ARM-8 代码还原动态调试 pstree 之 out_char

void out_char(char c)
{
  /*403370:	b00001c1 	adrp	x1, 43c000 <memcpy@GLIBC_2.17>
  403374:	9121c021 	add	x1, x1, #0x870  // x1=0x43c870
  403378:	12001c00 	and	w0, w0, #0xff   // w0=c&0xff
  40337c:	b9401022 	ldr	w2, [x1, #16]   // w2=[0x43c880]=charlen
  403380:	34000242 	cbz	w2, 4033c8 <ferror@plt+0x14e8>
    if (charlen == 0) {
        // cbz 4033c8
    }

  403384:	51000442 	sub	w2, w2, #0x1    // w2=charlen-1
  403388:	b00001c3 	adrp	x3, 43c000 <memcpy@GLIBC_2.17>
    charlen = charlen--;

  40338c:	910a2064 	add	x4, x3, #0x288  // x4=0x43c288=&cur_x
  403390:	b9001022 	str	w2, [x1, #16]   // charlen=charlen-1
    charlen = charlen;  // 通过这两处的不同执行点，看不到if ((c & 0x80) == 0) 和 charlen = 4 的代码逻辑，
                        // charlen--在c代码的逻辑中都会执行，但在汇编中执行流程被优化改变了。
                        // 无论(c & 0x80)等0或不等于0，charlen都等于1，c代码逻辑会执行charlen--，所以charlen=0，
                        // 而汇编代码是因(charlen == 0)才到 4033c8 执行流程的，没有了 charlen-- 操作，故而
                        // (c & 0x80)等0或不等于0 (不包含其它if时), charlen 都是 0。

  403394:	b9400481 	ldr	w1, [x4, #4]    // x1=[0x43c28c]=trunc
  403398:	34000121 	cbz	w1, 4033bc <ferror@plt+0x14dc>
  40339c:	b9428862 	ldr	w2, [x3, #648]  // w2=[0x43c288]=cur_x
  4033a0:	b9400881 	ldr	w1, [x4, #8]    // x1=[0x43c290]=output_width
  4033a4:	6b01005f 	cmp	w2, w1          // w2=cur_x, w1=output_width
  4033a8:	540000ad 	b.le	4033bc <ferror@plt+0x14dc>
    if (!trunc || cur_x <= output_width) {
        // b.le 4033bc
        putchar(c);
    }
    else {
  4033ac:	11000421 	add	w1, w1, #0x1    // w1=output_width+1
  4033b0:	6b01005f 	cmp	w2, w1          // cur_x ? output_width+1
  4033b4:	54000300 	b.eq	403414 <ferror@plt+0x1534>  // b.none
        if (trunc && (cur_x == output_width + 1) {
            // b.eq 403414
            putchar('+');
        }
    }
  4033b8:	d65f03c0 	ret*/
}

  /*4033bc:	b00001c1 	adrp	x1, 43c000 <memcpy@GLIBC_2.17>
  4033c0:	f9442821 	ldr	x1, [x1, #2128]     // x1=[0x43c850]
  4033c4:	17fff9bb 	b	401ab0 <putc@plt>

  4033c8:	373800c0 	tbnz	w0, #7, 4033e0 <ferror@plt+0x1500>
    if ((c & 0x80) != 0) {
        // tbnz 4033e0
    }

  4033cc:	b00001c3 	adrp	x3, 43c000 <memcpy@GLIBC_2.17>
  4033d0:	b9428864 	ldr	w4, [x3, #648]      // w4=[0x43c288]=cur_x
  4033d4:	11000484 	add	w4, w4, #0x1        // w4=cur_x+1
  4033d8:	b9028864 	str	w4, [x3, #648]      // [0x43c288]=cur_x+1
    cur_x++;
  4033dc:	17ffffec 	b	40338c <ferror@plt+0x14ac>

  4033e0:	121b0803 	and	w3, w0, #0xe0       // w3=c&0xe0
  4033e4:	52800022 	mov	w2, #0x1            // charlen=1
    charlen = 1;

  4033e8:	7103007f 	cmp	w3, #0xc0
  4033ec:	54ffff00 	b.eq	4033cc <ferror@plt+0x14ec>  // b.none
    if ((c & 0xe0) == 0xc0) {
        // b.eq 4033cc
    }

  4033f0:	121c0c03 	and	w3, w0, #0xf0       // w3=c*0xf0
  4033f4:	52800042 	mov	w2, #0x2            // charlen=2
    charlen = 2;

  4033f8:	7103807f 	cmp	w3, #0xe0
  4033fc:	54fffe80 	b.eq	4033cc <ferror@plt+0x14ec>  // b.none
    if ((c & 0xf0) == 0xe0) {
        // b.eq 4033cc
    }

  403400:	121d1003 	and	w3, w0, #0xf8       // w3=c*0xf8
  403404:	52800062 	mov	w2, #0x3            // charlen=3
    charlen = 3;

  403408:	7103c07f 	cmp	w3, #0xf0
  40340c:	1a8213e2 	csel	w2, wzr, w2, ne  // ne = any
    if ((c & 0xf8) == 0xf0) {
        charlen = 3;
    }

  403410:	17ffffef 	b	4033cc <ferror@plt+0x14ec>

  403414:	b00001c1 	adrp	x1, 43c000 <memcpy@GLIBC_2.17>
  403418:	52800560 	mov	w0, #0x2b                  	// #43='+'
  40341c:	f9442821 	ldr	x1, [x1, #2128]
  403420:	17fff9a4 	b	401ab0 <putc@plt>*/

static void out_char(char c)
{
    // 403370
    if (charlen == 0) {                /* "new" character */
        // 4033c8 -> 40338c
        if ((c & 0x80) == 0) {
            charlen = 1;        /* ASCII */
        // 4033e8
        } else if ((c & 0xe0) == 0xc0) {        /* 110.. 2 bytes */
            charlen = 2;
        // 4033f8
        } else if ((c & 0xf0) == 0xe0) {        /* 1110.. 3 bytes */
            charlen = 3;
        // 403408
        } else if ((c & 0xf8) == 0xf0) {        /* 11110.. 4 bytes */
            charlen = 4;
        } else {
            charlen = 1;
        }
        cur_x++;                /* count first byte of whatever it is only */
    }
    // 403384
    charlen--;          // 汇编代码有点难理解。
    // 403394
    if (!trunc || cur_x <= output_width)
        putchar(c);
    // 4033ac
    else {
        if (trunc && (cur_x == output_width + 1))
            putchar('+');
    }
}