XC32函数在向量x到大

对于初学者来说，你的调度函数中缺少一个ErET。



      以下为原文

    For starters you are missing an eret from your dispatch function.

 .ent __vector_dispatch_1
__vector_dispatch_1:
 jal soft_int_0
 nop
 eret
 .set reorder
 .end __vector_dispatch_1

你的右边，我从编译器输出中复制了一部分，所以第二个函数应该是：SoftTytIn0:ErET-但是仍然不能解决这个问题。有趣的是，我复制了编译器的程序集输出，为什么它在C中工作而不是直接在汇编中工作？我漏掉了像指令一样的东西吗？



      以下为原文

    Your right, i copied a part from the compiler output, so the second function should be:
soft_int_0:
    eret
 
But that still doesn't solve the problem.
The funny thing is that i copied the assembly output from the compiler, so why does it work in c and not directly in assembly?
Am i missing something like a directive?

向量间隔为8字节，SoftTytIn 0函数不适用于Vector 1节。把调度存根和ISR函数之间的方向移到主代码部分。然而，ISR已经被编译为正常函数，而调度存根将浪费$RA寄存器，所以我很怀疑它的工作原理，而不是偶然的。如果使用中断函数属性，或者向IP-ISR宏添加IPL说明符，会发生什么？



      以下为原文

    With a vector spacing of 8 bytes, the soft_int_0 function doesn't fit inside the .vector_1 section. Put the directive
  .textbetween the dispatch stub and ISR function to move it into the main code section.
 
However, as-is the ISR has been compiled as a normal function, and the dispatch stub will trash the $ra register, so I'm rather suspicious of the claim that it works other than by chance. What happens if you use the interrupt function attribute, or add the IPL specifier to the __ISR macro?

安德姆斯说WRT中断函数在.Text中。（我没有注意到，已经很晚了）同样，ISR的一个更好的存根形式是：-即跳转到中断例程和中断程序的ErET，而不是存根。节省必须跳回存根，不浪费RA。注意，如果你真的想要多功能中断（创建堆栈帧、推/ POP状态、EPC、重新启用中断、清除中断标志等），则缺少很多代码。



      以下为原文

    What andersm said wrt the interrupt function being in .text. (I didn't notice that, it was late.)
 
Also, a better form of stub for an isr is:-
     .globl    __vector_dispatch_1
    .section    .vector_1,code,keep
    .set    nomips16
    .set micromips
    .set noreorder
    .align    2
    .ent    __vector_dispatch_1
__vector_dispatch_1:
    j    _soft_int_0
    nop
    .set reorder
    .end    __vector_dispatch_1
    .size    __vector_dispatch_1, .-__vector_dispatch_1
    
    
    
    .text
    .set nomips16
    .set micromips
    .set noreorder
    .align 2
    .global _soft_int_0
    .ent    _soft_int_0
_soft_int_0:
    ;
    ;interrupt code here
    ;
    eret
    .set reorder
    .end    _soft_int_0
    .size    _soft_int_0, .-_soft_int_0

I.e. just jump to interrupt routine and eret from the interrupt routine, not the stub. Saves having to jump back to the stub and doesn't trash RA.
 
Note you are missing a lot of code if you actually want functioning multiple interrupts (Create stack frame, push/pop STATUS, EPC, re-enable interrupts, clear interrupt flag etc.)

除非你想完全放弃C，你也可以在C--yAtEngy（（vector（a CoeLeAuthReal00vector）））中声明IRQ，然后在ASM中创建一个具有相同名称的全局函数（in .text节），它将跳转到向量表中的ASM函数，WH。ICH要简单得多



      以下为原文

    unless you are trying to ditch C altogether, you can also just declare the irq in C-
__attribute((vector(_CORE_SOFTWARE_0_VECTOR))) void soft_int_0();
 
then in asm, create a global function with the same name (in .text section)
 
which will get the jump to your asm function placed in the vector table, which is much simpler
 

这确实是我忘记的文字部分，对我来说太晚了。谢谢。谢谢。



      以下为原文

    It was indeed the .text section i forgot, it was late for me too wink: .
 
Thanks