C-language function calls when the reference operation on the 32-bit x86 machine relies on the stack. But in the x86_64 machine uses some registers as auxiliary, but if the parameter is too many, the register is not enough to use, at this time also must use the stack operation to implement the parameter. Although the C language does not explicitly limit the number of function pass arguments ( Depending on the compiler implementation: Http://stackoverflow.com/questions/9034787/function-parameters-max-number), but too much parameter passing is bound to affect code execution efficiency.
Usually the C language function is very explicit, as in the following function:
int test (int a,float b,int *pointer);
Note: The following examples are described using the 32-bit X86,GCC compiler.
But what if the array or struct body is passed as a parameter? How many parameters are passed, how much stack space is occupied?
typedef struct list{
int A;
int b;
int C;
int D;
int e;
} list;
int test1 (int array[5]);
int test2 (list listtest);
First look at the array example:
/*passing arguments test:array*/
#include <stdio.h>
void Array (int tmp[5]) {
printf ("%d\n", tmp[2]) ;
}
int main (void) {
int test[]={1,2,3,4,5};
Array (test);
To compile the assembly code, first intercept the main function pass-parameter section:
MOVL $ -32 (%EBP)
movl $ -28 (%EBP)
movl $ -24 (%EBP)
movl $ -20 (%EBP)
movl $ -16 (%EBP)
Leal -32 (%EBP),%eax pushl%eax Call Array
You can see that in the main function the array element is first written to the array space, then the array address (that is, the address of the element A[0] is saved in the eax, then the EAX is pressed and the array function is finally called. And look at the array function:
Array:
. LFB0:
pushl %ebp
movl %esp,%EBP
subl $,%esp
movl 8 (%EBP),%eax
addl $ 8,%eax
movl (%eax),%eax
subl $,%esp pushl%eax $. LC0
call printf
addl $16,%esp
nop
leave
ret
The first is the old-fashioned operation: Save Ebp, Sync ebp, and esp,esp Move down to create a new function stack environment. Then take 8 (%EBP). What is it? In fact, the function execution went through 3 times the press stack: PUSHL%eax,call% Ebp. And each time the stack is 32 bits, that is, 4 bytes. So 0 (%EBP) is the value of PUSHL%EBP, 4 (%EBP) is the function return address, and 8 (%EBP) is the PUSHL%eax value, that is, the array address. Of course Addl $8,%eax is tmp[2. Address, no longer repeat. So much, in a word, array (test) just passed an array address, and did not pass the entire array element as a parameter to the child function.
Then look at the structure of the example:
#include <stdio.h>
typedef struct list{
int A;
int b;
int C;
int D;
} list;
void Test (list tmp) {
printf ("%d\n", tmp.b);
}
int main (void) {
list tmp={.a=10,.b=20,.c=30,.d=40};
Test (TMP);
}
Also intercepts the main function parameter pass fragment:
MOVL $ -24 (%EBP)
movl $, -20 (%EBP) movl $, -16 (%EBP)
MOVL $ -12 (%EBP)
PUSHL -12 (%EBP)
pushl -16 (%EBP)
pushl -20 (%EBP)
pushl -24 (%EBP)
Call Test
As you can see, this is actually putting the members of the structure in full pressure stack. Therefore, in open source project code, you often don't see the structure as a parameter, but instead pass a pointer to the structure body. So no matter how many members of the structure, the pressure stack is only pressed into a value. As the following code:
#include <stdio.h> typedef struct list{int A;
int b;
int C;
int D;
}list;
void Test (list *tmp) {printf ("%d\n", tmp->b);} int main (void) {list tmp={.a=10,.b=20,.c=30,.d=40};
Test (&TMP); }