1 /* SPDX-License-Identifier: GPL-2.0 */
11 /* Those of a delicate disposition might like to skip the next couple of
15  * memset(_, 0, _). However their five instances add at least a kilobyte
18  * Not a great fan of assembler for the sake of it, but I think
21  * a graphics driver. Question is whether some scheme with the blitter
22  * would be faster. I suspect not for simple text system - not much
28  * For more than 15 bytes, we try to align the write direction on a
30  * Unaligned read/write used requires 68020+ - think this is a problem?
55 		"	lsr.l	#1,%1 ; jcc 1f ; move.b %2,-(%0)\n"  in fb_memclear_small()
56 		"1:	lsr.l	#1,%1 ; jcc 1f ; move.w %2,-(%0)\n"  in fb_memclear_small()
57 		"1:	lsr.l	#1,%1 ; jcc 1f ; move.l %2,-(%0)\n"  in fb_memclear_small()
58 		"1:	lsr.l	#1,%1 ; jcc 1f ; move.l %2,-(%0) ; move.l %2,-(%0)\n"  in fb_memclear_small()
60 		: "=a" (s), "=d" (count)  in fb_memclear_small()
64 		"	jcs	3f\n"  in fb_memclear_small()
66 		"2:	movem.l	%2/%%d4/%%d5/%%d6,-(%0)\n"  in fb_memclear_small()
69 		: "=a" (s), "=d" (count)  in fb_memclear_small()
85 			"	lsr.l	#1,%1 ; jcc 1f ; clr.b (%0)+\n"  in fb_memclear()
86 			"1:	lsr.l	#1,%1 ; jcc 1f ; clr.w (%0)+\n"  in fb_memclear()
87 			"1:	lsr.l	#1,%1 ; jcc 1f ; clr.l (%0)+\n"  in fb_memclear()
88 			"1:	lsr.l	#1,%1 ; jcc 1f ; clr.l (%0)+ ; clr.l (%0)+\n"  in fb_memclear()
90 			: "=a" (s), "=d" (count)  in fb_memclear()
96 			"	lsr.l	#1,%2 ; jcc 1f ; clr.b (%0)+ ; subq.w #1,%1\n"  in fb_memclear()
97 			"	lsr.l	#1,%2 ; jcs 2f\n"  /* %0 increased=>bit 2 switched*/  in fb_memclear()
98 			"	clr.w	(%0)+  ; subq.w  #2,%1 ; jra 2f\n"  in fb_memclear()
99 			"1:	lsr.l	#1,%2 ; jcc 2f\n"  in fb_memclear()
101 			"2:	move.w	%1,%2; lsr.l #2,%1 ; jeq 6f\n"  in fb_memclear()
102 			"	lsr.l	#1,%1 ; jcc 3f ; clr.l (%0)+\n"  in fb_memclear()
103 			"3:	lsr.l	#1,%1 ; jcc 4f ; clr.l (%0)+ ; clr.l (%0)+\n"  in fb_memclear()
104 			"4:	subq.l	#1,%1 ; jcs 6f\n"  in fb_memclear()
107 			"6:	move.w	%2,%1; btst #1,%1 ; jeq 7f ; clr.w (%0)+\n"  in fb_memclear()
108 			"7:	btst	#0,%1 ; jeq 8f ; clr.b (%0)+\n"  in fb_memclear()
110 			: "=a" (s), "=d" (count), "=d" (tmp)  in fb_memclear()
124 		"	lsr.l	#1,%1 ; jcc 1f ; move.b %2,-(%0)\n"  in fb_memset255()
125 		"1:	lsr.l	#1,%1 ; jcc 1f ; move.w %2,-(%0)\n"  in fb_memset255()
126 		"1:	lsr.l	#1,%1 ; jcc 1f ; move.l %2,-(%0)\n"  in fb_memset255()
127 		"1:	lsr.l	#1,%1 ; jcc 1f ; move.l %2,-(%0) ; move.l %2,-(%0)\n"  in fb_memset255()
129 		: "=a" (s), "=d" (count)  in fb_memset255()
130 		: "d" (-1), "0" ((char *)s+count), "1" (count));  in fb_memset255()
132 		"	subq.l	#1,%1 ; jcs 3f\n"  in fb_memset255()
134 		"2:	movem.l	%2/%%d4/%%d5/%%d6,-(%0)\n"  in fb_memset255()
137 		: "=a" (s), "=d" (count)  in fb_memset255()
138 		: "d" (-1), "0" (s), "1" (count)  in fb_memset255()
150 				"	lsr.l	#1,%2 ; jcc 1f ; move.b (%1)+,(%0)+\n"  in fb_memmove()
151 				"1:	lsr.l	#1,%2 ; jcc 1f ; move.w (%1)+,(%0)+\n"  in fb_memmove()
152 				"1:	lsr.l	#1,%2 ; jcc 1f ; move.l (%1)+,(%0)+\n"  in fb_memmove()
153 				"1:	lsr.l	#1,%2 ; jcc 1f ; move.l (%1)+,(%0)+ ; move.l (%1)+,(%0)+\n"  in fb_memmove()
155 				: "=a" (d), "=a" (s), "=d" (count)  in fb_memmove()
161 				"	lsr.l	#1,%3 ; jcc 1f ; move.b (%1)+,(%0)+ ; subqw #1,%2\n"  in fb_memmove()
162 				"	lsr.l	#1,%3 ; jcs 2f\n"  /* %0 increased=>bit 2 switched*/  in fb_memmove()
163 				"	move.w	(%1)+,(%0)+  ; subqw  #2,%2 ; jra 2f\n"  in fb_memmove()
164 				"1:	lsr.l   #1,%3 ; jcc 2f\n"  in fb_memmove()
166 				"2:	move.w	%2,%-; lsr.l #2,%2 ; jeq 6f\n"  in fb_memmove()
167 				"	lsr.l	#1,%2 ; jcc 3f ; move.l (%1)+,(%0)+\n"  in fb_memmove()
168 				"3:	lsr.l	#1,%2 ; jcc 4f ; move.l (%1)+,(%0)+ ; move.l (%1)+,(%0)+\n"  in fb_memmove()
169 				"4:	subq.l	#1,%2 ; jcs 6f\n"  in fb_memmove()
173 				"6:	move.w	%+,%2; btst #1,%2 ; jeq 7f ; move.w (%1)+,(%0)+\n"  in fb_memmove()
174 				"7:	btst	#0,%2 ; jeq 8f ; move.b (%1)+,(%0)+\n"  in fb_memmove()
176 				: "=a" (d), "=a" (s), "=d" (count), "=d" (tmp)  in fb_memmove()
182 				"	lsr.l	#1,%2 ; jcc 1f ; move.b -(%1),-(%0)\n"  in fb_memmove()
183 				"1:	lsr.l	#1,%2 ; jcc 1f ; move.w -(%1),-(%0)\n"  in fb_memmove()
184 				"1:	lsr.l	#1,%2 ; jcc 1f ; move.l -(%1),-(%0)\n"  in fb_memmove()
185 				"1:	lsr.l	#1,%2 ; jcc 1f ; move.l -(%1),-(%0) ; move.l -(%1),-(%0)\n"  in fb_memmove()
187 				: "=a" (d), "=a" (s), "=d" (count)  in fb_memmove()
194 				"	lsr.l	#1,%3 ; jcc 1f ; move.b -(%1),-(%0) ; subqw #1,%2\n"  in fb_memmove()
195 				"	lsr.l	#1,%3 ; jcs 2f\n"  /* %0 increased=>bit 2 switched*/  in fb_memmove()
196 				"	move.w	-(%1),-(%0) ; subqw  #2,%2 ; jra 2f\n"  in fb_memmove()
197 				"1:	lsr.l	#1,%3 ; jcc 2f\n"  in fb_memmove()
198 				"	move.w	-(%1),-(%0) ; subqw  #2,%2\n"  in fb_memmove()
199 				"2:	move.w	%2,%-; lsr.l #2,%2 ; jeq 6f\n"  in fb_memmove()
200 				"	lsr.l	#1,%2 ; jcc 3f ; move.l -(%1),-(%0)\n"  in fb_memmove()
201 				"3:	lsr.l	#1,%2 ; jcc 4f ; move.l -(%1),-(%0) ; move.l -(%1),-(%0)\n"  in fb_memmove()
202 				"4:	subq.l	#1,%2 ; jcs 6f\n"  in fb_memmove()
203 				"5:	move.l	-(%1),-(%0); move.l -(%1),-(%0)\n"  in fb_memmove()
204 				"	move.l	-(%1),-(%0); move.l -(%1),-(%0)\n"  in fb_memmove()
206 				"6:	move.w	%+,%2; btst #1,%2 ; jeq 7f ; move.w -(%1),-(%0)\n"  in fb_memmove()
207 				"7:	btst	#0,%2 ; jeq 8f ; move.b -(%1),-(%0)\n"  in fb_memmove()
209 				: "=a" (d), "=a" (s), "=d" (count), "=d" (tmp)  in fb_memmove()
232 			: "=a" (src), "=a" (dst), "=d" (size)  in fast_memmove()
233 			: "0" (src), "1" (dst), "2" (size / 16 - 1)  in fast_memmove()
239 			"	movem.l	%%d0/%%d1/%%a0/%%a1,-(%1)\n"  in fast_memmove()
243 			: "=a" (src), "=a" (dst), "=d" (size)  in fast_memmove()
244 			: "0" (src + size), "1" (dst + size), "2" (size / 16 - 1)  in fast_memmove()
251  * This expands a up to 8 bit color into two longs
278  * set an 8bit value to a color
370 		rows--;  in fill16_col()
396         } while (--h);  in memmove32_col()