xref: /openbmc/linux/arch/mips/lib/memset.S (revision 1c2dd16a)
1/*
2 * This file is subject to the terms and conditions of the GNU General Public
3 * License.  See the file "COPYING" in the main directory of this archive
4 * for more details.
5 *
6 * Copyright (C) 1998, 1999, 2000 by Ralf Baechle
7 * Copyright (C) 1999, 2000 Silicon Graphics, Inc.
8 * Copyright (C) 2007 by Maciej W. Rozycki
9 * Copyright (C) 2011, 2012 MIPS Technologies, Inc.
10 */
11#include <asm/asm.h>
12#include <asm/asm-offsets.h>
13#include <asm/export.h>
14#include <asm/regdef.h>
15
16#if LONGSIZE == 4
17#define LONG_S_L swl
18#define LONG_S_R swr
19#else
20#define LONG_S_L sdl
21#define LONG_S_R sdr
22#endif
23
24#ifdef CONFIG_CPU_MICROMIPS
25#define STORSIZE (LONGSIZE * 2)
26#define STORMASK (STORSIZE - 1)
27#define FILL64RG t8
28#define FILLPTRG t7
29#undef  LONG_S
30#define LONG_S LONG_SP
31#else
32#define STORSIZE LONGSIZE
33#define STORMASK LONGMASK
34#define FILL64RG a1
35#define FILLPTRG t0
36#endif
37
38#define LEGACY_MODE 1
39#define EVA_MODE    2
40
41/*
42 * No need to protect it with EVA #ifdefery. The generated block of code
43 * will never be assembled if EVA is not enabled.
44 */
45#define __EVAFY(insn, reg, addr) __BUILD_EVA_INSN(insn##e, reg, addr)
46#define ___BUILD_EVA_INSN(insn, reg, addr) __EVAFY(insn, reg, addr)
47
48#define EX(insn,reg,addr,handler)			\
49	.if \mode == LEGACY_MODE;			\
509:		insn	reg, addr;			\
51	.else;						\
529:		___BUILD_EVA_INSN(insn, reg, addr);	\
53	.endif;						\
54	.section __ex_table,"a";			\
55	PTR	9b, handler;				\
56	.previous
57
58	.macro	f_fill64 dst, offset, val, fixup, mode
59	EX(LONG_S, \val, (\offset +  0 * STORSIZE)(\dst), \fixup)
60	EX(LONG_S, \val, (\offset +  1 * STORSIZE)(\dst), \fixup)
61	EX(LONG_S, \val, (\offset +  2 * STORSIZE)(\dst), \fixup)
62	EX(LONG_S, \val, (\offset +  3 * STORSIZE)(\dst), \fixup)
63#if ((defined(CONFIG_CPU_MICROMIPS) && (LONGSIZE == 4)) || !defined(CONFIG_CPU_MICROMIPS))
64	EX(LONG_S, \val, (\offset +  4 * STORSIZE)(\dst), \fixup)
65	EX(LONG_S, \val, (\offset +  5 * STORSIZE)(\dst), \fixup)
66	EX(LONG_S, \val, (\offset +  6 * STORSIZE)(\dst), \fixup)
67	EX(LONG_S, \val, (\offset +  7 * STORSIZE)(\dst), \fixup)
68#endif
69#if (!defined(CONFIG_CPU_MICROMIPS) && (LONGSIZE == 4))
70	EX(LONG_S, \val, (\offset +  8 * STORSIZE)(\dst), \fixup)
71	EX(LONG_S, \val, (\offset +  9 * STORSIZE)(\dst), \fixup)
72	EX(LONG_S, \val, (\offset + 10 * STORSIZE)(\dst), \fixup)
73	EX(LONG_S, \val, (\offset + 11 * STORSIZE)(\dst), \fixup)
74	EX(LONG_S, \val, (\offset + 12 * STORSIZE)(\dst), \fixup)
75	EX(LONG_S, \val, (\offset + 13 * STORSIZE)(\dst), \fixup)
76	EX(LONG_S, \val, (\offset + 14 * STORSIZE)(\dst), \fixup)
77	EX(LONG_S, \val, (\offset + 15 * STORSIZE)(\dst), \fixup)
78#endif
79	.endm
80
81	.set	noreorder
82	.align	5
83
84	/*
85	 * Macro to generate the __bzero{,_user} symbol
86	 * Arguments:
87	 * mode: LEGACY_MODE or EVA_MODE
88	 */
89	.macro __BUILD_BZERO mode
90	/* Initialize __memset if this is the first time we call this macro */
91	.ifnotdef __memset
92	.set __memset, 1
93	.hidden __memset /* Make sure it does not leak */
94	.endif
95
96	sltiu		t0, a2, STORSIZE	/* very small region? */
97	bnez		t0, .Lsmall_memset\@
98	andi		t0, a0, STORMASK	/* aligned? */
99
100#ifdef CONFIG_CPU_MICROMIPS
101	move		t8, a1			/* used by 'swp' instruction */
102	move		t9, a1
103#endif
104#ifndef CONFIG_CPU_DADDI_WORKAROUNDS
105	beqz		t0, 1f
106	PTR_SUBU	t0, STORSIZE		/* alignment in bytes */
107#else
108	.set		noat
109	li		AT, STORSIZE
110	beqz		t0, 1f
111	PTR_SUBU	t0, AT			/* alignment in bytes */
112	.set		at
113#endif
114
115#ifndef CONFIG_CPU_MIPSR6
116	R10KCBARRIER(0(ra))
117#ifdef __MIPSEB__
118	EX(LONG_S_L, a1, (a0), .Lfirst_fixup\@)	/* make word/dword aligned */
119#else
120	EX(LONG_S_R, a1, (a0), .Lfirst_fixup\@)	/* make word/dword aligned */
121#endif
122	PTR_SUBU	a0, t0			/* long align ptr */
123	PTR_ADDU	a2, t0			/* correct size */
124
125#else /* CONFIG_CPU_MIPSR6 */
126#define STORE_BYTE(N)				\
127	EX(sb, a1, N(a0), .Lbyte_fixup\@);	\
128	beqz		t0, 0f;			\
129	PTR_ADDU	t0, 1;
130
131	PTR_ADDU	a2, t0			/* correct size */
132	PTR_ADDU	t0, 1
133	STORE_BYTE(0)
134	STORE_BYTE(1)
135#if LONGSIZE == 4
136	EX(sb, a1, 2(a0), .Lbyte_fixup\@)
137#else
138	STORE_BYTE(2)
139	STORE_BYTE(3)
140	STORE_BYTE(4)
141	STORE_BYTE(5)
142	EX(sb, a1, 6(a0), .Lbyte_fixup\@)
143#endif
1440:
145	ori		a0, STORMASK
146	xori		a0, STORMASK
147	PTR_ADDIU	a0, STORSIZE
148#endif /* CONFIG_CPU_MIPSR6 */
1491:	ori		t1, a2, 0x3f		/* # of full blocks */
150	xori		t1, 0x3f
151	beqz		t1, .Lmemset_partial\@	/* no block to fill */
152	andi		t0, a2, 0x40-STORSIZE
153
154	PTR_ADDU	t1, a0			/* end address */
155	.set		reorder
1561:	PTR_ADDIU	a0, 64
157	R10KCBARRIER(0(ra))
158	f_fill64 a0, -64, FILL64RG, .Lfwd_fixup\@, \mode
159	bne		t1, a0, 1b
160	.set		noreorder
161
162.Lmemset_partial\@:
163	R10KCBARRIER(0(ra))
164	PTR_LA		t1, 2f			/* where to start */
165#ifdef CONFIG_CPU_MICROMIPS
166	LONG_SRL	t7, t0, 1
167#endif
168#if LONGSIZE == 4
169	PTR_SUBU	t1, FILLPTRG
170#else
171	.set		noat
172	LONG_SRL	AT, FILLPTRG, 1
173	PTR_SUBU	t1, AT
174	.set		at
175#endif
176	jr		t1
177	PTR_ADDU	a0, t0			/* dest ptr */
178
179	.set		push
180	.set		noreorder
181	.set		nomacro
182	/* ... but first do longs ... */
183	f_fill64 a0, -64, FILL64RG, .Lpartial_fixup\@, \mode
1842:	.set		pop
185	andi		a2, STORMASK		/* At most one long to go */
186
187	beqz		a2, 1f
188#ifndef CONFIG_CPU_MIPSR6
189	PTR_ADDU	a0, a2			/* What's left */
190	R10KCBARRIER(0(ra))
191#ifdef __MIPSEB__
192	EX(LONG_S_R, a1, -1(a0), .Llast_fixup\@)
193#else
194	EX(LONG_S_L, a1, -1(a0), .Llast_fixup\@)
195#endif
196#else
197	PTR_SUBU	t0, $0, a2
198	PTR_ADDIU	t0, 1
199	STORE_BYTE(0)
200	STORE_BYTE(1)
201#if LONGSIZE == 4
202	EX(sb, a1, 2(a0), .Lbyte_fixup\@)
203#else
204	STORE_BYTE(2)
205	STORE_BYTE(3)
206	STORE_BYTE(4)
207	STORE_BYTE(5)
208	EX(sb, a1, 6(a0), .Lbyte_fixup\@)
209#endif
2100:
211#endif
2121:	jr		ra
213	move		a2, zero
214
215.Lsmall_memset\@:
216	beqz		a2, 2f
217	PTR_ADDU	t1, a0, a2
218
2191:	PTR_ADDIU	a0, 1			/* fill bytewise */
220	R10KCBARRIER(0(ra))
221	bne		t1, a0, 1b
222	sb		a1, -1(a0)
223
2242:	jr		ra			/* done */
225	move		a2, zero
226	.if __memset == 1
227	END(memset)
228	.set __memset, 0
229	.hidden __memset
230	.endif
231
232#ifdef CONFIG_CPU_MIPSR6
233.Lbyte_fixup\@:
234	PTR_SUBU	a2, $0, t0
235	jr		ra
236	 PTR_ADDIU	a2, 1
237#endif /* CONFIG_CPU_MIPSR6 */
238
239.Lfirst_fixup\@:
240	jr	ra
241	nop
242
243.Lfwd_fixup\@:
244	PTR_L		t0, TI_TASK($28)
245	andi		a2, 0x3f
246	LONG_L		t0, THREAD_BUADDR(t0)
247	LONG_ADDU	a2, t1
248	jr		ra
249	LONG_SUBU	a2, t0
250
251.Lpartial_fixup\@:
252	PTR_L		t0, TI_TASK($28)
253	andi		a2, STORMASK
254	LONG_L		t0, THREAD_BUADDR(t0)
255	LONG_ADDU	a2, t1
256	jr		ra
257	LONG_SUBU	a2, t0
258
259.Llast_fixup\@:
260	jr		ra
261	andi		v1, a2, STORMASK
262
263	.endm
264
265/*
266 * memset(void *s, int c, size_t n)
267 *
268 * a0: start of area to clear
269 * a1: char to fill with
270 * a2: size of area to clear
271 */
272
273LEAF(memset)
274EXPORT_SYMBOL(memset)
275	beqz		a1, 1f
276	move		v0, a0			/* result */
277
278	andi		a1, 0xff		/* spread fillword */
279	LONG_SLL		t1, a1, 8
280	or		a1, t1
281	LONG_SLL		t1, a1, 16
282#if LONGSIZE == 8
283	or		a1, t1
284	LONG_SLL		t1, a1, 32
285#endif
286	or		a1, t1
2871:
288#ifndef CONFIG_EVA
289FEXPORT(__bzero)
290EXPORT_SYMBOL(__bzero)
291#else
292FEXPORT(__bzero_kernel)
293EXPORT_SYMBOL(__bzero_kernel)
294#endif
295	__BUILD_BZERO LEGACY_MODE
296
297#ifdef CONFIG_EVA
298LEAF(__bzero)
299EXPORT_SYMBOL(__bzero)
300	__BUILD_BZERO EVA_MODE
301END(__bzero)
302#endif
303