2 files changed, 349 insertions, 11 deletions
diff --git a/firmware/target/arm/sandisk/sansa-c200/lcd-as-c200.S b/firmware/target/arm/sandisk/sansa-c200/lcd-as-c200.S
index 9b0cd75bae..aaeb8439b6 100644
--- a/firmware/target/arm/sandisk/sansa-c200/lcd-as-c200.S
+++ b/firmware/target/arm/sandisk/sansa-c200/lcd-as-c200.S
@@ -241,3 +241,309 @@ lcd_write_yuv420_lines:
    .ltorg                              @ dump constant pool
    .size   lcd_write_yuv420_lines, .-lcd_write_yuv420_lines
+/****************************************************************************
+ * void lcd_write_yuv_420_lines_odither(unsigned char const * const src[3],
+ *                                      int width,
+ *                                      int stride,
+ *                                      int x_screen,
+ *                                      int y_screen);
+ *
+ *   |R|   |1.000000 -0.000001  1.402000| |Y'|
+ *   |G| = |1.000000 -0.334136 -0.714136| |Pb|
+ *   |B|   |1.000000  1.772000  0.000000| |Pr|
+ *   Red scaled at twice g & b but at same precision to place it in correct
+ *   bit position after multiply and leave instruction count lower.
+ *   |R|   |258   0  408| |Y' -  16|
+ *   |G| = |149 -49 -104| |Cb - 128|
+ *   |B|   |149 258    0| |Cr - 128|
+ *
+ * Write four RGB565 pixels in the following order on each loop:
+ * 1 3 + > down
+ * 2 4 \/ left
+ *
+ * Kernel pattern (raw|rotated|use order):
+ * 5 3 4 2   2 6 3 7     row0   row2          > down
+ * 1 7 0 6 | 4 0 5 1 | 2 4 6 0 3 5 7 1 col0     left
+ * 4 2 5 3 | 3 7 2 6 | 3 5 7 1 2 4 6 0 col2  \/
+ * 0 6 1 7   5 1 4 0
+ */
+    .section    .icode, "ax", %progbits
+    .align      2
+    .global     lcd_write_yuv420_lines_odither
+    .type       lcd_write_yuv420_lines_odither, %function
+lcd_write_yuv420_lines_odither:
+                                        @ r0 = yuv_src
+                                        @ r1 = width
+                                        @ r2 = stride
+                                        @ r3 = x_screen
+                                        @ [sp] = y_screen
+    stmfd       sp!, { r4-r12, lr }     @ save non-scratch
+    ldmia       r0, { r4, r5, r6 }      @ r4 = yuv_src[0] = Y'_p
+                                        @ r5 = yuv_src[1] = Cb_p
+                                        @ r6 = yuv_src[2] = Cr_p
+                                        @
+    sub         r2, r2, #1              @
+    ldr         r14, [sp, #40]          @ Line up pattern and kernel quadrant
+    eor         r14, r14, r3            @
+    and         r14, r14, #0x2          @
+    mov         r14, r14, lsl #6        @ 0x00 or 0x80
+    mov         r3, #0x70000000         @
+    orr         r3, r3, #0x3000         @ r3 = LCD1_BASE
+10: @ loop line                         @
+                                        @
+    ldrb        r7, [r4], #1            @ r7 = *Y'_p++;
+    ldrb        r8, [r5], #1            @ r8 = *Cb_p++;
+    ldrb        r9, [r6], #1            @ r9 = *Cr_p++;
+                                        @
+    eor         r14, r14, #0x80         @ flip pattern quadrant
+                                        @
+    sub         r7, r7, #16             @ r7 = Y = (Y' - 16)*149
+    add         r12, r7, r7, asl #2     @
+    add         r12, r12, r12, asl #4   @
+    add         r7, r12, r7, asl #6     @
+                                        @    
+    sub         r8, r8, #128            @ Cb -= 128
+    sub         r9, r9, #128            @ Cr -= 128
+                                        @
+    add         r10, r8, r8, asl #4     @ r10 = guv = Cr*104 + Cb*49
+    add         r10, r10, r8, asl #5    @
+    add         r10, r10, r9, asl #3    @
+    add         r10, r10, r9, asl #5    @
+    add         r10, r10, r9, asl #6    @
+                                        @
+    mov         r8, r8, asl #1          @ r8 = bu = Cb*258
+    add         r8, r8, r8, asl #7      @
+                                        @
+    add         r9, r9, r9, asl #1      @ r9 = rv = Cr*408
+    add         r9, r9, r9, asl #4      @
+    mov         r9, r9, asl #3          @
+                                        @
+                                        @ compute R, G, and B
+    add         r0, r8, r7              @ r0  = b' = Y + bu
+    add         r11, r9, r7, asl #1     @ r11 = r' = Y*2 + rv
+    rsb         r7, r10, r7             @ r7  = g' = Y + guv
+                                        @
+                                        @ r8 = bu, r9 = rv, r10 = guv
+                                        @
+    sub         r12, r0, r0, lsr #5     @ r0 = 31/32*b + b/256
+    add         r0, r12, r0, lsr #8     @
+                                        @
+    sub         r12, r11, r11, lsr #5   @ r11 = 31/32*r + r/256
+    add         r11, r12, r11, lsr #8   @
+                                        @
+    sub         r12, r7, r7, lsr #6     @ r7 = 63/64*g + g/256
+    add         r7, r12, r7, lsr #8     @
+                                        @
+    add         r12, r14, #0x100        @
+                                        @
+    add         r0, r0, r12             @ b = r0 + delta
+    add         r11, r11, r12, lsl #1   @ r = r11 + delta*2
+    add         r7, r7, r12, lsr #1     @ g = r7 + delta/2
+                                        @
+    orr         r12, r0, r11, asr #1    @ check if clamping is needed...
+    orr         r12, r12, r7            @ ...at all
+    movs        r12, r12, asr #15       @
+    beq         15f @ no clamp          @
+    movs        r12, r0, asr #15        @ clamp b
+    mvnne       r0, r12, lsr #15        @
+    andne       r0, r0, #0x7c00         @ mask b only if clamped
+    movs        r12, r11, asr #16       @ clamp r
+    mvnne       r11, r12, lsr #16       @
+    movs        r12, r7, asr #15        @ clamp g
+    mvnne       r7, r12, lsr #15        @
+15: @ no clamp                          @
+                                        @
+    ldrb        r12, [r4, r2]           @ r12 = Y' = *(Y'_p + stride)
+                                        @
+    and         r11, r11, #0xf800       @ pack pixel
+    mov         r11, r11, lsr #8
+    and         r7, r7, #0x7e00
+    orr         r11, r11, r7, lsr #12
+    mov         r7, r7, lsr#4
+    orr         r0, r7, r0, lsr #10
+1: @ busy                               @
+    ldr         r7, [r3]                @ r7 = LCD1_BASE
+    tst         r7, #LCD1_BUSY_MASK     @ bridge busy?
+    bne         1b                      @
+    str         r11, [r3, #0x10]        @ send MSB
+1: @ busy                               @
+    ldr         r7, [r3]                @ r7 = LCD1_BASE
+    tst         r7, #LCD1_BUSY_MASK     @ bridge busy?
+    bne         1b                      @
+    str         r0, [r3, #0x10]         @ send LSB
+                                                                               @
+    sub         r7, r12, #16            @ r7 = Y = (Y' - 16)*149
+    add         r12, r7, r7, asl #2     @
+    add         r12, r12, r12, asl #4   @
+    add         r7, r12, r7, asl #6     @
+                                        @ compute R, G, and B
+    add         r0, r8, r7              @ r0  = b' = Y + bu
+    add         r11, r9, r7, asl #1     @ r11 = r' = Y*2 + rv
+    rsb         r7, r10, r7             @ r7  = g' = Y + guv
+                                        @
+    sub         r12, r0, r0, lsr #5     @ r0  = 31/32*b' + b'/256
+    add         r0, r12, r0, lsr #8     @
+                                        @
+    sub         r12, r11, r11, lsr #5   @ r11 = 31/32*r' + r'/256
+    add         r11, r12, r11, lsr #8   @
+                                        @
+    sub         r12, r7, r7, lsr #6     @ r7  = 63/64*g' + g'/256
+    add         r7, r12, r7, lsr #8     @
+                                        @
+    add         r12, r14, #0x200        @
+                                        @
+    add         r0, r0, r12             @ b = r0 + delta
+    add         r11, r11, r12, lsl #1   @ r = r11 + delta*2
+    add         r7, r7, r12, lsr #1     @ g = r7 + delta/2
+                                        @
+    orr         r12, r0, r11, asr #1    @ check if clamping is needed...
+    orr         r12, r12, r7            @ ...at all
+    movs        r12, r12, asr #15       @
+    beq         15f @ no clamp          @
+    movs        r12, r0, asr #15        @ clamp b
+    mvnne       r0, r12, lsr #15        @
+    andne       r0, r0, #0x7c00         @ mask b only if clamped
+    movs        r12, r11, asr #16       @ clamp r
+    mvnne       r11, r12, lsr #16       @
+    movs        r12, r7, asr #15        @ clamp g
+    mvnne       r7, r12, lsr #15        @
+15: @ no clamp                          @
+                                        @
+    ldrb        r12, [r4], #1           @ r12 = Y' = *(Y'_p++)
+    and         r11, r11, #0xf800       @ pack pixel
+    mov         r11, r11, lsr #8
+    and         r7, r7, #0x7e00
+    orr         r11, r11, r7, lsr #12
+    mov         r7, r7, lsr#4
+    orr         r0, r7, r0, lsr #10
+1: @ busy                               @
+    ldr         r7, [r3]                @ r7 = LCD1_BASE
+    tst         r7, #LCD1_BUSY_MASK     @ bridge busy?
+    bne         1b                      @
+    str         r11, [r3, #0x10]        @ send MSB
+1: @ busy                               @
+    ldr         r7, [r3]                @ r7 = LCD1_BASE
+    tst         r7, #LCD1_BUSY_MASK     @ bridge busy?
+    bne         1b                      @
+    str         r0, [r3, #0x10]         @ send LSB
+    sub         r7, r12, #16            @ r7 = Y = (Y' - 16)*149
+    add         r12, r7, r7, asl #2     @
+    add         r12, r12, r12, asl #4   @
+    add         r7, r12, r7, asl #6     @
+                                        @ compute R, G, and B
+    add         r0, r8, r7              @ r0  = b' = Y + bu
+    add         r11, r9, r7, asl #1     @ r11 = r' = Y*2 + rv
+    rsb         r7, r10, r7             @ r7  = g' = Y + guv
+                                        @
+                                        @ r8 = bu, r9 = rv, r10 = guv
+                                        @
+    sub         r12, r0, r0, lsr #5     @ r0  = 31/32*b' + b'/256
+    add         r0, r12, r0, lsr #8     @
+                                        @
+    sub         r12, r11, r11, lsr #5   @ r11 = 31/32*r' + r'/256
+    add         r11, r12, r11, lsr #8   @
+                                        @
+    sub         r12, r7, r7, lsr #6     @ r7  = 63/64*g' + g'/256
+    add         r7, r12, r7, lsr #8     @
+                                        @
+    add         r12, r14, #0x300        @
+                                        @
+    add         r0, r0, r12             @ b = r0 + delta
+    add         r11, r11, r12, lsl #1   @ r = r11 + delta*2
+    add         r7, r7, r12, lsr #1     @ g = r7 + delta/2
+                                        @
+    orr         r12, r0, r11, asr #1    @ check if clamping is needed...
+    orr         r12, r12, r7            @ ...at all
+    movs        r12, r12, asr #15       @
+    beq         15f @ no clamp          @
+    movs        r12, r0, asr #15        @ clamp b
+    mvnne       r0, r12, lsr #15        @
+    andne       r0, r0, #0x7c00         @ mask b only if clamped
+    movs        r12, r11, asr #16       @ clamp r
+    mvnne       r11, r12, lsr #16       @
+    movs        r12, r7, asr #15        @ clamp g
+    mvnne       r7, r12, lsr #15        @
+15: @ no clamp                          @
+                                        @
+    ldrb        r12, [r4, r2]           @ r12 = Y' = *(Y'_p + stride)
+    and         r11, r11, #0xf800       @ pack pixel
+    mov         r11, r11, lsr #8
+    and         r7, r7, #0x7e00
+    orr         r11, r11, r7, lsr #12
+    mov         r7, r7, lsr#4
+    orr         r0, r7, r0, lsr #10
+1: @ busy                               @
+    ldr         r7, [r3]                @ r7 = LCD1_BASE
+    tst         r7, #LCD1_BUSY_MASK     @ bridge busy?
+    bne         1b                      @
+    str         r11, [r3, #0x10]        @ send MSB
+1: @ busy                               @
+    ldr         r7, [r3]                @ r7 = LCD1_BASE
+    tst         r7, #LCD1_BUSY_MASK     @ bridge busy?
+    bne         1b                      @
+    str         r0, [r3, #0x10]         @ send LSB
+    sub         r7, r12, #16            @ r7 = Y = (Y' - 16)*149
+    add         r12, r7, r7, asl #2     @
+    add         r12, r12, r12, asl #4   @
+    add         r7, r12, r7, asl #6     @
+                                        @ compute R, G, and B
+    add         r0, r8, r7              @ r0  = b' = Y + bu
+    add         r11, r9, r7, asl #1     @ r11 = r' = Y*2 + rv
+    rsb         r7, r10, r7             @ r7  = g' = Y + guv
+                                        @
+    sub         r12, r0, r0, lsr #5     @ r0 = 31/32*b + b/256
+    add         r0, r12, r0, lsr #8     @
+                                        @
+    sub         r12, r11, r11, lsr #5   @ r11 = 31/32*r + r/256
+    add         r11, r12, r11, lsr #8   @
+                                        @
+    sub         r12, r7, r7, lsr #6     @ r7 = 63/64*g + g/256
+    add         r7, r12, r7, lsr #8     @
+                                        @
+    @ This element is zero - use r14    @
+                                        @
+    add         r0, r0, r14             @ b = r0 + delta
+    add         r11, r11, r14, lsl #1   @ r = r11 + delta*2
+    add         r7, r7, r14, lsr #1     @ g = r7 + delta/2
+                                        @
+    orr         r12, r0, r11, asr #1    @ check if clamping is needed...
+    orr         r12, r12, r7            @ ...at all
+    movs        r12, r12, asr #15       @
+    beq         15f @ no clamp          @
+    movs        r12, r0, asr #15        @ clamp b
+    mvnne       r0, r12, lsr #15        @
+    andne       r0, r0, #0x7c00         @ mask b only if clamped
+    movs        r12, r11, asr #16       @ clamp r
+    mvnne       r11, r12, lsr #16       @
+    movs        r12, r7, asr #15        @ clamp g
+    mvnne       r7, r12, lsr #15        @
+15: @ no clamp                          @
+    and         r11, r11, #0xf800       @ pack pixel
+    mov         r11, r11, lsr #8
+    and         r7, r7, #0x7e00
+    orr         r11, r11, r7, lsr #12
+    mov         r7, r7, lsr#4
+    orr         r0, r7, r0, lsr #10
+1: @ busy                               @
+    ldr         r7, [r3]                @ r7 = LCD1_BASE
+    tst         r7, #LCD1_BUSY_MASK     @ bridge busy?
+    bne         1b                      @
+    str         r11, [r3, #0x10]        @ send MSB
+1: @ busy                               @
+    ldr         r7, [r3]                @ r7 = LCD1_BASE
+    tst         r7, #LCD1_BUSY_MASK     @ bridge busy?
+    bne         1b                      @
+    str         r0, [r3, #0x10]         @ send LSB
+    subs        r1, r1, #2              @ subtract block from width
+    bgt         10b @ loop line         @
+                                        @
+    ldmfd       sp!, { r4-r12, pc }     @ restore registers and return
+    .ltorg                              @ dump constant pool
+    .size   lcd_write_yuv420_lines_odither, .-lcd_write_yuv420_lines_odither
diff --git a/firmware/target/arm/sandisk/sansa-c200/lcd-c200.c b/firmware/target/arm/sandisk/sansa-c200/lcd-c200.c
index 5975ab159d..fa8581f286 100644
--- a/firmware/target/arm/sandisk/sansa-c200/lcd-c200.c
+++ b/firmware/target/arm/sandisk/sansa-c200/lcd-c200.c
@@ -22,6 +22,9 @@
 #include "kernel.h"
 #include "system.h"
+/* Display status */
+static unsigned lcd_yuv_options NOCACHEBSS_ATTR = 0;
 /* LCD command set for Samsung S6B33B2 */
 #define R_OSCILLATION_MODE     0x02
@@ -206,10 +209,20 @@ void lcd_blit(const fb_data* data, int x, int by, int width,
    (void)stride;
 }
+void lcd_yuv_set_options(unsigned options)
+{
+    lcd_yuv_options = options;
+}
 /* Line write helper function for lcd_yuv_blit. Write two lines of yuv420. */
 extern void lcd_write_yuv420_lines(unsigned char const * const src[3],
                                   int width,
                                   int stride);
+extern void lcd_write_yuv420_lines_odither(unsigned char const * const src[3],
+                                           int width,
+                                           int stride,
+                                           int x_screen, /* To align dither pattern */
+                                           int y_screen);
 /* Performance function to blit a YUV bitmap directly to the LCD */
 void lcd_yuv_blit(unsigned char * const src[3],
                  int src_x, int src_y, int stride,
@@ -236,19 +249,38 @@ void lcd_yuv_blit(unsigned char * const src[3],
    lcd_send_command(x);
    lcd_send_command(x + width - 1);
-    do
+    if (lcd_yuv_options & LCD_YUV_DITHER)
    {
-        lcd_send_command(R_Y_ADDR_AREA);
+        do
-        lcd_send_command(y);
+        {
-        lcd_send_command(y + 1);
+            lcd_send_command(R_Y_ADDR_AREA);
+            lcd_send_command(y);
-        lcd_write_yuv420_lines(yuv_src, width, stride);
+            lcd_send_command(y + 1);
-        yuv_src[0] += stride << 1; /* Skip down two luma lines */
-        yuv_src[1] += stride >> 1; /* Skip down one chroma line */
+            lcd_write_yuv420_lines_odither(yuv_src, width, stride, x, y);
-        yuv_src[2] += stride >> 1;
+            yuv_src[0] += stride << 1; /* Skip down two luma lines */
-        y += 2;
+            yuv_src[1] += stride >> 1; /* Skip down one chroma line */
+            yuv_src[2] += stride >> 1;
+            y += 2;
+        }
+         while (--height > 0);
+    }
+    else
+    {
+        do
+        {
+            lcd_send_command(R_Y_ADDR_AREA);
+            lcd_send_command(y);
+            lcd_send_command(y + 1);
+            lcd_write_yuv420_lines(yuv_src, width, stride);
+            yuv_src[0] += stride << 1; /* Skip down two luma lines */
+            yuv_src[1] += stride >> 1; /* Skip down one chroma line */
+            yuv_src[2] += stride >> 1;
+            y += 2;
+        }
+         while (--height > 0);
    }
-    while (--height > 0);
 }
 /* Update the display.

diff --git a/firmware/target/arm/sandisk/sansa-c200/lcd-as-c200.S b/firmware/target/arm/sandisk/sansa-c200/lcd-as-c200.S index 9b0cd75bae..aaeb8439b6 100644 --- a/firmware/target/arm/sandisk/sansa-c200/lcd-as-c200.S +++ b/firmware/target/arm/sandisk/sansa-c200/lcd-as-c200.S
@@ -241,3 +241,309 @@ lcd_write_yuv420_lines:
241	.ltorg @ dump constant pool	241	.ltorg @ dump constant pool
242	.size lcd_write_yuv420_lines, .-lcd_write_yuv420_lines	242	.size lcd_write_yuv420_lines, .-lcd_write_yuv420_lines
243		243
		244	/****************************************************************************
		245	* void lcd_write_yuv_420_lines_odither(unsigned char const * const src[3],
		246	* int width,
		247	* int stride,
		248	* int x_screen,
		249	* int y_screen);
		250	*
		251	* \|R\| \|1.000000 -0.000001 1.402000\| \|Y'\|
		252	* \|G\| = \|1.000000 -0.334136 -0.714136\| \|Pb\|
		253	* \|B\| \|1.000000 1.772000 0.000000\| \|Pr\|
		254	* Red scaled at twice g & b but at same precision to place it in correct
		255	* bit position after multiply and leave instruction count lower.
		256	* \|R\| \|258 0 408\| \|Y' - 16\|
		257	* \|G\| = \|149 -49 -104\| \|Cb - 128\|
		258	* \|B\| \|149 258 0\| \|Cr - 128\|
		259	*
		260	* Write four RGB565 pixels in the following order on each loop:
		261	* 1 3 + > down
		262	* 2 4 \/ left
		263	*
		264	* Kernel pattern (raw\|rotated\|use order):
		265	* 5 3 4 2 2 6 3 7 row0 row2 > down
		266	* 1 7 0 6 \| 4 0 5 1 \| 2 4 6 0 3 5 7 1 col0 left
		267	* 4 2 5 3 \| 3 7 2 6 \| 3 5 7 1 2 4 6 0 col2 \/
		268	* 0 6 1 7 5 1 4 0
		269	*/
		270	.section .icode, "ax", %progbits
		271	.align 2
		272	.global lcd_write_yuv420_lines_odither
		273	.type lcd_write_yuv420_lines_odither, %function
		274	lcd_write_yuv420_lines_odither:
		275	@ r0 = yuv_src
		276	@ r1 = width
		277	@ r2 = stride
		278	@ r3 = x_screen
		279	@ [sp] = y_screen
		280	stmfd sp!, { r4-r12, lr } @ save non-scratch
		281	ldmia r0, { r4, r5, r6 } @ r4 = yuv_src[0] = Y'_p
		282	@ r5 = yuv_src[1] = Cb_p
		283	@ r6 = yuv_src[2] = Cr_p
		284	@
		285	sub r2, r2, #1 @
		286	ldr r14, [sp, #40] @ Line up pattern and kernel quadrant
		287	eor r14, r14, r3 @
		288	and r14, r14, #0x2 @
		289	mov r14, r14, lsl #6 @ 0x00 or 0x80
		290	mov r3, #0x70000000 @
		291	orr r3, r3, #0x3000 @ r3 = LCD1_BASE
		292	10: @ loop line @
		293	@
		294	ldrb r7, [r4], #1 @ r7 = *Y'_p++;
		295	ldrb r8, [r5], #1 @ r8 = *Cb_p++;
		296	ldrb r9, [r6], #1 @ r9 = *Cr_p++;
		297	@
		298	eor r14, r14, #0x80 @ flip pattern quadrant
		299	@
		300	sub r7, r7, #16 @ r7 = Y = (Y' - 16)*149
		301	add r12, r7, r7, asl #2 @
		302	add r12, r12, r12, asl #4 @
		303	add r7, r12, r7, asl #6 @
		304	@
		305	sub r8, r8, #128 @ Cb -= 128
		306	sub r9, r9, #128 @ Cr -= 128
		307	@
		308	add r10, r8, r8, asl #4 @ r10 = guv = Cr104 + Cb49
		309	add r10, r10, r8, asl #5 @
		310	add r10, r10, r9, asl #3 @
		311	add r10, r10, r9, asl #5 @
		312	add r10, r10, r9, asl #6 @
		313	@
		314	mov r8, r8, asl #1 @ r8 = bu = Cb*258
		315	add r8, r8, r8, asl #7 @
		316	@
		317	add r9, r9, r9, asl #1 @ r9 = rv = Cr*408
		318	add r9, r9, r9, asl #4 @
		319	mov r9, r9, asl #3 @
		320	@
		321	@ compute R, G, and B
		322	add r0, r8, r7 @ r0 = b' = Y + bu
		323	add r11, r9, r7, asl #1 @ r11 = r' = Y*2 + rv
		324	rsb r7, r10, r7 @ r7 = g' = Y + guv
		325	@
		326	@ r8 = bu, r9 = rv, r10 = guv
		327	@
		328	sub r12, r0, r0, lsr #5 @ r0 = 31/32*b + b/256
		329	add r0, r12, r0, lsr #8 @
		330	@
		331	sub r12, r11, r11, lsr #5 @ r11 = 31/32*r + r/256
		332	add r11, r12, r11, lsr #8 @
		333	@
		334	sub r12, r7, r7, lsr #6 @ r7 = 63/64*g + g/256
		335	add r7, r12, r7, lsr #8 @
		336	@
		337	add r12, r14, #0x100 @
		338	@
		339	add r0, r0, r12 @ b = r0 + delta
		340	add r11, r11, r12, lsl #1 @ r = r11 + delta*2
		341	add r7, r7, r12, lsr #1 @ g = r7 + delta/2
		342	@
		343	orr r12, r0, r11, asr #1 @ check if clamping is needed...
		344	orr r12, r12, r7 @ ...at all
		345	movs r12, r12, asr #15 @
		346	beq 15f @ no clamp @
		347	movs r12, r0, asr #15 @ clamp b
		348	mvnne r0, r12, lsr #15 @
		349	andne r0, r0, #0x7c00 @ mask b only if clamped
		350	movs r12, r11, asr #16 @ clamp r
		351	mvnne r11, r12, lsr #16 @
		352	movs r12, r7, asr #15 @ clamp g
		353	mvnne r7, r12, lsr #15 @
		354	15: @ no clamp @
		355	@
		356	ldrb r12, [r4, r2] @ r12 = Y' = *(Y'_p + stride)
		357	@
		358
		359	and r11, r11, #0xf800 @ pack pixel
		360	mov r11, r11, lsr #8
		361	and r7, r7, #0x7e00
		362	orr r11, r11, r7, lsr #12
		363	mov r7, r7, lsr#4
		364	orr r0, r7, r0, lsr #10
		365	1: @ busy @
		366	ldr r7, [r3] @ r7 = LCD1_BASE
		367	tst r7, #LCD1_BUSY_MASK @ bridge busy?
		368	bne 1b @
		369	str r11, [r3, #0x10] @ send MSB
		370	1: @ busy @
		371	ldr r7, [r3] @ r7 = LCD1_BASE
		372	tst r7, #LCD1_BUSY_MASK @ bridge busy?
		373	bne 1b @
		374	str r0, [r3, #0x10] @ send LSB
		375	@
		376	sub r7, r12, #16 @ r7 = Y = (Y' - 16)*149
		377	add r12, r7, r7, asl #2 @
		378	add r12, r12, r12, asl #4 @
		379	add r7, r12, r7, asl #6 @
		380	@ compute R, G, and B
		381	add r0, r8, r7 @ r0 = b' = Y + bu
		382	add r11, r9, r7, asl #1 @ r11 = r' = Y*2 + rv
		383	rsb r7, r10, r7 @ r7 = g' = Y + guv
		384	@
		385	sub r12, r0, r0, lsr #5 @ r0 = 31/32*b' + b'/256
		386	add r0, r12, r0, lsr #8 @
		387	@
		388	sub r12, r11, r11, lsr #5 @ r11 = 31/32*r' + r'/256
		389	add r11, r12, r11, lsr #8 @
		390	@
		391	sub r12, r7, r7, lsr #6 @ r7 = 63/64*g' + g'/256
		392	add r7, r12, r7, lsr #8 @
		393	@
		394	add r12, r14, #0x200 @
		395	@
		396	add r0, r0, r12 @ b = r0 + delta
		397	add r11, r11, r12, lsl #1 @ r = r11 + delta*2
		398	add r7, r7, r12, lsr #1 @ g = r7 + delta/2
		399	@
		400	orr r12, r0, r11, asr #1 @ check if clamping is needed...
		401	orr r12, r12, r7 @ ...at all
		402	movs r12, r12, asr #15 @
		403	beq 15f @ no clamp @
		404	movs r12, r0, asr #15 @ clamp b
		405	mvnne r0, r12, lsr #15 @
		406	andne r0, r0, #0x7c00 @ mask b only if clamped
		407	movs r12, r11, asr #16 @ clamp r
		408	mvnne r11, r12, lsr #16 @
		409	movs r12, r7, asr #15 @ clamp g
		410	mvnne r7, r12, lsr #15 @
		411	15: @ no clamp @
		412	@
		413	ldrb r12, [r4], #1 @ r12 = Y' = *(Y'_p++)
		414
		415	and r11, r11, #0xf800 @ pack pixel
		416	mov r11, r11, lsr #8
		417	and r7, r7, #0x7e00
		418	orr r11, r11, r7, lsr #12
		419	mov r7, r7, lsr#4
		420	orr r0, r7, r0, lsr #10
		421	1: @ busy @
		422	ldr r7, [r3] @ r7 = LCD1_BASE
		423	tst r7, #LCD1_BUSY_MASK @ bridge busy?
		424	bne 1b @
		425	str r11, [r3, #0x10] @ send MSB
		426	1: @ busy @
		427	ldr r7, [r3] @ r7 = LCD1_BASE
		428	tst r7, #LCD1_BUSY_MASK @ bridge busy?
		429	bne 1b @
		430	str r0, [r3, #0x10] @ send LSB
		431
		432	sub r7, r12, #16 @ r7 = Y = (Y' - 16)*149
		433	add r12, r7, r7, asl #2 @
		434	add r12, r12, r12, asl #4 @
		435	add r7, r12, r7, asl #6 @
		436	@ compute R, G, and B
		437	add r0, r8, r7 @ r0 = b' = Y + bu
		438	add r11, r9, r7, asl #1 @ r11 = r' = Y*2 + rv
		439	rsb r7, r10, r7 @ r7 = g' = Y + guv
		440	@
		441	@ r8 = bu, r9 = rv, r10 = guv
		442	@
		443	sub r12, r0, r0, lsr #5 @ r0 = 31/32*b' + b'/256
		444	add r0, r12, r0, lsr #8 @
		445	@
		446	sub r12, r11, r11, lsr #5 @ r11 = 31/32*r' + r'/256
		447	add r11, r12, r11, lsr #8 @
		448	@
		449	sub r12, r7, r7, lsr #6 @ r7 = 63/64*g' + g'/256
		450	add r7, r12, r7, lsr #8 @
		451	@
		452	add r12, r14, #0x300 @
		453	@
		454	add r0, r0, r12 @ b = r0 + delta
		455	add r11, r11, r12, lsl #1 @ r = r11 + delta*2
		456	add r7, r7, r12, lsr #1 @ g = r7 + delta/2
		457	@
		458	orr r12, r0, r11, asr #1 @ check if clamping is needed...
		459	orr r12, r12, r7 @ ...at all
		460	movs r12, r12, asr #15 @
		461	beq 15f @ no clamp @
		462	movs r12, r0, asr #15 @ clamp b
		463	mvnne r0, r12, lsr #15 @
		464	andne r0, r0, #0x7c00 @ mask b only if clamped
		465	movs r12, r11, asr #16 @ clamp r
		466	mvnne r11, r12, lsr #16 @
		467	movs r12, r7, asr #15 @ clamp g
		468	mvnne r7, r12, lsr #15 @
		469	15: @ no clamp @
		470	@
		471	ldrb r12, [r4, r2] @ r12 = Y' = *(Y'_p + stride)
		472
		473	and r11, r11, #0xf800 @ pack pixel
		474	mov r11, r11, lsr #8
		475	and r7, r7, #0x7e00
		476	orr r11, r11, r7, lsr #12
		477	mov r7, r7, lsr#4
		478	orr r0, r7, r0, lsr #10
		479	1: @ busy @
		480	ldr r7, [r3] @ r7 = LCD1_BASE
		481	tst r7, #LCD1_BUSY_MASK @ bridge busy?
		482	bne 1b @
		483	str r11, [r3, #0x10] @ send MSB
		484	1: @ busy @
		485	ldr r7, [r3] @ r7 = LCD1_BASE
		486	tst r7, #LCD1_BUSY_MASK @ bridge busy?
		487	bne 1b @
		488	str r0, [r3, #0x10] @ send LSB
		489
		490	sub r7, r12, #16 @ r7 = Y = (Y' - 16)*149
		491	add r12, r7, r7, asl #2 @
		492	add r12, r12, r12, asl #4 @
		493	add r7, r12, r7, asl #6 @
		494	@ compute R, G, and B
		495	add r0, r8, r7 @ r0 = b' = Y + bu
		496	add r11, r9, r7, asl #1 @ r11 = r' = Y*2 + rv
		497	rsb r7, r10, r7 @ r7 = g' = Y + guv
		498	@
		499	sub r12, r0, r0, lsr #5 @ r0 = 31/32*b + b/256
		500	add r0, r12, r0, lsr #8 @
		501	@
		502	sub r12, r11, r11, lsr #5 @ r11 = 31/32*r + r/256
		503	add r11, r12, r11, lsr #8 @
		504	@
		505	sub r12, r7, r7, lsr #6 @ r7 = 63/64*g + g/256
		506	add r7, r12, r7, lsr #8 @
		507	@
		508	@ This element is zero - use r14 @
		509	@
		510	add r0, r0, r14 @ b = r0 + delta
		511	add r11, r11, r14, lsl #1 @ r = r11 + delta*2
		512	add r7, r7, r14, lsr #1 @ g = r7 + delta/2
		513	@
		514	orr r12, r0, r11, asr #1 @ check if clamping is needed...
		515	orr r12, r12, r7 @ ...at all
		516	movs r12, r12, asr #15 @
		517	beq 15f @ no clamp @
		518	movs r12, r0, asr #15 @ clamp b
		519	mvnne r0, r12, lsr #15 @
		520	andne r0, r0, #0x7c00 @ mask b only if clamped
		521	movs r12, r11, asr #16 @ clamp r
		522	mvnne r11, r12, lsr #16 @
		523	movs r12, r7, asr #15 @ clamp g
		524	mvnne r7, r12, lsr #15 @
		525	15: @ no clamp @
		526
		527	and r11, r11, #0xf800 @ pack pixel
		528	mov r11, r11, lsr #8
		529	and r7, r7, #0x7e00
		530	orr r11, r11, r7, lsr #12
		531	mov r7, r7, lsr#4
		532	orr r0, r7, r0, lsr #10
		533	1: @ busy @
		534	ldr r7, [r3] @ r7 = LCD1_BASE
		535	tst r7, #LCD1_BUSY_MASK @ bridge busy?
		536	bne 1b @
		537	str r11, [r3, #0x10] @ send MSB
		538	1: @ busy @
		539	ldr r7, [r3] @ r7 = LCD1_BASE
		540	tst r7, #LCD1_BUSY_MASK @ bridge busy?
		541	bne 1b @
		542	str r0, [r3, #0x10] @ send LSB
		543
		544	subs r1, r1, #2 @ subtract block from width
		545	bgt 10b @ loop line @
		546	@
		547	ldmfd sp!, { r4-r12, pc } @ restore registers and return
		548	.ltorg @ dump constant pool
		549	.size lcd_write_yuv420_lines_odither, .-lcd_write_yuv420_lines_odither


diff --git a/firmware/target/arm/sandisk/sansa-c200/lcd-c200.c b/firmware/target/arm/sandisk/sansa-c200/lcd-c200.c index 5975ab159d..fa8581f286 100644 --- a/firmware/target/arm/sandisk/sansa-c200/lcd-c200.c +++ b/firmware/target/arm/sandisk/sansa-c200/lcd-c200.c
@@ -22,6 +22,9 @@
22	#include "kernel.h"	22	#include "kernel.h"
23	#include "system.h"	23	#include "system.h"
24		24
		25	/* Display status */
		26	static unsigned lcd_yuv_options NOCACHEBSS_ATTR = 0;
		27
25	/* LCD command set for Samsung S6B33B2 */	28	/* LCD command set for Samsung S6B33B2 */
26		29
27	#define R_OSCILLATION_MODE 0x02	30	#define R_OSCILLATION_MODE 0x02
@@ -206,10 +209,20 @@ void lcd_blit(const fb_data* data, int x, int by, int width,
206	(void)stride;	209	(void)stride;
207	}	210	}
208		211
		212	void lcd_yuv_set_options(unsigned options)
		213	{
		214	lcd_yuv_options = options;
		215	}
		216
209	/* Line write helper function for lcd_yuv_blit. Write two lines of yuv420. */	217	/* Line write helper function for lcd_yuv_blit. Write two lines of yuv420. */
210	extern void lcd_write_yuv420_lines(unsigned char const * const src[3],	218	extern void lcd_write_yuv420_lines(unsigned char const * const src[3],
211	int width,	219	int width,
212	int stride);	220	int stride);
		221	extern void lcd_write_yuv420_lines_odither(unsigned char const * const src[3],
		222	int width,
		223	int stride,
		224	int x_screen, /* To align dither pattern */
		225	int y_screen);
213	/* Performance function to blit a YUV bitmap directly to the LCD */	226	/* Performance function to blit a YUV bitmap directly to the LCD */
214	void lcd_yuv_blit(unsigned char * const src[3],	227	void lcd_yuv_blit(unsigned char * const src[3],
215	int src_x, int src_y, int stride,	228	int src_x, int src_y, int stride,
@@ -236,19 +249,38 @@ void lcd_yuv_blit(unsigned char * const src[3],
236	lcd_send_command(x);	249	lcd_send_command(x);
237	lcd_send_command(x + width - 1);	250	lcd_send_command(x + width - 1);
238		251
239	do	252	if (lcd_yuv_options & LCD_YUV_DITHER)
240	{	253	{
241	lcd_send_command(R_Y_ADDR_AREA);	254	do
242	lcd_send_command(y);	255	{
243	lcd_send_command(y + 1);	256	lcd_send_command(R_Y_ADDR_AREA);
244		257	lcd_send_command(y);
245	lcd_write_yuv420_lines(yuv_src, width, stride);	258	lcd_send_command(y + 1);
246	yuv_src[0] += stride << 1; /* Skip down two luma lines */	259
247	yuv_src[1] += stride >> 1; /* Skip down one chroma line */	260	lcd_write_yuv420_lines_odither(yuv_src, width, stride, x, y);
248	yuv_src[2] += stride >> 1;	261	yuv_src[0] += stride << 1; /* Skip down two luma lines */
249	y += 2;	262	yuv_src[1] += stride >> 1; /* Skip down one chroma line */
		263	yuv_src[2] += stride >> 1;
		264	y += 2;
		265	}
		266	while (--height > 0);
		267	}
		268	else
		269	{
		270	do
		271	{
		272	lcd_send_command(R_Y_ADDR_AREA);
		273	lcd_send_command(y);
		274	lcd_send_command(y + 1);
		275
		276	lcd_write_yuv420_lines(yuv_src, width, stride);
		277	yuv_src[0] += stride << 1; /* Skip down two luma lines */
		278	yuv_src[1] += stride >> 1; /* Skip down one chroma line */
		279	yuv_src[2] += stride >> 1;
		280	y += 2;
		281	}
		282	while (--height > 0);
250	}	283	}
251	while (--height > 0);
252	}	284	}
253		285
254	/* Update the display.	286	/* Update the display.