Lines Matching +full:4 +full:f

29  * This function assumes 2- or 4-byte alignment.  Other alignments will fail!
35 	bgeu	sum, val, 99f	; \
44 	 * is aligned on either a 2-byte or 4-byte boundary.
48 	bnez	a5, 8f		/* branch if 2-byte aligned */
49 	/* Fall-through on common case, 4-byte alignment */
53 	loopgtz	a5, 2f
55 	beqz	a5, 2f
61 	l32i	a7, a2, 4
76 	addi	a2, a2, 4*8
81 	extui	a5, a3, 2, 3	/* remaining 4-byte chunks */
83 	loopgtz	a5, 3f
85 	beqz	a5, 3f
87 	add	a5, a5, a2	/* a5 = end of last 4-byte chunk */
92 	addi	a2, a2, 4
97 	_bbci.l	a3, 1, 5f	/* remaining 2-byte chunk */
102 	_bbci.l	a3, 0, 7f	/* remaining 1-byte chunk */
118 	bnez	a5, 8f		/* branch if 1-byte aligned */
124 	j	1b		/* now buf is 4-byte aligned */
131 	srli	a5, a3, 2	/* 4-byte chunks */
133 	loopgtz	a5, 2f
135 	beqz	a5, 2f
137 	add	a5, a5, a2	/* a5 = end of last 4-byte chunk */
152 	addi	a2, a2, 4
157 	_bbci.l	a3, 1, 3f	/* remaining 2-byte chunk, still odd addr */
187     This function is optimized for 4-byte aligned addresses.  Other
197 	/* We optimize the following alignment tests for the 4-byte
205 	beqz	a9, 1f		/* branch if both are 4-byte aligned */
206 	bbsi.l	a10, 0, 5f	/* branch if one address is odd */
207 	j	3f		/* one address is 2-byte aligned */
209 /*	_bbsi.l	a10, 0, 5f */	/* branch if odd address */
210 /*	_bbsi.l	a10, 1, 3f */	/* branch if 2-byte-aligned address */
213 	/* src and dst are both 4-byte aligned */
216 	loopgtz	a10, 2f
218 	beqz	a10, 2f
223 EX(10f)	l32i	a9, a2, 0
224 EX(10f)	l32i	a8, a2, 4
225 EX(10f)	s32i	a9, a3, 0
226 EX(10f)	s32i	a8, a3, 4
229 EX(10f)	l32i	a9, a2, 8
230 EX(10f)	l32i	a8, a2, 12
231 EX(10f)	s32i	a9, a3, 8
232 EX(10f)	s32i	a8, a3, 12
235 EX(10f)	l32i	a9, a2, 16
236 EX(10f)	l32i	a8, a2, 20
237 EX(10f)	s32i	a9, a3, 16
238 EX(10f)	s32i	a8, a3, 20
241 EX(10f)	l32i	a9, a2, 24
242 EX(10f)	l32i	a8, a2, 28
243 EX(10f)	s32i	a9, a3, 24
244 EX(10f)	s32i	a8, a3, 28
253 	extui	a10, a4, 2, 3	/* remaining 4-byte chunks */
256 	loopgtz	a10, 3f
258 	beqz	a10, 3f
260 	add	a10, a10, a2	/* a10 = end of last 4-byte src chunk */
263 EX(10f)	l32i	a9, a2, 0
264 EX(10f)	s32i	a9, a3, 0
266 	addi	a2, a2, 4
267 	addi	a3, a3, 4
274 	to here from the 4-byte alignment case to process, at most,
278 	inefficient, so align your addresses to 4-byte boundaries.
287 	loopgtz	a10, 4f
289 	beqz	a10, 4f
294 EX(10f)	l16ui	a9, a2, 0
295 EX(10f)	s16i	a9, a3, 0
302 4:
304 	_bbci.l	a4, 0, 8f	/* 1-byte chunk */
305 EX(10f)	l8ui	a9, a2, 0
306 EX(10f)	s8i	a9, a3, 0
322 	loopgtz	a10, 6f
324 	beqz	a10, 6f
329 EX(10f)	l8ui	a9, a2, 0
330 EX(10f)	l8ui	a8, a2, 1
331 EX(10f)	s8i	a9, a3, 0
332 EX(10f)	s8i	a8, a3, 1
346 	j	4b		/* process the possible trailing odd byte */