xref: /openbmc/linux/arch/arm/mm/proc-arm1020e.S (revision 9c1f8594)
1/*
2 *  linux/arch/arm/mm/proc-arm1020e.S: MMU functions for ARM1020
3 *
4 *  Copyright (C) 2000 ARM Limited
5 *  Copyright (C) 2000 Deep Blue Solutions Ltd.
6 *  hacked for non-paged-MM by Hyok S. Choi, 2003.
7 *
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License as published by
10 * the Free Software Foundation; either version 2 of the License, or
11 * (at your option) any later version.
12 *
13 * This program is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
16 * GNU General Public License for more details.
17 *
18 * You should have received a copy of the GNU General Public License
19 * along with this program; if not, write to the Free Software
20 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
21 *
22 *
23 * These are the low level assembler for performing cache and TLB
24 * functions on the arm1020e.
25 *
26 *  CONFIG_CPU_ARM1020_CPU_IDLE -> nohlt
27 */
28#include <linux/linkage.h>
29#include <linux/init.h>
30#include <asm/assembler.h>
31#include <asm/asm-offsets.h>
32#include <asm/hwcap.h>
33#include <asm/pgtable-hwdef.h>
34#include <asm/pgtable.h>
35#include <asm/ptrace.h>
36
37#include "proc-macros.S"
38
39/*
40 * This is the maximum size of an area which will be invalidated
41 * using the single invalidate entry instructions.  Anything larger
42 * than this, and we go for the whole cache.
43 *
44 * This value should be chosen such that we choose the cheapest
45 * alternative.
46 */
47#define MAX_AREA_SIZE	32768
48
49/*
50 * The size of one data cache line.
51 */
52#define CACHE_DLINESIZE	32
53
54/*
55 * The number of data cache segments.
56 */
57#define CACHE_DSEGMENTS	16
58
59/*
60 * The number of lines in a cache segment.
61 */
62#define CACHE_DENTRIES	64
63
64/*
65 * This is the size at which it becomes more efficient to
66 * clean the whole cache, rather than using the individual
67 * cache line maintenance instructions.
68 */
69#define CACHE_DLIMIT	32768
70
71	.text
72/*
73 * cpu_arm1020e_proc_init()
74 */
75ENTRY(cpu_arm1020e_proc_init)
76	mov	pc, lr
77
78/*
79 * cpu_arm1020e_proc_fin()
80 */
81ENTRY(cpu_arm1020e_proc_fin)
82	mrc	p15, 0, r0, c1, c0, 0		@ ctrl register
83	bic	r0, r0, #0x1000 		@ ...i............
84	bic	r0, r0, #0x000e 		@ ............wca.
85	mcr	p15, 0, r0, c1, c0, 0		@ disable caches
86	mov	pc, lr
87
88/*
89 * cpu_arm1020e_reset(loc)
90 *
91 * Perform a soft reset of the system.	Put the CPU into the
92 * same state as it would be if it had been reset, and branch
93 * to what would be the reset vector.
94 *
95 * loc: location to jump to for soft reset
96 */
97	.align	5
98ENTRY(cpu_arm1020e_reset)
99	mov	ip, #0
100	mcr	p15, 0, ip, c7, c7, 0		@ invalidate I,D caches
101	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
102#ifdef CONFIG_MMU
103	mcr	p15, 0, ip, c8, c7, 0		@ invalidate I & D TLBs
104#endif
105	mrc	p15, 0, ip, c1, c0, 0		@ ctrl register
106	bic	ip, ip, #0x000f 		@ ............wcam
107	bic	ip, ip, #0x1100 		@ ...i...s........
108	mcr	p15, 0, ip, c1, c0, 0		@ ctrl register
109	mov	pc, r0
110
111/*
112 * cpu_arm1020e_do_idle()
113 */
114	.align	5
115ENTRY(cpu_arm1020e_do_idle)
116	mcr	p15, 0, r0, c7, c0, 4		@ Wait for interrupt
117	mov	pc, lr
118
119/* ================================= CACHE ================================ */
120
121	.align	5
122
123/*
124 *	flush_icache_all()
125 *
126 *	Unconditionally clean and invalidate the entire icache.
127 */
128ENTRY(arm1020e_flush_icache_all)
129#ifndef CONFIG_CPU_ICACHE_DISABLE
130	mov	r0, #0
131	mcr	p15, 0, r0, c7, c5, 0		@ invalidate I cache
132#endif
133	mov	pc, lr
134ENDPROC(arm1020e_flush_icache_all)
135
136/*
137 *	flush_user_cache_all()
138 *
139 *	Invalidate all cache entries in a particular address
140 *	space.
141 */
142ENTRY(arm1020e_flush_user_cache_all)
143	/* FALLTHROUGH */
144/*
145 *	flush_kern_cache_all()
146 *
147 *	Clean and invalidate the entire cache.
148 */
149ENTRY(arm1020e_flush_kern_cache_all)
150	mov	r2, #VM_EXEC
151	mov	ip, #0
152__flush_whole_cache:
153#ifndef CONFIG_CPU_DCACHE_DISABLE
154	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
155	mov	r1, #(CACHE_DSEGMENTS - 1) << 5	@ 16 segments
1561:	orr	r3, r1, #(CACHE_DENTRIES - 1) << 26 @ 64 entries
1572:	mcr	p15, 0, r3, c7, c14, 2		@ clean+invalidate D index
158	subs	r3, r3, #1 << 26
159	bcs	2b				@ entries 63 to 0
160	subs	r1, r1, #1 << 5
161	bcs	1b				@ segments 15 to 0
162#endif
163	tst	r2, #VM_EXEC
164#ifndef CONFIG_CPU_ICACHE_DISABLE
165	mcrne	p15, 0, ip, c7, c5, 0		@ invalidate I cache
166#endif
167	mcrne	p15, 0, ip, c7, c10, 4		@ drain WB
168	mov	pc, lr
169
170/*
171 *	flush_user_cache_range(start, end, flags)
172 *
173 *	Invalidate a range of cache entries in the specified
174 *	address space.
175 *
176 *	- start	- start address (inclusive)
177 *	- end	- end address (exclusive)
178 *	- flags	- vm_flags for this space
179 */
180ENTRY(arm1020e_flush_user_cache_range)
181	mov	ip, #0
182	sub	r3, r1, r0			@ calculate total size
183	cmp	r3, #CACHE_DLIMIT
184	bhs	__flush_whole_cache
185
186#ifndef CONFIG_CPU_DCACHE_DISABLE
1871:	mcr	p15, 0, r0, c7, c14, 1		@ clean+invalidate D entry
188	add	r0, r0, #CACHE_DLINESIZE
189	cmp	r0, r1
190	blo	1b
191#endif
192	tst	r2, #VM_EXEC
193#ifndef CONFIG_CPU_ICACHE_DISABLE
194	mcrne	p15, 0, ip, c7, c5, 0		@ invalidate I cache
195#endif
196	mcrne	p15, 0, ip, c7, c10, 4		@ drain WB
197	mov	pc, lr
198
199/*
200 *	coherent_kern_range(start, end)
201 *
202 *	Ensure coherency between the Icache and the Dcache in the
203 *	region described by start.  If you have non-snooping
204 *	Harvard caches, you need to implement this function.
205 *
206 *	- start	- virtual start address
207 *	- end	- virtual end address
208 */
209ENTRY(arm1020e_coherent_kern_range)
210	/* FALLTHROUGH */
211/*
212 *	coherent_user_range(start, end)
213 *
214 *	Ensure coherency between the Icache and the Dcache in the
215 *	region described by start.  If you have non-snooping
216 *	Harvard caches, you need to implement this function.
217 *
218 *	- start	- virtual start address
219 *	- end	- virtual end address
220 */
221ENTRY(arm1020e_coherent_user_range)
222	mov	ip, #0
223	bic	r0, r0, #CACHE_DLINESIZE - 1
2241:
225#ifndef CONFIG_CPU_DCACHE_DISABLE
226	mcr	p15, 0, r0, c7, c10, 1		@ clean D entry
227#endif
228#ifndef CONFIG_CPU_ICACHE_DISABLE
229	mcr	p15, 0, r0, c7, c5, 1		@ invalidate I entry
230#endif
231	add	r0, r0, #CACHE_DLINESIZE
232	cmp	r0, r1
233	blo	1b
234	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
235	mov	pc, lr
236
237/*
238 *	flush_kern_dcache_area(void *addr, size_t size)
239 *
240 *	Ensure no D cache aliasing occurs, either with itself or
241 *	the I cache
242 *
243 *	- addr	- kernel address
244 *	- size	- region size
245 */
246ENTRY(arm1020e_flush_kern_dcache_area)
247	mov	ip, #0
248#ifndef CONFIG_CPU_DCACHE_DISABLE
249	add	r1, r0, r1
2501:	mcr	p15, 0, r0, c7, c14, 1		@ clean+invalidate D entry
251	add	r0, r0, #CACHE_DLINESIZE
252	cmp	r0, r1
253	blo	1b
254#endif
255	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
256	mov	pc, lr
257
258/*
259 *	dma_inv_range(start, end)
260 *
261 *	Invalidate (discard) the specified virtual address range.
262 *	May not write back any entries.  If 'start' or 'end'
263 *	are not cache line aligned, those lines must be written
264 *	back.
265 *
266 *	- start	- virtual start address
267 *	- end	- virtual end address
268 *
269 * (same as v4wb)
270 */
271arm1020e_dma_inv_range:
272	mov	ip, #0
273#ifndef CONFIG_CPU_DCACHE_DISABLE
274	tst	r0, #CACHE_DLINESIZE - 1
275	bic	r0, r0, #CACHE_DLINESIZE - 1
276	mcrne	p15, 0, r0, c7, c10, 1		@ clean D entry
277	tst	r1, #CACHE_DLINESIZE - 1
278	mcrne	p15, 0, r1, c7, c10, 1		@ clean D entry
2791:	mcr	p15, 0, r0, c7, c6, 1		@ invalidate D entry
280	add	r0, r0, #CACHE_DLINESIZE
281	cmp	r0, r1
282	blo	1b
283#endif
284	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
285	mov	pc, lr
286
287/*
288 *	dma_clean_range(start, end)
289 *
290 *	Clean the specified virtual address range.
291 *
292 *	- start	- virtual start address
293 *	- end	- virtual end address
294 *
295 * (same as v4wb)
296 */
297arm1020e_dma_clean_range:
298	mov	ip, #0
299#ifndef CONFIG_CPU_DCACHE_DISABLE
300	bic	r0, r0, #CACHE_DLINESIZE - 1
3011:	mcr	p15, 0, r0, c7, c10, 1		@ clean D entry
302	add	r0, r0, #CACHE_DLINESIZE
303	cmp	r0, r1
304	blo	1b
305#endif
306	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
307	mov	pc, lr
308
309/*
310 *	dma_flush_range(start, end)
311 *
312 *	Clean and invalidate the specified virtual address range.
313 *
314 *	- start	- virtual start address
315 *	- end	- virtual end address
316 */
317ENTRY(arm1020e_dma_flush_range)
318	mov	ip, #0
319#ifndef CONFIG_CPU_DCACHE_DISABLE
320	bic	r0, r0, #CACHE_DLINESIZE - 1
3211:	mcr	p15, 0, r0, c7, c14, 1		@ clean+invalidate D entry
322	add	r0, r0, #CACHE_DLINESIZE
323	cmp	r0, r1
324	blo	1b
325#endif
326	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
327	mov	pc, lr
328
329/*
330 *	dma_map_area(start, size, dir)
331 *	- start	- kernel virtual start address
332 *	- size	- size of region
333 *	- dir	- DMA direction
334 */
335ENTRY(arm1020e_dma_map_area)
336	add	r1, r1, r0
337	cmp	r2, #DMA_TO_DEVICE
338	beq	arm1020e_dma_clean_range
339	bcs	arm1020e_dma_inv_range
340	b	arm1020e_dma_flush_range
341ENDPROC(arm1020e_dma_map_area)
342
343/*
344 *	dma_unmap_area(start, size, dir)
345 *	- start	- kernel virtual start address
346 *	- size	- size of region
347 *	- dir	- DMA direction
348 */
349ENTRY(arm1020e_dma_unmap_area)
350	mov	pc, lr
351ENDPROC(arm1020e_dma_unmap_area)
352
353	@ define struct cpu_cache_fns (see <asm/cacheflush.h> and proc-macros.S)
354	define_cache_functions arm1020e
355
356	.align	5
357ENTRY(cpu_arm1020e_dcache_clean_area)
358#ifndef CONFIG_CPU_DCACHE_DISABLE
359	mov	ip, #0
3601:	mcr	p15, 0, r0, c7, c10, 1		@ clean D entry
361	add	r0, r0, #CACHE_DLINESIZE
362	subs	r1, r1, #CACHE_DLINESIZE
363	bhi	1b
364#endif
365	mov	pc, lr
366
367/* =============================== PageTable ============================== */
368
369/*
370 * cpu_arm1020e_switch_mm(pgd)
371 *
372 * Set the translation base pointer to be as described by pgd.
373 *
374 * pgd: new page tables
375 */
376	.align	5
377ENTRY(cpu_arm1020e_switch_mm)
378#ifdef CONFIG_MMU
379#ifndef CONFIG_CPU_DCACHE_DISABLE
380	mcr	p15, 0, r3, c7, c10, 4
381	mov	r1, #0xF			@ 16 segments
3821:	mov	r3, #0x3F			@ 64 entries
3832:	mov	ip, r3, LSL #26 		@ shift up entry
384	orr	ip, ip, r1, LSL #5		@ shift in/up index
385	mcr	p15, 0, ip, c7, c14, 2		@ Clean & Inval DCache entry
386	mov	ip, #0
387	subs	r3, r3, #1
388	cmp	r3, #0
389	bge	2b				@ entries 3F to 0
390	subs	r1, r1, #1
391	cmp	r1, #0
392	bge	1b				@ segments 15 to 0
393
394#endif
395	mov	r1, #0
396#ifndef CONFIG_CPU_ICACHE_DISABLE
397	mcr	p15, 0, r1, c7, c5, 0		@ invalidate I cache
398#endif
399	mcr	p15, 0, r1, c7, c10, 4		@ drain WB
400	mcr	p15, 0, r0, c2, c0, 0		@ load page table pointer
401	mcr	p15, 0, r1, c8, c7, 0		@ invalidate I & D TLBs
402#endif
403	mov	pc, lr
404
405/*
406 * cpu_arm1020e_set_pte(ptep, pte)
407 *
408 * Set a PTE and flush it out
409 */
410	.align	5
411ENTRY(cpu_arm1020e_set_pte_ext)
412#ifdef CONFIG_MMU
413	armv3_set_pte_ext
414	mov	r0, r0
415#ifndef CONFIG_CPU_DCACHE_DISABLE
416	mcr	p15, 0, r0, c7, c10, 1		@ clean D entry
417#endif
418#endif /* CONFIG_MMU */
419	mov	pc, lr
420
421	__CPUINIT
422
423	.type	__arm1020e_setup, #function
424__arm1020e_setup:
425	mov	r0, #0
426	mcr	p15, 0, r0, c7, c7		@ invalidate I,D caches on v4
427	mcr	p15, 0, r0, c7, c10, 4		@ drain write buffer on v4
428#ifdef CONFIG_MMU
429	mcr	p15, 0, r0, c8, c7		@ invalidate I,D TLBs on v4
430#endif
431	adr	r5, arm1020e_crval
432	ldmia	r5, {r5, r6}
433	mrc	p15, 0, r0, c1, c0		@ get control register v4
434	bic	r0, r0, r5
435	orr	r0, r0, r6
436#ifdef CONFIG_CPU_CACHE_ROUND_ROBIN
437	orr	r0, r0, #0x4000 		@ .R.. .... .... ....
438#endif
439	mov	pc, lr
440	.size	__arm1020e_setup, . - __arm1020e_setup
441
442	/*
443	 *  R
444	 * .RVI ZFRS BLDP WCAM
445	 * .011 1001 ..11 0101
446	 */
447	.type	arm1020e_crval, #object
448arm1020e_crval:
449	crval	clear=0x00007f3f, mmuset=0x00003935, ucset=0x00001930
450
451	__INITDATA
452	@ define struct processor (see <asm/proc-fns.h> and proc-macros.S)
453	define_processor_functions arm1020e, dabort=v4t_early_abort, pabort=legacy_pabort
454
455	.section ".rodata"
456
457	string	cpu_arch_name, "armv5te"
458	string	cpu_elf_name, "v5"
459	string	cpu_arm1020e_name, "ARM1020E"
460
461	.align
462
463	.section ".proc.info.init", #alloc, #execinstr
464
465	.type	__arm1020e_proc_info,#object
466__arm1020e_proc_info:
467	.long	0x4105a200			@ ARM 1020TE (Architecture v5TE)
468	.long	0xff0ffff0
469	.long   PMD_TYPE_SECT | \
470		PMD_BIT4 | \
471		PMD_SECT_AP_WRITE | \
472		PMD_SECT_AP_READ
473	.long   PMD_TYPE_SECT | \
474		PMD_BIT4 | \
475		PMD_SECT_AP_WRITE | \
476		PMD_SECT_AP_READ
477	b	__arm1020e_setup
478	.long	cpu_arch_name
479	.long	cpu_elf_name
480	.long	HWCAP_SWP | HWCAP_HALF | HWCAP_THUMB | HWCAP_EDSP
481	.long	cpu_arm1020e_name
482	.long	arm1020e_processor_functions
483	.long	v4wbi_tlb_fns
484	.long	v4wb_user_fns
485	.long	arm1020e_cache_fns
486	.size	__arm1020e_proc_info, . - __arm1020e_proc_info
487