0

我是 linux 新手，在 Ubuntu 虚拟盒环境中使用终端。我无法弄清楚这些错误发生的原因和原因，而且它们似乎与缺少的 ';' 不匹配 或#define 冲突。

以下是我从 gcc 编译器得到的错误：

my_predictor.h:82:42: error: template argument 2 is invalid
my_predictor.h:82:42: error: template argument 4 is invalid
my_predictor.h:84:9: error: expected unqualified-id before ‘for’
my_predictor.h:84:23: error: ‘g’ does not name a type
my_predictor.h:84:44: error: ‘g’ does not name a type
my_predictor.h:91:42: error: template argument 2 is invalid
my_predictor.h:91:42: error: template argument 4 is invalid
my_predictor.h:93:9: error: expected unqualified-id before ‘for’
my_predictor.h:93:23: error: ‘f’ does not name a type
my_predictor.h:93:39: error: ‘f’ does not name a type
my_predictor.h: In member function ‘virtual branch_update* local_predictor::predict(branch_info&)’:
my_predictor.h:107:57: error: invalid types ‘int[int]’ for array subscript
my_predictor.h:111:40: error: invalid types ‘int[int]’ for array subscript
my_predictor.h: In member function ‘virtual void local_predictor::update(branch_update*, bool, unsigned int)’:
my_predictor.h:121:62: error: invalid types ‘int[int]’ for array subscript
my_predictor.h:124:46: error: invalid types ‘int[int]’ for array subscript
my_predictor.h:125:33: error: invalid types ‘int[int]’ for array subscript
my_predictor.h:127:41: error: invalid types ‘int[int]’ for array subscript
my_predictor.h:128:29: error: invalid types ‘int[int]’ for array subscript
my_predictor.h:131:33: error: invalid types ‘int[int]’ for array subscript
my_predictor.h:132:30: error: invalid types ‘int[int]’ for array subscript
my_predictor.h:135:33: error: invalid types ‘int[int]’ for array subscript
my_predictor.h:136:30: error: invalid types ‘int[int]’ for array subscript

我认为关键问题在于 Map 构造的声明和初始化，其中包含一个索引键，然后是一个 bitset 对象，它是一串位。似乎使用数组 [] 操作创建了后来的错误（从我在 cplusplus.com/reference/map 站点上找到的内容应该是有效的）。

这是我的代码：

1:// my_predictor.h
// This file contains a sample gshare_predictor class.
// It is a simple 32,768-entry gshare with a history length of 15.

5:#include <bitset>
#include <map>
7:using namespace std;

64:class local_predictor : public branch_predictor {
65:public:
#define LHBITLEN 10
#define PREDCNTR 2
#define LOCHISTTABLERNG 4096
#define LOCPREDRNG 1024
70:     local_update u;
        branch_info bcopy;

        // otherwise ints, where each will be multiplied by 10, then add 1 if taken/true. Initial test case to change if 10 bits long already

75:     // ?correct location? Bit array of length 10 for local history table value entries
        bitset<LHBITLEN> lhthistval; 
        bitset<PREDCNTR> lpcounter;

        // initialize to 0's 
80:       // lhthistval.reset();

        std::map<int, (bitset<LHBITLEN>) > lochisttab;
        //map<int, (bitset<10>) > lht;
        for (int g=0; g < LOCHISTTABLERNG; g++) {
85:            //const int j = g;
            lochisttab[g] = lhthistval.reset();
            //lht.insert(pair<int, (bitset<LHBITLEN>)>(j, lhthistval.reset()));
        }

90:
    std::map<int, (bitset<PREDCNTR>) > locprediction;
    //map<int, (bitset<2>) > locpredict;
    for (int f=0; f < LOCPREDRNG; f++) {
        //const int j = f;
95:        //locpredict.insert(pair<int, (bitset<PREDCNTR>)>(j,lpcounter.reset()));
        locprediction[f] = lpcounter.reset();
    }


100:    local_predictor (void) {
    }

    branch_update *predict (branch_info & b) {
        bcopy = b;
105:        // address for locpredict based on value by modulus (remainder) of branch address divide by 4096
        int braddr = static_cast<int>(b.address % LOCHISTTABLERNG);
        bitset<LHBITLEN> address = lochisttab[braddr];
        // safe and compiler allow since it is only 10 bits long max?
        int indx = static_cast<int>(address.to_ulong());
110:        // use the MSB or Pos 1 in 2 bit array to set boolean Take/Don't
        bool take = locprediction[indx].test(1);
        u.direction_prediction (take);
        u.target_prediction (0);
        return &u;
115:    }

    void update (branch_update *u, bool taken, unsigned int target) {

        if (bcopy.br_flags & BR_CONDITIONAL) {
120:            int lhtaddress = bcopy.address % LOCHISTTABLERNG;
            bitset<LHBITLEN> addr = lochisttab[lhtaddress];
            int indx = static_cast<int>(addr.to_ulong());
            for (int i=1; i < LHBITLEN; i++) {
                bool prval = lochisttab[lhtaddress].test(i-1);
125:                lochisttab[lhtaddress].set(i,prval);
            }
            bool prev = locprediction[indx].test(0);
            locprediction[indx].set(1,prev);

130:            if (taken) {
                lochisttab[lhtaddress].set(0);
                locprediction[indx].set(0);
            }
            else {
135:                lochisttab[lhtaddress].reset(0);
                locprediction[indx].reset(0);
            }
        }
    }

};

编辑：我最初对 map 声明的 bitset 参数没有任何附加条件。编译器给出了第一个错误“'for'之前的预期不合格ID”。模板参数错误来自括号（但我认为这是结果，因为最终可能理解了 bitset 数据类型）。

4

2 回答 2

1

我猜周围的虚假括号(bitset<LHBITLEN>)82 行的虚假括号是问题所在——它们可能会导致编译器尝试将其解析为模板值参数而不是类型参数，然后在尝试恢复时感到困惑。

像这样的一连串错误通常是这种情况，只有第一个是相关的，并告诉您问题所在。后面的那些是由于编译器在尝试恢复有意义的上下文以继续进行时丢弃了太多的标记，对它正在查看的内容感到困惑。它可能扔掉了>and;并认为当它看到时它仍在尝试解析模板参数列表for

于 2013-04-18T23:01:12.410 回答
0

这是您需要的语法：

`map<int, bitset<10> > intBitsetMap;  //Note the space between the two: > >`
于 2013-04-18T23:11:35.207 回答