Greyscale library: Changed the internal data format once more (separated pixel values and phases), allowing for further optimisation of drawing, scrolling etc. * Optimised grey phase blitting in the core reduces CPU load on all architectures, most significantly on coldfire. Previous version was too slow to keep up at 45MHz, leading to unwanted graininess (update frequency was halved). Also fixed screendump on 2bpp targets with vertical pixel packing.

git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16043 a1c6a512-1295-4272-9138-f99709370657
This commit is contained in:
Jens Arnold 2008-01-09 23:48:26 +00:00
parent 75380fd27d
commit 6a56c14e17
13 changed files with 462 additions and 313 deletions

View file

@ -119,12 +119,12 @@
#define PLUGIN_MAGIC 0x526F634B /* RocK */ #define PLUGIN_MAGIC 0x526F634B /* RocK */
/* increase this every time the api struct changes */ /* increase this every time the api struct changes */
#define PLUGIN_API_VERSION 95 #define PLUGIN_API_VERSION 96
/* update this to latest version if a change to the api struct breaks /* update this to latest version if a change to the api struct breaks
backwards compatibility (and please take the opportunity to sort in any backwards compatibility (and please take the opportunity to sort in any
new function which are "waiting" at the end of the function table) */ new function which are "waiting" at the end of the function table) */
#define PLUGIN_MIN_API_VERSION 95 #define PLUGIN_MIN_API_VERSION 96
/* plugin return codes */ /* plugin return codes */
enum plugin_status { enum plugin_status {
@ -268,8 +268,9 @@ struct plugin_api {
int height); int height);
#endif #endif
#if defined(HAVE_LCD_BITMAP) && (LCD_DEPTH < 4) && !defined(SIMULATOR) #if defined(HAVE_LCD_BITMAP) && (LCD_DEPTH < 4) && !defined(SIMULATOR)
void (*lcd_grey_phase_blit)(const struct grey_data *data, int bx, int by, void (*lcd_grey_phase_blit)(unsigned char *values, unsigned char *phases,
int bwidth, int bheight, int stride); int bx, int by, int bwidth, int bheight,
int stride);
#endif #endif
#if defined(HAVE_LCD_COLOR) #if defined(HAVE_LCD_COLOR)
void (*lcd_yuv_blit)(unsigned char * const src[3], void (*lcd_yuv_blit)(unsigned char * const src[3],

View file

@ -121,12 +121,12 @@ void grey_ub_scroll_down(int count);
#endif #endif
#if LCD_PIXELFORMAT == HORIZONTAL_PACKING #if LCD_PIXELFORMAT == HORIZONTAL_PACKING
#define _GREY_X_ADVANCE sizeof(struct grey_data) #define _GREY_X_ADVANCE 1
#else #else
#if LCD_DEPTH == 1 #if LCD_DEPTH == 1
#define _GREY_X_ADVANCE (8*sizeof(struct grey_data)) #define _GREY_X_ADVANCE 8
#elif LCD_DEPTH == 2 #elif LCD_DEPTH == 2
#define _GREY_X_ADVANCE (4*sizeof(struct grey_data)) #define _GREY_X_ADVANCE 4
#endif #endif
#endif /* LCD_PIXELFORMAT */ #endif /* LCD_PIXELFORMAT */
@ -146,7 +146,8 @@ struct _grey_info
#endif #endif
unsigned long flags; /* various flags, see #defines */ unsigned long flags; /* various flags, see #defines */
#ifndef SIMULATOR #ifndef SIMULATOR
struct grey_data *data; /* start of greyscale display data */ unsigned char *values; /* start of greyscale pixel values */
unsigned char *phases; /* start of greyscale pixel phases */
#endif #endif
unsigned char *buffer; /* start of chunky pixel buffer (for buffered mode) */ unsigned char *buffer; /* start of chunky pixel buffer (for buffered mode) */
unsigned char gvalue[256]; /* calculated brightness -> greyvalue table */ unsigned char gvalue[256]; /* calculated brightness -> greyvalue table */

View file

@ -222,11 +222,13 @@ static inline void _deferred_update(void)
static void _timer_isr(void) static void _timer_isr(void)
{ {
#if LCD_PIXELFORMAT == HORIZONTAL_PACKING #if LCD_PIXELFORMAT == HORIZONTAL_PACKING
_grey_rb->lcd_grey_phase_blit(_grey_info.data, _grey_info.bx, _grey_info.y, _grey_rb->lcd_grey_phase_blit(_grey_info.values, _grey_info.phases,
_grey_info.bx, _grey_info.y,
_grey_info.bwidth, _grey_info.height, _grey_info.bwidth, _grey_info.height,
_grey_info.width); _grey_info.width);
#else #else
_grey_rb->lcd_grey_phase_blit(_grey_info.data, _grey_info.x, _grey_info.by, _grey_rb->lcd_grey_phase_blit(_grey_info.values, _grey_info.phases,
_grey_info.x, _grey_info.by,
_grey_info.width, _grey_info.bheight, _grey_info.width, _grey_info.bheight,
_grey_info.width); _grey_info.width);
#endif #endif
@ -321,7 +323,7 @@ bool grey_init(struct plugin_api* newrb, unsigned char *gbuf, long gbuf_size,
long plane_size, buftaken; long plane_size, buftaken;
unsigned data; unsigned data;
#ifndef SIMULATOR #ifndef SIMULATOR
struct grey_data *grey_data, *grey_end; unsigned *dst, *end;
#endif #endif
_grey_rb = newrb; _grey_rb = newrb;
@ -343,35 +345,41 @@ bool grey_init(struct plugin_api* newrb, unsigned char *gbuf, long gbuf_size,
#endif #endif
#endif #endif
/* the buffer has to be long aligned */
buftaken = (-(long)gbuf) & 3;
gbuf += buftaken;
plane_size = _GREY_MULUQ(width, height); plane_size = _GREY_MULUQ(width, height);
#ifdef CPU_COLDFIRE
plane_size += (-plane_size) & 0xf; /* All buffers should be line aligned */
buftaken = (-(long)gbuf) & 0xf;
#else
buftaken = (-(long)gbuf) & 3; /* All buffers must be long aligned. */
#endif
gbuf += buftaken;
if (buffered) /* chunky buffer */ if (buffered) /* chunky buffer */
{ {
buftaken += plane_size;
_grey_info.buffer = gbuf; _grey_info.buffer = gbuf;
gbuf += plane_size; gbuf += plane_size;
buftaken += plane_size;
} }
buftaken += sizeof(struct grey_data) * plane_size;
if (buftaken > gbuf_size)
return false;
#ifdef SIMULATOR #ifdef SIMULATOR
_grey_info.buffer = gbuf; _grey_info.buffer = gbuf;
#else #else
grey_data = (struct grey_data *)gbuf; _grey_info.values = gbuf;
grey_end = grey_data + plane_size; gbuf += plane_size;
_grey_info.data = grey_data; _grey_info.phases = gbuf;
#endif
buftaken += 2 * plane_size;
while (grey_data < grey_end) if (buftaken > gbuf_size)
{ return false;
grey_data->phase = _grey_rb->rand() & 0xff;
grey_data->value = 128; /* init to white */ #ifndef SIMULATOR
grey_data++; _grey_rb->memset(_grey_info.values, 0x80, plane_size);
} dst = (unsigned*)(_grey_info.phases);
end = (unsigned*)(_grey_info.phases + plane_size);
do
*dst++ = _grey_rb->rand();
while (dst < end);
#endif #endif
_grey_info.x = 0; _grey_info.x = 0;
@ -393,7 +401,6 @@ bool grey_init(struct plugin_api* newrb, unsigned char *gbuf, long gbuf_size,
_grey_info.drawmode = DRMODE_SOLID; _grey_info.drawmode = DRMODE_SOLID;
_grey_info.curfont = FONT_SYSFIXED; _grey_info.curfont = FONT_SYSFIXED;
/* precalculate the value -> pattern index conversion table, taking /* precalculate the value -> pattern index conversion table, taking
linearisation and gamma correction into account */ linearisation and gamma correction into account */
for (i = 0; i < 256; i++) for (i = 0; i < 256; i++)
@ -532,7 +539,7 @@ void grey_update_rect(int x, int y, int width, int height)
int idx = _GREY_MULUQ(_grey_info.width, y & ~3) + (x << 2) + (~y & 3); int idx = _GREY_MULUQ(_grey_info.width, y & ~3) + (x << 2) + (~y & 3);
#endif #endif
#endif /* LCD_PIXELFORMAT */ #endif /* LCD_PIXELFORMAT */
unsigned char *dst_row = &_grey_info.data[idx].value; unsigned char *dst_row = _grey_info.values + idx;
unsigned char *src_row = src; unsigned char *src_row = src;
unsigned char *src_end = src + width; unsigned char *src_end = src + width;
@ -684,8 +691,8 @@ static void grey_screendump_hook(int fd)
for (i = 0; i < 4; i++) for (i = 0; i < 4; i++)
linebuf[x + i] = BMP_FIXEDCOLORS + *src++; linebuf[x + i] = BMP_FIXEDCOLORS + *src++;
#else #else
unsigned char *src = &_grey_info.data[_GREY_MULUQ(_grey_info.width, unsigned char *src = _grey_info.values
gy) + gx].value; + _GREY_MULUQ(_grey_info.width, gy) + gx;
for (i = 0; i < 4; i++) for (i = 0; i < 4; i++)
{ {
linebuf[x + i] = BMP_FIXEDCOLORS + *src; linebuf[x + i] = BMP_FIXEDCOLORS + *src;
@ -722,8 +729,8 @@ static void grey_screendump_hook(int fd)
gy) + gx]; gy) + gx];
#else #else
linebuf[x] = BMP_FIXEDCOLORS linebuf[x] = BMP_FIXEDCOLORS
+ _grey_info.data[_GREY_MULUQ(_grey_info.width, + _grey_info.values[_GREY_MULUQ(_grey_info.width,
gy & ~7) + (gx << 3) + (~gy & 7)].value; gy & ~7) + (gx << 3) + (~gy & 7)];
#endif #endif
} }
else else
@ -749,8 +756,8 @@ static void grey_screendump_hook(int fd)
gy) + gx]; gy) + gx];
#else #else
linebuf[x] = BMP_FIXEDCOLORS linebuf[x] = BMP_FIXEDCOLORS
+ _grey_info.data[_GREY_MULUQ(_grey_info.width, + _grey_info.values[_GREY_MULUQ(_grey_info.width,
gy & ~3) + (gx << 2) + (~gy & 7)].value; gy & ~3) + (gx << 2) + (~gy & 3)];
#endif #endif
} }
else else

View file

@ -602,16 +602,9 @@ void grey_ub_clear_display(void)
{ {
int value = (_grey_info.drawmode & DRMODE_INVERSEVID) ? int value = (_grey_info.drawmode & DRMODE_INVERSEVID) ?
_grey_info.fg_val : _grey_info.bg_val; _grey_info.fg_val : _grey_info.bg_val;
unsigned char *dst = &_grey_info.data[0].value;
unsigned char *dst_end = dst + sizeof(struct grey_data)
* _GREY_MULUQ(_grey_info.width, _grey_info.height);
do _grey_rb->memset(_grey_info.values, value,
{ _GREY_MULUQ(_grey_info.width, _grey_info.height));
*dst = value;
dst += sizeof(struct grey_data);
}
while (dst < dst_end);
} }
/* Draw a partial greyscale bitmap, canonical format */ /* Draw a partial greyscale bitmap, canonical format */
@ -654,7 +647,7 @@ void grey_ub_gray_bitmap_part(const unsigned char *src, int src_x, int src_y,
int idx = _GREY_MULUQ(_grey_info.width, y & ~3) + (x << 2) + (~y & 3); int idx = _GREY_MULUQ(_grey_info.width, y & ~3) + (x << 2) + (~y & 3);
#endif #endif
#endif /* LCD_PIXELFORMAT */ #endif /* LCD_PIXELFORMAT */
unsigned char *dst_row = &_grey_info.data[idx].value; unsigned char *dst_row = _grey_info.values + idx;
const unsigned char *src_row = src; const unsigned char *src_row = src;
const unsigned char *src_end = src + width; const unsigned char *src_end = src + width;

View file

@ -169,7 +169,7 @@ void grey_ub_scroll_left(int count)
idx = _GREY_MULUQ(_grey_info.width, y & ~3) + (~y & 3); idx = _GREY_MULUQ(_grey_info.width, y & ~3) + (~y & 3);
#endif #endif
#endif /* LCD_PIXELFORMAT */ #endif /* LCD_PIXELFORMAT */
dst = &_grey_info.data[idx].value; dst = _grey_info.values + idx;
src = dst + count * _GREY_X_ADVANCE; src = dst + count * _GREY_X_ADVANCE;
end = dst + _grey_info.width * _GREY_X_ADVANCE; end = dst + _grey_info.width * _GREY_X_ADVANCE;
@ -213,7 +213,7 @@ void grey_ub_scroll_right(int count)
idx = _GREY_MULUQ(_grey_info.width, y & ~3) + (~y & 3); idx = _GREY_MULUQ(_grey_info.width, y & ~3) + (~y & 3);
#endif #endif
#endif /* LCD_PIXELFORMAT */ #endif /* LCD_PIXELFORMAT */
start = &_grey_info.data[idx].value; start = _grey_info.values + idx;
dst = start + _grey_info.width * _GREY_X_ADVANCE; dst = start + _grey_info.width * _GREY_X_ADVANCE;
src = dst - count * _GREY_X_ADVANCE; src = dst - count * _GREY_X_ADVANCE;
@ -259,8 +259,8 @@ void grey_ub_scroll_up(int count)
is = _GREY_MULUQ(_grey_info.width, ys & ~3) + (~ys & 3); is = _GREY_MULUQ(_grey_info.width, ys & ~3) + (~ys & 3);
#endif #endif
#endif /* LCD_PIXELFORMAT */ #endif /* LCD_PIXELFORMAT */
dst = &_grey_info.data[id].value; dst = _grey_info.values + id;
src = &_grey_info.data[is].value; src = _grey_info.values + is;
dst_end = dst + _grey_info.width * _GREY_X_ADVANCE; dst_end = dst + _grey_info.width * _GREY_X_ADVANCE;
do do
@ -282,7 +282,7 @@ void grey_ub_scroll_up(int count)
id = _GREY_MULUQ(_grey_info.width, yd & ~3) + (~yd & 3); id = _GREY_MULUQ(_grey_info.width, yd & ~3) + (~yd & 3);
#endif #endif
#endif /* LCD_PIXELFORMAT */ #endif /* LCD_PIXELFORMAT */
dst = &_grey_info.data[id].value; dst = _grey_info.values + id;
dst_end = dst + _grey_info.width * _GREY_X_ADVANCE; dst_end = dst + _grey_info.width * _GREY_X_ADVANCE;
do do
@ -320,8 +320,8 @@ void grey_ub_scroll_down(int count)
is = _GREY_MULUQ(_grey_info.width, ys & ~3) + (~ys & 3); is = _GREY_MULUQ(_grey_info.width, ys & ~3) + (~ys & 3);
#endif #endif
#endif /* LCD_PIXELFORMAT */ #endif /* LCD_PIXELFORMAT */
dst = &_grey_info.data[id].value; dst = _grey_info.values + id;
src = &_grey_info.data[is].value; src = _grey_info.values + is;
dst_end = dst + _grey_info.width * _GREY_X_ADVANCE; dst_end = dst + _grey_info.width * _GREY_X_ADVANCE;
do do
@ -343,7 +343,7 @@ void grey_ub_scroll_down(int count)
id = _GREY_MULUQ(_grey_info.width, yd & ~3) + (~yd & 3); id = _GREY_MULUQ(_grey_info.width, yd & ~3) + (~yd & 3);
#endif #endif
#endif /* LCD_PIXELFORMAT */ #endif /* LCD_PIXELFORMAT */
dst = &_grey_info.data[id].value; dst = _grey_info.values + id;
dst_end = dst + _grey_info.width * _GREY_X_ADVANCE; dst_end = dst + _grey_info.width * _GREY_X_ADVANCE;
do do

View file

@ -131,13 +131,9 @@ extern void lcd_yuv_blit(unsigned char * const src[3],
int src_x, int src_y, int stride, int src_x, int src_y, int stride,
int x, int y, int width, int height); int x, int y, int width, int height);
#else #else
struct grey_data { extern void lcd_grey_phase_blit(unsigned char *values, unsigned char *phases,
unsigned char phase; /* SH1 uses it signed (doesn't matter for high level) */ int bx, int by, int bwidth, int bheight,
unsigned char value; /* 0..128 are allowed */ int stride);
} __attribute__((packed));
extern void lcd_grey_data(const struct grey_data *data, int count); /* private */
extern void lcd_grey_phase_blit(const struct grey_data *data, int bx, int by,
int bwidth, int bheight, int stride);
#endif #endif
/* performance function */ /* performance function */

View file

@ -301,118 +301,88 @@ void lcd_blit(const unsigned char* data, int bx, int y, int bwidth,
/* Performance function that works with an external buffer /* Performance function that works with an external buffer
note that bx and bwidth are in 8-pixel units! */ note that bx and bwidth are in 8-pixel units! */
void lcd_grey_phase_blit(const struct grey_data *data, int bx, int y, void lcd_grey_phase_blit(unsigned char *values, unsigned char *phases,
int bwidth, int height, int stride) int bx, int y, int bwidth, int height, int stride)
{ {
const struct grey_data *addr; unsigned char *val, *ph;
int width; int bw;
while (height--) { while (height--) {
lcd_cmd_and_data(R_RAM_ADDR_SET, (y++ << 5) + addr_offset - bx); lcd_cmd_and_data(R_RAM_ADDR_SET, (y++ << 5) + addr_offset - bx);
lcd_prepare_cmd(R_RAM_DATA); lcd_prepare_cmd(R_RAM_DATA);
addr = data; val = values;
width = bwidth; ph = phases;
bw = bwidth;
asm volatile ( asm volatile (
"10: \n" "10: \n"
"ldmia %[addr]!, {r0-r3} \n" /* r0 = v1p1v0p0 ... */ "ldmia %[ph], {r0-r1} \n" /* Fetch 8 pixel phases */
"ldmia %[val]!, {r2-r3} \n" /* Fetch 8 pixel values */
#ifdef IPOD_MINI2G #ifdef IPOD_MINI2G
"mov r5, #0x7600 \n" "mov r4, #0x7600 \n"
#else #else
"mov r5, #0 \n" "mov r4, #0 \n"
#endif #endif
"tst r0, #0x80 \n"
"and r4, r0, %[mask] \n" /* r4 = --p1--p0 */ "orreq r4, r4, #0xc0 \n"
"and r0, %[mask], r0, lsr #8 \n" /* r0 = --v1--v0 */ "tst r0, #0x8000 \n"
"orreq r4, r4, #0x30 \n"
"tst r4, #0x80 \n" "tst r0, #0x800000 \n"
"orreq r5, r5, #0xc0 \n" "orreq r4, r4, #0x0c \n"
"tst r4, #0x800000 \n" "tst r0, #0x80000000 \n"
"orreq r5, r5, #0x30 \n" "orreq r4, r4, #0x03 \n"
"bic r4, r4, %[clbt] \n" "bic r0, r0, %[clbt] \n"
"add r0, r0, r2 \n"
"add r4, r0, r4 \n" /* p0 += v0; p1 += v1; */
"strb r4, [%[addr], #-16] \n"
"mov r4, r4, lsr #16 \n"
"strb r4, [%[addr], #-14] \n"
"and r4, r1, %[mask] \n"
"and r1, %[mask], r1, lsr #8 \n"
"tst r4, #0x80 \n"
"orreq r5, r5, #0x0c \n"
"tst r4, #0x800000 \n"
"orreq r5, r5, #0x03 \n"
"bic r4, r4, %[clbt] \n"
"add r4, r1, r4 \n"
"strb r4, [%[addr], #-12] \n"
"mov r4, r4, lsr #16 \n"
"strb r4, [%[addr], #-10] \n"
#ifdef IPOD_MINI2G #ifdef IPOD_MINI2G
"mov r5, r5, lsl #8 \n" "mov r4, r4, lsl #8 \n"
#else #else
"1: \n" "1: \n"
"ldr r4, [%[lcdb]] \n" "ldr r2, [%[lcdb]] \n"
"tst r4, #0x8000 \n" "tst r2, #0x8000 \n"
"bne 1b \n" "bne 1b \n"
"str r5, [%[lcdb], #0x10] \n" "str r4, [%[lcdb], #0x10] \n"
"mov r5, #0 \n" "mov r4, #0 \n"
#endif #endif
"and r4, r2, %[mask] \n" "tst r1, #0x80 \n"
"and r2, %[mask], r2, lsr #8 \n" "orreq r4, r4, #0xc0 \n"
"tst r1, #0x8000 \n"
"orreq r4, r4, #0x30 \n"
"tst r1, #0x800000 \n"
"orreq r4, r4, #0x0c \n"
"tst r1, #0x80000000 \n"
"orreq r4, r4, #0x03 \n"
"bic r1, r1, %[clbt] \n"
"add r1, r1, r3 \n"
"tst r4, #0x80 \n" "stmia %[ph]!, {r0-r1} \n"
"orreq r5, r5, #0xc0 \n"
"tst r4, #0x800000 \n"
"orreq r5, r5, #0x30 \n"
"bic r4, r4, %[clbt] \n"
"add r4, r2, r4 \n"
"strb r4, [%[addr], #-8] \n"
"mov r4, r4, lsr #16 \n"
"strb r4, [%[addr], #-6] \n"
"and r4, r3, %[mask] \n"
"and r3, %[mask], r3, lsr #8 \n"
"tst r4, #0x80 \n"
"orreq r5, r5, #0x0c \n"
"tst r4, #0x800000 \n"
"orreq r5, r5, #0x03 \n"
"bic r4, r4, %[clbt] \n"
"add r4, r3, r4 \n"
"strb r4, [%[addr], #-4] \n"
"mov r4, r4, lsr #16 \n"
"strb r4, [%[addr], #-2] \n"
"1: \n" "1: \n"
"ldr r4, [%[lcdb]] \n" "ldr r2, [%[lcdb]] \n"
"tst r4, #0x8000 \n" "tst r2, #0x8000 \n"
"bne 1b \n" "bne 1b \n"
#ifdef IPOD_MINI2G #ifdef IPOD_MINI2G
"str r5, [%[lcdb], #0x08] \n" "str r4, [%[lcdb], #0x08] \n"
#else #else
"str r5, [%[lcdb], #0x10] \n" "str r4, [%[lcdb], #0x10] \n"
#endif #endif
"subs %[wdth], %[wdth], #1 \n" "subs %[bw], %[bw], #1 \n"
"bne 10b \n" "bne 10b \n"
: /* outputs */ : /* outputs */
[addr]"+r"(addr), [val]"+r"(val),
[wdth]"+r"(width) [ph] "+r"(ph),
[bw] "+r"(bw)
: /* inputs */ : /* inputs */
[mask]"r"(0x00ff00ff), [clbt]"r"(0x80808080),
[clbt]"r"(0x00800080),
[lcdb]"r"(LCD1_BASE) [lcdb]"r"(LCD1_BASE)
: /* clobbers */ : /* clobbers */
"r0", "r1", "r2", "r3", "r4", "r5" "r0", "r1", "r2", "r3", "r4"
); );
data += stride; values += stride;
phases += stride;
} }
} }

View file

@ -88,57 +88,146 @@ lcd_write_data:
.type lcd_grey_data,@function .type lcd_grey_data,@function
lcd_grey_data: lcd_grey_data:
lea.l (-4*4, %sp), %sp lea.l (-9*4, %sp), %sp
movem.l %d2-%d5, (%sp) movem.l %d2-%d5/%a2-%a6, (%sp) /* free some registers */
movem.l (4*4+4, %sp), %a0-%a1 /* Data pointer */ movem.l (9*4+4, %sp), %a0-%a2 /* values, phases, length */
move.l %a1, %d0 /* Length */ lea.l (%a1, %a2.l*4), %a2 /* end address */
lea 0xf0008002, %a1 /* LCD data port */ lea 0xf0008002, %a3 /* LCD data port */
move.l #0xff00ff00, %d2 /* mask for splitting value/phase pairs */
.greyloop: moveq.l #15, %d3
movem.l (%a0), %d4-%d5 /* fetch 4 pixel phase/value pairs at once */ add.l %a1, %d3
/* %d4 = p0v0p1v1, %d5 = p2v2p3v3 */ and.l #0xfffffff0, %d3 /* first line bound */
move.l %d2, %d3 /* copy mask */ move.l %a2, %d1
and.l %d4, %d3 /* %d3 = p0--p1-- */ and.l #0xfffffff0, %d1 /* last line bound */
eor.l %d3, %d4 /* %d4 = --v0--v1 */ cmp.l %d3, %d1
lsr.l #8, %d3 /* %d3 = --p0--p1 */ bls.w .g_tloop /* no lines to copy - jump to tail loop */
cmp.l %a1, %d0
bls.s .g_lloop /* no head blocks - jump to line loop */
bclr.l #23, %d3 /* Z = !(p0 & 0x80); p0 &= ~0x80; */ .g_hloop:
seq.b %d1 /* %d1 = ........................00000000 */ move.l (%a1), %d2 /* fetch 4 pixel phases */
lsl.l #2, %d1 /* %d1 = ......................00000000.. */
bclr.l #7, %d3 /* Z = !(p1 & 0x80); p1 &= ~0x80; */
seq.b %d1 /* %d1 = ......................0011111111 */
lsl.l #2, %d1 /* %d1 = ....................0011111111.. */
add.l %d4, %d3 /* p0 += v0; p1 += v1; */ bclr.l #31, %d2 /* Z = !(p0 & 0x80); p0 &= ~0x80; */
move.b %d3, (2, %a0) /* store p1 */ seq.b %d0 /* %d0 = ........................00000000 */
swap %d3 lsl.l #2, %d0 /* %d0 = ......................00000000.. */
move.b %d3, (%a0) /* store p0 */ bclr.l #23, %d2 /* Z = !(p1 & 0x80); p1 &= ~0x80; */
seq.b %d0 /* %d0 = ......................0011111111 */
lsl.l #2, %d0 /* %d0 = ....................0011111111.. */
bclr.l #15, %d2 /* Z = !(p2 & 0x80); p2 &= ~0x80; */
seq.b %d0 /* %d0 = ....................001122222222 */
lsl.l #2, %d0 /* %d0 = ..................001122222222.. */
bclr.l #7, %d2 /* Z = !(p3 & 0x80); p3 &= ~0x80; */
seq.b %d0 /* %d0 = ..................00112233333333 */
lsr.l #6, %d0 /* %d0 = ........................00112233 */
move.w %d0, (%a3) /* write pixel block */
move.l %d2, %d3 /* copy mask */ add.l (%a0)+, %d2 /* add 4 pixel values to the phases */
and.l %d5, %d3 /* %d3 = p2--p3-- */ move.l %d2, (%a1)+ /* store new phases, advance pointer */
eor.l %d3, %d5 /* %d5 = --v2--v3 */
lsr.l #8, %d3 /* %d3 = --p2--p3 */
bclr.l #23, %d3 /* Z = !(p2 & 0x80); p2 &= ~0x80; */ cmp.l %a1, %d3 /* go up to first line bound */
seq.b %d1 /* %d1 = ....................001122222222 */ bhi.s .g_hloop
lsl.l #2, %d1 /* %d1 = ..................001122222222.. */
bclr.l #7, %d3 /* Z = !(p3 & 0x80); p3 &= ~0x80; */
seq.b %d1 /* %d1 = ..................00112233333333 */
lsr.l #6, %d1 /* %d1 = ........................00112233 */
add.l %d5, %d3 /* p2 += v2; p3 += v3; */ .g_lloop:
move.b %d3, (6, %a0) /* store p3 */ movem.l (%a1), %d2-%d5
swap %d3
move.b %d3, (4, %a0) /* store p2 */
move.w %d1, (%a1) /* write pixel block */ bclr.l #31, %d2
addq.l #8, %a0 /* advance address pointer */ seq.b %d0
subq.l #1, %d0 /* any blocks left? */ lsl.l #2, %d0
bne.b .greyloop bclr.l #23, %d2
seq.b %d0
lsl.l #2, %d0
bclr.l #15, %d2
seq.b %d0
lsl.l #2, %d0
bclr.l #7, %d2
seq.b %d0
lsr.l #6, %d0
move.w %d0, (%a3)
movem.l (%sp), %d2-%d5 bclr.l #31, %d3
lea.l (4*4, %sp), %sp seq.b %d0
lsl.l #2, %d0
bclr.l #23, %d3
seq.b %d0
lsl.l #2, %d0
bclr.l #15, %d3
seq.b %d0
lsl.l #2, %d0
bclr.l #7, %d3
seq.b %d0
lsr.l #6, %d0
move.w %d0, (%a3)
bclr.l #31, %d4
seq.b %d0
lsl.l #2, %d0
bclr.l #23, %d4
seq.b %d0
lsl.l #2, %d0
bclr.l #15, %d4
seq.b %d0
lsl.l #2, %d0
bclr.l #7, %d4
seq.b %d0
lsr.l #6, %d0
move.w %d0, (%a3)
bclr.l #31, %d5
seq.b %d0
lsl.l #2, %d0
bclr.l #23, %d5
seq.b %d0
lsl.l #2, %d0
bclr.l #15, %d5
seq.b %d0
lsl.l #2, %d0
bclr.l #7, %d5
seq.b %d0
lsr.l #6, %d0
move.w %d0, (%a3)
movem.l (%a0), %d0/%a4-%a6
lea.l (16, %a0), %a0
add.l %d0, %d2
add.l %a4, %d3
add.l %a5, %d4
add.l %a6, %d5
movem.l %d2-%d5, (%a1)
lea.l (16, %a1), %a1
cmp.l %a1, %d1 /* go up to last line bound */
bhi.w .g_lloop
cmp.l %a1, %a2
bls.s .g_no_tail
.g_tloop:
move.l (%a1), %d2
bclr.l #31, %d2
seq.b %d0
lsl.l #2, %d0
bclr.l #23, %d2
seq.b %d0
lsl.l #2, %d0
bclr.l #15, %d2
seq.b %d0
lsl.l #2, %d0
bclr.l #7, %d2
seq.b %d0
lsr.l #6, %d0
move.w %d0, (%a3)
add.l (%a0)+, %d2 /* go up to end address */
move.l %d2, (%a1)+
cmp.l %a1, %a2
bhi.s .g_tloop
.g_no_tail:
movem.l (%sp), %d2-%d5/%a2-%a6 /* restore registers */
lea.l (9*4, %sp), %sp
rts rts
.gd_end: .gd_end:
.size lcd_grey_data,.gd_end-lcd_grey_data .size lcd_grey_data,.gd_end-lcd_grey_data

View file

@ -171,10 +171,13 @@ void lcd_blit(const unsigned char* data, int x, int by, int width,
} }
} }
/* Helper function for lcd_grey_phase_blit(). */
void lcd_grey_data(unsigned char *values, unsigned char *phases, int count);
/* Performance function that works with an external buffer /* Performance function that works with an external buffer
note that by and bheight are in 4-pixel units! */ note that by and bheight are in 4-pixel units! */
void lcd_grey_phase_blit(const struct grey_data *data, int x, int by, void lcd_grey_phase_blit(unsigned char *values, unsigned char *phases,
int width, int bheight, int stride) int x, int by, int width, int bheight, int stride)
{ {
stride <<= 2; /* 4 pixels per block */ stride <<= 2; /* 4 pixels per block */
while (bheight--) while (bheight--)
@ -182,8 +185,9 @@ void lcd_grey_phase_blit(const struct grey_data *data, int x, int by,
lcd_write_command_ex(LCD_CNTL_PAGE, by++, -1); lcd_write_command_ex(LCD_CNTL_PAGE, by++, -1);
lcd_write_command_ex(LCD_CNTL_COLUMN, x, -1); lcd_write_command_ex(LCD_CNTL_COLUMN, x, -1);
lcd_write_command(LCD_CNTL_DATA_WRITE); lcd_write_command(LCD_CNTL_DATA_WRITE);
lcd_grey_data(data, width); lcd_grey_data(values, phases, width);
data += stride; values += stride;
phases += stride;
} }
} }

View file

@ -100,59 +100,148 @@ lcd_write_data:
.type lcd_grey_data,@function .type lcd_grey_data,@function
lcd_grey_data: lcd_grey_data:
lea.l (-4*4, %sp), %sp lea.l (-9*4, %sp), %sp
movem.l %d2-%d5, (%sp) movem.l %d2-%d5/%a2-%a6, (%sp) /* free some registers */
movem.l (4*4+4, %sp), %a0-%a1 /* Data pointer */ movem.l (9*4+4, %sp), %a0-%a2 /* values, phases, length */
move.l %a1, %d0 /* Length */ lea.l (%a1, %a2.l*4), %a2 /* end address */
moveq #8, %d1 moveq #8, %d1
or.l %d1, (MBAR2+0xb4) /* A0 = 1 (data) */ or.l %d1, (MBAR2+0xb4) /* A0 = 1 (data) */
lea 0xf0000000, %a1 /* LCD data port */ lea 0xf0000000, %a3 /* LCD data port */
move.l #0xff00ff00, %d2 /* mask for splitting value/phase pairs */
.greyloop: moveq.l #15, %d3
movem.l (%a0), %d4-%d5 /* fetch 4 pixel phase/value pairs at once */ add.l %a1, %d3
/* %d4 = p0v0p1v1, %d5 = p2v2p3v3 */ and.l #0xfffffff0, %d3 /* first line bound */
move.l %d2, %d3 /* copy mask */ move.l %a2, %d1
and.l %d4, %d3 /* %d3 = p0--p1-- */ and.l #0xfffffff0, %d1 /* last line bound */
eor.l %d3, %d4 /* %d4 = --v0--v1 */ cmp.l %d3, %d1
lsr.l #8, %d3 /* %d3 = --p0--p1 */ bls.w .g_tloop /* no lines to copy - jump to tail loop */
cmp.l %a1, %d0
bls.s .g_lloop /* no head blocks - jump to line loop */
bclr.l #23, %d3 /* Z = !(p0 & 0x80); p0 &= ~0x80; */ .g_hloop:
seq.b %d1 /* %d1 = ........................00000000 */ move.l (%a1), %d2 /* fetch 4 pixel phases */
lsl.l #2, %d1 /* %d1 = ......................00000000.. */
bclr.l #7, %d3 /* Z = !(p1 & 0x80); p1 &= ~0x80; */
seq.b %d1 /* %d1 = ......................0011111111 */
lsl.l #2, %d1 /* %d1 = ....................0011111111.. */
add.l %d4, %d3 /* p0 += v0; p1 += v1; */ bclr.l #31, %d2 /* Z = !(p0 & 0x80); p0 &= ~0x80; */
move.b %d3, (2, %a0) /* store p1 */ seq.b %d0 /* %d0 = ........................00000000 */
swap %d3 lsl.l #2, %d0 /* %d0 = ......................00000000.. */
move.b %d3, (%a0) /* store p0 */ bclr.l #23, %d2 /* Z = !(p1 & 0x80); p1 &= ~0x80; */
seq.b %d0 /* %d0 = ......................0011111111 */
lsl.l #2, %d0 /* %d0 = ....................0011111111.. */
bclr.l #15, %d2 /* Z = !(p2 & 0x80); p2 &= ~0x80; */
seq.b %d0 /* %d0 = ....................001122222222 */
lsl.l #2, %d0 /* %d0 = ..................001122222222.. */
bclr.l #7, %d2 /* Z = !(p3 & 0x80); p3 &= ~0x80; */
seq.b %d0 /* %d0 = ..................00112233333333 */
lsr.l #6, %d0 /* %d0 = ........................00112233 */
move.w %d0, (%a3) /* write pixel block */
move.l %d2, %d3 /* copy mask */ add.l (%a0)+, %d2 /* add 4 pixel values to the phases */
and.l %d5, %d3 /* %d3 = p2--p3-- */ move.l %d2, (%a1)+ /* store new phases, advance pointer */
eor.l %d3, %d5 /* %d5 = --v2--v3 */
lsr.l #8, %d3 /* %d3 = --p2--p3 */
bclr.l #23, %d3 /* Z = !(p2 & 0x80); p2 &= ~0x80; */ cmp.l %a1, %d3 /* go up to first line bound */
seq.b %d1 /* %d1 = ....................001122222222 */ bhi.s .g_hloop
lsl.l #2, %d1 /* %d1 = ..................001122222222.. */
bclr.l #7, %d3 /* Z = !(p3 & 0x80); p3 &= ~0x80; */
seq.b %d1 /* %d1 = ..................00112233333333 */
lsr.l #6, %d1 /* %d1 = ........................00112233 */
add.l %d5, %d3 /* p2 += v2; p3 += v3; */ .g_lloop:
move.b %d3, (6, %a0) /* store p3 */ movem.l (%a1), %d2-%d5
swap %d3
move.b %d3, (4, %a0) /* store p2 */
move.w %d1, (%a1) /* write pixel block */ bclr.l #31, %d2
addq.l #8, %a0 /* advance address pointer */ seq.b %d0
subq.l #1, %d0 /* any blocks left? */ lsl.l #2, %d0
bne.b .greyloop bclr.l #23, %d2
seq.b %d0
lsl.l #2, %d0
bclr.l #15, %d2
seq.b %d0
lsl.l #2, %d0
bclr.l #7, %d2
seq.b %d0
lsr.l #6, %d0
move.w %d0, (%a3)
movem.l (%sp), %d2-%d5 bclr.l #31, %d3
lea.l (4*4, %sp), %sp seq.b %d0
lsl.l #2, %d0
bclr.l #23, %d3
seq.b %d0
lsl.l #2, %d0
bclr.l #15, %d3
seq.b %d0
lsl.l #2, %d0
bclr.l #7, %d3
seq.b %d0
lsr.l #6, %d0
move.w %d0, (%a3)
bclr.l #31, %d4
seq.b %d0
lsl.l #2, %d0
bclr.l #23, %d4
seq.b %d0
lsl.l #2, %d0
bclr.l #15, %d4
seq.b %d0
lsl.l #2, %d0
bclr.l #7, %d4
seq.b %d0
lsr.l #6, %d0
move.w %d0, (%a3)
bclr.l #31, %d5
seq.b %d0
lsl.l #2, %d0
bclr.l #23, %d5
seq.b %d0
lsl.l #2, %d0
bclr.l #15, %d5
seq.b %d0
lsl.l #2, %d0
bclr.l #7, %d5
seq.b %d0
lsr.l #6, %d0
move.w %d0, (%a3)
movem.l (%a0), %d0/%a4-%a6
lea.l (16, %a0), %a0
add.l %d0, %d2
add.l %a4, %d3
add.l %a5, %d4
add.l %a6, %d5
movem.l %d2-%d5, (%a1)
lea.l (16, %a1), %a1
cmp.l %a1, %d1 /* go up to last line bound */
bhi.w .g_lloop
cmp.l %a1, %a2
bls.s .g_no_tail
.g_tloop:
move.l (%a1), %d2
bclr.l #31, %d2
seq.b %d0
lsl.l #2, %d0
bclr.l #23, %d2
seq.b %d0
lsl.l #2, %d0
bclr.l #15, %d2
seq.b %d0
lsl.l #2, %d0
bclr.l #7, %d2
seq.b %d0
lsr.l #6, %d0
move.w %d0, (%a3)
add.l (%a0)+, %d2
move.l %d2, (%a1)+
cmp.l %a1, %a2 /* go up to end address */
bhi.s .g_tloop
.g_no_tail:
movem.l (%sp), %d2-%d5/%a2-%a6 /* restore registers */
lea.l (9*4, %sp), %sp
rts rts
.gd_end: .gd_end:
.size lcd_grey_data,.gd_end-lcd_grey_data .size lcd_grey_data,.gd_end-lcd_grey_data

View file

@ -180,10 +180,13 @@ void lcd_blit(const unsigned char* data, int x, int by, int width,
} }
} }
/* Helper function for lcd_grey_phase_blit(). */
void lcd_grey_data(unsigned char *values, unsigned char *phases, int count);
/* Performance function that works with an external buffer /* Performance function that works with an external buffer
note that by and bheight are in 4-pixel units! */ note that by and bheight are in 4-pixel units! */
void lcd_grey_phase_blit(const struct grey_data *data, int x, int by, void lcd_grey_phase_blit(unsigned char *values, unsigned char *phases,
int width, int bheight, int stride) int x, int by, int width, int bheight, int stride)
{ {
stride <<= 2; /* 4 pixels per block */ stride <<= 2; /* 4 pixels per block */
while (bheight--) while (bheight--)
@ -191,8 +194,9 @@ void lcd_grey_phase_blit(const struct grey_data *data, int x, int by,
lcd_write_command_ex(LCD_CNTL_PAGE, by++, -1); lcd_write_command_ex(LCD_CNTL_PAGE, by++, -1);
lcd_write_command_ex(LCD_CNTL_COLUMN, x, -1); lcd_write_command_ex(LCD_CNTL_COLUMN, x, -1);
lcd_write_command(LCD_CNTL_DATA_WRITE); lcd_write_command(LCD_CNTL_DATA_WRITE);
lcd_grey_data(data, width); lcd_grey_data(values, phases, width);
data += stride; values += stride;
phases += stride;
} }
} }

View file

@ -155,10 +155,13 @@ void lcd_blit(const unsigned char* data, int x, int by, int width,
} }
} }
/* Helper function for lcd_grey_phase_blit(). */
void lcd_grey_data(unsigned char *values, unsigned char *phases, int count);
/* Performance function that works with an external buffer /* Performance function that works with an external buffer
note that by and bheight are in 8-pixel units! */ note that by and bheight are in 8-pixel units! */
void lcd_grey_phase_blit(const struct grey_data *data, int x, int by, void lcd_grey_phase_blit(unsigned char *values, unsigned char *phases,
int width, int bheight, int stride) int x, int by, int width, int bheight, int stride)
{ {
stride <<= 3; /* 8 pixels per block */ stride <<= 3; /* 8 pixels per block */
while (bheight--) while (bheight--)
@ -167,8 +170,9 @@ void lcd_grey_phase_blit(const struct grey_data *data, int x, int by,
lcd_write_command (LCD_CNTL_HIGHCOL | (((x+xoffset)>>4) & 0xf)); lcd_write_command (LCD_CNTL_HIGHCOL | (((x+xoffset)>>4) & 0xf));
lcd_write_command (LCD_CNTL_LOWCOL | ((x+xoffset) & 0xf)); lcd_write_command (LCD_CNTL_LOWCOL | ((x+xoffset) & 0xf));
lcd_grey_data(data, width); lcd_grey_data(values, phases, width);
data += stride; values += stride;
phases += stride;
} }
} }

View file

@ -210,28 +210,35 @@ _lcd_write_data:
* one or multiple pixels. * one or multiple pixels.
* *
* Arguments: * Arguments:
* r4 - data address, (phase,value)-pairs * r4 - pixel value data address
* r5 - pixel block count * r5 - pixel phase data address
* r6 - pixel block count
* *
* Register usage: * Register usage:
* r0 - current pixel value * r0 - scratch / phase signs mask
* r1 - scratch * r1 - scratch
* r2 - precalculated port value (CS and SC low, DS and SD high), * r2 - precalculated port value (CS and SC low, DS and SD high),
* negated (neg)! * negated (neg)!
* r3 - lcd port address * r3 - lcd port address
* r5 - end address * r4 - current value address
* r6/r7 - current/next pixel phase * r5 - current phase address
* r8 - current block address (for writing back phase) * r6 - end address
* r9 - 0x80 (for phase modification) * r7/r8 - current/next pixel phase
* r9 - current pixel value
* r10 - 0x00000080 \
* r11 - 0x00008000 > for phase sign check
* r12 - 0x00800000 /
*/ */
_lcd_grey_data: _lcd_grey_data:
mov.l r8, @-r15 /* save r8 */ mov.l r8, @-r15 /* save r8 */
shll2 r5 /* v */ shll2 r6 /* v */
mov.l r9, @-r15 /* save r9 */ mov.l r9, @-r15 /* save r9 */
shll2 r5 /* r5 *= 16; (8 pixel per block * 2 bytes/pixel) */ shll r6 /* r6 *= 8; (8 pixels per block) */
mov.l r10, @-r15 /* save r10 */
add r4, r6 /* end address */
mov.l .lcdr, r3 /* put lcd data port address in r3 */ mov.l .lcdr, r3 /* put lcd data port address in r3 */
add r4, r5 /* end address */ nop /* keep alignment */
/* This code will fail if an interrupt changes the contents of PBDRL. /* This code will fail if an interrupt changes the contents of PBDRL.
* If so, we must disable the interrupt here. If disabling interrupts * If so, we must disable the interrupt here. If disabling interrupts
@ -240,110 +247,90 @@ _lcd_grey_data:
* this would significantly decrease performance. */ * this would significantly decrease performance. */
mov.b @r3, r0 /* r0 = PBDRL */ mov.b @r3, r0 /* r0 = PBDRL */
mov #0x80, r9 /* for phase modification - "or #imm,xx" only allows r0 */
mov.b @r4+, r6 /* fetch first pixel phase */
or #(LCD_DS|LCD_SD), r0 /* r0 |= LCD_DS|LCD_SD */ or #(LCD_DS|LCD_SD), r0 /* r0 |= LCD_DS|LCD_SD */
mov.l r11, @-r15 /* save r11 */
and #(~(LCD_CS|LCD_SC)), r0 /* r0 &= ~(LCD_CS|LCD_SC) */ and #(~(LCD_CS|LCD_SC)), r0 /* r0 &= ~(LCD_CS|LCD_SC) */
mov.l r12, @-r15 /* save r12 */
neg r0, r2 /* r2 = 0 - r0 */ neg r0, r2 /* r2 = 0 - r0 */
mov #-3, r0 /* offset for storing phase */
/* loop exploits that SD is on bit 0 for recorders and Ondios */ /* loop exploits that SD is on bit 0 for recorders and Ondios */
mov.w .ptest, r10
swap.b r10, r11
mov.l @r5, r7
swap.w r10, r12
mov.l .pmask, r0
.greyloop: .greyloop:
cmp/pz r6 /* phase non-negative? */
mov.b @r4+, r8 /* fetch pixel value */
negc r2, r1 /* T -> SD, SC low */
mov.b r1, @r3 /* set port */
or r9, r6 /* r6 -= (r6 >= 0) ? 128 : 0; */
mov.b @r4+, r7 /* fetch next pixel phase */
add #(LCD_SC), r1 /* rise SC */
mov.b r1, @r3 /* set port */
add r8, r6 /* calculate new phase */
mov.b r6, @(r0,r4) /* store phase */
cmp/pz r7 cmp/pz r7
mov.b @r4+, r8 mov.l @r4+, r9
negc r2, r1 negc r2, r1
mov.b r1, @r3 mov.b r1, @r3
or r9, r7
mov.b @r4+, r6
add #(LCD_SC), r1 add #(LCD_SC), r1
mov.b r1, @r3 mov.b r1, @r3
add r8, r7
mov.b r7, @(r0,r4)
cmp/pz r6 tst r12, r7
mov.b @r4+, r8 mov.l @(4,r5), r8
negc r2, r1 negc r2, r1
mov.b r1, @r3 mov.b r1, @r3
or r9, r6
mov.b @r4+, r7
add #(LCD_SC), r1 add #(LCD_SC), r1
mov.b r1, @r3 mov.b r1, @r3
add r8, r6
mov.b r6, @(r0,r4)
cmp/pz r7 tst r11, r7
mov.b @r4+, r8 negc r2, r1
tst r10, r7
mov.b r1, @r3
add #(LCD_SC), r1
mov.b r1, @r3
or r0, r7
sub r9, r7
negc r2, r1 negc r2, r1
mov.b r1, @r3 mov.b r1, @r3
or r9, r7
mov.b @r4+, r6
add #(LCD_SC), r1 add #(LCD_SC), r1
mov.b r1, @r3 mov.b r1, @r3
add r8, r7
mov.b r7, @(r0,r4)
cmp/pz r6 cmp/pz r8
mov.b @r4+, r8 mov.l r7, @r5
negc r2, r1 negc r2, r1
mov.b r1, @r3 mov.b r1, @r3
or r9, r6
mov.b @r4+, r7
add #(LCD_SC), r1 add #(LCD_SC), r1
mov.b r1, @r3 mov.b r1, @r3
add r8, r6
mov.b r6, @(r0,r4)
cmp/pz r7 tst r12, r8
mov.b @r4+, r8 mov.l @r4+, r9
negc r2, r1 negc r2, r1
mov.b r1, @r3 mov.b r1, @r3
or r9, r7
mov.b @r4+, r6
add #(LCD_SC), r1 add #(LCD_SC), r1
mov.b r1, @r3 mov.b r1, @r3
add r8, r7
mov.b r7, @(r0,r4)
cmp/pz r6 tst r11, r8
mov.b @r4+, r8 mov.l @(8,r5), r7
negc r2, r1 negc r2, r1
mov.b r1, @r3 mov.b r1, @r3
or r9, r6
mov.b @r4+, r7
add #(LCD_SC), r1 add #(LCD_SC), r1
mov.b r1, @r3 mov.b r1, @r3
add r8, r6
mov.b r6, @(r0,r4)
cmp/pz r7 tst r10, r8
mov.b @r4+, r8 or r0, r8
negc r2, r1 negc r2, r1
mov.b r1, @r3 mov.b r1, @r3
or r9, r7
mov.b @r4+, r6
add #(LCD_SC), r1 add #(LCD_SC), r1
mov.b r1, @r3 mov.b r1, @r3
add r8, r7
mov.b r7, @(r0,r4)
cmp/hi r4, r5 /* some blocks left? */ sub r9, r8
mov.l r8, @(4,r5)
add #8, r5
cmp/hi r4, r6
bt .greyloop bt .greyloop
mov.l @r15+, r12 /* restore r12 */
mov #(LCD_CS|LCD_DS|LCD_SD|LCD_SC), r0 mov #(LCD_CS|LCD_DS|LCD_SD|LCD_SC), r0
mov.l @r15+, r9 /* restore r9 */ mov.l @r15+, r11 /* restore r11 */
or r0, r1 /* restore port */ or r0, r1 /* restore port */
mov.l @r15+, r10 /* restore r10 */
mov.l @r15+, r9 /* restore r9 */
mov.l @r15+, r8 /* restore r8 */ mov.l @r15+, r8 /* restore r8 */
rts rts
mov.b r1, @r3 mov.b r1, @r3
@ -351,7 +338,11 @@ _lcd_grey_data:
/* This is the place to reenable the interrupts, if we have disabled /* This is the place to reenable the interrupts, if we have disabled
* them. See above. */ * them. See above. */
.ptest:
.short 0x0080
.align 2 .align 2
.lcdr: .lcdr:
.long LCDR .long LCDR
.pmask:
.long 0x80808080