pixman-mmx.c

嵌入式图形库

		vdest = pack565 (in_over(vsrc, vsrca, load8888(m), vdest), vdest, 0);
		*q = ULLONG(vdest);
	    }

	    twidth--;
	    p++;
	    q++;
	}

	maskLine += maskStride;
	dstLine += dstStride;
    }

    _mm_empty ();
}

void
fbCompositeIn_nx8x8mmx (pixman_op_t op,
			pixman_image_t * pSrc,
			pixman_image_t * pMask,
			pixman_image_t * pDst,
			int16_t      xSrc,
			int16_t      ySrc,
			int16_t      xMask,
			int16_t      yMask,
			int16_t      xDst,
			int16_t      yDst,
			uint16_t     width,
			uint16_t     height)
{
    uint8_t	*dstLine, *dst;
    uint8_t	*maskLine, *mask;
    int	dstStride, maskStride;
    uint16_t	w;
    uint32_t	src;
    uint8_t	sa;
    __m64	vsrc, vsrca;

    fbComposeGetStart (pDst, xDst, yDst, uint8_t, dstStride, dstLine, 1);
    fbComposeGetStart (pMask, xMask, yMask, uint8_t, maskStride, maskLine, 1);

    fbComposeGetSolid(pSrc, src, pDst->bits.format);

    sa = src >> 24;
    if (sa == 0)
	return;

    vsrc = load8888(src);
    vsrca = expand_alpha(vsrc);

    while (height--)
    {
	dst = dstLine;
	dstLine += dstStride;
	mask = maskLine;
	maskLine += maskStride;
	w = width;

	if ((((unsigned long)pDst & 3) == 0) &&
	    (((unsigned long)pSrc & 3) == 0))
	{
	    while (w >= 4)
	    {
		uint32_t m;
		__m64 vmask;
		__m64 vdest;

		m = 0;

		vmask = load8888 (*(uint32_t *)mask);
		vdest = load8888 (*(uint32_t *)dst);

		*(uint32_t *)dst = store8888 (in (in (vsrca, vmask), vdest));

		dst += 4;
		mask += 4;
		w -= 4;
	    }
	}

	while (w--)
	{
	    uint16_t	tmp;
	    uint8_t	a;
	    uint32_t	m, d;
	    uint32_t	r;

	    a = *mask++;
	    d = *dst;

	    m = FbInU (sa, 0, a, tmp);
	    r = FbInU (m, 0, d, tmp);

	    *dst++ = r;
	}
    }

    _mm_empty();
}

void
fbCompositeIn_8x8mmx (pixman_op_t op,
		      pixman_image_t * pSrc,
		      pixman_image_t * pMask,
		      pixman_image_t * pDst,
		      int16_t      xSrc,
		      int16_t      ySrc,
		      int16_t      xMask,
		      int16_t      yMask,
		      int16_t      xDst,
		      int16_t      yDst,
		      uint16_t     width,
		      uint16_t     height)
{
    uint8_t	*dstLine, *dst;
    uint8_t	*srcLine, *src;
    int	srcStride, dstStride;
    uint16_t	w;

    fbComposeGetStart (pDst, xDst, yDst, uint8_t, dstStride, dstLine, 1);
    fbComposeGetStart (pSrc, xSrc, ySrc, uint8_t, srcStride, srcLine, 1);

    while (height--)
    {
	dst = dstLine;
	dstLine += dstStride;
	src = srcLine;
	srcLine += srcStride;
	w = width;

	if ((((unsigned long)pDst & 3) == 0) &&
	    (((unsigned long)pSrc & 3) == 0))
	{
	    while (w >= 4)
	    {
		uint32_t *s = (uint32_t *)src;
		uint32_t *d = (uint32_t *)dst;

		*d = store8888 (in (load8888 (*s), load8888 (*d)));

		w -= 4;
		dst += 4;
		src += 4;
	    }
	}

	while (w--)
	{
	    uint8_t s, d;
	    uint16_t tmp;

	    s = *src;
	    d = *dst;

	    *dst = FbInU (s, 0, d, tmp);

	    src++;
	    dst++;
	}
    }

    _mm_empty ();
}

void
fbCompositeSrcAdd_8888x8x8mmx (pixman_op_t op,
			       pixman_image_t * pSrc,
			       pixman_image_t * pMask,
			       pixman_image_t * pDst,
			       int16_t      xSrc,
			       int16_t      ySrc,
			       int16_t      xMask,
			       int16_t      yMask,
			       int16_t      xDst,
			       int16_t      yDst,
			       uint16_t     width,
			       uint16_t     height)
{
    uint8_t	*dstLine, *dst;
    uint8_t	*maskLine, *mask;
    int	dstStride, maskStride;
    uint16_t	w;
    uint32_t	src;
    uint8_t	sa;
    __m64	vsrc, vsrca;

    fbComposeGetStart (pDst, xDst, yDst, uint8_t, dstStride, dstLine, 1);
    fbComposeGetStart (pMask, xMask, yMask, uint8_t, maskStride, maskLine, 1);

    fbComposeGetSolid(pSrc, src, pDst->bits.format);

    sa = src >> 24;
    if (sa == 0)
	return;

    vsrc = load8888(src);
    vsrca = expand_alpha(vsrc);

    while (height--)
    {
	dst = dstLine;
	dstLine += dstStride;
	mask = maskLine;
	maskLine += maskStride;
	w = width;

	if ((((unsigned long)pMask & 3) == 0) &&
	    (((unsigned long)pDst  & 3) == 0))
	{
	    while (w >= 4)
	    {
		__m64 vmask = load8888 (*(uint32_t *)mask);
		__m64 vdest = load8888 (*(uint32_t *)dst);

		*(uint32_t *)dst = store8888 (_mm_adds_pu8 (in (vsrca, vmask), vdest));

		w -= 4;
		dst += 4;
		mask += 4;
	    }
	}

	while (w--)
	{
	    uint16_t	tmp;
	    uint16_t	a;
	    uint32_t	m, d;
	    uint32_t	r;

	    a = *mask++;
	    d = *dst;

	    m = FbInU (sa, 0, a, tmp);
	    r = FbAdd (m, d, 0, tmp);

	    *dst++ = r;
	}
    }

    _mm_empty();
}

void
fbCompositeSrcAdd_8000x8000mmx (pixman_op_t op,
				pixman_image_t * pSrc,
				pixman_image_t * pMask,
				pixman_image_t * pDst,
				int16_t      xSrc,
				int16_t      ySrc,
				int16_t      xMask,
				int16_t      yMask,
				int16_t      xDst,
				int16_t      yDst,
				uint16_t     width,
				uint16_t     height)
{
    uint8_t	*dstLine, *dst;
    uint8_t	*srcLine, *src;
    int	dstStride, srcStride;
    uint16_t	w;
    uint8_t	s, d;
    uint16_t	t;

    CHECKPOINT();

    fbComposeGetStart (pSrc, xSrc, ySrc, uint8_t, srcStride, srcLine, 1);
    fbComposeGetStart (pDst, xDst, yDst, uint8_t, dstStride, dstLine, 1);

    while (height--)
    {
	dst = dstLine;
	dstLine += dstStride;
	src = srcLine;
	srcLine += srcStride;
	w = width;

	while (w && (unsigned long)dst & 7)
	{
	    s = *src;
	    d = *dst;
	    t = d + s;
	    s = t | (0 - (t >> 8));
	    *dst = s;

	    dst++;
	    src++;
	    w--;
	}

	while (w >= 8)
	{
	    *(__m64*)dst = _mm_adds_pu8(*(__m64*)src, *(__m64*)dst);
	    dst += 8;
	    src += 8;
	    w -= 8;
	}

	while (w)
	{
	    s = *src;
	    d = *dst;
	    t = d + s;
	    s = t | (0 - (t >> 8));
	    *dst = s;

	    dst++;
	    src++;
	    w--;
	}
    }

    _mm_empty();
}

void
fbCompositeSrcAdd_8888x8888mmx (pixman_op_t 	op,
				pixman_image_t *	pSrc,
				pixman_image_t *	pMask,
				pixman_image_t *	 pDst,
				int16_t		 xSrc,
				int16_t      ySrc,
				int16_t      xMask,
				int16_t      yMask,
				int16_t      xDst,
				int16_t      yDst,
				uint16_t     width,
				uint16_t     height)
{
    __m64 dst64;
    uint32_t	*dstLine, *dst;
    uint32_t	*srcLine, *src;
    int	dstStride, srcStride;
    uint16_t	w;

    CHECKPOINT();

    fbComposeGetStart (pSrc, xSrc, ySrc, uint32_t, srcStride, srcLine, 1);
    fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);

    while (height--)
    {
	dst = dstLine;
	dstLine += dstStride;
	src = srcLine;
	srcLine += srcStride;
	w = width;

	while (w && (unsigned long)dst & 7)
	{
	    *dst = _mm_cvtsi64_si32(_mm_adds_pu8(_mm_cvtsi32_si64(*src),
						 _mm_cvtsi32_si64(*dst)));
	    dst++;
	    src++;
	    w--;
	}

	while (w >= 2)
	{
	    dst64 = _mm_adds_pu8(*(__m64*)src, *(__m64*)dst);
	    *(ullong*)dst = ULLONG(dst64);
	    dst += 2;
	    src += 2;
	    w -= 2;
	}

	if (w)
	{
	    *dst = _mm_cvtsi64_si32(_mm_adds_pu8(_mm_cvtsi32_si64(*src),
						 _mm_cvtsi32_si64(*dst)));

	}
    }

    _mm_empty();
}

pixman_bool_t
pixman_blt_mmx (uint32_t *src_bits,
		uint32_t *dst_bits,
		int src_stride,
		int dst_stride,
		int src_bpp,
		int dst_bpp,
		int src_x, int src_y,
		int dst_x, int dst_y,
		int width, int height)
{
    uint8_t *	src_bytes;
    uint8_t *	dst_bytes;
    int		byte_width;

    if (src_bpp != dst_bpp)
	return FALSE;

    if (src_bpp == 16)
    {
	src_stride = src_stride * (int) sizeof (uint32_t) / 2;
	dst_stride = dst_stride * (int) sizeof (uint32_t) / 2;
	src_bytes = (uint8_t *)(((uint16_t *)src_bits) + src_stride * (src_y) + (src_x));
	dst_bytes = (uint8_t *)(((uint16_t *)dst_bits) + dst_stride * (dst_y) + (dst_x));
	byte_width = 2 * width;
	src_stride *= 2;
	dst_stride *= 2;
    } else if (src_bpp == 32) {
	src_stride = src_stride * (int) sizeof (uint32_t) / 4;
	dst_stride = dst_stride * (int) sizeof (uint32_t) / 4;
	src_bytes = (uint8_t *)(((uint32_t *)src_bits) + src_stride * (src_y) + (src_x));
	dst_bytes = (uint8_t *)(((uint32_t *)dst_bits) + dst_stride * (dst_y) + (dst_x));
	byte_width = 4 * width;
	src_stride *= 4;
	dst_stride *= 4;
    } else {
	return FALSE;
    }

    while (height--)
    {
	int w;
	uint8_t *s = src_bytes;
	uint8_t *d = dst_bytes;
	src_bytes += src_stride;
	dst_bytes += dst_stride;
	w = byte_width;

	while (w >= 2 && ((unsigned long)d & 3))
	{
	    *(uint16_t *)d = *(uint16_t *)s;
	    w -= 2;
	    s += 2;
	    d += 2;
	}

	while (w >= 4 && ((unsigned long)d & 7))
	{
	    *(uint32_t *)d = *(uint32_t *)s;

	    w -= 4;
	    s += 4;
	    d += 4;
	}

	while (w >= 64)
	{
#ifdef __GNUC__
	    __asm__ (
		"movq	  (%1),	  %%mm0\n"
		"movq	 8(%1),	  %%mm1\n"
		"movq	16(%1),	  %%mm2\n"
		"movq	24(%1),	  %%mm3\n"
		"movq	32(%1),	  %%mm4\n"
		"movq	40(%1),	  %%mm5\n"
		"movq	48(%1),	  %%mm6\n"
		"movq	56(%1),	  %%mm7\n"

		"movq	%%mm0,	  (%0)\n"
		"movq	%%mm1,	 8(%0)\n"
		"movq	%%mm2,	16(%0)\n"
		"movq	%%mm3,	24(%0)\n"
		"movq	%%mm4,	32(%0)\n"
		"movq	%%mm5,	40(%0)\n"
		"movq	%%mm6,	48(%0)\n"
		"movq	%%mm7,	56(%0)\n"
		:
		: "r" (d), "r" (s)
		: "memory",
		  "%mm0", "%mm1", "%mm2", "%mm3",
		  "%mm4", "%mm5", "%mm6", "%mm7");
#else
	    __m64 v0 = *(__m64 *)(s + 0);
	    __m64 v1 = *(__m64 *)(s + 8);
	    __m64 v2 = *(__m64 *)(s + 16);
	    __m64 v3 = *(__m64 *)(s + 24);
	    __m64 v4 = *(__m64 *)(s + 32);
	    __m64 v5 = *(__m64 *)(s + 40);
	    __m64 v6 = *(__m64 *)(s + 48);
	    __m64 v7 = *(__m64 *)(s + 56);
	    *(__m64 *)(d + 0)  = v0;
	    *(__m64 *)(d + 8)  = v1;
	    *(__m64 *)(d + 16) = v2;
	    *(__m64 *)(d + 24) = v3;
	    *(__m64 *)(d + 32) = v4;
	    *(__m64 *)(d + 40) = v5;
	    *(__m64 *)(d + 48) = v6;
	    *(__m64 *)(d + 56) = v7;
#endif

	    w -= 64;
	    s += 64;
	    d += 64;
	}
	while (w >= 4)
	{
	    *(uint32_t *)d = *(uint32_t *)s;

	    w -= 4;
	    s += 4;
	    d += 4;
	}
	if (w >= 2)
	{
	    *(uint16_t *)d = *(uint16_t *)s;
	    w -= 2;
	    s += 2;
	    d += 2;
	}
    }

    _mm_empty();

    return TRUE;
}

void
fbCompositeCopyAreammx (pixman_op_t       op,
			pixman_image_t *	pSrc,
			pixman_image_t *	pMask,
			pixman_image_t *	pDst,
			int16_t		xSrc,
			int16_t		ySrc,
			int16_t		xMask,
			int16_t		yMask,
			int16_t		xDst,
			int16_t		yDst,
			uint16_t		width,
			uint16_t		height)
{
    pixman_blt_mmx (pSrc->bits.bits,
		    pDst->bits.bits,
		    pSrc->bits.rowstride,
		    pDst->bits.rowstride,
		    PIXMAN_FORMAT_BPP (pSrc->bits.format),
		    PIXMAN_FORMAT_BPP (pDst->bits.format),
		    xSrc, ySrc, xDst, yDst, width, height);
}

void
fbCompositeOver_x888x8x8888mmx (pixman_op_t      op,
				pixman_image_t * pSrc,
				pixman_image_t * pMask,
				pixman_image_t * pDst,
				int16_t      xSrc,
				int16_t      ySrc,
				int16_t      xMask,
				int16_t      yMask,
				int16_t      xDst,
				int16_t      yDst,
				uint16_t     width,
				uint16_t     height)
{
    uint32_t	*src, *srcLine;
    uint32_t    *dst, *dstLine;
    uint8_t	*mask, *maskLine;
    int		 srcStride, maskStride, dstStride;
    uint16_t w;

    fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
    fbComposeGetStart (pMask, xMask, yMask, uint8_t, maskStride, maskLine, 1);
    fbComposeGetStart (pSrc, xSrc, ySrc, uint32_t, srcStride, srcLine, 1);

    while (height--)
    {
	src = srcLine;
	srcLine += srcStride;
	dst = dstLine;
	dstLine += dstStride;
	mask = maskLine;
	maskLine += maskStride;

	w = width;

	while (w--)
	{
	    ullong m = *mask;

	    if (m)
	    {
		__m64 s = load8888 (*src | 0xff000000);

		if (m == 0xff)
		    *dst = store8888 (s);
		else
		{
		    __m64 sa = expand_alpha (s);
		    __m64 vm = expand_alpha_rev (M64(m));
		    __m64 vdest = in_over(s, sa, vm, load8888 (*dst));

		    *dst = store8888 (vdest);
		}
	    }

	    mask++;
	    dst++;
	    src++;
	}
    }

    _mm_empty();
}



#endif /* USE_MMX */