MSVCRT strcpy以及strcat

        page    ,132
        title   strcat - concatenate (append) one string to another
;***
;strcat.asm - contains strcat() and strcpy() routines
;
;       Copyright (c) 1985-1997, Microsoft Corporation. All rights reserved.
;
;Purpose:
;       STRCAT concatenates (appends) a copy of the source string to the
;       end of the destination string, returning the destination string.
;
;*******************************************************************************

        .xlist
        include cruntime.inc
        .list


page
;***
;char *strcat(dst, src) - concatenate (append) one string to another
;
;Purpose:
;       Concatenates src onto the end of dest.  Assumes enough
;       space in dest.
;
;       Algorithm:
;       char * strcat (char * dst, char * src)
;       {
;           char * cp = dst;
;
;           while( *cp )
;                   ++cp;           /* Find end of dst */
;           while( *cp++ = *src++ )
;                   ;               /* Copy src to end of dst */
;           return( dst );
;       }
;
;Entry:
;       char *dst - string to which "src" is to be appended
;       const char *src - string to be appended to the end of "dst"
;
;Exit:
;       The address of "dst" in EAX
;
;Uses:
;       EAX, ECX
;
;Exceptions:
;
;*******************************************************************************

page
;***
;char *strcpy(dst, src) - copy one string over another
;
;Purpose:
;       Copies the string src into the spot specified by
;       dest; assumes enough room.
;
;       Algorithm:
;       char * strcpy (char * dst, char * src)
;       {
;           char * cp = dst;
;
;           while( *cp++ = *src++ )
;                   ;               /* Copy src over dst */
;           return( dst );
;       }
;
;Entry:
;       char * dst - string over which "src" is to be copied
;       const char * src - string to be copied over "dst"
;
;Exit:
;       The address of "dst" in EAX
;
;Uses:
;       EAX, ECX
;
;Exceptions:
;*******************************************************************************


        CODESEG

%       public  strcat, strcpy      ; make both functions available
strcpy  proc
        push    edi                 ; preserve edi
        mov     edi,[esp+8]         ; edi points to dest string 堆栈
        jmp     short copy_start

strcpy  endp

        align   16

strcat  proc

        .FPO    ( 0, 2, 0, 0, 0, 0 )

        mov     ecx,[esp+4]         ; ecx -> dest string
        push    edi                 ; preserve edi
        test    ecx,3               ; test if string is aligned on 32 bits
        je      short find_end_of_dest_string_loop

dest_misaligned:                    ; simple byte loop until string is aligned
        mov     al,byte ptr [ecx]
        inc     ecx
        test    al,al
        je      short start_byte_3
        test    ecx,3
        jne     short dest_misaligned

        align   4

find_end_of_dest_string_loop:
        mov     eax,dword ptr [ecx] ; read 4 bytes
        mov     edx,7efefeffh
        add     edx,eax
        xor     eax,-1
        xor     eax,edx
        add     ecx,4
        test    eax,81010100h
        je      short find_end_of_dest_string_loop
        ; found zero byte in the loop
        mov     eax,[ecx - 4]
        test    al,al               ; is it byte 0
        je      short start_byte_0
        test    ah,ah               ; is it byte 1
        je      short start_byte_1
        test    eax,00ff0000h       ; is it byte 2
        je      short start_byte_2
        test    eax,0ff000000h      ; is it byte 3
        je      short start_byte_3
        jmp     short find_end_of_dest_string_loop
                                    ; taken if bits 24-30 are clear and bit
                                    ; 31 is set
start_byte_3:
        lea     edi,[ecx - 1]
        jmp     short copy_start
start_byte_2:
        lea     edi,[ecx - 2]
        jmp     short copy_start
start_byte_1:
        lea     edi,[ecx - 3]
        jmp     short copy_start
start_byte_0:
        lea     edi,[ecx - 4]
;       jmp     short copy_start

;       edi points to the end of dest string.

copy_start::
        mov     ecx,[esp+0ch]       ; ecx -> sorc string
        test    ecx,3               ; test if string is aligned on 32 bits
        je      short main_loop_entrance

src_misaligned:                     ; simple byte loop until string is aligned
        mov     dl,byte ptr [ecx]
        inc     ecx
        test    dl,dl
        je      short byte_0
        mov     [edi],dl
        inc     edi
        test    ecx,3
        jne     short src_misaligned ;continue aligned loop
        jmp     short main_loop_entrance

main_loop:                          ; edx contains first dword of sorc string
        mov     [edi],edx           ; store one more dword
        add     edi,4               ; kick dest pointer
main_loop_entrance:
        mov     edx,7efefeffh
        mov     eax,dword ptr [ecx] ; read 4 bytes

        add     edx,eax             ; plus edx magic_bits
        xor     eax,-1              ; eax = ~eax

        xor     eax,edx             ;  
        mov     edx,[ecx]           ; it's in cache now

        add     ecx,4               ; kick dest pointer
        test    eax,81010100h

        je      short main_loop
        ; found zero byte in the loop
; main_loop_end:
        test    dl,dl               ; is it byte 0
        je      short byte_0
        test    dh,dh               ; is it byte 1
        je      short byte_1
        test    edx,00ff0000h       ; is it byte 2
        je      short byte_2
        test    edx,0ff000000h      ; is it byte 3
        je      short byte_3
        jmp     short main_loop     ; taken if bits 24-30 are clear and bit
                                    ; 31 is set
byte_3:
        mov     [edi],edx
        mov     eax,[esp+8]         ; return in eax pointer to dest string
        pop     edi
        ret
byte_2:
        mov     [edi],dx
        mov     eax,[esp+8]         ; return in eax pointer to dest string
        mov     byte ptr [edi+2],0
        pop     edi
        ret
byte_1:
        mov     [edi],dx
        mov     eax,[esp+8]         ; return in eax pointer to dest string
        pop     edi
        ret
byte_0:
        mov     [edi],dl
        mov     eax,[esp+8]         ; return in eax pointer to dest string
        pop     edi
        ret

strcat  endp

        end

char* strcpy(char*pdest,const char*psrc)
{
	const char *char_psrc;
	const ulong *longword_psrc,*longword_pdest;
	register ulong longword, magic_bits;
	char *char_pest=pdest;

	for (	char_psrc = psrc; //初始位置
			((ulong)char_src& (sizeof(ulong) - 1)) != 0; //边界对齐判断 //地址的size以下全部是0
			*char_pdest++ = *char_psrc++) //copy...
	{
		if (*char_psrc == '\0') //结束...
			{
				*char_pdest='\0';
				return pdest;
			}
	}
	longword_psrc =  (ulong*)char_psrc; //从自然边界开始
	longword_pdest = (ulong*)char_pdest;
	magic_bits = 0x7efefeffL;

	while (1)
	{
		longword = *longword_psrc++;
		if ((((longword + magic_bits) ^ ~longword) & ~magic_bits) != 0)
		{
			const char *cp = (const char*)(longword_psrc - 1);
			if (cp[0] == 0)  
			{
				*(char*)longword_pdest = '\0';
				return pdest;
			}
			if (cp[1] == 0)
			{
				*(WORD*)longword_pdest = *(WORD*)longword_word;
				return pdest;
			}
			if (cp[2] == 0)  
			{
				*(WORD*)longword_pdest = *(WORD*)longword_word;
				*((char*)longword_pdest+2) = '\0';
				return pdest;
			}
            if (cp[3] == 0)  
			{
				*longword_pdest = *longword_word;
				return pdest;
			}
		}
		*longword_pdest++ = *longword;
	}
}

char* strcat(char*pdest,const char*psrc)
{
	const char *char_psrc;
	const ulong *longword_psrc,*longword_pdest;
	register ulong longword, magic_bits;
	char *char_pest=pdest;
	while(1) //查找源字符串终止位置
	{
		longword = *longword_psrc++;
		if ((((longword + magic_bits) ^ ~longword) & ~magic_bits) != 0)
		{
			const char *cp = (const char*)(longword_psrc - 1);
			if (cp[0] == 0)
			{
				char_pdest = (char*)longword;
				break;
			}
			if (cp[1] == 0)
			{
				char_pdest = (char*)longword + 1;
				break;
			}
			if (cp[2] == 0)
			{
				char_pdest = (char*)longword + 2;
				break;
			}
			if (cp[3] == 0)
			{
				char_pdest = (char*)longword + 3;
				break;
			}
		}
	}

//strcpy
	for (	char_psrc = psrc;
			((ulong)char_src& (sizeof(ulong) - 1)) != 0;
			*char_pdest++ = *char_psrc++)
	{
		if (*char_psrc == '\0')
			{
				*char_pdest='\0';
				return pdest;
			}
	}

	//对齐后的strcpy
	longword_psrc =  (ulong*)char_psrc;
	longword_pdest = (ulong*)char_pdest;
	magic_bits = 0x7efefeffL;

	while (1)
	{
		longword = *longword_psrc++;
		if ((((longword + magic_bits) ^ ~longword) & ~magic_bits) != 0)
		{//判断是否有0
			const char *cp = (const char*)(longword_psrc - 1);
			if (cp[0] == 0)
			{
				*(char*)longword_pdest = '\0';
				return pdest;
			}
			if (cp[1] == 0)
			{
				*(WORD*)longword_pdest = *(WORD*)longword_word;
				return pdest;
			}
			if (cp[2] == 0)
			{
				*(WORD*)longword_pdest = *(WORD*)longword_word;
				*((char*)longword_pdest+2) = '\0';
				return pdest;
			}
			if (cp[3] == 0)
			{
				*longword_pdest = *longword_word;
				return pdest;
			}
		}
		*longword_pdest++ = *longword;
	}
}