xref: /openbmc/linux/arch/sparc/lib/copy_user.S (revision 55fd7e02) 
1/* SPDX-License-Identifier: GPL-2.0 */
2/* copy_user.S: Sparc optimized copy_from_user and copy_to_user code.
3 *
4 *  Copyright(C) 1995 Linus Torvalds
5 *  Copyright(C) 1996 David S. Miller
6 *  Copyright(C) 1996 Eddie C. Dost
7 *  Copyright(C) 1996,1998 Jakub Jelinek
8 *
9 * derived from:
10 *	e-mail between David and Eddie.
11 *
12 * Returns 0 if successful, otherwise count of bytes not copied yet
13 */
14
15#include <asm/ptrace.h>
16#include <asm/asmmacro.h>
17#include <asm/page.h>
18#include <asm/thread_info.h>
19#include <asm/export.h>
20
21/* Work around cpp -rob */
22#define ALLOC #alloc
23#define EXECINSTR #execinstr
24#define EX(x,y,a,b) 				\
2598: 	x,y;					\
26	.section .fixup,ALLOC,EXECINSTR;	\
27	.align	4;				\
2899:	ba fixupretl;				\
29	 a, b, %g3;				\
30	.section __ex_table,ALLOC;		\
31	.align	4;				\
32	.word	98b, 99b;			\
33	.text;					\
34	.align	4
35
36#define EX2(x,y,c,d,e,a,b) 			\
3798: 	x,y;					\
38	.section .fixup,ALLOC,EXECINSTR;	\
39	.align	4;				\
4099:	c, d, e;				\
41	ba fixupretl;				\
42	 a, b, %g3;				\
43	.section __ex_table,ALLOC;		\
44	.align	4;				\
45	.word	98b, 99b;			\
46	.text;					\
47	.align	4
48
49#define EXO2(x,y) 				\
5098: 	x, y;					\
51	.section __ex_table,ALLOC;		\
52	.align	4;				\
53	.word	98b, 97f;			\
54	.text;					\
55	.align	4
56
57#define EXT(start,end,handler)			\
58	.section __ex_table,ALLOC;		\
59	.align	4;				\
60	.word	start, 0, end, handler;		\
61	.text;					\
62	.align	4
63
64/* Please do not change following macros unless you change logic used
65 * in .fixup at the end of this file as well
66 */
67
68/* Both these macros have to start with exactly the same insn */
69#define MOVE_BIGCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \
70	ldd	[%src + (offset) + 0x00], %t0; \
71	ldd	[%src + (offset) + 0x08], %t2; \
72	ldd	[%src + (offset) + 0x10], %t4; \
73	ldd	[%src + (offset) + 0x18], %t6; \
74	st	%t0, [%dst + (offset) + 0x00]; \
75	st	%t1, [%dst + (offset) + 0x04]; \
76	st	%t2, [%dst + (offset) + 0x08]; \
77	st	%t3, [%dst + (offset) + 0x0c]; \
78	st	%t4, [%dst + (offset) + 0x10]; \
79	st	%t5, [%dst + (offset) + 0x14]; \
80	st	%t6, [%dst + (offset) + 0x18]; \
81	st	%t7, [%dst + (offset) + 0x1c];
82
83#define MOVE_BIGALIGNCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \
84	ldd	[%src + (offset) + 0x00], %t0; \
85	ldd	[%src + (offset) + 0x08], %t2; \
86	ldd	[%src + (offset) + 0x10], %t4; \
87	ldd	[%src + (offset) + 0x18], %t6; \
88	std	%t0, [%dst + (offset) + 0x00]; \
89	std	%t2, [%dst + (offset) + 0x08]; \
90	std	%t4, [%dst + (offset) + 0x10]; \
91	std	%t6, [%dst + (offset) + 0x18];
92
93#define MOVE_LASTCHUNK(src, dst, offset, t0, t1, t2, t3) \
94	ldd	[%src - (offset) - 0x10], %t0; \
95	ldd	[%src - (offset) - 0x08], %t2; \
96	st	%t0, [%dst - (offset) - 0x10]; \
97	st	%t1, [%dst - (offset) - 0x0c]; \
98	st	%t2, [%dst - (offset) - 0x08]; \
99	st	%t3, [%dst - (offset) - 0x04];
100
101#define MOVE_HALFCHUNK(src, dst, offset, t0, t1, t2, t3) \
102	lduh	[%src + (offset) + 0x00], %t0; \
103	lduh	[%src + (offset) + 0x02], %t1; \
104	lduh	[%src + (offset) + 0x04], %t2; \
105	lduh	[%src + (offset) + 0x06], %t3; \
106	sth	%t0, [%dst + (offset) + 0x00]; \
107	sth	%t1, [%dst + (offset) + 0x02]; \
108	sth	%t2, [%dst + (offset) + 0x04]; \
109	sth	%t3, [%dst + (offset) + 0x06];
110
111#define MOVE_SHORTCHUNK(src, dst, offset, t0, t1) \
112	ldub	[%src - (offset) - 0x02], %t0; \
113	ldub	[%src - (offset) - 0x01], %t1; \
114	stb	%t0, [%dst - (offset) - 0x02]; \
115	stb	%t1, [%dst - (offset) - 0x01];
116
117	.text
118	.align	4
119
120	.globl  __copy_user_begin
121__copy_user_begin:
122
123	.globl	__copy_user
124	EXPORT_SYMBOL(__copy_user)
125dword_align:
126	andcc	%o1, 1, %g0
127	be	4f
128	 andcc	%o1, 2, %g0
129
130	EXO2(ldub [%o1], %g2)
131	add	%o1, 1, %o1
132	EXO2(stb %g2, [%o0])
133	sub	%o2, 1, %o2
134	bne	3f
135	 add	%o0, 1, %o0
136
137	EXO2(lduh [%o1], %g2)
138	add	%o1, 2, %o1
139	EXO2(sth %g2, [%o0])
140	sub	%o2, 2, %o2
141	b	3f
142	 add	%o0, 2, %o0
1434:
144	EXO2(lduh [%o1], %g2)
145	add	%o1, 2, %o1
146	EXO2(sth %g2, [%o0])
147	sub	%o2, 2, %o2
148	b	3f
149	 add	%o0, 2, %o0
150
151__copy_user:	/* %o0=dst %o1=src %o2=len */
152	xor	%o0, %o1, %o4
1531:
154	andcc	%o4, 3, %o5
1552:
156	bne	cannot_optimize
157	 cmp	%o2, 15
158
159	bleu	short_aligned_end
160	 andcc	%o1, 3, %g0
161
162	bne	dword_align
1633:
164	 andcc	%o1, 4, %g0
165
166	be	2f
167	 mov	%o2, %g1
168
169	EXO2(ld [%o1], %o4)
170	sub	%g1, 4, %g1
171	EXO2(st %o4, [%o0])
172	add	%o1, 4, %o1
173	add	%o0, 4, %o0
1742:
175	andcc	%g1, 0xffffff80, %g7
176	be	3f
177	 andcc	%o0, 4, %g0
178
179	be	ldd_std + 4
1805:
181	MOVE_BIGCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5)
182	MOVE_BIGCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5)
183	MOVE_BIGCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5)
184	MOVE_BIGCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5)
18580:
186	EXT(5b, 80b, 50f)
187	subcc	%g7, 128, %g7
188	add	%o1, 128, %o1
189	bne	5b
190	 add	%o0, 128, %o0
1913:
192	andcc	%g1, 0x70, %g7
193	be	copy_user_table_end
194	 andcc	%g1, 8, %g0
195
196	sethi	%hi(copy_user_table_end), %o5
197	srl	%g7, 1, %o4
198	add	%g7, %o4, %o4
199	add	%o1, %g7, %o1
200	sub	%o5, %o4, %o5
201	jmpl	%o5 + %lo(copy_user_table_end), %g0
202	 add	%o0, %g7, %o0
203
204copy_user_table:
205	MOVE_LASTCHUNK(o1, o0, 0x60, g2, g3, g4, g5)
206	MOVE_LASTCHUNK(o1, o0, 0x50, g2, g3, g4, g5)
207	MOVE_LASTCHUNK(o1, o0, 0x40, g2, g3, g4, g5)
208	MOVE_LASTCHUNK(o1, o0, 0x30, g2, g3, g4, g5)
209	MOVE_LASTCHUNK(o1, o0, 0x20, g2, g3, g4, g5)
210	MOVE_LASTCHUNK(o1, o0, 0x10, g2, g3, g4, g5)
211	MOVE_LASTCHUNK(o1, o0, 0x00, g2, g3, g4, g5)
212copy_user_table_end:
213	EXT(copy_user_table, copy_user_table_end, 51f)
214	be	copy_user_last7
215	 andcc	%g1, 4, %g0
216
217	EX(ldd	[%o1], %g2, and %g1, 0xf)
218	add	%o0, 8, %o0
219	add	%o1, 8, %o1
220	EX(st	%g2, [%o0 - 0x08], and %g1, 0xf)
221	EX2(st	%g3, [%o0 - 0x04], and %g1, 0xf, %g1, sub %g1, 4)
222copy_user_last7:
223	be	1f
224	 andcc	%g1, 2, %g0
225
226	EX(ld	[%o1], %g2, and %g1, 7)
227	add	%o1, 4, %o1
228	EX(st	%g2, [%o0], and %g1, 7)
229	add	%o0, 4, %o0
2301:
231	be	1f
232	 andcc	%g1, 1, %g0
233
234	EX(lduh	[%o1], %g2, and %g1, 3)
235	add	%o1, 2, %o1
236	EX(sth	%g2, [%o0], and %g1, 3)
237	add	%o0, 2, %o0
2381:
239	be	1f
240	 nop
241
242	EX(ldub	[%o1], %g2, add %g0, 1)
243	EX(stb	%g2, [%o0], add %g0, 1)
2441:
245	retl
246 	 clr	%o0
247
248ldd_std:
249	MOVE_BIGALIGNCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5)
250	MOVE_BIGALIGNCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5)
251	MOVE_BIGALIGNCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5)
252	MOVE_BIGALIGNCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5)
25381:
254	EXT(ldd_std, 81b, 52f)
255	subcc	%g7, 128, %g7
256	add	%o1, 128, %o1
257	bne	ldd_std
258	 add	%o0, 128, %o0
259
260	andcc	%g1, 0x70, %g7
261	be	copy_user_table_end
262	 andcc	%g1, 8, %g0
263
264	sethi	%hi(copy_user_table_end), %o5
265	srl	%g7, 1, %o4
266	add	%g7, %o4, %o4
267	add	%o1, %g7, %o1
268	sub	%o5, %o4, %o5
269	jmpl	%o5 + %lo(copy_user_table_end), %g0
270	 add	%o0, %g7, %o0
271
272cannot_optimize:
273	bleu	short_end
274	 cmp	%o5, 2
275
276	bne	byte_chunk
277	 and	%o2, 0xfffffff0, %o3
278
279	andcc	%o1, 1, %g0
280	be	10f
281	 nop
282
283	EXO2(ldub [%o1], %g2)
284	add	%o1, 1, %o1
285	EXO2(stb %g2, [%o0])
286	sub	%o2, 1, %o2
287	andcc	%o2, 0xfffffff0, %o3
288	be	short_end
289	 add	%o0, 1, %o0
29010:
291	MOVE_HALFCHUNK(o1, o0, 0x00, g2, g3, g4, g5)
292	MOVE_HALFCHUNK(o1, o0, 0x08, g2, g3, g4, g5)
29382:
294	EXT(10b, 82b, 53f)
295	subcc	%o3, 0x10, %o3
296	add	%o1, 0x10, %o1
297	bne	10b
298	 add	%o0, 0x10, %o0
299	b	2f
300	 and	%o2, 0xe, %o3
301
302byte_chunk:
303	MOVE_SHORTCHUNK(o1, o0, -0x02, g2, g3)
304	MOVE_SHORTCHUNK(o1, o0, -0x04, g2, g3)
305	MOVE_SHORTCHUNK(o1, o0, -0x06, g2, g3)
306	MOVE_SHORTCHUNK(o1, o0, -0x08, g2, g3)
307	MOVE_SHORTCHUNK(o1, o0, -0x0a, g2, g3)
308	MOVE_SHORTCHUNK(o1, o0, -0x0c, g2, g3)
309	MOVE_SHORTCHUNK(o1, o0, -0x0e, g2, g3)
310	MOVE_SHORTCHUNK(o1, o0, -0x10, g2, g3)
31183:
312	EXT(byte_chunk, 83b, 54f)
313	subcc	%o3, 0x10, %o3
314	add	%o1, 0x10, %o1
315	bne	byte_chunk
316	 add	%o0, 0x10, %o0
317
318short_end:
319	and	%o2, 0xe, %o3
3202:
321	sethi	%hi(short_table_end), %o5
322	sll	%o3, 3, %o4
323	add	%o0, %o3, %o0
324	sub	%o5, %o4, %o5
325	add	%o1, %o3, %o1
326	jmpl	%o5 + %lo(short_table_end), %g0
327	 andcc	%o2, 1, %g0
32884:
329	MOVE_SHORTCHUNK(o1, o0, 0x0c, g2, g3)
330	MOVE_SHORTCHUNK(o1, o0, 0x0a, g2, g3)
331	MOVE_SHORTCHUNK(o1, o0, 0x08, g2, g3)
332	MOVE_SHORTCHUNK(o1, o0, 0x06, g2, g3)
333	MOVE_SHORTCHUNK(o1, o0, 0x04, g2, g3)
334	MOVE_SHORTCHUNK(o1, o0, 0x02, g2, g3)
335	MOVE_SHORTCHUNK(o1, o0, 0x00, g2, g3)
336short_table_end:
337	EXT(84b, short_table_end, 55f)
338	be	1f
339	 nop
340	EX(ldub	[%o1], %g2, add %g0, 1)
341	EX(stb	%g2, [%o0], add %g0, 1)
3421:
343	retl
344 	 clr	%o0
345
346short_aligned_end:
347	bne	short_end
348	 andcc	%o2, 8, %g0
349
350	be	1f
351	 andcc	%o2, 4, %g0
352
353	EXO2(ld	[%o1 + 0x00], %g2)
354	EXO2(ld	[%o1 + 0x04], %g3)
355	add	%o1, 8, %o1
356	EXO2(st	%g2, [%o0 + 0x00])
357	EX(st	%g3, [%o0 + 0x04], sub %o2, 4)
358	add	%o0, 8, %o0
3591:
360	b	copy_user_last7
361	 mov	%o2, %g1
362
363	.section .fixup,#alloc,#execinstr
364	.align	4
36597:
366	mov	%o2, %g3
367fixupretl:
368	retl
369	 mov	%g3, %o0
370
371/* exception routine sets %g2 to (broken_insn - first_insn)>>2 */
37250:
373/* This magic counts how many bytes are left when crash in MOVE_BIGCHUNK
374 * happens. This is derived from the amount ldd reads, st stores, etc.
375 * x = g2 % 12;
376 * g3 = g1 + g7 - ((g2 / 12) * 32 + (x < 4) ? 0 : (x - 4) * 4);
377 * o0 += (g2 / 12) * 32;
378 */
379	cmp	%g2, 12
380	add	%o0, %g7, %o0
381	bcs	1f
382	 cmp	%g2, 24
383	bcs	2f
384	 cmp	%g2, 36
385	bcs	3f
386	 nop
387	sub	%g2, 12, %g2
388	sub	%g7, 32, %g7
3893:	sub	%g2, 12, %g2
390	sub	%g7, 32, %g7
3912:	sub	%g2, 12, %g2
392	sub	%g7, 32, %g7
3931:	cmp	%g2, 4
394	bcs,a	60f
395	 clr	%g2
396	sub	%g2, 4, %g2
397	sll	%g2, 2, %g2
39860:	and	%g1, 0x7f, %g3
399	sub	%o0, %g7, %o0
400	add	%g3, %g7, %g3
401	ba	fixupretl
402	 sub	%g3, %g2, %g3
40351:
404/* i = 41 - g2; j = i % 6;
405 * g3 = (g1 & 15) + (i / 6) * 16 + (j < 4) ? (j + 1) * 4 : 16;
406 * o0 -= (i / 6) * 16 + 16;
407 */
408	neg	%g2
409	and	%g1, 0xf, %g1
410	add	%g2, 41, %g2
411	add	%o0, %g1, %o0
4121:	cmp	%g2, 6
413	bcs,a	2f
414	 cmp	%g2, 4
415	add	%g1, 16, %g1
416	b	1b
417	 sub	%g2, 6, %g2
4182:	bcc,a	2f
419	 mov	16, %g2
420	inc	%g2
421	sll	%g2, 2, %g2
4222:	add	%g1, %g2, %g3
423	ba	fixupretl
424	 sub	%o0, %g3, %o0
42552:
426/* g3 = g1 + g7 - (g2 / 8) * 32 + (g2 & 4) ? (g2 & 3) * 8 : 0;
427   o0 += (g2 / 8) * 32 */
428	andn	%g2, 7, %g4
429	add	%o0, %g7, %o0
430	andcc	%g2, 4, %g0
431	and	%g2, 3, %g2
432	sll	%g4, 2, %g4
433	sll	%g2, 3, %g2
434	bne	60b
435	 sub	%g7, %g4, %g7
436	ba	60b
437	 clr	%g2
43853:
439/* g3 = o3 + (o2 & 15) - (g2 & 8) - (g2 & 4) ? (g2 & 3) * 2 : 0;
440   o0 += (g2 & 8) */
441	and	%g2, 3, %g4
442	andcc	%g2, 4, %g0
443	and	%g2, 8, %g2
444	sll	%g4, 1, %g4
445	be	1f
446	 add	%o0, %g2, %o0
447	add	%g2, %g4, %g2
4481:	and	%o2, 0xf, %g3
449	add	%g3, %o3, %g3
450	ba	fixupretl
451	 sub	%g3, %g2, %g3
45254:
453/* g3 = o3 + (o2 & 15) - (g2 / 4) * 2 - (g2 & 2) ? (g2 & 1) : 0;
454   o0 += (g2 / 4) * 2 */
455	srl	%g2, 2, %o4
456	and	%g2, 1, %o5
457	srl	%g2, 1, %g2
458	add	%o4, %o4, %o4
459	and	%o5, %g2, %o5
460	and	%o2, 0xf, %o2
461	add	%o0, %o4, %o0
462	sub	%o3, %o5, %o3
463	sub	%o2, %o4, %o2
464	ba	fixupretl
465	 add	%o2, %o3, %g3
46655:
467/* i = 27 - g2;
468   g3 = (o2 & 1) + i / 4 * 2 + !(i & 3);
469   o0 -= i / 4 * 2 + 1 */
470	neg	%g2
471	and	%o2, 1, %o2
472	add	%g2, 27, %g2
473	srl	%g2, 2, %o5
474	andcc	%g2, 3, %g0
475	mov	1, %g2
476	add	%o5, %o5, %o5
477	be,a	1f
478	 clr	%g2
4791:	add	%g2, %o5, %g3
480	sub	%o0, %g3, %o0
481	ba	fixupretl
482	 add	%g3, %o2, %g3
483
484	.globl  __copy_user_end
485__copy_user_end:
486