/**
 ** shiftscl.c ---- shift and copy an array (scanline)
 **                 for 1bpp and 4bpp frame driver blit operations
 **
 ** Copyright (c) 1998 Hartmut Schirmer
 **
 ** This file is part of the GRX graphics library.
 **
 ** The GRX graphics library is free software; you can redistribute it
 ** and/or modify it under some conditions; see the "copying.grx" file
 ** for details.
 **
 ** This library is distributed in the hope that it will be useful,
 ** but WITHOUT ANY WARRANTY; without even the implied warranty of
 ** MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
 **
 **/

#include "libgrx.h"
#include "highlow.h"

void _GR_shift_scanline(GR_int8u **dst,
			GR_int8u **src,
			int ws, int shift, int planes) {
  int plane;

  GRX_ENTER();
  if (shift <= 0) {
    shift = -shift;
    for (plane = 0; plane < planes; ++plane) {
      GR_int8u *s = *(src++) + ws;
      GR_int8u *d = *(dst++) + ws;
#     if defined(__GNUC__) && defined(__i386__)
	int _dummy_, w = ws;
	/* sad but true: the x86 bytesex forces this inefficient code :( */
	asm volatile ("\n"
	  "   movb    (%0),%%ch    \n"
	  "   jmp     1f           \n"
	  "   .align  4,0x90       \n"
	  "1: decl    %0           \n"
	  "   movb    %%ch,%%al    \n"
	  "   movb    (%0),%%ah    \n"
	  "   movb    %%ah,%%ch    \n"
	  "   shrl    %%cl,%%eax   \n"
	  "   movb    %%al,(%1)    \n"
	  "   decl    %1           \n"
	  "   decl    %2           \n"
	  "   jne     1b           \n"
	  "   shrb    %%cl,%%ch    \n"
	  "   movb    %%ch,(%1)      "
	  : "=r" ((void *)s), "=r" ((void *)d), "=r" ((int)w), "=c" (_dummy_)
	  : "0" ((void *)s), "1" ((void *)d), "2" ((int)w), "3" ((int)shift)
 	  : "ax"
	);
#     else
	int w = ws;
	do {
	  --s;
	  *(d--) = highlowP(s)>>shift;
	} while (--w);
	*d = *s >> shift;
#     endif
    }
  } else {
    shift = 8-shift;
    for (plane = 0; plane < planes; ++plane) {
      GR_int8u *s = *(src++);
      GR_int8u *d = *(dst++);
#     if defined(__GNUC__) && defined(__i386__)
	int _dummy_, w = ws;
	asm volatile ("\n"
	  "   movb    (%0),%%ch    \n"
	  "   jmp     1f           \n"
	  "   .align  4,0x90       \n"
	  "1: incl    %0           \n"
	  "   movb    %%ch,%%ah    \n"
	  "   movb    (%0),%%al    \n"
	  "   movb    %%al,%%ch    \n"
	  "   shrl    %%cl,%%eax   \n"
	  "   movb    %%al,(%1)    \n"
	  "   incl    %1           \n"
	  "   decl    %2           \n"
	  "   jne     1b             "
	  : "=r" ((void *)s), "=r" ((void *)d), "=r" ((int)w), "=c" (_dummy_)
	  : "0" ((void *)s), "1" ((void *)d), "2" ((int)w), "3" ((int)shift)
	  : "ax"
	);
#     else
	int w = ws;
	do {
	  *(d++) = highlowP(s)>>shift /* sh */;
	  s++;
	} while (--w);
#     endif
    }
  }
  GRX_LEAVE();
}