+ xmm_dst = load_128_aligned ((__m128i*)dst);
+
+ unpack_128_2x128 (xmm_dst, &xmm_dst_lo, &xmm_dst_hi);
+ expand_alpha_2x128 (xmm_dst_lo, xmm_dst_hi, &xmm_dsta_lo, &xmm_dsta_hi);
+
+ tmp_lo = xmm_src;
+ tmp_hi = xmm_src;
+
+ over_2x128 (&xmm_dst_lo, &xmm_dst_hi,
+ &xmm_dsta_lo, &xmm_dsta_hi,
+ &tmp_lo, &tmp_hi);
+
+ save_128_aligned (
+ (__m128i*)dst, pack_2x128_128 (tmp_lo, tmp_hi));
+
+ w -= 4;
+ dst += 4;
+ }
+
+ while (w)
+ {
+ __m128i vd;
+
+ vd = unpack_32_1x128 (*dst);
+
+ *dst = pack_1x128_32 (over_1x128 (vd, expand_alpha_1x128 (vd),
+ xmm_src));
+ w--;
+ dst++;
+ }
+
+ }
+
+}
+
+static void
+sse2_composite_over_8888_8888_8888 (pixman_implementation_t *imp,
+ pixman_composite_info_t *info)
+{
+ PIXMAN_COMPOSITE_ARGS (info);
+ uint32_t *src, *src_line, s;
+ uint32_t *dst, *dst_line, d;
+ uint32_t *mask, *mask_line;
+ uint32_t m;
+ int src_stride, mask_stride, dst_stride;
+ int32_t w;
+
+ __m128i xmm_src, xmm_src_lo, xmm_src_hi, xmm_srca_lo, xmm_srca_hi;
+ __m128i xmm_dst, xmm_dst_lo, xmm_dst_hi;
+ __m128i xmm_mask, xmm_mask_lo, xmm_mask_hi;
+
+ PIXMAN_IMAGE_GET_LINE (
+ dest_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1);
+ PIXMAN_IMAGE_GET_LINE (
+ mask_image, mask_x, mask_y, uint32_t, mask_stride, mask_line, 1);
+ PIXMAN_IMAGE_GET_LINE (
+ src_image, src_x, src_y, uint32_t, src_stride, src_line, 1);
+
+ while (height--)
+ {
+ src = src_line;
+ src_line += src_stride;
+ dst = dst_line;
+ dst_line += dst_stride;
+ mask = mask_line;
+ mask_line += mask_stride;
+
+ w = width;
+
+ while (w && (unsigned long)dst & 15)
+ {
+ uint32_t sa;
+
+ s = *src++;
+ m = (*mask++) >> 24;
+ d = *dst;
+
+ sa = s >> 24;
+
+ if (m)
+ {
+ if (sa == 0xff && m == 0xff)
+ {
+ *dst = s;
+ }
+ else
+ {
+ __m128i ms, md, ma, msa;
+
+ ma = expand_alpha_rev_1x128 (load_32_1x128 (m));
+ ms = unpack_32_1x128 (s);
+ md = unpack_32_1x128 (d);
+
+ msa = expand_alpha_rev_1x128 (load_32_1x128 (sa));
+
+ *dst = pack_1x128_32 (in_over_1x128 (&ms, &msa, &ma, &md));
+ }
+ }
+
+ dst++;
+ w--;
+ }
+
+ while (w >= 4)
+ {
+ xmm_mask = load_128_unaligned ((__m128i*)mask);
+
+ if (!is_transparent (xmm_mask))
+ {
+ xmm_src = load_128_unaligned ((__m128i*)src);
+
+ if (is_opaque (xmm_mask) && is_opaque (xmm_src))
+ {
+ save_128_aligned ((__m128i *)dst, xmm_src);
+ }
+ else
+ {
+ xmm_dst = load_128_aligned ((__m128i *)dst);
+
+ unpack_128_2x128 (xmm_src, &xmm_src_lo, &xmm_src_hi);
+ unpack_128_2x128 (xmm_mask, &xmm_mask_lo, &xmm_mask_hi);
+ unpack_128_2x128 (xmm_dst, &xmm_dst_lo, &xmm_dst_hi);
+
+ expand_alpha_2x128 (xmm_src_lo, xmm_src_hi, &xmm_srca_lo, &xmm_srca_hi);
+ expand_alpha_2x128 (xmm_mask_lo, xmm_mask_hi, &xmm_mask_lo, &xmm_mask_hi);
+
+ in_over_2x128 (&xmm_src_lo, &xmm_src_hi, &xmm_srca_lo, &xmm_srca_hi,
+ &xmm_mask_lo, &xmm_mask_hi, &xmm_dst_lo, &xmm_dst_hi);
+
+ save_128_aligned ((__m128i*)dst, pack_2x128_128 (xmm_dst_lo, xmm_dst_hi));
+ }
+ }
+
+ src += 4;
+ dst += 4;
+ mask += 4;
+ w -= 4;
+ }
+
+ while (w)
+ {
+ uint32_t sa;
+
+ s = *src++;
+ m = (*mask++) >> 24;
+ d = *dst;
+
+ sa = s >> 24;
+
+ if (m)
+ {
+ if (sa == 0xff && m == 0xff)
+ {
+ *dst = s;
+ }
+ else
+ {
+ __m128i ms, md, ma, msa;
+
+ ma = expand_alpha_rev_1x128 (load_32_1x128 (m));
+ ms = unpack_32_1x128 (s);
+ md = unpack_32_1x128 (d);
+
+ msa = expand_alpha_rev_1x128 (load_32_1x128 (sa));
+
+ *dst = pack_1x128_32 (in_over_1x128 (&ms, &msa, &ma, &md));
+ }
+ }
+
+ dst++;
+ w--;
+ }
+ }
+
+}
+
+/* A variant of 'sse2_combine_over_u' with minor tweaks */
+static force_inline void
+scaled_nearest_scanline_sse2_8888_8888_OVER (uint32_t* pd,
+ const uint32_t* ps,
+ int32_t w,
+ pixman_fixed_t vx,
+ pixman_fixed_t unit_x,
+ pixman_fixed_t max_vx,
+ pixman_bool_t fully_transparent_src)
+{
+ uint32_t s, d;
+ const uint32_t* pm = NULL;
+
+ __m128i xmm_dst_lo, xmm_dst_hi;
+ __m128i xmm_src_lo, xmm_src_hi;
+ __m128i xmm_alpha_lo, xmm_alpha_hi;
+
+ if (fully_transparent_src)
+ return;
+
+ /* Align dst on a 16-byte boundary */
+ while (w && ((unsigned long)pd & 15))
+ {
+ d = *pd;
+ s = combine1 (ps + (vx >> 16), pm);
+ vx += unit_x;
+
+ *pd++ = core_combine_over_u_pixel_sse2 (s, d);
+ if (pm)
+ pm++;
+ w--;
+ }
+
+ while (w >= 4)
+ {
+ __m128i tmp;
+ uint32_t tmp1, tmp2, tmp3, tmp4;
+
+ tmp1 = ps[vx >> 16];
+ vx += unit_x;
+ tmp2 = ps[vx >> 16];
+ vx += unit_x;
+ tmp3 = ps[vx >> 16];
+ vx += unit_x;
+ tmp4 = ps[vx >> 16];
+ vx += unit_x;
+
+ tmp = _mm_set_epi32 (tmp4, tmp3, tmp2, tmp1);
+
+ xmm_src_hi = combine4 ((__m128i*)&tmp, (__m128i*)pm);
+
+ if (is_opaque (xmm_src_hi))
+ {
+ save_128_aligned ((__m128i*)pd, xmm_src_hi);
+ }
+ else if (!is_zero (xmm_src_hi))
+ {
+ xmm_dst_hi = load_128_aligned ((__m128i*) pd);
+
+ unpack_128_2x128 (xmm_src_hi, &xmm_src_lo, &xmm_src_hi);
+ unpack_128_2x128 (xmm_dst_hi, &xmm_dst_lo, &xmm_dst_hi);
+
+ expand_alpha_2x128 (
+ xmm_src_lo, xmm_src_hi, &xmm_alpha_lo, &xmm_alpha_hi);
+
+ over_2x128 (&xmm_src_lo, &xmm_src_hi,
+ &xmm_alpha_lo, &xmm_alpha_hi,
+ &xmm_dst_lo, &xmm_dst_hi);
+
+ /* rebuid the 4 pixel data and save*/
+ save_128_aligned ((__m128i*)pd,
+ pack_2x128_128 (xmm_dst_lo, xmm_dst_hi));
+ }
+
+ w -= 4;
+ pd += 4;
+ if (pm)
+ pm += 4;
+ }
+
+ while (w)
+ {
+ d = *pd;
+ s = combine1 (ps + (vx >> 16), pm);
+ vx += unit_x;
+
+ *pd++ = core_combine_over_u_pixel_sse2 (s, d);
+ if (pm)
+ pm++;
+
+ w--;
+ }
+}
+
+FAST_NEAREST_MAINLOOP (sse2_8888_8888_cover_OVER,
+ scaled_nearest_scanline_sse2_8888_8888_OVER,
+ uint32_t, uint32_t, COVER)
+FAST_NEAREST_MAINLOOP (sse2_8888_8888_none_OVER,
+ scaled_nearest_scanline_sse2_8888_8888_OVER,
+ uint32_t, uint32_t, NONE)
+FAST_NEAREST_MAINLOOP (sse2_8888_8888_pad_OVER,
+ scaled_nearest_scanline_sse2_8888_8888_OVER,
+ uint32_t, uint32_t, PAD)
+
+static force_inline void
+scaled_nearest_scanline_sse2_8888_n_8888_OVER (const uint32_t * mask,
+ uint32_t * dst,
+ const uint32_t * src,
+ int32_t w,
+ pixman_fixed_t vx,
+ pixman_fixed_t unit_x,
+ pixman_fixed_t max_vx,
+ pixman_bool_t zero_src)
+{
+ __m128i xmm_mask;
+ __m128i xmm_src, xmm_src_lo, xmm_src_hi;
+ __m128i xmm_dst, xmm_dst_lo, xmm_dst_hi;
+ __m128i xmm_alpha_lo, xmm_alpha_hi;
+
+ if (zero_src || (*mask >> 24) == 0)
+ return;
+
+ xmm_mask = create_mask_16_128 (*mask >> 24);
+
+ while (w && (unsigned long)dst & 15)
+ {
+ uint32_t s = src[pixman_fixed_to_int (vx)];
+ vx += unit_x;
+
+ if (s)
+ {
+ uint32_t d = *dst;
+
+ __m128i ms = unpack_32_1x128 (s);
+ __m128i alpha = expand_alpha_1x128 (ms);
+ __m128i dest = xmm_mask;
+ __m128i alpha_dst = unpack_32_1x128 (d);
+
+ *dst = pack_1x128_32 (
+ in_over_1x128 (&ms, &alpha, &dest, &alpha_dst));
+ }
+ dst++;
+ w--;
+ }
+
+ while (w >= 4)
+ {
+ uint32_t tmp1, tmp2, tmp3, tmp4;
+
+ tmp1 = src[pixman_fixed_to_int (vx)];
+ vx += unit_x;
+ tmp2 = src[pixman_fixed_to_int (vx)];
+ vx += unit_x;
+ tmp3 = src[pixman_fixed_to_int (vx)];
+ vx += unit_x;
+ tmp4 = src[pixman_fixed_to_int (vx)];
+ vx += unit_x;
+
+ xmm_src = _mm_set_epi32 (tmp4, tmp3, tmp2, tmp1);
+
+ if (!is_zero (xmm_src))
+ {
+ xmm_dst = load_128_aligned ((__m128i*)dst);
+
+ unpack_128_2x128 (xmm_src, &xmm_src_lo, &xmm_src_hi);
+ unpack_128_2x128 (xmm_dst, &xmm_dst_lo, &xmm_dst_hi);
+ expand_alpha_2x128 (xmm_src_lo, xmm_src_hi,
+ &xmm_alpha_lo, &xmm_alpha_hi);
+
+ in_over_2x128 (&xmm_src_lo, &xmm_src_hi,
+ &xmm_alpha_lo, &xmm_alpha_hi,
+ &xmm_mask, &xmm_mask,
+ &xmm_dst_lo, &xmm_dst_hi);
+
+ save_128_aligned (
+ (__m128i*)dst, pack_2x128_128 (xmm_dst_lo, xmm_dst_hi));
+ }
+
+ dst += 4;
+ w -= 4;
+ }
+
+ while (w)
+ {
+ uint32_t s = src[pixman_fixed_to_int (vx)];
+ vx += unit_x;
+
+ if (s)
+ {
+ uint32_t d = *dst;
+
+ __m128i ms = unpack_32_1x128 (s);
+ __m128i alpha = expand_alpha_1x128 (ms);
+ __m128i mask = xmm_mask;
+ __m128i dest = unpack_32_1x128 (d);
+
+ *dst = pack_1x128_32 (
+ in_over_1x128 (&ms, &alpha, &mask, &dest));
+ }
+
+ dst++;
+ w--;
+ }
+
+}
+
+FAST_NEAREST_MAINLOOP_COMMON (sse2_8888_n_8888_cover_OVER,
+ scaled_nearest_scanline_sse2_8888_n_8888_OVER,
+ uint32_t, uint32_t, uint32_t, COVER, TRUE, TRUE)
+FAST_NEAREST_MAINLOOP_COMMON (sse2_8888_n_8888_pad_OVER,
+ scaled_nearest_scanline_sse2_8888_n_8888_OVER,
+ uint32_t, uint32_t, uint32_t, PAD, TRUE, TRUE)
+FAST_NEAREST_MAINLOOP_COMMON (sse2_8888_n_8888_none_OVER,
+ scaled_nearest_scanline_sse2_8888_n_8888_OVER,
+ uint32_t, uint32_t, uint32_t, NONE, TRUE, TRUE)
+
+#define BILINEAR_DECLARE_VARIABLES \
+ const __m128i xmm_wt = _mm_set_epi16 (wt, wt, wt, wt, wt, wt, wt, wt); \
+ const __m128i xmm_wb = _mm_set_epi16 (wb, wb, wb, wb, wb, wb, wb, wb); \
+ const __m128i xmm_xorc = _mm_set_epi16 (0, 0, 0, 0, 0xff, 0xff, 0xff, 0xff);\
+ const __m128i xmm_addc = _mm_set_epi16 (0, 0, 0, 0, 1, 1, 1, 1); \
+ const __m128i xmm_ux = _mm_set_epi16 (unit_x, unit_x, unit_x, unit_x, \
+ unit_x, unit_x, unit_x, unit_x); \
+ const __m128i xmm_zero = _mm_setzero_si128 (); \
+ __m128i xmm_x = _mm_set_epi16 (vx, vx, vx, vx, vx, vx, vx, vx)
+
+#define BILINEAR_INTERPOLATE_ONE_PIXEL(pix) \
+do { \
+ __m128i xmm_wh, xmm_lo, xmm_hi, a; \
+ /* fetch 2x2 pixel block into sse2 register */ \
+ uint32_t tl = src_top [pixman_fixed_to_int (vx)]; \
+ uint32_t tr = src_top [pixman_fixed_to_int (vx) + 1]; \
+ uint32_t bl = src_bottom [pixman_fixed_to_int (vx)]; \
+ uint32_t br = src_bottom [pixman_fixed_to_int (vx) + 1]; \
+ a = _mm_set_epi32 (tr, tl, br, bl); \
+ vx += unit_x; \
+ /* vertical interpolation */ \
+ a = _mm_add_epi16 (_mm_mullo_epi16 (_mm_unpackhi_epi8 (a, xmm_zero), \
+ xmm_wt), \
+ _mm_mullo_epi16 (_mm_unpacklo_epi8 (a, xmm_zero), \
+ xmm_wb)); \
+ /* calculate horizontal weights */ \
+ xmm_wh = _mm_add_epi16 (xmm_addc, \
+ _mm_xor_si128 (xmm_xorc, \
+ _mm_srli_epi16 (xmm_x, 8))); \
+ xmm_x = _mm_add_epi16 (xmm_x, xmm_ux); \
+ /* horizontal interpolation */ \
+ xmm_lo = _mm_mullo_epi16 (a, xmm_wh); \
+ xmm_hi = _mm_mulhi_epu16 (a, xmm_wh); \
+ a = _mm_add_epi32 (_mm_unpacklo_epi16 (xmm_lo, xmm_hi), \
+ _mm_unpackhi_epi16 (xmm_lo, xmm_hi)); \
+ /* shift and pack the result */ \
+ a = _mm_srli_epi32 (a, 16); \
+ a = _mm_packs_epi32 (a, a); \
+ a = _mm_packus_epi16 (a, a); \
+ pix = _mm_cvtsi128_si32 (a); \
+} while (0)
+
+#define BILINEAR_SKIP_ONE_PIXEL() \
+do { \
+ vx += unit_x; \
+ xmm_x = _mm_add_epi16 (xmm_x, xmm_ux); \
+} while(0)
+
+static force_inline void
+scaled_bilinear_scanline_sse2_8888_8888_SRC (uint32_t * dst,
+ const uint32_t * mask,
+ const uint32_t * src_top,
+ const uint32_t * src_bottom,
+ int32_t w,
+ int wt,
+ int wb,
+ pixman_fixed_t vx,
+ pixman_fixed_t unit_x,
+ pixman_fixed_t max_vx,
+ pixman_bool_t zero_src)
+{
+ BILINEAR_DECLARE_VARIABLES;
+ uint32_t pix1, pix2, pix3, pix4;
+
+ while ((w -= 4) >= 0)
+ {
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix1);
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix2);
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix3);
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix4);
+ *dst++ = pix1;
+ *dst++ = pix2;
+ *dst++ = pix3;
+ *dst++ = pix4;
+ }
+
+ if (w & 2)
+ {
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix1);
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix2);
+ *dst++ = pix1;
+ *dst++ = pix2;
+ }
+
+ if (w & 1)
+ {
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix1);
+ *dst = pix1;
+ }
+
+}
+
+FAST_BILINEAR_MAINLOOP_COMMON (sse2_8888_8888_cover_SRC,
+ scaled_bilinear_scanline_sse2_8888_8888_SRC,
+ uint32_t, uint32_t, uint32_t,
+ COVER, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (sse2_8888_8888_pad_SRC,
+ scaled_bilinear_scanline_sse2_8888_8888_SRC,
+ uint32_t, uint32_t, uint32_t,
+ PAD, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (sse2_8888_8888_none_SRC,
+ scaled_bilinear_scanline_sse2_8888_8888_SRC,
+ uint32_t, uint32_t, uint32_t,
+ NONE, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (sse2_8888_8888_normal_SRC,
+ scaled_bilinear_scanline_sse2_8888_8888_SRC,
+ uint32_t, uint32_t, uint32_t,
+ NORMAL, FLAG_NONE)
+
+static force_inline void
+scaled_bilinear_scanline_sse2_8888_8888_OVER (uint32_t * dst,
+ const uint32_t * mask,
+ const uint32_t * src_top,
+ const uint32_t * src_bottom,
+ int32_t w,
+ int wt,
+ int wb,
+ pixman_fixed_t vx,
+ pixman_fixed_t unit_x,
+ pixman_fixed_t max_vx,
+ pixman_bool_t zero_src)
+{
+ BILINEAR_DECLARE_VARIABLES;
+ uint32_t pix1, pix2, pix3, pix4;
+
+ while (w && ((unsigned long)dst & 15))
+ {
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix1);
+
+ if (pix1)
+ {
+ pix2 = *dst;
+ *dst = core_combine_over_u_pixel_sse2 (pix1, pix2);
+ }
+
+ w--;
+ dst++;
+ }
+
+ while (w >= 4)
+ {
+ __m128i xmm_src;
+ __m128i xmm_src_hi, xmm_src_lo, xmm_dst_hi, xmm_dst_lo;
+ __m128i xmm_alpha_hi, xmm_alpha_lo;
+
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix1);
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix2);
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix3);
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix4);
+
+ xmm_src = _mm_set_epi32 (pix4, pix3, pix2, pix1);
+
+ if (!is_zero (xmm_src))
+ {
+ if (is_opaque (xmm_src))
+ {
+ save_128_aligned ((__m128i *)dst, xmm_src);
+ }
+ else
+ {
+ __m128i xmm_dst = load_128_aligned ((__m128i *)dst);
+
+ unpack_128_2x128 (xmm_src, &xmm_src_lo, &xmm_src_hi);
+ unpack_128_2x128 (xmm_dst, &xmm_dst_lo, &xmm_dst_hi);
+
+ expand_alpha_2x128 (xmm_src_lo, xmm_src_hi, &xmm_alpha_lo, &xmm_alpha_hi);
+ over_2x128 (&xmm_src_lo, &xmm_src_hi, &xmm_alpha_lo, &xmm_alpha_hi,
+ &xmm_dst_lo, &xmm_dst_hi);
+
+ save_128_aligned ((__m128i *)dst, pack_2x128_128 (xmm_dst_lo, xmm_dst_hi));
+ }
+ }
+
+ w -= 4;
+ dst += 4;
+ }
+
+ while (w)
+ {
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix1);
+
+ if (pix1)
+ {
+ pix2 = *dst;
+ *dst = core_combine_over_u_pixel_sse2 (pix1, pix2);
+ }
+
+ w--;
+ dst++;
+ }
+}
+
+FAST_BILINEAR_MAINLOOP_COMMON (sse2_8888_8888_cover_OVER,
+ scaled_bilinear_scanline_sse2_8888_8888_OVER,
+ uint32_t, uint32_t, uint32_t,
+ COVER, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (sse2_8888_8888_pad_OVER,
+ scaled_bilinear_scanline_sse2_8888_8888_OVER,
+ uint32_t, uint32_t, uint32_t,
+ PAD, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (sse2_8888_8888_none_OVER,
+ scaled_bilinear_scanline_sse2_8888_8888_OVER,
+ uint32_t, uint32_t, uint32_t,
+ NONE, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (sse2_8888_8888_normal_OVER,
+ scaled_bilinear_scanline_sse2_8888_8888_OVER,
+ uint32_t, uint32_t, uint32_t,
+ NORMAL, FLAG_NONE)
+
+static force_inline void
+scaled_bilinear_scanline_sse2_8888_8_8888_OVER (uint32_t * dst,
+ const uint8_t * mask,
+ const uint32_t * src_top,
+ const uint32_t * src_bottom,
+ int32_t w,
+ int wt,
+ int wb,
+ pixman_fixed_t vx,
+ pixman_fixed_t unit_x,
+ pixman_fixed_t max_vx,
+ pixman_bool_t zero_src)
+{
+ BILINEAR_DECLARE_VARIABLES;
+ uint32_t pix1, pix2, pix3, pix4;
+ uint32_t m;
+
+ while (w && ((unsigned long)dst & 15))
+ {
+ uint32_t sa;
+
+ m = (uint32_t) *mask++;
+
+ if (m)
+ {
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix1);
+ sa = pix1 >> 24;
+
+ if (sa == 0xff && m == 0xff)
+ {
+ *dst = pix1;
+ }
+ else
+ {
+ __m128i ms, md, ma, msa;
+
+ pix2 = *dst;
+ ma = expand_alpha_rev_1x128 (load_32_1x128 (m));
+ ms = unpack_32_1x128 (pix1);
+ md = unpack_32_1x128 (pix2);
+
+ msa = expand_alpha_rev_1x128 (load_32_1x128 (sa));
+
+ *dst = pack_1x128_32 (in_over_1x128 (&ms, &msa, &ma, &md));
+ }
+ }
+ else
+ {
+ BILINEAR_SKIP_ONE_PIXEL ();
+ }
+
+ w--;
+ dst++;
+ }
+
+ while (w >= 4)
+ {
+ __m128i xmm_src, xmm_src_lo, xmm_src_hi, xmm_srca_lo, xmm_srca_hi;
+ __m128i xmm_dst, xmm_dst_lo, xmm_dst_hi;
+ __m128i xmm_mask, xmm_mask_lo, xmm_mask_hi;
+
+ m = *(uint32_t*)mask;
+
+ if (m)
+ {
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix1);
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix2);
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix3);
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix4);
+
+ xmm_src = _mm_set_epi32 (pix4, pix3, pix2, pix1);
+
+ if (m == 0xffffffff && is_opaque (xmm_src))
+ {
+ save_128_aligned ((__m128i *)dst, xmm_src);
+ }
+ else
+ {
+ xmm_dst = load_128_aligned ((__m128i *)dst);
+
+ xmm_mask = _mm_unpacklo_epi16 (unpack_32_1x128 (m), _mm_setzero_si128());
+
+ unpack_128_2x128 (xmm_src, &xmm_src_lo, &xmm_src_hi);
+ unpack_128_2x128 (xmm_mask, &xmm_mask_lo, &xmm_mask_hi);
+ unpack_128_2x128 (xmm_dst, &xmm_dst_lo, &xmm_dst_hi);
+
+ expand_alpha_2x128 (xmm_src_lo, xmm_src_hi, &xmm_srca_lo, &xmm_srca_hi);
+ expand_alpha_rev_2x128 (xmm_mask_lo, xmm_mask_hi, &xmm_mask_lo, &xmm_mask_hi);
+
+ in_over_2x128 (&xmm_src_lo, &xmm_src_hi, &xmm_srca_lo, &xmm_srca_hi,
+ &xmm_mask_lo, &xmm_mask_hi, &xmm_dst_lo, &xmm_dst_hi);
+
+ save_128_aligned ((__m128i*)dst, pack_2x128_128 (xmm_dst_lo, xmm_dst_hi));
+ }
+ }
+ else
+ {
+ BILINEAR_SKIP_ONE_PIXEL ();
+ BILINEAR_SKIP_ONE_PIXEL ();
+ BILINEAR_SKIP_ONE_PIXEL ();
+ BILINEAR_SKIP_ONE_PIXEL ();
+ }
+
+ w -= 4;
+ dst += 4;
+ mask += 4;
+ }
+
+ while (w)
+ {
+ uint32_t sa;
+
+ m = (uint32_t) *mask++;
+
+ if (m)
+ {
+ BILINEAR_INTERPOLATE_ONE_PIXEL (pix1);
+ sa = pix1 >> 24;
+
+ if (sa == 0xff && m == 0xff)
+ {
+ *dst = pix1;
+ }
+ else
+ {
+ __m128i ms, md, ma, msa;
+
+ pix2 = *dst;
+ ma = expand_alpha_rev_1x128 (load_32_1x128 (m));
+ ms = unpack_32_1x128 (pix1);
+ md = unpack_32_1x128 (pix2);
+
+ msa = expand_alpha_rev_1x128 (load_32_1x128 (sa));
+
+ *dst = pack_1x128_32 (in_over_1x128 (&ms, &msa, &ma, &md));
+ }
+ }
+ else
+ {
+ BILINEAR_SKIP_ONE_PIXEL ();
+ }
+
+ w--;
+ dst++;
+ }
+}
+
+FAST_BILINEAR_MAINLOOP_COMMON (sse2_8888_8_8888_cover_OVER,
+ scaled_bilinear_scanline_sse2_8888_8_8888_OVER,
+ uint32_t, uint8_t, uint32_t,
+ COVER, FLAG_HAVE_NON_SOLID_MASK)
+FAST_BILINEAR_MAINLOOP_COMMON (sse2_8888_8_8888_pad_OVER,
+ scaled_bilinear_scanline_sse2_8888_8_8888_OVER,
+ uint32_t, uint8_t, uint32_t,
+ PAD, FLAG_HAVE_NON_SOLID_MASK)
+FAST_BILINEAR_MAINLOOP_COMMON (sse2_8888_8_8888_none_OVER,
+ scaled_bilinear_scanline_sse2_8888_8_8888_OVER,
+ uint32_t, uint8_t, uint32_t,
+ NONE, FLAG_HAVE_NON_SOLID_MASK)
+FAST_BILINEAR_MAINLOOP_COMMON (sse2_8888_8_8888_normal_OVER,
+ scaled_bilinear_scanline_sse2_8888_8_8888_OVER,
+ uint32_t, uint8_t, uint32_t,
+ NORMAL, FLAG_HAVE_NON_SOLID_MASK)
+
+static const pixman_fast_path_t sse2_fast_paths[] =
+{
+ /* PIXMAN_OP_OVER */
+ PIXMAN_STD_FAST_PATH (OVER, solid, a8, r5g6b5, sse2_composite_over_n_8_0565),
+ PIXMAN_STD_FAST_PATH (OVER, solid, a8, b5g6r5, sse2_composite_over_n_8_0565),
+ PIXMAN_STD_FAST_PATH (OVER, solid, null, a8r8g8b8, sse2_composite_over_n_8888),
+ PIXMAN_STD_FAST_PATH (OVER, solid, null, x8r8g8b8, sse2_composite_over_n_8888),
+ PIXMAN_STD_FAST_PATH (OVER, solid, null, r5g6b5, sse2_composite_over_n_0565),
+ PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, a8r8g8b8, sse2_composite_over_8888_8888),
+ PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, x8r8g8b8, sse2_composite_over_8888_8888),
+ PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, a8b8g8r8, sse2_composite_over_8888_8888),
+ PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, x8b8g8r8, sse2_composite_over_8888_8888),
+ PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, r5g6b5, sse2_composite_over_8888_0565),
+ PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, b5g6r5, sse2_composite_over_8888_0565),
+ PIXMAN_STD_FAST_PATH (OVER, solid, a8, a8r8g8b8, sse2_composite_over_n_8_8888),
+ PIXMAN_STD_FAST_PATH (OVER, solid, a8, x8r8g8b8, sse2_composite_over_n_8_8888),
+ PIXMAN_STD_FAST_PATH (OVER, solid, a8, a8b8g8r8, sse2_composite_over_n_8_8888),
+ PIXMAN_STD_FAST_PATH (OVER, solid, a8, x8b8g8r8, sse2_composite_over_n_8_8888),
+ PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, a8r8g8b8, sse2_composite_over_8888_8888_8888),
+ PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, a8, x8r8g8b8, sse2_composite_over_8888_8_8888),
+ PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, a8, a8r8g8b8, sse2_composite_over_8888_8_8888),
+ PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, a8, x8b8g8r8, sse2_composite_over_8888_8_8888),
+ PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, a8, a8b8g8r8, sse2_composite_over_8888_8_8888),
+ PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, a8, x8r8g8b8, sse2_composite_over_x888_8_8888),
+ PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, a8, a8r8g8b8, sse2_composite_over_x888_8_8888),
+ PIXMAN_STD_FAST_PATH (OVER, x8b8g8r8, a8, x8b8g8r8, sse2_composite_over_x888_8_8888),
+ PIXMAN_STD_FAST_PATH (OVER, x8b8g8r8, a8, a8b8g8r8, sse2_composite_over_x888_8_8888),
+ PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, solid, a8r8g8b8, sse2_composite_over_x888_n_8888),
+ PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, solid, x8r8g8b8, sse2_composite_over_x888_n_8888),
+ PIXMAN_STD_FAST_PATH (OVER, x8b8g8r8, solid, a8b8g8r8, sse2_composite_over_x888_n_8888),
+ PIXMAN_STD_FAST_PATH (OVER, x8b8g8r8, solid, x8b8g8r8, sse2_composite_over_x888_n_8888),
+ PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, solid, a8r8g8b8, sse2_composite_over_8888_n_8888),
+ PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, solid, x8r8g8b8, sse2_composite_over_8888_n_8888),
+ PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, solid, a8b8g8r8, sse2_composite_over_8888_n_8888),
+ PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, solid, x8b8g8r8, sse2_composite_over_8888_n_8888),
+ PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, a8r8g8b8, sse2_composite_over_n_8888_8888_ca),
+ PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, x8r8g8b8, sse2_composite_over_n_8888_8888_ca),
+ PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, a8b8g8r8, sse2_composite_over_n_8888_8888_ca),
+ PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, x8b8g8r8, sse2_composite_over_n_8888_8888_ca),
+ PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, r5g6b5, sse2_composite_over_n_8888_0565_ca),
+ PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, b5g6r5, sse2_composite_over_n_8888_0565_ca),
+ PIXMAN_STD_FAST_PATH (OVER, pixbuf, pixbuf, a8r8g8b8, sse2_composite_over_pixbuf_8888),
+ PIXMAN_STD_FAST_PATH (OVER, pixbuf, pixbuf, x8r8g8b8, sse2_composite_over_pixbuf_8888),
+ PIXMAN_STD_FAST_PATH (OVER, rpixbuf, rpixbuf, a8b8g8r8, sse2_composite_over_pixbuf_8888),
+ PIXMAN_STD_FAST_PATH (OVER, rpixbuf, rpixbuf, x8b8g8r8, sse2_composite_over_pixbuf_8888),
+ PIXMAN_STD_FAST_PATH (OVER, pixbuf, pixbuf, r5g6b5, sse2_composite_over_pixbuf_0565),
+ PIXMAN_STD_FAST_PATH (OVER, rpixbuf, rpixbuf, b5g6r5, sse2_composite_over_pixbuf_0565),
+ PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, null, x8r8g8b8, sse2_composite_copy_area),
+ PIXMAN_STD_FAST_PATH (OVER, x8b8g8r8, null, x8b8g8r8, sse2_composite_copy_area),
+
+ /* PIXMAN_OP_OVER_REVERSE */
+ PIXMAN_STD_FAST_PATH (OVER_REVERSE, solid, null, a8r8g8b8, sse2_composite_over_reverse_n_8888),
+ PIXMAN_STD_FAST_PATH (OVER_REVERSE, solid, null, a8b8g8r8, sse2_composite_over_reverse_n_8888),
+
+ /* PIXMAN_OP_ADD */
+ PIXMAN_STD_FAST_PATH_CA (ADD, solid, a8r8g8b8, a8r8g8b8, sse2_composite_add_n_8888_8888_ca),
+ PIXMAN_STD_FAST_PATH (ADD, a8, null, a8, sse2_composite_add_8_8),
+ PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, null, a8r8g8b8, sse2_composite_add_8888_8888),
+ PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, null, a8b8g8r8, sse2_composite_add_8888_8888),
+ PIXMAN_STD_FAST_PATH (ADD, solid, a8, a8, sse2_composite_add_n_8_8),
+ PIXMAN_STD_FAST_PATH (ADD, solid, null, a8, sse2_composite_add_n_8),
+
+ /* PIXMAN_OP_SRC */
+ PIXMAN_STD_FAST_PATH (SRC, solid, a8, a8r8g8b8, sse2_composite_src_n_8_8888),
+ PIXMAN_STD_FAST_PATH (SRC, solid, a8, x8r8g8b8, sse2_composite_src_n_8_8888),
+ PIXMAN_STD_FAST_PATH (SRC, solid, a8, a8b8g8r8, sse2_composite_src_n_8_8888),
+ PIXMAN_STD_FAST_PATH (SRC, solid, a8, x8b8g8r8, sse2_composite_src_n_8_8888),
+ PIXMAN_STD_FAST_PATH (SRC, x8r8g8b8, null, a8r8g8b8, sse2_composite_src_x888_8888),
+ PIXMAN_STD_FAST_PATH (SRC, x8b8g8r8, null, a8b8g8r8, sse2_composite_src_x888_8888),
+ PIXMAN_STD_FAST_PATH (SRC, a8r8g8b8, null, a8r8g8b8, sse2_composite_copy_area),
+ PIXMAN_STD_FAST_PATH (SRC, a8b8g8r8, null, a8b8g8r8, sse2_composite_copy_area),
+ PIXMAN_STD_FAST_PATH (SRC, a8r8g8b8, null, x8r8g8b8, sse2_composite_copy_area),
+ PIXMAN_STD_FAST_PATH (SRC, a8b8g8r8, null, x8b8g8r8, sse2_composite_copy_area),
+ PIXMAN_STD_FAST_PATH (SRC, x8r8g8b8, null, x8r8g8b8, sse2_composite_copy_area),
+ PIXMAN_STD_FAST_PATH (SRC, x8b8g8r8, null, x8b8g8r8, sse2_composite_copy_area),
+ PIXMAN_STD_FAST_PATH (SRC, r5g6b5, null, r5g6b5, sse2_composite_copy_area),
+ PIXMAN_STD_FAST_PATH (SRC, b5g6r5, null, b5g6r5, sse2_composite_copy_area),
+
+ /* PIXMAN_OP_IN */
+ PIXMAN_STD_FAST_PATH (IN, a8, null, a8, sse2_composite_in_8_8),
+ PIXMAN_STD_FAST_PATH (IN, solid, a8, a8, sse2_composite_in_n_8_8),
+ PIXMAN_STD_FAST_PATH (IN, solid, null, a8, sse2_composite_in_n_8),
+
+ SIMPLE_NEAREST_FAST_PATH_COVER (OVER, a8r8g8b8, x8r8g8b8, sse2_8888_8888),
+ SIMPLE_NEAREST_FAST_PATH_COVER (OVER, a8b8g8r8, x8b8g8r8, sse2_8888_8888),
+ SIMPLE_NEAREST_FAST_PATH_COVER (OVER, a8r8g8b8, a8r8g8b8, sse2_8888_8888),
+ SIMPLE_NEAREST_FAST_PATH_COVER (OVER, a8b8g8r8, a8b8g8r8, sse2_8888_8888),
+ SIMPLE_NEAREST_FAST_PATH_NONE (OVER, a8r8g8b8, x8r8g8b8, sse2_8888_8888),
+ SIMPLE_NEAREST_FAST_PATH_NONE (OVER, a8b8g8r8, x8b8g8r8, sse2_8888_8888),
+ SIMPLE_NEAREST_FAST_PATH_NONE (OVER, a8r8g8b8, a8r8g8b8, sse2_8888_8888),
+ SIMPLE_NEAREST_FAST_PATH_NONE (OVER, a8b8g8r8, a8b8g8r8, sse2_8888_8888),
+ SIMPLE_NEAREST_FAST_PATH_PAD (OVER, a8r8g8b8, x8r8g8b8, sse2_8888_8888),
+ SIMPLE_NEAREST_FAST_PATH_PAD (OVER, a8b8g8r8, x8b8g8r8, sse2_8888_8888),
+ SIMPLE_NEAREST_FAST_PATH_PAD (OVER, a8r8g8b8, a8r8g8b8, sse2_8888_8888),
+ SIMPLE_NEAREST_FAST_PATH_PAD (OVER, a8b8g8r8, a8b8g8r8, sse2_8888_8888),
+
+ SIMPLE_NEAREST_SOLID_MASK_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, sse2_8888_n_8888),
+ SIMPLE_NEAREST_SOLID_MASK_FAST_PATH (OVER, a8b8g8r8, a8b8g8r8, sse2_8888_n_8888),
+ SIMPLE_NEAREST_SOLID_MASK_FAST_PATH (OVER, a8r8g8b8, x8r8g8b8, sse2_8888_n_8888),
+ SIMPLE_NEAREST_SOLID_MASK_FAST_PATH (OVER, a8b8g8r8, x8b8g8r8, sse2_8888_n_8888),
+
+ SIMPLE_BILINEAR_FAST_PATH (SRC, a8r8g8b8, a8r8g8b8, sse2_8888_8888),
+ SIMPLE_BILINEAR_FAST_PATH (SRC, a8r8g8b8, x8r8g8b8, sse2_8888_8888),
+ SIMPLE_BILINEAR_FAST_PATH (SRC, x8r8g8b8, x8r8g8b8, sse2_8888_8888),
+
+ SIMPLE_BILINEAR_FAST_PATH (OVER, a8r8g8b8, x8r8g8b8, sse2_8888_8888),
+ SIMPLE_BILINEAR_FAST_PATH (OVER, a8b8g8r8, x8b8g8r8, sse2_8888_8888),
+ SIMPLE_BILINEAR_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, sse2_8888_8888),
+ SIMPLE_BILINEAR_FAST_PATH (OVER, a8b8g8r8, a8b8g8r8, sse2_8888_8888),
+
+ SIMPLE_BILINEAR_A8_MASK_FAST_PATH (OVER, a8r8g8b8, x8r8g8b8, sse2_8888_8_8888),
+ SIMPLE_BILINEAR_A8_MASK_FAST_PATH (OVER, a8b8g8r8, x8b8g8r8, sse2_8888_8_8888),
+ SIMPLE_BILINEAR_A8_MASK_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, sse2_8888_8_8888),
+ SIMPLE_BILINEAR_A8_MASK_FAST_PATH (OVER, a8b8g8r8, a8b8g8r8, sse2_8888_8_8888),
+
+ { PIXMAN_OP_NONE },
+};
+
+static pixman_bool_t
+sse2_blt (pixman_implementation_t *imp,
+ 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 dest_x,
+ int dest_y,
+ int width,
+ int height)
+{
+ if (!pixman_blt_sse2 (
+ src_bits, dst_bits, src_stride, dst_stride, src_bpp, dst_bpp,
+ src_x, src_y, dest_x, dest_y, width, height))
+
+ {
+ return _pixman_implementation_blt (
+ imp->delegate,
+ src_bits, dst_bits, src_stride, dst_stride, src_bpp, dst_bpp,
+ src_x, src_y, dest_x, dest_y, width, height);
+ }
+
+ return TRUE;
+}
+
+#if defined(__GNUC__) && !defined(__x86_64__) && !defined(__amd64__)
+__attribute__((__force_align_arg_pointer__))
+#endif
+static pixman_bool_t
+sse2_fill (pixman_implementation_t *imp,
+ uint32_t * bits,
+ int stride,
+ int bpp,
+ int x,
+ int y,
+ int width,
+ int height,
+ uint32_t xor)
+{
+ if (!pixman_fill_sse2 (bits, stride, bpp, x, y, width, height, xor))
+ {
+ return _pixman_implementation_fill (
+ imp->delegate, bits, stride, bpp, x, y, width, height, xor);
+ }
+
+ return TRUE;
+}
+
+static uint32_t *
+sse2_fetch_x8r8g8b8 (pixman_iter_t *iter, const uint32_t *mask)
+{
+ int w = iter->width;
+ __m128i ff000000 = mask_ff000000;
+ uint32_t *dst = iter->buffer;
+ uint32_t *src = (uint32_t *)iter->bits;
+
+ iter->bits += iter->stride;
+
+ while (w && ((unsigned long)dst) & 0x0f)
+ {
+ *dst++ = (*src++) | 0xff000000;
+ w--;
+ }
+
+ while (w >= 4)
+ {
+ save_128_aligned (
+ (__m128i *)dst, _mm_or_si128 (
+ load_128_unaligned ((__m128i *)src), ff000000));
+
+ dst += 4;
+ src += 4;
+ w -= 4;
+ }
+
+ while (w)
+ {
+ *dst++ = (*src++) | 0xff000000;
+ w--;
+ }
+
+ return iter->buffer;
+}
+
+static uint32_t *
+sse2_fetch_r5g6b5 (pixman_iter_t *iter, const uint32_t *mask)
+{
+ int w = iter->width;
+ uint32_t *dst = iter->buffer;
+ uint16_t *src = (uint16_t *)iter->bits;
+ __m128i ff000000 = mask_ff000000;
+
+ iter->bits += iter->stride;
+
+ while (w && ((unsigned long)dst) & 0x0f)
+ {
+ uint16_t s = *src++;
+
+ *dst++ = CONVERT_0565_TO_8888 (s);
+ w--;
+ }
+
+ while (w >= 8)
+ {
+ __m128i lo, hi, s;
+
+ s = _mm_loadu_si128 ((__m128i *)src);
+
+ lo = unpack_565_to_8888 (_mm_unpacklo_epi16 (s, _mm_setzero_si128 ()));
+ hi = unpack_565_to_8888 (_mm_unpackhi_epi16 (s, _mm_setzero_si128 ()));
+
+ save_128_aligned ((__m128i *)(dst + 0), _mm_or_si128 (lo, ff000000));
+ save_128_aligned ((__m128i *)(dst + 4), _mm_or_si128 (hi, ff000000));
+
+ dst += 8;
+ src += 8;
+ w -= 8;
+ }
+
+ while (w)
+ {
+ uint16_t s = *src++;
+
+ *dst++ = CONVERT_0565_TO_8888 (s);
+ w--;
+ }
+
+ return iter->buffer;
+}
+
+static uint32_t *
+sse2_fetch_a8 (pixman_iter_t *iter, const uint32_t *mask)
+{
+ int w = iter->width;
+ uint32_t *dst = iter->buffer;
+ uint8_t *src = iter->bits;
+ __m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6;
+
+ iter->bits += iter->stride;
+
+ while (w && (((unsigned long)dst) & 15))
+ {
+ *dst++ = *(src++) << 24;
+ w--;
+ }
+
+ while (w >= 16)
+ {
+ xmm0 = _mm_loadu_si128((__m128i *)src);
+
+ xmm1 = _mm_unpacklo_epi8 (_mm_setzero_si128(), xmm0);
+ xmm2 = _mm_unpackhi_epi8 (_mm_setzero_si128(), xmm0);
+ xmm3 = _mm_unpacklo_epi16 (_mm_setzero_si128(), xmm1);
+ xmm4 = _mm_unpackhi_epi16 (_mm_setzero_si128(), xmm1);
+ xmm5 = _mm_unpacklo_epi16 (_mm_setzero_si128(), xmm2);
+ xmm6 = _mm_unpackhi_epi16 (_mm_setzero_si128(), xmm2);
+
+ _mm_store_si128(((__m128i *)(dst + 0)), xmm3);
+ _mm_store_si128(((__m128i *)(dst + 4)), xmm4);
+ _mm_store_si128(((__m128i *)(dst + 8)), xmm5);
+ _mm_store_si128(((__m128i *)(dst + 12)), xmm6);
+
+ dst += 16;
+ src += 16;
+ w -= 16;
+ }
+
+ while (w)
+ {
+ *dst++ = *(src++) << 24;
+ w--;
+ }
+
+ return iter->buffer;
+}
+
+typedef struct
+{
+ pixman_format_code_t format;
+ pixman_iter_get_scanline_t get_scanline;
+} fetcher_info_t;
+
+static const fetcher_info_t fetchers[] =
+{
+ { PIXMAN_x8r8g8b8, sse2_fetch_x8r8g8b8 },
+ { PIXMAN_r5g6b5, sse2_fetch_r5g6b5 },
+ { PIXMAN_a8, sse2_fetch_a8 },
+ { PIXMAN_null }
+};
+
+static void
+sse2_src_iter_init (pixman_implementation_t *imp, pixman_iter_t *iter)
+{
+ pixman_image_t *image = iter->image;
+ int x = iter->x;
+ int y = iter->y;
+ int width = iter->width;
+ int height = iter->height;
+
+#define FLAGS \
+ (FAST_PATH_STANDARD_FLAGS | FAST_PATH_ID_TRANSFORM | FAST_PATH_BITS_IMAGE)
+
+ if ((iter->flags & ITER_NARROW) &&
+ (image->common.flags & FLAGS) == FLAGS &&
+ x >= 0 && y >= 0 &&
+ x + width <= image->bits.width &&
+ y + height <= image->bits.height)
+ {
+ const fetcher_info_t *f;
+
+ for (f = &fetchers[0]; f->format != PIXMAN_null; f++)
+ {
+ if (image->common.extended_format_code == f->format)
+ {
+ uint8_t *b = (uint8_t *)image->bits.bits;
+ int s = image->bits.rowstride * 4;
+
+ iter->bits = b + s * iter->y + x * PIXMAN_FORMAT_BPP (f->format) / 8;
+ iter->stride = s;
+
+ iter->get_scanline = f->get_scanline;
+ return;
+ }
+ }
+ }
+
+ imp->delegate->src_iter_init (imp->delegate, iter);
+}
+
+#if defined(__GNUC__) && !defined(__x86_64__) && !defined(__amd64__)
+__attribute__((__force_align_arg_pointer__))
+#endif
+pixman_implementation_t *
+_pixman_implementation_create_sse2 (pixman_implementation_t *fallback)
+{
+ pixman_implementation_t *imp = _pixman_implementation_create (fallback, sse2_fast_paths);
+
+ /* SSE2 constants */
+ mask_565_r = create_mask_2x32_128 (0x00f80000, 0x00f80000);
+ mask_565_g1 = create_mask_2x32_128 (0x00070000, 0x00070000);
+ mask_565_g2 = create_mask_2x32_128 (0x000000e0, 0x000000e0);
+ mask_565_b = create_mask_2x32_128 (0x0000001f, 0x0000001f);
+ mask_red = create_mask_2x32_128 (0x00f80000, 0x00f80000);
+ mask_green = create_mask_2x32_128 (0x0000fc00, 0x0000fc00);
+ mask_blue = create_mask_2x32_128 (0x000000f8, 0x000000f8);
+ mask_565_fix_rb = create_mask_2x32_128 (0x00e000e0, 0x00e000e0);
+ mask_565_fix_g = create_mask_2x32_128 (0x0000c000, 0x0000c000);
+ mask_0080 = create_mask_16_128 (0x0080);
+ mask_00ff = create_mask_16_128 (0x00ff);
+ mask_0101 = create_mask_16_128 (0x0101);
+ mask_ffff = create_mask_16_128 (0xffff);
+ mask_ff000000 = create_mask_2x32_128 (0xff000000, 0xff000000);
+ mask_alpha = create_mask_2x32_128 (0x00ff0000, 0x00000000);
+
+ /* Set up function pointers */
+ imp->combine_32[PIXMAN_OP_OVER] = sse2_combine_over_u;
+ imp->combine_32[PIXMAN_OP_OVER_REVERSE] = sse2_combine_over_reverse_u;
+ imp->combine_32[PIXMAN_OP_IN] = sse2_combine_in_u;
+ imp->combine_32[PIXMAN_OP_IN_REVERSE] = sse2_combine_in_reverse_u;
+ imp->combine_32[PIXMAN_OP_OUT] = sse2_combine_out_u;
+ imp->combine_32[PIXMAN_OP_OUT_REVERSE] = sse2_combine_out_reverse_u;
+ imp->combine_32[PIXMAN_OP_ATOP] = sse2_combine_atop_u;
+ imp->combine_32[PIXMAN_OP_ATOP_REVERSE] = sse2_combine_atop_reverse_u;
+ imp->combine_32[PIXMAN_OP_XOR] = sse2_combine_xor_u;
+ imp->combine_32[PIXMAN_OP_ADD] = sse2_combine_add_u;
+
+ imp->combine_32[PIXMAN_OP_SATURATE] = sse2_combine_saturate_u;
+
+ imp->combine_32_ca[PIXMAN_OP_SRC] = sse2_combine_src_ca;
+ imp->combine_32_ca[PIXMAN_OP_OVER] = sse2_combine_over_ca;
+ imp->combine_32_ca[PIXMAN_OP_OVER_REVERSE] = sse2_combine_over_reverse_ca;
+ imp->combine_32_ca[PIXMAN_OP_IN] = sse2_combine_in_ca;
+ imp->combine_32_ca[PIXMAN_OP_IN_REVERSE] = sse2_combine_in_reverse_ca;
+ imp->combine_32_ca[PIXMAN_OP_OUT] = sse2_combine_out_ca;
+ imp->combine_32_ca[PIXMAN_OP_OUT_REVERSE] = sse2_combine_out_reverse_ca;
+ imp->combine_32_ca[PIXMAN_OP_ATOP] = sse2_combine_atop_ca;
+ imp->combine_32_ca[PIXMAN_OP_ATOP_REVERSE] = sse2_combine_atop_reverse_ca;
+ imp->combine_32_ca[PIXMAN_OP_XOR] = sse2_combine_xor_ca;
+ imp->combine_32_ca[PIXMAN_OP_ADD] = sse2_combine_add_ca;
+
+ imp->blt = sse2_blt;
+ imp->fill = sse2_fill;
+
+ imp->src_iter_init = sse2_src_iter_init;
+
+ return imp;
+}