xref: /linux/arch/arm/mm/proc-arm1020e.S (revision 3932b9ca55b0be314a36d3e84faff3e823c081f5)
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	ret	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	ret	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
98	.pushsection	.idmap.text, "ax"
99ENTRY(cpu_arm1020e_reset)
100	mov	ip, #0
101	mcr	p15, 0, ip, c7, c7, 0		@ invalidate I,D caches
102	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
103#ifdef CONFIG_MMU
104	mcr	p15, 0, ip, c8, c7, 0		@ invalidate I & D TLBs
105#endif
106	mrc	p15, 0, ip, c1, c0, 0		@ ctrl register
107	bic	ip, ip, #0x000f 		@ ............wcam
108	bic	ip, ip, #0x1100 		@ ...i...s........
109	mcr	p15, 0, ip, c1, c0, 0		@ ctrl register
110	ret	r0
111ENDPROC(cpu_arm1020e_reset)
112	.popsection
113
114/*
115 * cpu_arm1020e_do_idle()
116 */
117	.align	5
118ENTRY(cpu_arm1020e_do_idle)
119	mcr	p15, 0, r0, c7, c0, 4		@ Wait for interrupt
120	ret	lr
121
122/* ================================= CACHE ================================ */
123
124	.align	5
125
126/*
127 *	flush_icache_all()
128 *
129 *	Unconditionally clean and invalidate the entire icache.
130 */
131ENTRY(arm1020e_flush_icache_all)
132#ifndef CONFIG_CPU_ICACHE_DISABLE
133	mov	r0, #0
134	mcr	p15, 0, r0, c7, c5, 0		@ invalidate I cache
135#endif
136	ret	lr
137ENDPROC(arm1020e_flush_icache_all)
138
139/*
140 *	flush_user_cache_all()
141 *
142 *	Invalidate all cache entries in a particular address
143 *	space.
144 */
145ENTRY(arm1020e_flush_user_cache_all)
146	/* FALLTHROUGH */
147/*
148 *	flush_kern_cache_all()
149 *
150 *	Clean and invalidate the entire cache.
151 */
152ENTRY(arm1020e_flush_kern_cache_all)
153	mov	r2, #VM_EXEC
154	mov	ip, #0
155__flush_whole_cache:
156#ifndef CONFIG_CPU_DCACHE_DISABLE
157	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
158	mov	r1, #(CACHE_DSEGMENTS - 1) << 5	@ 16 segments
1591:	orr	r3, r1, #(CACHE_DENTRIES - 1) << 26 @ 64 entries
1602:	mcr	p15, 0, r3, c7, c14, 2		@ clean+invalidate D index
161	subs	r3, r3, #1 << 26
162	bcs	2b				@ entries 63 to 0
163	subs	r1, r1, #1 << 5
164	bcs	1b				@ segments 15 to 0
165#endif
166	tst	r2, #VM_EXEC
167#ifndef CONFIG_CPU_ICACHE_DISABLE
168	mcrne	p15, 0, ip, c7, c5, 0		@ invalidate I cache
169#endif
170	mcrne	p15, 0, ip, c7, c10, 4		@ drain WB
171	ret	lr
172
173/*
174 *	flush_user_cache_range(start, end, flags)
175 *
176 *	Invalidate a range of cache entries in the specified
177 *	address space.
178 *
179 *	- start	- start address (inclusive)
180 *	- end	- end address (exclusive)
181 *	- flags	- vm_flags for this space
182 */
183ENTRY(arm1020e_flush_user_cache_range)
184	mov	ip, #0
185	sub	r3, r1, r0			@ calculate total size
186	cmp	r3, #CACHE_DLIMIT
187	bhs	__flush_whole_cache
188
189#ifndef CONFIG_CPU_DCACHE_DISABLE
1901:	mcr	p15, 0, r0, c7, c14, 1		@ clean+invalidate D entry
191	add	r0, r0, #CACHE_DLINESIZE
192	cmp	r0, r1
193	blo	1b
194#endif
195	tst	r2, #VM_EXEC
196#ifndef CONFIG_CPU_ICACHE_DISABLE
197	mcrne	p15, 0, ip, c7, c5, 0		@ invalidate I cache
198#endif
199	mcrne	p15, 0, ip, c7, c10, 4		@ drain WB
200	ret	lr
201
202/*
203 *	coherent_kern_range(start, end)
204 *
205 *	Ensure coherency between the Icache and the Dcache in the
206 *	region described by start.  If you have non-snooping
207 *	Harvard caches, you need to implement this function.
208 *
209 *	- start	- virtual start address
210 *	- end	- virtual end address
211 */
212ENTRY(arm1020e_coherent_kern_range)
213	/* FALLTHROUGH */
214/*
215 *	coherent_user_range(start, end)
216 *
217 *	Ensure coherency between the Icache and the Dcache in the
218 *	region described by start.  If you have non-snooping
219 *	Harvard caches, you need to implement this function.
220 *
221 *	- start	- virtual start address
222 *	- end	- virtual end address
223 */
224ENTRY(arm1020e_coherent_user_range)
225	mov	ip, #0
226	bic	r0, r0, #CACHE_DLINESIZE - 1
2271:
228#ifndef CONFIG_CPU_DCACHE_DISABLE
229	mcr	p15, 0, r0, c7, c10, 1		@ clean D entry
230#endif
231#ifndef CONFIG_CPU_ICACHE_DISABLE
232	mcr	p15, 0, r0, c7, c5, 1		@ invalidate I entry
233#endif
234	add	r0, r0, #CACHE_DLINESIZE
235	cmp	r0, r1
236	blo	1b
237	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
238	mov	r0, #0
239	ret	lr
240
241/*
242 *	flush_kern_dcache_area(void *addr, size_t size)
243 *
244 *	Ensure no D cache aliasing occurs, either with itself or
245 *	the I cache
246 *
247 *	- addr	- kernel address
248 *	- size	- region size
249 */
250ENTRY(arm1020e_flush_kern_dcache_area)
251	mov	ip, #0
252#ifndef CONFIG_CPU_DCACHE_DISABLE
253	add	r1, r0, r1
2541:	mcr	p15, 0, r0, c7, c14, 1		@ clean+invalidate D entry
255	add	r0, r0, #CACHE_DLINESIZE
256	cmp	r0, r1
257	blo	1b
258#endif
259	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
260	ret	lr
261
262/*
263 *	dma_inv_range(start, end)
264 *
265 *	Invalidate (discard) the specified virtual address range.
266 *	May not write back any entries.  If 'start' or 'end'
267 *	are not cache line aligned, those lines must be written
268 *	back.
269 *
270 *	- start	- virtual start address
271 *	- end	- virtual end address
272 *
273 * (same as v4wb)
274 */
275arm1020e_dma_inv_range:
276	mov	ip, #0
277#ifndef CONFIG_CPU_DCACHE_DISABLE
278	tst	r0, #CACHE_DLINESIZE - 1
279	bic	r0, r0, #CACHE_DLINESIZE - 1
280	mcrne	p15, 0, r0, c7, c10, 1		@ clean D entry
281	tst	r1, #CACHE_DLINESIZE - 1
282	mcrne	p15, 0, r1, c7, c10, 1		@ clean D entry
2831:	mcr	p15, 0, r0, c7, c6, 1		@ invalidate D entry
284	add	r0, r0, #CACHE_DLINESIZE
285	cmp	r0, r1
286	blo	1b
287#endif
288	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
289	ret	lr
290
291/*
292 *	dma_clean_range(start, end)
293 *
294 *	Clean the specified virtual address range.
295 *
296 *	- start	- virtual start address
297 *	- end	- virtual end address
298 *
299 * (same as v4wb)
300 */
301arm1020e_dma_clean_range:
302	mov	ip, #0
303#ifndef CONFIG_CPU_DCACHE_DISABLE
304	bic	r0, r0, #CACHE_DLINESIZE - 1
3051:	mcr	p15, 0, r0, c7, c10, 1		@ clean D entry
306	add	r0, r0, #CACHE_DLINESIZE
307	cmp	r0, r1
308	blo	1b
309#endif
310	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
311	ret	lr
312
313/*
314 *	dma_flush_range(start, end)
315 *
316 *	Clean and invalidate the specified virtual address range.
317 *
318 *	- start	- virtual start address
319 *	- end	- virtual end address
320 */
321ENTRY(arm1020e_dma_flush_range)
322	mov	ip, #0
323#ifndef CONFIG_CPU_DCACHE_DISABLE
324	bic	r0, r0, #CACHE_DLINESIZE - 1
3251:	mcr	p15, 0, r0, c7, c14, 1		@ clean+invalidate D entry
326	add	r0, r0, #CACHE_DLINESIZE
327	cmp	r0, r1
328	blo	1b
329#endif
330	mcr	p15, 0, ip, c7, c10, 4		@ drain WB
331	ret	lr
332
333/*
334 *	dma_map_area(start, size, dir)
335 *	- start	- kernel virtual start address
336 *	- size	- size of region
337 *	- dir	- DMA direction
338 */
339ENTRY(arm1020e_dma_map_area)
340	add	r1, r1, r0
341	cmp	r2, #DMA_TO_DEVICE
342	beq	arm1020e_dma_clean_range
343	bcs	arm1020e_dma_inv_range
344	b	arm1020e_dma_flush_range
345ENDPROC(arm1020e_dma_map_area)
346
347/*
348 *	dma_unmap_area(start, size, dir)
349 *	- start	- kernel virtual start address
350 *	- size	- size of region
351 *	- dir	- DMA direction
352 */
353ENTRY(arm1020e_dma_unmap_area)
354	ret	lr
355ENDPROC(arm1020e_dma_unmap_area)
356
357	.globl	arm1020e_flush_kern_cache_louis
358	.equ	arm1020e_flush_kern_cache_louis, arm1020e_flush_kern_cache_all
359
360	@ define struct cpu_cache_fns (see <asm/cacheflush.h> and proc-macros.S)
361	define_cache_functions arm1020e
362
363	.align	5
364ENTRY(cpu_arm1020e_dcache_clean_area)
365#ifndef CONFIG_CPU_DCACHE_DISABLE
366	mov	ip, #0
3671:	mcr	p15, 0, r0, c7, c10, 1		@ clean D entry
368	add	r0, r0, #CACHE_DLINESIZE
369	subs	r1, r1, #CACHE_DLINESIZE
370	bhi	1b
371#endif
372	ret	lr
373
374/* =============================== PageTable ============================== */
375
376/*
377 * cpu_arm1020e_switch_mm(pgd)
378 *
379 * Set the translation base pointer to be as described by pgd.
380 *
381 * pgd: new page tables
382 */
383	.align	5
384ENTRY(cpu_arm1020e_switch_mm)
385#ifdef CONFIG_MMU
386#ifndef CONFIG_CPU_DCACHE_DISABLE
387	mcr	p15, 0, r3, c7, c10, 4
388	mov	r1, #0xF			@ 16 segments
3891:	mov	r3, #0x3F			@ 64 entries
3902:	mov	ip, r3, LSL #26 		@ shift up entry
391	orr	ip, ip, r1, LSL #5		@ shift in/up index
392	mcr	p15, 0, ip, c7, c14, 2		@ Clean & Inval DCache entry
393	mov	ip, #0
394	subs	r3, r3, #1
395	cmp	r3, #0
396	bge	2b				@ entries 3F to 0
397	subs	r1, r1, #1
398	cmp	r1, #0
399	bge	1b				@ segments 15 to 0
400
401#endif
402	mov	r1, #0
403#ifndef CONFIG_CPU_ICACHE_DISABLE
404	mcr	p15, 0, r1, c7, c5, 0		@ invalidate I cache
405#endif
406	mcr	p15, 0, r1, c7, c10, 4		@ drain WB
407	mcr	p15, 0, r0, c2, c0, 0		@ load page table pointer
408	mcr	p15, 0, r1, c8, c7, 0		@ invalidate I & D TLBs
409#endif
410	ret	lr
411
412/*
413 * cpu_arm1020e_set_pte(ptep, pte)
414 *
415 * Set a PTE and flush it out
416 */
417	.align	5
418ENTRY(cpu_arm1020e_set_pte_ext)
419#ifdef CONFIG_MMU
420	armv3_set_pte_ext
421	mov	r0, r0
422#ifndef CONFIG_CPU_DCACHE_DISABLE
423	mcr	p15, 0, r0, c7, c10, 1		@ clean D entry
424#endif
425#endif /* CONFIG_MMU */
426	ret	lr
427
428	.type	__arm1020e_setup, #function
429__arm1020e_setup:
430	mov	r0, #0
431	mcr	p15, 0, r0, c7, c7		@ invalidate I,D caches on v4
432	mcr	p15, 0, r0, c7, c10, 4		@ drain write buffer on v4
433#ifdef CONFIG_MMU
434	mcr	p15, 0, r0, c8, c7		@ invalidate I,D TLBs on v4
435#endif
436	adr	r5, arm1020e_crval
437	ldmia	r5, {r5, r6}
438	mrc	p15, 0, r0, c1, c0		@ get control register v4
439	bic	r0, r0, r5
440	orr	r0, r0, r6
441#ifdef CONFIG_CPU_CACHE_ROUND_ROBIN
442	orr	r0, r0, #0x4000 		@ .R.. .... .... ....
443#endif
444	ret	lr
445	.size	__arm1020e_setup, . - __arm1020e_setup
446
447	/*
448	 *  R
449	 * .RVI ZFRS BLDP WCAM
450	 * .011 1001 ..11 0101
451	 */
452	.type	arm1020e_crval, #object
453arm1020e_crval:
454	crval	clear=0x00007f3f, mmuset=0x00003935, ucset=0x00001930
455
456	__INITDATA
457	@ define struct processor (see <asm/proc-fns.h> and proc-macros.S)
458	define_processor_functions arm1020e, dabort=v4t_early_abort, pabort=legacy_pabort
459
460	.section ".rodata"
461
462	string	cpu_arch_name, "armv5te"
463	string	cpu_elf_name, "v5"
464	string	cpu_arm1020e_name, "ARM1020E"
465
466	.align
467
468	.section ".proc.info.init", #alloc, #execinstr
469
470	.type	__arm1020e_proc_info,#object
471__arm1020e_proc_info:
472	.long	0x4105a200			@ ARM 1020TE (Architecture v5TE)
473	.long	0xff0ffff0
474	.long   PMD_TYPE_SECT | \
475		PMD_BIT4 | \
476		PMD_SECT_AP_WRITE | \
477		PMD_SECT_AP_READ
478	.long   PMD_TYPE_SECT | \
479		PMD_BIT4 | \
480		PMD_SECT_AP_WRITE | \
481		PMD_SECT_AP_READ
482	b	__arm1020e_setup
483	.long	cpu_arch_name
484	.long	cpu_elf_name
485	.long	HWCAP_SWP | HWCAP_HALF | HWCAP_THUMB | HWCAP_EDSP
486	.long	cpu_arm1020e_name
487	.long	arm1020e_processor_functions
488	.long	v4wbi_tlb_fns
489	.long	v4wb_user_fns
490	.long	arm1020e_cache_fns
491	.size	__arm1020e_proc_info, . - __arm1020e_proc_info
492