PC = 0) {
   
     RI->InitMCRegisterInfo(X86RegDesc, 292, RA, PC, X86MCRegisterClasses, 126,
                            X86RegUnitRoots, 173, X86RegDiffLists, X86LaneMaskLists,
                            X86RegStrings, X86RegClassStrings,
                            X86SubRegIdxLists, 11, X86SubRegIdxRanges, X86RegEncodingTable);
     ...
}

Reg 与 RegUnit

Reg

Reg就是我们平常在后端td文件中定义的寄存器 （X86: X86RegisterInfo.td）
如RCX，其主要通过X86Reg类来定义。

lib/Target/X86/X86RegisterInfo.td
164 let SubRegIndices = [sub_32bit] in {
   
def RCX : X86Reg<"rcx", 1, [ECX]>, DwarfRegNum<[2, -2, -2]>;
}

Reg的enum 值通按照其字符串加数字的排列顺序确定：

namespace llvm {
   
namespace X86 {
   
enum {
   
  NoRegister,
  AH = 1,
  AL = 2, .