[gpsp.git] / arm / video_blend.S

.align 2\r
\r
.global expand_blend\r
.global expand_normal\r
\r
@ Input:\r
@   r0 = screen_src_ptr\r
@   r1 = screen_dest_ptr\r
@   r2 = start\r
@   r3 = end\r
\r
6:\r
  .word io_registers\r
  .word palette_ram_converted\r
  .word 0x04000200                @ combine test mask\r
  .word 0x07E0F81F                @ clamp mask\r
  .word 0x000003FE                @ palette index mask\r
  .word 0x08010020                @ saturation mask\r
\r
expand_blend:\r
  stmdb sp!, { r4, r5, r6, r9, r10, r11, r14 }\r
\r
  add r0, r0, r2, lsl #2          @ screen_src_ptr += start\r
  add r1, r1, r2, lsl #1          @ screen_dest_ptr += start\r
  sub r2, r3, r2                  @ r2 = end - start\r
  ldr r3, 6b                      @ r3 = io_registers\r
  ldrh r3, [r3, #0x52]            @ r3 = bldalpha\r
  mov r4, r3, lsr #8              @ r4 = bldalpha >> 8\r
  and r3, r3, #0x1F               @ r3 = blend_a\r
  and r4, r4, #0x1F               @ r4 = blend_b\r
  cmp r3, #16                     @ if(blend_a > 16)\r
  movgt r3, #16                   @   blend_a = 16\r
  cmp r4, #16                     @ if(blend_b > 16)\r
  movgt r4, #16                   @   blend_b = 16\r
\r
  ldr r14, 6b + 4                 @ r14 = palette_ram_converted\r
  ldr r12, 6b + 8                 @ r12 = 0x04000200\r
  ldr r11, 6b + 12                @ r11 = 0x07E0F81F\r
  ldr r10, 6b + 16                @ r10 = 0x000003FE\r
\r
  add r5, r3, r4                  @ r5 = blend_a + blend_b\r
  cmp r5, #16                     @ if((blend_a + blend_b) > 16)\r
  bgt 3f                          @   goto loop w/saturation\r
\r
\r
  @ loop w/o saturation\r
1:\r
  ldr r5, [r0], #4                @ r5 = pixel_pair, screen_src_ptr++\r
  and r6, r5, r12                 @ r6 = r5 & 0x04000200\r
  cmp r6, r12                     @ if(r6 != 0x4000200)\r
  bne 2f                          @   goto no_blend\r
\r
  and r6, r10, r5, lsl #1         @ r6 = (pixel_pair & 0x1FF) << 1\r
  ldrh r6, [r14, r6]              @ r6 = pixel_top\r
  orr r6, r6, r6, lsl #16         @ r6 = pixel_top | (pixel_top << 16)\r
  and r6, r6, r11                 @ r6 = pixel_top_dilated\r
\r
  and r5, r10, r5, lsr #15        @ r5 = ((pixel_pair >> 16) & 0x1FF) << 1\r
  ldrh r5, [r14, r5]              @ r5 = pixel_bottom\r
  orr r5, r5, r5, lsl #16         @ r5 = pixel_bottom | (pixel_bottom << 16)\r
  and r5, r5, r11                 @ r5 = pixel_bottom_dilated\r
\r
  mul r5, r4, r5                  @ r5 = pixel_bottom * blend_b = bottom_mul\r
  mla r5, r3, r6, r5              @ r5 = (pixel_top * blend_a) + bottom_mul\r
\r
  and r5, r11, r5, lsr #4         @ r5 = (color_dilated >> 4) & 0x07E0F81F\r
  orr r5, r5, r5, lsr #16         @ r5 = color_dilated | (color_dilated >> 16)\r
\r
  strh r5, [r1], #2               @ *screen_dest_ptr = r5, screen_dest_ptr++\r
  subs r2, r2, #1                 @ counter--\r
  bne 1b                          @ go again\r
\r
  ldmia sp!, { r4, r5, r6, r9, r10, r11, pc }\r
\r
2:\r
  and r5, r10, r5, lsl #1         @ r5 = (pixel_pair & 0x1FF) << 1\r
  ldrh r5, [r14, r5]              @ r5 = pixel_top\r
  strh r5, [r1], #2               @ *screen_dest_ptr = r5, screen_dest_ptr++\r
\r
  subs r2, r2, #1                 @ counter--\r
  bne 1b                          @ go again\r
\r
  ldmia sp!, { r4, r5, r6, r9, r10, r11, pc }\r
\r
@ loop w/saturation\r
\r
3:\r
  ldr r9, 6b + 20                 @ r9 = 0x08010020\r
\r
4:\r
  ldr r5, [r0], #4                @ r5 = pixel_pair, screen_src_ptr++\r
  and r6, r5, r12                 @ r6 = r5 & 0x04000200\r
  cmp r6, r12                     @ if(r6 != 0x4000200)\r
  bne 5f                          @   goto no_blend\r
\r
  and r6, r10, r5, lsl #1         @ r6 = (pixel_pair & 0x1FF) << 1\r
  ldrh r6, [r14, r6]              @ r6 = pixel_top\r
  orr r6, r6, r6, lsl #16         @ r6 = pixel_top | (pixel_top << 16)\r
  and r6, r6, r11                 @ r6 = pixel_top_dilated\r
\r
  and r5, r10, r5, lsr #15        @ r5 = ((pixel_pair >> 16) & 0x1FF) << 1\r
  ldrh r5, [r14, r5]              @ r5 = pixel_bottom\r
  orr r5, r5, r5, lsl #16         @ r5 = pixel_bottom | (pixel_bottom << 16)\r
  and r5, r5, r11                 @ r5 = pixel_bottom_dilated\r
\r
  mul r5, r4, r5                  @ r5 = pixel_bottom * blend_b = bottom_mul\r
  mla r5, r3, r6, r5              @ r5 = (pixel_top * blend_a) + bottom_mul\r
\r
  and r6, r9, r5, lsr #4          @ r6 = saturation bits\r
  orr r6, r6, r6, lsr #1          @ propogate saturation down msb\r
  orr r6, r6, r6, lsr #2          @ propogate down next two bits\r
  orr r6, r6, r6, lsr #3          @ propogate down next three bits\r
  orr r5, r6, r5, lsr #4          @ mask over result w/saturation\r
\r
  and r5, r11, r5                 @ r5 = (color_dilated >> 4) & 0x07E0F81F\r
  orr r5, r5, r5, lsr #16         @ r5 = color_dilated | (color_dilated >> 16)\r
  strh r5, [r1], #2               @ *screen_dest_ptr = r5, screen_dest_ptr++\r
\r
  subs r2, r2, #1                 @ counter--\r
  bne 4b                          @ go again\r
\r
  ldmia sp!, { r4, r5, r6, r9, r10, r11, pc }\r
\r
5:\r
  and r5, r10, r5, lsl #1         @ r5 = (pixel_pair & 0x1FF) << 1\r
  ldrh r5, [r14, r5]              @ r5 = pixel_top\r
  strh r5, [r1], #2               @ *screen_dest_ptr = r5, screen_dest_ptr++\r
\r
  subs r2, r2, #1                 @ counter--\r
  bne 4b                          @ go again\r
\r
  ldmia sp!, { r4, r5, r6, r9, r10, r11, pc }\r
\r
\r
\r
@ The following function isn't complete (only works on run multiples of 8),\r
@ but unfortunately I don't see much potential for actually being able to\r
@ use it..\r
\r
#define expand_pixel_pair(reg, temp)                                         ;\\r
  and temp, r3, reg, lsr #15                                                 ;\\r
  ldrh temp, [r2, temp]                                                      ;\\r
                                                                             ;\\r
  and reg, r3, reg, lsl #1                                                   ;\\r
  ldrh reg, [r2, reg]                                                        ;\\r
                                                                             ;\\r
  orr reg, reg, temp, lsl #16                                                ;\\r
\r
\r
@ Input:\r
@   r0 = screen_ptr\r
@   r1 = start\r
@   r2 = end\r
\r
1:\r
  .word palette_ram_converted\r
  .word 0x3FE\r
\r
expand_normal:\r
  stmdb sp!, { r4, r5, r6, r7, r14 }\r
\r
  add r0, r0, r1, lsl #1          @ screen_ptr += start\r
  sub r1, r2, r1                  @ r1 = end - start\r
  ldr r2, 1b                      @ r2 = palette_ram_converted\r
  ldr r3, 1b + 4                  @ r3 = 0x3FE\r
\r
2:\r
  ldmia r0, { r4, r5, r6, r7 }\r
\r
  expand_pixel_pair(r4, r14)\r
  expand_pixel_pair(r5, r14)\r
  expand_pixel_pair(r6, r14)\r
  expand_pixel_pair(r7, r14)\r
\r
  stmia r0!, { r4, r5, r6, r7 }\r
\r
  subs r1, r1, #8\r
  bne 2b\r
\r
  ldmia sp!, { r4, r5, r6, r7, pc }\r
\r
Commit	Line	Data
2823a4c8	1	.align 2\r
	2	\r
	3	.global expand_blend\r
	4	.global expand_normal\r
	5	\r
	6	@ Input:\r
	7	@ r0 = screen_src_ptr\r
	8	@ r1 = screen_dest_ptr\r
	9	@ r2 = start\r
	10	@ r3 = end\r
	11	\r
	12	6:\r
	13	.word io_registers\r
	14	.word palette_ram_converted\r
	15	.word 0x04000200 @ combine test mask\r
	16	.word 0x07E0F81F @ clamp mask\r
	17	.word 0x000003FE @ palette index mask\r
	18	.word 0x08010020 @ saturation mask\r
	19	\r
	20	expand_blend:\r
	21	stmdb sp!, { r4, r5, r6, r9, r10, r11, r14 }\r
	22	\r
	23	add r0, r0, r2, lsl #2 @ screen_src_ptr += start\r
	24	add r1, r1, r2, lsl #1 @ screen_dest_ptr += start\r
	25	sub r2, r3, r2 @ r2 = end - start\r
	26	ldr r3, 6b @ r3 = io_registers\r
fe113069	27	ldrh r3, [r3, #0x52] @ r3 = bldalpha\r
2823a4c8	28	mov r4, r3, lsr #8 @ r4 = bldalpha >> 8\r
	29	and r3, r3, #0x1F @ r3 = blend_a\r
	30	and r4, r4, #0x1F @ r4 = blend_b\r
	31	cmp r3, #16 @ if(blend_a > 16)\r
	32	movgt r3, #16 @ blend_a = 16\r
	33	cmp r4, #16 @ if(blend_b > 16)\r
c95affa7	34	movgt r4, #16 @ blend_b = 16\r
2823a4c8	35	\r
	36	ldr r14, 6b + 4 @ r14 = palette_ram_converted\r
	37	ldr r12, 6b + 8 @ r12 = 0x04000200\r
	38	ldr r11, 6b + 12 @ r11 = 0x07E0F81F\r
	39	ldr r10, 6b + 16 @ r10 = 0x000003FE\r
	40	\r
	41	add r5, r3, r4 @ r5 = blend_a + blend_b\r
	42	cmp r5, #16 @ if((blend_a + blend_b) > 16)\r
	43	bgt 3f @ goto loop w/saturation\r
	44	\r
	45	\r
	46	@ loop w/o saturation\r
	47	1:\r
	48	ldr r5, [r0], #4 @ r5 = pixel_pair, screen_src_ptr++\r
	49	and r6, r5, r12 @ r6 = r5 & 0x04000200\r
	50	cmp r6, r12 @ if(r6 != 0x4000200)\r
	51	bne 2f @ goto no_blend\r
	52	\r
	53	and r6, r10, r5, lsl #1 @ r6 = (pixel_pair & 0x1FF) << 1\r
	54	ldrh r6, [r14, r6] @ r6 = pixel_top\r
	55	orr r6, r6, r6, lsl #16 @ r6 = pixel_top \| (pixel_top << 16)\r
	56	and r6, r6, r11 @ r6 = pixel_top_dilated\r
	57	\r
	58	and r5, r10, r5, lsr #15 @ r5 = ((pixel_pair >> 16) & 0x1FF) << 1\r
	59	ldrh r5, [r14, r5] @ r5 = pixel_bottom\r
	60	orr r5, r5, r5, lsl #16 @ r5 = pixel_bottom \| (pixel_bottom << 16)\r
	61	and r5, r5, r11 @ r5 = pixel_bottom_dilated\r
	62	\r
	63	mul r5, r4, r5 @ r5 = pixel_bottom * blend_b = bottom_mul\r
	64	mla r5, r3, r6, r5 @ r5 = (pixel_top * blend_a) + bottom_mul\r
	65	\r
	66	and r5, r11, r5, lsr #4 @ r5 = (color_dilated >> 4) & 0x07E0F81F\r
	67	orr r5, r5, r5, lsr #16 @ r5 = color_dilated \| (color_dilated >> 16)\r
	68	\r
	69	strh r5, [r1], #2 @ *screen_dest_ptr = r5, screen_dest_ptr++\r
	70	subs r2, r2, #1 @ counter--\r
	71	bne 1b @ go again\r
	72	\r
	73	ldmia sp!, { r4, r5, r6, r9, r10, r11, pc }\r
	74	\r
	75	2:\r
	76	and r5, r10, r5, lsl #1 @ r5 = (pixel_pair & 0x1FF) << 1\r
	77	ldrh r5, [r14, r5] @ r5 = pixel_top\r
	78	strh r5, [r1], #2 @ *screen_dest_ptr = r5, screen_dest_ptr++\r
	79	\r
	80	subs r2, r2, #1 @ counter--\r
	81	bne 1b @ go again\r
	82	\r
	83	ldmia sp!, { r4, r5, r6, r9, r10, r11, pc }\r
	84	\r
	85	@ loop w/saturation\r
	86	\r
	87	3:\r
	88	ldr r9, 6b + 20 @ r9 = 0x08010020\r
	89	\r
	90	4:\r
	91	ldr r5, [r0], #4 @ r5 = pixel_pair, screen_src_ptr++\r
	92	and r6, r5, r12 @ r6 = r5 & 0x04000200\r
	93	cmp r6, r12 @ if(r6 != 0x4000200)\r
	94	bne 5f @ goto no_blend\r
	95	\r
	96	and r6, r10, r5, lsl #1 @ r6 = (pixel_pair & 0x1FF) << 1\r
	97	ldrh r6, [r14, r6] @ r6 = pixel_top\r
	98	orr r6, r6, r6, lsl #16 @ r6 = pixel_top \| (pixel_top << 16)\r
99	and r6, r6, r11 @ r6 = pixel_top_dilated\r
100	\r
101	and r5, r10, r5, lsr #15 @ r5 = ((pixel_pair >> 16) & 0x1FF) << 1\r
102	ldrh r5, [r14, r5] @ r5 = pixel_bottom\r
103	orr r5, r5, r5, lsl #16 @ r5 = pixel_bottom \| (pixel_bottom << 16)\r
104	and r5, r5, r11 @ r5 = pixel_bottom_dilated\r
105	\r
106	mul r5, r4, r5 @ r5 = pixel_bottom * blend_b = bottom_mul\r
107	mla r5, r3, r6, r5 @ r5 = (pixel_top * blend_a) + bottom_mul\r
108	\r
109	and r6, r9, r5, lsr #4 @ r6 = saturation bits\r
110	orr r6, r6, r6, lsr #1 @ propogate saturation down msb\r
111	orr r6, r6, r6, lsr #2 @ propogate down next two bits\r
112	orr r6, r6, r6, lsr #3 @ propogate down next three bits\r
113	orr r5, r6, r5, lsr #4 @ mask over result w/saturation\r
114	\r
115	and r5, r11, r5 @ r5 = (color_dilated >> 4) & 0x07E0F81F\r
116	orr r5, r5, r5, lsr #16 @ r5 = color_dilated \| (color_dilated >> 16)\r
117	strh r5, [r1], #2 @ *screen_dest_ptr = r5, screen_dest_ptr++\r
118	\r
119	subs r2, r2, #1 @ counter--\r
120	bne 4b @ go again\r
121	\r
122	ldmia sp!, { r4, r5, r6, r9, r10, r11, pc }\r
123	\r
124	5:\r
125	and r5, r10, r5, lsl #1 @ r5 = (pixel_pair & 0x1FF) << 1\r
126	ldrh r5, [r14, r5] @ r5 = pixel_top\r
127	strh r5, [r1], #2 @ *screen_dest_ptr = r5, screen_dest_ptr++\r
128	\r
129	subs r2, r2, #1 @ counter--\r
130	bne 4b @ go again\r
131	\r
132	ldmia sp!, { r4, r5, r6, r9, r10, r11, pc }\r
133	\r
134	\r
135	\r
136	@ The following function isn't complete (only works on run multiples of 8),\r
137	@ but unfortunately I don't see much potential for actually being able to\r
138	@ use it..\r
139	\r
140	#define expand_pixel_pair(reg, temp) ;\\r
141	and temp, r3, reg, lsr #15 ;\\r
142	ldrh temp, [r2, temp] ;\\r
143	;\\r
144	and reg, r3, reg, lsl #1 ;\\r
145	ldrh reg, [r2, reg] ;\\r
146	;\\r
147	orr reg, reg, temp, lsl #16 ;\\r
148	\r
149	\r
150	@ Input:\r
151	@ r0 = screen_ptr\r
152	@ r1 = start\r
153	@ r2 = end\r
154	\r
155	1:\r
156	.word palette_ram_converted\r
157	.word 0x3FE\r
158	\r
159	expand_normal:\r
160	stmdb sp!, { r4, r5, r6, r7, r14 }\r
161	\r
162	add r0, r0, r1, lsl #1 @ screen_ptr += start\r
163	sub r1, r2, r1 @ r1 = end - start\r
164	ldr r2, 1b @ r2 = palette_ram_converted\r
165	ldr r3, 1b + 4 @ r3 = 0x3FE\r
166	\r
167	2:\r
168	ldmia r0, { r4, r5, r6, r7 }\r
169	\r
170	expand_pixel_pair(r4, r14)\r
171	expand_pixel_pair(r5, r14)\r
172	expand_pixel_pair(r6, r14)\r
173	expand_pixel_pair(r7, r14)\r
174	\r
175	stmia r0!, { r4, r5, r6, r7 }\r
176	\r
177	subs r1, r1, #8\r
178	bne 2b\r
179	\r
180	ldmia sp!, { r4, r5, r6, r7, pc }\r
181	\r