xref: /openbmc/linux/arch/sparc/lib/U1memcpy.S (revision fe998f4c) 
1/* SPDX-License-Identifier: GPL-2.0 */
2/* U1memcpy.S: UltraSPARC-I/II/IIi/IIe optimized memcpy.
3 *
4 * Copyright (C) 1997, 2004 David S. Miller (davem@redhat.com)
5 * Copyright (C) 1996, 1997, 1998, 1999 Jakub Jelinek (jj@ultra.linux.cz)
6 */
7
8#ifdef __KERNEL__
9#include <linux/export.h>
10#include <linux/linkage.h>
11#include <asm/visasm.h>
12#include <asm/asi.h>
13#define GLOBAL_SPARE	g7
14#else
15#define GLOBAL_SPARE	g5
16#define ASI_BLK_P 0xf0
17#define FPRS_FEF  0x04
18#ifdef MEMCPY_DEBUG
19#define VISEntry rd %fprs, %o5; wr %g0, FPRS_FEF, %fprs; \
20		 clr %g1; clr %g2; clr %g3; subcc %g0, %g0, %g0;
21#define VISExit and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
22#else
23#define VISEntry rd %fprs, %o5; wr %g0, FPRS_FEF, %fprs
24#define VISExit and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
25#endif
26#endif
27
28#ifndef EX_LD
29#define EX_LD(x,y)	x
30#endif
31#ifndef EX_LD_FP
32#define EX_LD_FP(x,y)	x
33#endif
34
35#ifndef EX_ST
36#define EX_ST(x,y)	x
37#endif
38#ifndef EX_ST_FP
39#define EX_ST_FP(x,y)	x
40#endif
41
42#ifndef LOAD
43#define LOAD(type,addr,dest)	type [addr], dest
44#endif
45
46#ifndef LOAD_BLK
47#define LOAD_BLK(addr,dest)	ldda [addr] ASI_BLK_P, dest
48#endif
49
50#ifndef STORE
51#define STORE(type,src,addr)	type src, [addr]
52#endif
53
54#ifndef STORE_BLK
55#define STORE_BLK(src,addr)	stda src, [addr] ASI_BLK_P
56#endif
57
58#ifndef FUNC_NAME
59#define FUNC_NAME	memcpy
60#endif
61
62#ifndef PREAMBLE
63#define PREAMBLE
64#endif
65
66#ifndef XCC
67#define XCC xcc
68#endif
69
70#define FREG_FROB(f1, f2, f3, f4, f5, f6, f7, f8, f9)		\
71	faligndata		%f1, %f2, %f48;			\
72	faligndata		%f2, %f3, %f50;			\
73	faligndata		%f3, %f4, %f52;			\
74	faligndata		%f4, %f5, %f54;			\
75	faligndata		%f5, %f6, %f56;			\
76	faligndata		%f6, %f7, %f58;			\
77	faligndata		%f7, %f8, %f60;			\
78	faligndata		%f8, %f9, %f62;
79
80#define MAIN_LOOP_CHUNK(src, dest, fdest, fsrc, jmptgt)			\
81	EX_LD_FP(LOAD_BLK(%src, %fdest), U1_gs_80_fp);			\
82	EX_ST_FP(STORE_BLK(%fsrc, %dest), U1_gs_80_fp);			\
83	add			%src, 0x40, %src;			\
84	subcc			%GLOBAL_SPARE, 0x40, %GLOBAL_SPARE;	\
85	be,pn			%xcc, jmptgt;				\
86	 add			%dest, 0x40, %dest;			\
87
88#define LOOP_CHUNK1(src, dest, branch_dest)		\
89	MAIN_LOOP_CHUNK(src, dest, f0,  f48, branch_dest)
90#define LOOP_CHUNK2(src, dest, branch_dest)		\
91	MAIN_LOOP_CHUNK(src, dest, f16, f48, branch_dest)
92#define LOOP_CHUNK3(src, dest, branch_dest)		\
93	MAIN_LOOP_CHUNK(src, dest, f32, f48, branch_dest)
94
95#define DO_SYNC			membar	#Sync;
96#define STORE_SYNC(dest, fsrc)				\
97	EX_ST_FP(STORE_BLK(%fsrc, %dest), U1_gs_80_fp);	\
98	add			%dest, 0x40, %dest;	\
99	DO_SYNC
100
101#define STORE_JUMP(dest, fsrc, target)			\
102	EX_ST_FP(STORE_BLK(%fsrc, %dest), U1_gs_40_fp);	\
103	add			%dest, 0x40, %dest;	\
104	ba,pt			%xcc, target;		\
105	 nop;
106
107#define FINISH_VISCHUNK(dest, f0, f1)			\
108	subcc			%g3, 8, %g3;		\
109	bl,pn			%xcc, 95f;		\
110	 faligndata		%f0, %f1, %f48;		\
111	EX_ST_FP(STORE(std, %f48, %dest), U1_g3_8_fp);	\
112	add			%dest, 8, %dest;
113
114#define UNEVEN_VISCHUNK_LAST(dest, f0, f1)	\
115	subcc			%g3, 8, %g3;	\
116	bl,pn			%xcc, 95f;	\
117	 fsrc2			%f0, %f1;
118
119#define UNEVEN_VISCHUNK(dest, f0, f1)		\
120	UNEVEN_VISCHUNK_LAST(dest, f0, f1)	\
121	ba,a,pt			%xcc, 93f;
122
123	.register	%g2,#scratch
124	.register	%g3,#scratch
125
126	.text
127#ifndef EX_RETVAL
128#define EX_RETVAL(x)	x
129ENTRY(U1_g1_1_fp)
130	VISExitHalf
131	add		%g1, 1, %g1
132	add		%g1, %g2, %g1
133	retl
134	 add		%g1, %o2, %o0
135ENDPROC(U1_g1_1_fp)
136ENTRY(U1_g2_0_fp)
137	VISExitHalf
138	retl
139	 add		%g2, %o2, %o0
140ENDPROC(U1_g2_0_fp)
141ENTRY(U1_g2_8_fp)
142	VISExitHalf
143	add		%g2, 8, %g2
144	retl
145	 add		%g2, %o2, %o0
146ENDPROC(U1_g2_8_fp)
147ENTRY(U1_gs_0_fp)
148	VISExitHalf
149	add		%GLOBAL_SPARE, %g3, %o0
150	retl
151	 add		%o0, %o2, %o0
152ENDPROC(U1_gs_0_fp)
153ENTRY(U1_gs_80_fp)
154	VISExitHalf
155	add		%GLOBAL_SPARE, 0x80, %GLOBAL_SPARE
156	add		%GLOBAL_SPARE, %g3, %o0
157	retl
158	 add		%o0, %o2, %o0
159ENDPROC(U1_gs_80_fp)
160ENTRY(U1_gs_40_fp)
161	VISExitHalf
162	add		%GLOBAL_SPARE, 0x40, %GLOBAL_SPARE
163	add		%GLOBAL_SPARE, %g3, %o0
164	retl
165	 add		%o0, %o2, %o0
166ENDPROC(U1_gs_40_fp)
167ENTRY(U1_g3_0_fp)
168	VISExitHalf
169	retl
170	 add		%g3, %o2, %o0
171ENDPROC(U1_g3_0_fp)
172ENTRY(U1_g3_8_fp)
173	VISExitHalf
174	add		%g3, 8, %g3
175	retl
176	 add		%g3, %o2, %o0
177ENDPROC(U1_g3_8_fp)
178ENTRY(U1_o2_0_fp)
179	VISExitHalf
180	retl
181	 mov		%o2, %o0
182ENDPROC(U1_o2_0_fp)
183ENTRY(U1_o2_1_fp)
184	VISExitHalf
185	retl
186	 add		%o2, 1, %o0
187ENDPROC(U1_o2_1_fp)
188ENTRY(U1_gs_0)
189	VISExitHalf
190	retl
191	 add		%GLOBAL_SPARE, %o2, %o0
192ENDPROC(U1_gs_0)
193ENTRY(U1_gs_8)
194	VISExitHalf
195	add		%GLOBAL_SPARE, %o2, %GLOBAL_SPARE
196	retl
197	 add		%GLOBAL_SPARE, 0x8, %o0
198ENDPROC(U1_gs_8)
199ENTRY(U1_gs_10)
200	VISExitHalf
201	add		%GLOBAL_SPARE, %o2, %GLOBAL_SPARE
202	retl
203	 add		%GLOBAL_SPARE, 0x10, %o0
204ENDPROC(U1_gs_10)
205ENTRY(U1_o2_0)
206	retl
207	 mov		%o2, %o0
208ENDPROC(U1_o2_0)
209ENTRY(U1_o2_8)
210	retl
211	 add		%o2, 8, %o0
212ENDPROC(U1_o2_8)
213ENTRY(U1_o2_4)
214	retl
215	 add		%o2, 4, %o0
216ENDPROC(U1_o2_4)
217ENTRY(U1_o2_1)
218	retl
219	 add		%o2, 1, %o0
220ENDPROC(U1_o2_1)
221ENTRY(U1_g1_0)
222	retl
223	 add		%g1, %o2, %o0
224ENDPROC(U1_g1_0)
225ENTRY(U1_g1_1)
226	add		%g1, 1, %g1
227	retl
228	 add		%g1, %o2, %o0
229ENDPROC(U1_g1_1)
230ENTRY(U1_gs_0_o2_adj)
231	and		%o2, 7, %o2
232	retl
233	 add		%GLOBAL_SPARE, %o2, %o0
234ENDPROC(U1_gs_0_o2_adj)
235ENTRY(U1_gs_8_o2_adj)
236	and		%o2, 7, %o2
237	add		%GLOBAL_SPARE, 8, %GLOBAL_SPARE
238	retl
239	 add		%GLOBAL_SPARE, %o2, %o0
240ENDPROC(U1_gs_8_o2_adj)
241#endif
242
243	.align		64
244
245	.globl		FUNC_NAME
246	.type		FUNC_NAME,#function
247FUNC_NAME:		/* %o0=dst, %o1=src, %o2=len */
248	srlx		%o2, 31, %g2
249	cmp		%g2, 0
250	tne		%xcc, 5
251	PREAMBLE
252	mov		%o0, %o4
253	cmp		%o2, 0
254	be,pn		%XCC, 85f
255	 or		%o0, %o1, %o3
256	cmp		%o2, 16
257	blu,a,pn	%XCC, 80f
258	 or		%o3, %o2, %o3
259
260	cmp		%o2, (5 * 64)
261	blu,pt		%XCC, 70f
262	 andcc		%o3, 0x7, %g0
263
264	/* Clobbers o5/g1/g2/g3/g7/icc/xcc.  */
265	VISEntry
266
267	/* Is 'dst' already aligned on an 64-byte boundary? */
268	andcc		%o0, 0x3f, %g2
269	be,pt		%XCC, 2f
270
271	/* Compute abs((dst & 0x3f) - 0x40) into %g2.  This is the number
272	 * of bytes to copy to make 'dst' 64-byte aligned.  We pre-
273	 * subtract this from 'len'.
274	 */
275	 sub		%o0, %o1, %GLOBAL_SPARE
276	sub		%g2, 0x40, %g2
277	sub		%g0, %g2, %g2
278	sub		%o2, %g2, %o2
279	andcc		%g2, 0x7, %g1
280	be,pt		%icc, 2f
281	 and		%g2, 0x38, %g2
282
2831:	subcc		%g1, 0x1, %g1
284	EX_LD_FP(LOAD(ldub, %o1 + 0x00, %o3), U1_g1_1_fp)
285	EX_ST_FP(STORE(stb, %o3, %o1 + %GLOBAL_SPARE), U1_g1_1_fp)
286	bgu,pt		%XCC, 1b
287	 add		%o1, 0x1, %o1
288
289	add		%o1, %GLOBAL_SPARE, %o0
290
2912:	cmp		%g2, 0x0
292	and		%o1, 0x7, %g1
293	be,pt		%icc, 3f
294	 alignaddr	%o1, %g0, %o1
295
296	EX_LD_FP(LOAD(ldd, %o1, %f4), U1_g2_0_fp)
2971:	EX_LD_FP(LOAD(ldd, %o1 + 0x8, %f6), U1_g2_0_fp)
298	add		%o1, 0x8, %o1
299	subcc		%g2, 0x8, %g2
300	faligndata	%f4, %f6, %f0
301	EX_ST_FP(STORE(std, %f0, %o0), U1_g2_8_fp)
302	be,pn		%icc, 3f
303	 add		%o0, 0x8, %o0
304
305	EX_LD_FP(LOAD(ldd, %o1 + 0x8, %f4), U1_g2_0_fp)
306	add		%o1, 0x8, %o1
307	subcc		%g2, 0x8, %g2
308	faligndata	%f6, %f4, %f0
309	EX_ST_FP(STORE(std, %f0, %o0), U1_g2_8_fp)
310	bne,pt		%icc, 1b
311	 add		%o0, 0x8, %o0
312
313	/* Destination is 64-byte aligned.  */
3143:
315	membar		  #LoadStore | #StoreStore | #StoreLoad
316
317	subcc		%o2, 0x40, %GLOBAL_SPARE
318	add		%o1, %g1, %g1
319	andncc		%GLOBAL_SPARE, (0x40 - 1), %GLOBAL_SPARE
320	srl		%g1, 3, %g2
321	sub		%o2, %GLOBAL_SPARE, %g3
322	andn		%o1, (0x40 - 1), %o1
323	and		%g2, 7, %g2
324	andncc		%g3, 0x7, %g3
325	fsrc2		%f0, %f2
326	sub		%g3, 0x8, %g3
327	sub		%o2, %GLOBAL_SPARE, %o2
328
329	add		%g1, %GLOBAL_SPARE, %g1
330	subcc		%o2, %g3, %o2
331
332	EX_LD_FP(LOAD_BLK(%o1, %f0), U1_gs_0_fp)
333	add		%o1, 0x40, %o1
334	add		%g1, %g3, %g1
335	EX_LD_FP(LOAD_BLK(%o1, %f16), U1_gs_0_fp)
336	add		%o1, 0x40, %o1
337	sub		%GLOBAL_SPARE, 0x80, %GLOBAL_SPARE
338	EX_LD_FP(LOAD_BLK(%o1, %f32), U1_gs_80_fp)
339	add		%o1, 0x40, %o1
340
341	/* There are 8 instances of the unrolled loop,
342	 * one for each possible alignment of the
343	 * source buffer.  Each loop instance is 452
344	 * bytes.
345	 */
346	sll		%g2, 3, %o3
347	sub		%o3, %g2, %o3
348	sllx		%o3, 4, %o3
349	add		%o3, %g2, %o3
350	sllx		%o3, 2, %g2
3511:	rd		%pc, %o3
352	add		%o3, %lo(1f - 1b), %o3
353	jmpl		%o3 + %g2, %g0
354	 nop
355
356	.align		64
3571:	FREG_FROB(f0, f2, f4, f6, f8, f10,f12,f14,f16)
358	LOOP_CHUNK1(o1, o0, 1f)
359	FREG_FROB(f16,f18,f20,f22,f24,f26,f28,f30,f32)
360	LOOP_CHUNK2(o1, o0, 2f)
361	FREG_FROB(f32,f34,f36,f38,f40,f42,f44,f46,f0)
362	LOOP_CHUNK3(o1, o0, 3f)
363	ba,pt		%xcc, 1b+4
364	 faligndata	%f0, %f2, %f48
3651:	FREG_FROB(f16,f18,f20,f22,f24,f26,f28,f30,f32)
366	STORE_SYNC(o0, f48)
367	FREG_FROB(f32,f34,f36,f38,f40,f42,f44,f46,f0)
368	STORE_JUMP(o0, f48, 40f)
3692:	FREG_FROB(f32,f34,f36,f38,f40,f42,f44,f46,f0)
370	STORE_SYNC(o0, f48)
371	FREG_FROB(f0, f2, f4, f6, f8, f10,f12,f14,f16)
372	STORE_JUMP(o0, f48, 48f)
3733:	FREG_FROB(f0, f2, f4, f6, f8, f10,f12,f14,f16)
374	STORE_SYNC(o0, f48)
375	FREG_FROB(f16,f18,f20,f22,f24,f26,f28,f30,f32)
376	STORE_JUMP(o0, f48, 56f)
377
3781:	FREG_FROB(f2, f4, f6, f8, f10,f12,f14,f16,f18)
379	LOOP_CHUNK1(o1, o0, 1f)
380	FREG_FROB(f18,f20,f22,f24,f26,f28,f30,f32,f34)
381	LOOP_CHUNK2(o1, o0, 2f)
382	FREG_FROB(f34,f36,f38,f40,f42,f44,f46,f0, f2)
383	LOOP_CHUNK3(o1, o0, 3f)
384	ba,pt		%xcc, 1b+4
385	 faligndata	%f2, %f4, %f48
3861:	FREG_FROB(f18,f20,f22,f24,f26,f28,f30,f32,f34)
387	STORE_SYNC(o0, f48)
388	FREG_FROB(f34,f36,f38,f40,f42,f44,f46,f0, f2)
389	STORE_JUMP(o0, f48, 41f)
3902:	FREG_FROB(f34,f36,f38,f40,f42,f44,f46,f0, f2)
391	STORE_SYNC(o0, f48)
392	FREG_FROB(f2, f4, f6, f8, f10,f12,f14,f16,f18)
393	STORE_JUMP(o0, f48, 49f)
3943:	FREG_FROB(f2, f4, f6, f8, f10,f12,f14,f16,f18)
395	STORE_SYNC(o0, f48)
396	FREG_FROB(f18,f20,f22,f24,f26,f28,f30,f32,f34)
397	STORE_JUMP(o0, f48, 57f)
398
3991:	FREG_FROB(f4, f6, f8, f10,f12,f14,f16,f18,f20)
400	LOOP_CHUNK1(o1, o0, 1f)
401	FREG_FROB(f20,f22,f24,f26,f28,f30,f32,f34,f36)
402	LOOP_CHUNK2(o1, o0, 2f)
403	FREG_FROB(f36,f38,f40,f42,f44,f46,f0, f2, f4)
404	LOOP_CHUNK3(o1, o0, 3f)
405	ba,pt		%xcc, 1b+4
406	 faligndata	%f4, %f6, %f48
4071:	FREG_FROB(f20,f22,f24,f26,f28,f30,f32,f34,f36)
408	STORE_SYNC(o0, f48)
409	FREG_FROB(f36,f38,f40,f42,f44,f46,f0, f2, f4)
410	STORE_JUMP(o0, f48, 42f)
4112:	FREG_FROB(f36,f38,f40,f42,f44,f46,f0, f2, f4)
412	STORE_SYNC(o0, f48)
413	FREG_FROB(f4, f6, f8, f10,f12,f14,f16,f18,f20)
414	STORE_JUMP(o0, f48, 50f)
4153:	FREG_FROB(f4, f6, f8, f10,f12,f14,f16,f18,f20)
416	STORE_SYNC(o0, f48)
417	FREG_FROB(f20,f22,f24,f26,f28,f30,f32,f34,f36)
418	STORE_JUMP(o0, f48, 58f)
419
4201:	FREG_FROB(f6, f8, f10,f12,f14,f16,f18,f20,f22)
421	LOOP_CHUNK1(o1, o0, 1f)
422	FREG_FROB(f22,f24,f26,f28,f30,f32,f34,f36,f38)
423	LOOP_CHUNK2(o1, o0, 2f)
424	FREG_FROB(f38,f40,f42,f44,f46,f0, f2, f4, f6)
425	LOOP_CHUNK3(o1, o0, 3f)
426	ba,pt		%xcc, 1b+4
427	 faligndata	%f6, %f8, %f48
4281:	FREG_FROB(f22,f24,f26,f28,f30,f32,f34,f36,f38)
429	STORE_SYNC(o0, f48)
430	FREG_FROB(f38,f40,f42,f44,f46,f0, f2, f4, f6)
431	STORE_JUMP(o0, f48, 43f)
4322:	FREG_FROB(f38,f40,f42,f44,f46,f0, f2, f4, f6)
433	STORE_SYNC(o0, f48)
434	FREG_FROB(f6, f8, f10,f12,f14,f16,f18,f20,f22)
435	STORE_JUMP(o0, f48, 51f)
4363:	FREG_FROB(f6, f8, f10,f12,f14,f16,f18,f20,f22)
437	STORE_SYNC(o0, f48)
438	FREG_FROB(f22,f24,f26,f28,f30,f32,f34,f36,f38)
439	STORE_JUMP(o0, f48, 59f)
440
4411:	FREG_FROB(f8, f10,f12,f14,f16,f18,f20,f22,f24)
442	LOOP_CHUNK1(o1, o0, 1f)
443	FREG_FROB(f24,f26,f28,f30,f32,f34,f36,f38,f40)
444	LOOP_CHUNK2(o1, o0, 2f)
445	FREG_FROB(f40,f42,f44,f46,f0, f2, f4, f6, f8)
446	LOOP_CHUNK3(o1, o0, 3f)
447	ba,pt		%xcc, 1b+4
448	 faligndata	%f8, %f10, %f48
4491:	FREG_FROB(f24,f26,f28,f30,f32,f34,f36,f38,f40)
450	STORE_SYNC(o0, f48)
451	FREG_FROB(f40,f42,f44,f46,f0, f2, f4, f6, f8)
452	STORE_JUMP(o0, f48, 44f)
4532:	FREG_FROB(f40,f42,f44,f46,f0, f2, f4, f6, f8)
454	STORE_SYNC(o0, f48)
455	FREG_FROB(f8, f10,f12,f14,f16,f18,f20,f22,f24)
456	STORE_JUMP(o0, f48, 52f)
4573:	FREG_FROB(f8, f10,f12,f14,f16,f18,f20,f22,f24)
458	STORE_SYNC(o0, f48)
459	FREG_FROB(f24,f26,f28,f30,f32,f34,f36,f38,f40)
460	STORE_JUMP(o0, f48, 60f)
461
4621:	FREG_FROB(f10,f12,f14,f16,f18,f20,f22,f24,f26)
463	LOOP_CHUNK1(o1, o0, 1f)
464	FREG_FROB(f26,f28,f30,f32,f34,f36,f38,f40,f42)
465	LOOP_CHUNK2(o1, o0, 2f)
466	FREG_FROB(f42,f44,f46,f0, f2, f4, f6, f8, f10)
467	LOOP_CHUNK3(o1, o0, 3f)
468	ba,pt		%xcc, 1b+4
469	 faligndata	%f10, %f12, %f48
4701:	FREG_FROB(f26,f28,f30,f32,f34,f36,f38,f40,f42)
471	STORE_SYNC(o0, f48)
472	FREG_FROB(f42,f44,f46,f0, f2, f4, f6, f8, f10)
473	STORE_JUMP(o0, f48, 45f)
4742:	FREG_FROB(f42,f44,f46,f0, f2, f4, f6, f8, f10)
475	STORE_SYNC(o0, f48)
476	FREG_FROB(f10,f12,f14,f16,f18,f20,f22,f24,f26)
477	STORE_JUMP(o0, f48, 53f)
4783:	FREG_FROB(f10,f12,f14,f16,f18,f20,f22,f24,f26)
479	STORE_SYNC(o0, f48)
480	FREG_FROB(f26,f28,f30,f32,f34,f36,f38,f40,f42)
481	STORE_JUMP(o0, f48, 61f)
482
4831:	FREG_FROB(f12,f14,f16,f18,f20,f22,f24,f26,f28)
484	LOOP_CHUNK1(o1, o0, 1f)
485	FREG_FROB(f28,f30,f32,f34,f36,f38,f40,f42,f44)
486	LOOP_CHUNK2(o1, o0, 2f)
487	FREG_FROB(f44,f46,f0, f2, f4, f6, f8, f10,f12)
488	LOOP_CHUNK3(o1, o0, 3f)
489	ba,pt		%xcc, 1b+4
490	 faligndata	%f12, %f14, %f48
4911:	FREG_FROB(f28,f30,f32,f34,f36,f38,f40,f42,f44)
492	STORE_SYNC(o0, f48)
493	FREG_FROB(f44,f46,f0, f2, f4, f6, f8, f10,f12)
494	STORE_JUMP(o0, f48, 46f)
4952:	FREG_FROB(f44,f46,f0, f2, f4, f6, f8, f10,f12)
496	STORE_SYNC(o0, f48)
497	FREG_FROB(f12,f14,f16,f18,f20,f22,f24,f26,f28)
498	STORE_JUMP(o0, f48, 54f)
4993:	FREG_FROB(f12,f14,f16,f18,f20,f22,f24,f26,f28)
500	STORE_SYNC(o0, f48)
501	FREG_FROB(f28,f30,f32,f34,f36,f38,f40,f42,f44)
502	STORE_JUMP(o0, f48, 62f)
503
5041:	FREG_FROB(f14,f16,f18,f20,f22,f24,f26,f28,f30)
505	LOOP_CHUNK1(o1, o0, 1f)
506	FREG_FROB(f30,f32,f34,f36,f38,f40,f42,f44,f46)
507	LOOP_CHUNK2(o1, o0, 2f)
508	FREG_FROB(f46,f0, f2, f4, f6, f8, f10,f12,f14)
509	LOOP_CHUNK3(o1, o0, 3f)
510	ba,pt		%xcc, 1b+4
511	 faligndata	%f14, %f16, %f48
5121:	FREG_FROB(f30,f32,f34,f36,f38,f40,f42,f44,f46)
513	STORE_SYNC(o0, f48)
514	FREG_FROB(f46,f0, f2, f4, f6, f8, f10,f12,f14)
515	STORE_JUMP(o0, f48, 47f)
5162:	FREG_FROB(f46,f0, f2, f4, f6, f8, f10,f12,f14)
517	STORE_SYNC(o0, f48)
518	FREG_FROB(f14,f16,f18,f20,f22,f24,f26,f28,f30)
519	STORE_JUMP(o0, f48, 55f)
5203:	FREG_FROB(f14,f16,f18,f20,f22,f24,f26,f28,f30)
521	STORE_SYNC(o0, f48)
522	FREG_FROB(f30,f32,f34,f36,f38,f40,f42,f44,f46)
523	STORE_JUMP(o0, f48, 63f)
524
52540:	FINISH_VISCHUNK(o0, f0,  f2)
52641:	FINISH_VISCHUNK(o0, f2,  f4)
52742:	FINISH_VISCHUNK(o0, f4,  f6)
52843:	FINISH_VISCHUNK(o0, f6,  f8)
52944:	FINISH_VISCHUNK(o0, f8,  f10)
53045:	FINISH_VISCHUNK(o0, f10, f12)
53146:	FINISH_VISCHUNK(o0, f12, f14)
53247:	UNEVEN_VISCHUNK(o0, f14, f0)
53348:	FINISH_VISCHUNK(o0, f16, f18)
53449:	FINISH_VISCHUNK(o0, f18, f20)
53550:	FINISH_VISCHUNK(o0, f20, f22)
53651:	FINISH_VISCHUNK(o0, f22, f24)
53752:	FINISH_VISCHUNK(o0, f24, f26)
53853:	FINISH_VISCHUNK(o0, f26, f28)
53954:	FINISH_VISCHUNK(o0, f28, f30)
54055:	UNEVEN_VISCHUNK(o0, f30, f0)
54156:	FINISH_VISCHUNK(o0, f32, f34)
54257:	FINISH_VISCHUNK(o0, f34, f36)
54358:	FINISH_VISCHUNK(o0, f36, f38)
54459:	FINISH_VISCHUNK(o0, f38, f40)
54560:	FINISH_VISCHUNK(o0, f40, f42)
54661:	FINISH_VISCHUNK(o0, f42, f44)
54762:	FINISH_VISCHUNK(o0, f44, f46)
54863:	UNEVEN_VISCHUNK_LAST(o0, f46, f0)
549
55093:	EX_LD_FP(LOAD(ldd, %o1, %f2), U1_g3_0_fp)
551	add		%o1, 8, %o1
552	subcc		%g3, 8, %g3
553	faligndata	%f0, %f2, %f8
554	EX_ST_FP(STORE(std, %f8, %o0), U1_g3_8_fp)
555	bl,pn		%xcc, 95f
556	 add		%o0, 8, %o0
557	EX_LD_FP(LOAD(ldd, %o1, %f0), U1_g3_0_fp)
558	add		%o1, 8, %o1
559	subcc		%g3, 8, %g3
560	faligndata	%f2, %f0, %f8
561	EX_ST_FP(STORE(std, %f8, %o0), U1_g3_8_fp)
562	bge,pt		%xcc, 93b
563	 add		%o0, 8, %o0
564
56595:	brz,pt		%o2, 2f
566	 mov		%g1, %o1
567
5681:	EX_LD_FP(LOAD(ldub, %o1, %o3), U1_o2_0_fp)
569	add		%o1, 1, %o1
570	subcc		%o2, 1, %o2
571	EX_ST_FP(STORE(stb, %o3, %o0), U1_o2_1_fp)
572	bne,pt		%xcc, 1b
573	 add		%o0, 1, %o0
574
5752:	membar		#StoreLoad | #StoreStore
576	VISExit
577	retl
578	 mov		EX_RETVAL(%o4), %o0
579
580	.align		64
58170:	/* 16 < len <= (5 * 64) */
582	bne,pn		%XCC, 75f
583	 sub		%o0, %o1, %o3
584
58572:	andn		%o2, 0xf, %GLOBAL_SPARE
586	and		%o2, 0xf, %o2
5871:	EX_LD(LOAD(ldx, %o1 + 0x00, %o5), U1_gs_0)
588	EX_LD(LOAD(ldx, %o1 + 0x08, %g1), U1_gs_0)
589	subcc		%GLOBAL_SPARE, 0x10, %GLOBAL_SPARE
590	EX_ST(STORE(stx, %o5, %o1 + %o3), U1_gs_10)
591	add		%o1, 0x8, %o1
592	EX_ST(STORE(stx, %g1, %o1 + %o3), U1_gs_8)
593	bgu,pt		%XCC, 1b
594	 add		%o1, 0x8, %o1
59573:	andcc		%o2, 0x8, %g0
596	be,pt		%XCC, 1f
597	 nop
598	EX_LD(LOAD(ldx, %o1, %o5), U1_o2_0)
599	sub		%o2, 0x8, %o2
600	EX_ST(STORE(stx, %o5, %o1 + %o3), U1_o2_8)
601	add		%o1, 0x8, %o1
6021:	andcc		%o2, 0x4, %g0
603	be,pt		%XCC, 1f
604	 nop
605	EX_LD(LOAD(lduw, %o1, %o5), U1_o2_0)
606	sub		%o2, 0x4, %o2
607	EX_ST(STORE(stw, %o5, %o1 + %o3), U1_o2_4)
608	add		%o1, 0x4, %o1
6091:	cmp		%o2, 0
610	be,pt		%XCC, 85f
611	 nop
612	ba,pt		%xcc, 90f
613	 nop
614
61575:	andcc		%o0, 0x7, %g1
616	sub		%g1, 0x8, %g1
617	be,pn		%icc, 2f
618	 sub		%g0, %g1, %g1
619	sub		%o2, %g1, %o2
620
6211:	EX_LD(LOAD(ldub, %o1, %o5), U1_g1_0)
622	subcc		%g1, 1, %g1
623	EX_ST(STORE(stb, %o5, %o1 + %o3), U1_g1_1)
624	bgu,pt		%icc, 1b
625	 add		%o1, 1, %o1
626
6272:	add		%o1, %o3, %o0
628	andcc		%o1, 0x7, %g1
629	bne,pt		%icc, 8f
630	 sll		%g1, 3, %g1
631
632	cmp		%o2, 16
633	bgeu,pt		%icc, 72b
634	 nop
635	ba,a,pt		%xcc, 73b
636
6378:	mov		64, %o3
638	andn		%o1, 0x7, %o1
639	EX_LD(LOAD(ldx, %o1, %g2), U1_o2_0)
640	sub		%o3, %g1, %o3
641	andn		%o2, 0x7, %GLOBAL_SPARE
642	sllx		%g2, %g1, %g2
6431:	EX_LD(LOAD(ldx, %o1 + 0x8, %g3), U1_gs_0_o2_adj)
644	subcc		%GLOBAL_SPARE, 0x8, %GLOBAL_SPARE
645	add		%o1, 0x8, %o1
646	srlx		%g3, %o3, %o5
647	or		%o5, %g2, %o5
648	EX_ST(STORE(stx, %o5, %o0), U1_gs_8_o2_adj)
649	add		%o0, 0x8, %o0
650	bgu,pt		%icc, 1b
651	 sllx		%g3, %g1, %g2
652
653	srl		%g1, 3, %g1
654	andcc		%o2, 0x7, %o2
655	be,pn		%icc, 85f
656	 add		%o1, %g1, %o1
657	ba,pt		%xcc, 90f
658	 sub		%o0, %o1, %o3
659
660	.align		64
66180:	/* 0 < len <= 16 */
662	andcc		%o3, 0x3, %g0
663	bne,pn		%XCC, 90f
664	 sub		%o0, %o1, %o3
665
6661:	EX_LD(LOAD(lduw, %o1, %g1), U1_o2_0)
667	subcc		%o2, 4, %o2
668	EX_ST(STORE(stw, %g1, %o1 + %o3), U1_o2_4)
669	bgu,pt		%XCC, 1b
670	 add		%o1, 4, %o1
671
67285:	retl
673	 mov		EX_RETVAL(%o4), %o0
674
675	.align		32
67690:	EX_LD(LOAD(ldub, %o1, %g1), U1_o2_0)
677	subcc		%o2, 1, %o2
678	EX_ST(STORE(stb, %g1, %o1 + %o3), U1_o2_1)
679	bgu,pt		%XCC, 90b
680	 add		%o1, 1, %o1
681	retl
682	 mov		EX_RETVAL(%o4), %o0
683
684	.size		FUNC_NAME, .-FUNC_NAME
685EXPORT_SYMBOL(FUNC_NAME)
686