ARM: added 'neon_composite_add_8888_8888' fast path

diff --git a/pixman/pixman-arm-neon-asm.S b/pixman/pixman-arm-neon-asm.S
index 49d840c..81f0b92 100644 (file)
--- a/pixman/pixman-arm-neon-asm.S
+++ b/pixman/pixman-arm-neon-asm.S
@@ -427,6 +427,38 @@ generate_composite_function \
 
 /******************************************************************************/
 
+.macro pixman_composite_add_8888_8888_process_pixblock_tail_head
+    vld1.8      {d0, d1, d2, d3}, [SRC]!
+                                    PF add PF_X, PF_X, #8
+                                    PF tst PF_CTL, #0xF
+    vld1.8      {d4, d5, d6, d7}, [DST_R, :128]!
+                                    PF addne PF_X, PF_X, #8
+                                    PF subne PF_CTL, PF_CTL, #1
+        vst1.8      {d28, d29, d30, d31}, [DST_W, :128]!
+                                    PF cmp PF_X, ORIG_W
+                                    PF pld, [PF_SRC, PF_X, lsl #src_bpp_shift]
+                                    PF pld, [PF_DST, PF_X, lsl #dst_bpp_shift]
+                                    PF subge PF_X, PF_X, ORIG_W
+                                    PF subges PF_CTL, PF_CTL, #0x10
+    vqadd.u8    q14, q0, q2
+                                    PF ldrgeb DUMMY, [PF_SRC, SRC_STRIDE, lsl #src_bpp_shift]!
+                                    PF ldrgeb DUMMY, [PF_DST, DST_STRIDE, lsl #dst_bpp_shift]!
+    vqadd.u8    q15, q1, q3
+.endm
+
+generate_composite_function \
+    pixman_composite_add_8888_8888_asm_neon, 32, 0, 32, \
+    FLAG_DST_READWRITE, \
+    8, /* number of pixels, processed in a single block */ \
+    10, /* prefetch distance */ \
+    default_init, \
+    default_cleanup, \
+    pixman_composite_add_8000_8000_process_pixblock_head, \
+    pixman_composite_add_8000_8000_process_pixblock_tail, \
+    pixman_composite_add_8888_8888_process_pixblock_tail_head
+
+/******************************************************************************/
+
 .macro pixman_composite_over_8888_8888_process_pixblock_head
     vmvn.8      d24, d3  /* get inverted alpha */
     /* do alpha blending */

diff --git a/pixman/pixman-arm-neon.c b/pixman/pixman-arm-neon.c
index 675b64f..fe444f2 100644 (file)
--- a/pixman/pixman-arm-neon.c
+++ b/pixman/pixman-arm-neon.c
@@ -215,6 +215,7 @@ BIND_SRC_NULL_DST(src_0565_0565, uint16_t, 1, uint16_t, 1)
 BIND_SRC_NULL_DST(src_0888_0888, uint8_t, 3, uint8_t, 3)
 BIND_SRC_NULL_DST(src_8888_0565, uint32_t, 1, uint16_t, 1)
 BIND_SRC_NULL_DST(add_8000_8000, uint8_t, 1, uint8_t, 1)
+BIND_SRC_NULL_DST(add_8888_8888, uint32_t, 1, uint32_t, 1)
 
 BIND_SRC_NULL_DST(over_8888_0565, uint32_t, 1, uint16_t, 1)
 BIND_SRC_NULL_DST(over_8888_8888, uint32_t, 1, uint32_t, 1)
@@ -369,6 +370,8 @@ static const pixman_fast_path_t arm_neon_fast_path_array[] =
     { PIXMAN_OP_ADD,  PIXMAN_solid,    PIXMAN_a8,       PIXMAN_a8,       neon_composite_add_n_8_8        },
     { PIXMAN_OP_ADD,  PIXMAN_a8,       PIXMAN_a8,       PIXMAN_a8,       neon_composite_add_8_8_8        },
     { PIXMAN_OP_ADD,  PIXMAN_a8,       PIXMAN_null,     PIXMAN_a8,       neon_composite_add_8000_8000    },
+    { PIXMAN_OP_ADD,  PIXMAN_a8r8g8b8, PIXMAN_null,     PIXMAN_a8r8g8b8, neon_composite_add_8888_8888    },
+    { PIXMAN_OP_ADD,  PIXMAN_a8b8g8r8, PIXMAN_null,     PIXMAN_a8b8g8r8, neon_composite_add_8888_8888    },
     { PIXMAN_OP_NONE },
 };