如何使用 gcc 内联汇编代码访问成员变量

Question

所以我最近偶然发现了这篇博文 NeoPixels Revealed: How to (not need to) generate precisely timed signals 并支持 github project，现在我正在尝试将此代码的核心封装到 C++ 类中，以便我可以从多个 arduino uno 数字引脚访问各种新像素字符串。

为此，我创建了一个受保护的成员变量 (m_PixelChannel)，它存储访问灯串所需的引脚。但是，我无法让汇编代码识别成员变量。下面是我正在尝试制作的代码（这或多或少是来自 github 项目的原始代码的直接复制粘贴，并在其前添加了一个类名）：

// send a bit to the string. We must to drop to asm to enusre that the complier does
// not reorder things and make it so the delay happens in the wrong place.
inline void fastNeoPixels::sendBit(bool bitVal) {

  if (bitVal) { // 0 bit
    asm volatile(
        "sbi %[port], %[bit] \n\t" // Set the output bit
        ".rept %[onCycles] \n\t"   // Execute NOPs to delay exactly the specified number of cycles
        "nop \n\t"
        ".endr \n\t"
        "cbi %[port], %[bit] \n\t" // Clear the output bit
        ".rept %[offCycles] \n\t"  // Execute NOPs to delay exactly the specified number of cycles
        "nop \n\t"
        ".endr \n\t" ::
        [port] "I"(_SFR_IO_ADDR(PIXEL_PORT)),
        [bit] "r"(m_PixelChannel),
        // [bit] "I" (PIXEL_STRING0),
        [onCycles] "I"(NS_TO_CYCLES(T1H) - 2), // 1-bit width less overhead  for the actual bit setting, note that this delay could be longer and everything would still work
        [offCycles] "I"(NS_TO_CYCLES(T1L) - 2) // Minimum interbit delay. Note that we probably don't need this at all since the loop overhead will be enough, but here for correctness
    );
  } else { // 1 bit
    // **************************************************************************
    // This line is really the only tight goldilocks timing in the whole program!
    // **************************************************************************
    asm volatile(
        "sbi %[port], %[bit] \n\t" // Set the output bit
        ".rept %[onCycles] \n\t"   // Now timing actually matters. The 0-bit must be long enough to be detected but not too long or it will be a 1-bit
        "nop \n\t"                 // Execute NOPs to delay exactly the specified number of cycles
        ".endr \n\t"
        "cbi %[port], %[bit] \n\t" // Clear the output bit
        ".rept %[offCycles] \n\t"  // Execute NOPs to delay exactly the specified number of cycles
        "nop \n\t"
        ".endr \n\t" ::
        [port] "I"(_SFR_IO_ADDR(PIXEL_PORT)),
        [bit] "r" (m_PixelChannel),
        // [bit] "I" (PIXEL_STRING0),
        [onCycles] "I"(NS_TO_CYCLES(T0H) - 2),
        [offCycles] "I"(NS_TO_CYCLES(T0L) - 2)
    );
  }  // if (bitVal)...

  // Note that the inter-bit gap can be as long as you want as long as it doesn't exceed the 5us reset timeout (which is A long time)
  // Here I have been generous and not tried to squeeze the gap tight but instead erred on the side of lots of extra time.
  // This has thenice side effect of avoid glitches on very long strings becuase
}

我确信是 m_PixelChannel 变量导致了问题；与我想的约束有关，因为我可以通过取消注释 PIXEL_STRING0 代码行来让它再次工作。或者，我可以将值作为参数传递给方法并使用“n”约束代码使其工作（正如我已经成功完成的那样），但我认为我不应该将参数传递给具有已经访问了该值...

我尝试了以下约束代码，但没有成功：“n”、“o”、“I”、“m”、“+m”、“r”和“g”。

显然我错过了一些东西。有人可以为我指出正确的方向来完成这项工作吗？

Answer 1

问题在于 SBI 指令的操作数必须是常量（立即值）。所以唯一有效的约束是I，并且值必须是一个常数。无法设置变量位。

如果你想设置一个变量位，你必须使用类似 switch 语句来选择 8 个不同指令中的一个。

Answer 2

我决定我真的不喜欢我在评论 Chris Dodd 的回复时使用的模板 方法。因此，经过多次迭代，我能够弄清楚如何使其工作......

  void sendBit( bool bitVal ) {
    volatile uint8_t _hi   = *m_PixelPORT |  m_PinMask;
    volatile uint8_t _lo   = *m_PixelPORT & ~m_PinMask;
    if ( bitVal ) {
      asm volatile (
        "st %a[port], %[hi]"  "\n\t"        // Set the output bit
        ".rept %[onCycles]"  "\n\t"   // Execute NOPs to delay exactly the specified number of cycles
        "nop"                "\n\t"   
        ".endr"              "\n\t"  
        "st %a[port], %[lo]"  "\n\t"   // Clear the output bit
        ".rept %[offCycles]" "\n\t"   // Execute NOPs to delay exactly the specified number of cycles
        "nop"                "\n\t"
        ".endr"              "\n\t"
        : [port] "+e" (m_PixelPORT)
        : [hi] "r" (_hi),
          [lo] "r" (_lo),
          [onCycles] "I" (NS_TO_CYCLES(T1H) - 2), // 1-bit width less overhead  for the actual bit setting, note that this delay could be longer and everything would still work
          [offCycles] "I" (NS_TO_CYCLES(T1L) - 2) // Minimum interbit delay. Note that we probably don't need this at all since the loop overhead will be enough, but here for correctness
      );
    } else {
      asm volatile (
        "st %a[port], %[hi]"  "\n\t"        // Set the output bit
        ".rept %[onCycles]"  "\n\t"   // Execute NOPs to delay exactly the specified number of cycles
        "nop"                "\n\t"   
        ".endr"              "\n\t"  
        "st %a[port], %[lo]"  "\n\t"   // Clear the output bit
        ".rept %[offCycles]" "\n\t"   // Execute NOPs to delay exactly the specified number of cycles
        "nop"                "\n\t"
        ".endr"              "\n\t"
        : [port] "+e" (m_PixelPORT)
        : [hi] "r" (_hi),
          [lo] "r" (_lo),
          [onCycles] "I" (NS_TO_CYCLES(T0H) - 2),
          [offCycles]   "I" (NS_TO_CYCLES(T0L) - 2)
      );
    }
  }  

其中m_PinMask = _BV(digital_pin);

请注意，对 sbi/cbi 的调用已替换为对 st 的调用，并且约束类型。

应用这些更改后，代码可以完全按照我想要的方式执行，同时保持在 bit-bang 过程的时序要求内。

再次感谢 Chris 为我指明了正确的方向！