变量值在结构赋值后打印和没有结构赋值打印时是不同的

Question

我不知道如何用合适的标题来说明问题。因此，如果您有更丰富的标题要编辑，请随意。

为了理解这个问题，让我解释一下我在做什么。

我创建了一个结构如下：

typedef union __attribute__ ((__packed__)) adcs_measurements_t
{
    unsigned char raw[72];
    struct __attribute__ ((__packed__)) //191
    {
        int magneticFieldX : 16;
        int magneticFieldY : 16;
        int magneticFieldZ : 16;
        int coarseSunX : 16;
        int coarseSunY : 16;
        int coarseSunZ : 16;
        int sunX : 16;
        int sunY : 16;
        int sunZ : 16;
        int nadirX : 16;
        int nadirY : 16;
        int nadirZ : 16;
        int XAngularRate : 16;
        int YAngularRate : 16;
        int ZAngularRate : 16;
        int XWheelSpeed : 16;
        int YWheelSpeed : 16;
        int ZWheelSpeed : 16;
        int star1BX : 16;
        int star1BY : 16;
        int star1BZ : 16;
        int star1OX : 16;
        int star1OY : 16;
        int star1OZ : 16;
        int star2BX : 16;
        int star2BY : 16;
        int star2BZ : 16;
        int star2OX : 16;
        int star2OY : 16;
        int star2OZ : 16;
        int star3BX : 16;
        int star3BY : 16;
        int star3BZ : 16;
        int star3OX : 16;
        int star3OY : 16;
        int star3OZ : 16;
    } fields;
} adcs_measurements_t;

我通过调用如下函数来填充结构：

void adcsTM191_measurements(adcs_measurements_t* dataOut)
{
    int pass;
    unsigned char TMID = 191;
    unsigned char readBuff[72] = {0};
    pass = I2C_write(ADCS_ADDR, &TMID, 1);
    if(pass != 0)
    {
        printf("write error %d\n", pass);
    }
    pass = I2C_read(ADCS_ADDR, readBuff, 72);
    if(pass != 0)
    {
        printf("read error %d\n", pass);
    }

    dataOut->fields.magneticFieldX = (readBuff[1] & 0x00FF) << 8 | (readBuff[0] & 0x00FF);
    dataOut->fields.magneticFieldY = (readBuff[3] & 0x00FF) << 8 | (readBuff[2] & 0x00FF);
    dataOut->fields.magneticFieldZ = (readBuff[5] & 0x00FF) << 8 | (readBuff[4] & 0x00FF);
    dataOut->fields.coarseSunX = (readBuff[7] & 0x00FF) << 8 | (readBuff[6] & 0x00FF);
    dataOut->fields.coarseSunY = (readBuff[9] & 0x00FF) << 8 | (readBuff[8] & 0x00FF);
    dataOut->fields.coarseSunZ = (readBuff[11] & 0x00FF) << 8 | (readBuff[10] & 0x00FF);
    dataOut->fields.sunX = (readBuff[13] & 0x00FF) << 8 | (readBuff[12] & 0x00FF);
    dataOut->fields.sunY = (readBuff[15] & 0x00FF) << 8 | (readBuff[14] & 0x00FF);
    dataOut->fields.sunZ = (readBuff[17] & 0x00FF) << 8 | (readBuff[16] & 0x00FF);
    dataOut->fields.nadirX = (readBuff[19] & 0x00FF) << 8 | (readBuff[18] & 0x00FF);
    dataOut->fields.nadirY = (readBuff[21] & 0x00FF) << 8 | (readBuff[20] & 0x00FF);
    dataOut->fields.nadirZ = (readBuff[23] & 0x00FF) << 8 | (readBuff[22] & 0x00FF);
    dataOut->fields.XAngularRate = (readBuff[25] & 0x00FF) << 8 | (readBuff[24] & 0x00FF);
    dataOut->fields.YAngularRate = (readBuff[27] & 0x00FF) << 8 | (readBuff[26] & 0x00FF);
    dataOut->fields.ZAngularRate = (readBuff[29] & 0x00FF) << 8 | (readBuff[28] & 0x00FF);
    dataOut->fields.XWheelSpeed = (readBuff[31] & 0x00FF) << 8 | (readBuff[30] & 0x00FF);
    dataOut->fields.YWheelSpeed = (readBuff[33] & 0x00FF) << 8 | (readBuff[32] & 0x00FF);
    dataOut->fields.ZWheelSpeed = (readBuff[35] & 0x00FF) << 8 | (readBuff[34] & 0x00FF);
    dataOut->fields.star1BX = (readBuff[37] & 0x00FF) << 8 | (readBuff[36] & 0x00FF);
    dataOut->fields.star1BY = (readBuff[39] & 0x00FF) << 8 | (readBuff[38] & 0x00FF);
    dataOut->fields.star1BZ = (readBuff[41] & 0x00FF) << 8 | (readBuff[40] & 0x00FF);
    dataOut->fields.star1OX = (readBuff[43] & 0x00FF) << 8 | (readBuff[42] & 0x00FF);
    dataOut->fields.star1OY = (readBuff[45] & 0x00FF) << 8 | (readBuff[44] & 0x00FF);
    dataOut->fields.star1OZ = (readBuff[47] & 0x00FF) << 8 | (readBuff[46] & 0x00FF);
    dataOut->fields.star2BX = (readBuff[49] & 0x00FF) << 8 | (readBuff[48] & 0x00FF);
    dataOut->fields.star2BY = (readBuff[51] & 0x00FF) << 8 | (readBuff[50] & 0x00FF);
    dataOut->fields.star2BZ = (readBuff[53] & 0x00FF) << 8 | (readBuff[52] & 0x00FF);
    dataOut->fields.star2OX = (readBuff[55] & 0x00FF) << 8 | (readBuff[54] & 0x00FF);
    dataOut->fields.star2OY = (readBuff[57] & 0x00FF) << 8 | (readBuff[56] & 0x00FF);
    dataOut->fields.star2OZ = (readBuff[59] & 0x00FF) << 8 | (readBuff[58] & 0x00FF);
    dataOut->fields.star3BX = (readBuff[61] & 0x00FF) << 8 | (readBuff[60] & 0x00FF);
    dataOut->fields.star3BY = (readBuff[63] & 0x00FF) << 8 | (readBuff[62] & 0x00FF);
    dataOut->fields.star3BZ = (readBuff[65] & 0x00FF) << 8 | (readBuff[64] & 0x00FF);
    dataOut->fields.star3OX = (readBuff[67] & 0x00FF) << 8 | (readBuff[66] & 0x00FF);
    dataOut->fields.star3OY = (readBuff[69] & 0x00FF) << 8 | (readBuff[68] & 0x00FF);
    dataOut->fields.star3OZ = (readBuff[71] & 0x00FF) << 8 | (readBuff[70] & 0x00FF);

}

最后我打印出来，比如YWheelSpeed。

adcsTM191_measurements(&temp);  
printf("structure y wheel speed is: %d \n", temp.fields.YWheelSpeed);

这个值应该打印一个负值，它确实是：

structure y wheel speed is: -97

现在事情是这样的，如果我打印 (readBuff[27] & 0x00FF) << 8 | (readBuff[26] & 0x00FF)，它对应于 Y 轮速度变量中填充的内容，adcsTM191_measurements(adcs_measurements_t* dataOut) 中的任何地方都不会打印这个负值。相反，它打印无符号字符的最大值 (65,535)。

int y = (int) (readBuff[33] & 0x00FF) << 8 | (readBuff[32] & 0x00FF);
printf("inside struct y is: %d", y);

我期望存储在结构内部会进行一种隐式转换，因此它会按预期打印负值。它是怎么做到的？如何在不使用结构的情况下打印正确的值？

Answer 1

您可能有 32 位 int，因此初始化从不设置符号位。但是结构字段只有16位，在调用printf()时转换为int时会被符号扩展。

Answer 2

根据 C 2018 脚注 128，用 int 定义的位字段是有符号的还是无符号的，由实现定义。由于您的实现显示负值，推测它是有符号的，因此，作为 16 位有符号整数，它可以表示从 −32,768 到 32,767 的值。

我们还可以推断出 int 在您的实现中超过 16 位，可能是 32 位（根据事实是，在 int y 打印有“ %d”）。

考虑这个作业：

dataOut->fields.YWheelSpeed = (readBuff[33] & 0x00FF) << 8 | (readBuff[32] & 0x00FF);`

在此表达式中，readBuff[33] 和 readBuff[32] 通过 常规提升 转换为 int。 0x00FF 也是一个 int.

如果我们假设readBuff[33]是255，readBufff[32]是159（即2⁸−97），那么表达式上的值= 的右侧是 65,439（即 2¹⁶−97）。在赋值中，右操作数被转换为左操作数的类型，即 16 位有符号整数。在这种情况下，值 65,439 不能用 16 位有符号整数表示。 C 2018 6.3.1.3 3 告诉我们“要么结果是实现定义的，要么引发了实现定义的信号。”

此转换的常见实现方式是生成对 2¹⁶ 取模的结果，或者等效地，将 int 的 16 位低位重新解释为二进制补码16 位整数。这会产生 −97。由于您的实现随后显示了 −97 的值，因此大概这就是您的实现所做的。

因此，dataOut->fields.YWheelSpeed 被赋值为 −97。稍后打印时：

printf("structure y wheel speed is: %d \n", temp.fields.YWheelSpeed);

然后 默认参数提升 ，其中包括通常的整数提升，将 temp.fields.YWheelSpeed 从值为 −97 的有符号 16 位整数转换为 int 的值为 −97，并打印出“-97”。

相反，假设 (readBuff[33] & 0x00FF) << 8 | (readBuff[32] & 0x00FF) 打印为 %d。正如我们在上面看到的，这个表达式的值为 65,439，因此应该打印“65439”。

问题说明：

Now here is the thing, if I print (readBuff[27] & 0x00FF) << 8 | (readBuff[26] & 0x00FF), which corresponds to what was populated inside the Y wheel speed variable,… it prints the maximum value of an unsigned char (65,535‬).

但是，(readBuff[27] & 0x00FF) << 8 | (readBuff[26] & 0x00FF)并不是赋给YWheelSpeed的值，推测是“Y轮速度变量”。 YWheelSpeed 是从 readBuff 元素 32 和 33 赋值的，而不是 26 和 27。因此我们不应该对打印出一些不同的值而不是 65,439 感到惊讶。

变量值在结构赋值后打印和没有结构赋值打印时是不同的

A variable value is different when it is printed after structural assignment and when printed without the structural assignment

c

printf

struct

casting

shift