[net-next-2.6.git] / arch / powerpc / mm / ppc_mmu_32.c

/*
 * This file contains the routines for handling the MMU on those
 * PowerPC implementations where the MMU substantially follows the
 * architecture specification.  This includes the 6xx, 7xx, 7xxx,
 * 8260, and POWER3 implementations but excludes the 8xx and 4xx.
 *  -- paulus
 *
 *  Derived from arch/ppc/mm/init.c:
 *    Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
 *
 *  Modifications by Paul Mackerras (PowerMac) (paulus@cs.anu.edu.au)
 *  and Cort Dougan (PReP) (cort@cs.nmt.edu)
 *    Copyright (C) 1996 Paul Mackerras
 *
 *  Derived from "arch/i386/mm/init.c"
 *    Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
 *
 *  This program is free software; you can redistribute it and/or
 *  modify it under the terms of the GNU General Public License
 *  as published by the Free Software Foundation; either version
 *  2 of the License, or (at your option) any later version.
 *
 */

#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/init.h>
#include <linux/highmem.h>
#include <linux/memblock.h>

#include <asm/prom.h>
#include <asm/mmu.h>
#include <asm/machdep.h>

#include "mmu_decl.h"

struct hash_pte *Hash, *Hash_end;
unsigned long Hash_size, Hash_mask;
unsigned long _SDR1;

struct ppc_bat BATS[8][2];	/* 8 pairs of IBAT, DBAT */

struct batrange {		/* stores address ranges mapped by BATs */
	unsigned long start;
	unsigned long limit;
	phys_addr_t phys;
} bat_addrs[8];

/*
 * Return PA for this VA if it is mapped by a BAT, or 0
 */
phys_addr_t v_mapped_by_bats(unsigned long va)
{
	int b;
	for (b = 0; b < 4; ++b)
		if (va >= bat_addrs[b].start && va < bat_addrs[b].limit)
			return bat_addrs[b].phys + (va - bat_addrs[b].start);
	return 0;
}

/*
 * Return VA for a given PA or 0 if not mapped
 */
unsigned long p_mapped_by_bats(phys_addr_t pa)
{
	int b;
	for (b = 0; b < 4; ++b)
		if (pa >= bat_addrs[b].phys
	    	    && pa < (bat_addrs[b].limit-bat_addrs[b].start)
		              +bat_addrs[b].phys)
			return bat_addrs[b].start+(pa-bat_addrs[b].phys);
	return 0;
}

unsigned long __init mmu_mapin_ram(unsigned long top)
{
	unsigned long tot, bl, done;
	unsigned long max_size = (256<<20);

	if (__map_without_bats) {
		printk(KERN_DEBUG "RAM mapped without BATs\n");
		return 0;
	}

	/* Set up BAT2 and if necessary BAT3 to cover RAM. */

	/* Make sure we don't map a block larger than the
	   smallest alignment of the physical address. */
	tot = top;
	for (bl = 128<<10; bl < max_size; bl <<= 1) {
		if (bl * 2 > tot)
			break;
	}

	setbat(2, PAGE_OFFSET, 0, bl, PAGE_KERNEL_X);
	done = (unsigned long)bat_addrs[2].limit - PAGE_OFFSET + 1;
	if ((done < tot) && !bat_addrs[3].limit) {
		/* use BAT3 to cover a bit more */
		tot -= done;
		for (bl = 128<<10; bl < max_size; bl <<= 1)
			if (bl * 2 > tot)
				break;
		setbat(3, PAGE_OFFSET+done, done, bl, PAGE_KERNEL_X);
		done = (unsigned long)bat_addrs[3].limit - PAGE_OFFSET + 1;
	}

	return done;
}

/*
 * Set up one of the I/D BAT (block address translation) register pairs.
 * The parameters are not checked; in particular size must be a power
 * of 2 between 128k and 256M.
 */
void __init setbat(int index, unsigned long virt, phys_addr_t phys,
		   unsigned int size, int flags)
{
	unsigned int bl;
	int wimgxpp;
	struct ppc_bat *bat = BATS[index];

	if ((flags & _PAGE_NO_CACHE) ||
	    (cpu_has_feature(CPU_FTR_NEED_COHERENT) == 0))
		flags &= ~_PAGE_COHERENT;

	bl = (size >> 17) - 1;
	if (PVR_VER(mfspr(SPRN_PVR)) != 1) {
		/* 603, 604, etc. */
		/* Do DBAT first */
		wimgxpp = flags & (_PAGE_WRITETHRU | _PAGE_NO_CACHE
				   | _PAGE_COHERENT | _PAGE_GUARDED);
		wimgxpp |= (flags & _PAGE_RW)? BPP_RW: BPP_RX;
		bat[1].batu = virt | (bl << 2) | 2; /* Vs=1, Vp=0 */
		bat[1].batl = BAT_PHYS_ADDR(phys) | wimgxpp;
		if (flags & _PAGE_USER)
			bat[1].batu |= 1; 	/* Vp = 1 */
		if (flags & _PAGE_GUARDED) {
			/* G bit must be zero in IBATs */
			bat[0].batu = bat[0].batl = 0;
		} else {
			/* make IBAT same as DBAT */
			bat[0] = bat[1];
		}
	} else {
		/* 601 cpu */
		if (bl > BL_8M)
			bl = BL_8M;
		wimgxpp = flags & (_PAGE_WRITETHRU | _PAGE_NO_CACHE
				   | _PAGE_COHERENT);
		wimgxpp |= (flags & _PAGE_RW)?
			((flags & _PAGE_USER)? PP_RWRW: PP_RWXX): PP_RXRX;
		bat->batu = virt | wimgxpp | 4;	/* Ks=0, Ku=1 */
		bat->batl = phys | bl | 0x40;	/* V=1 */
	}

	bat_addrs[index].start = virt;
	bat_addrs[index].limit = virt + ((bl + 1) << 17) - 1;
	bat_addrs[index].phys = phys;
}

/*
 * Preload a translation in the hash table
 */
void hash_preload(struct mm_struct *mm, unsigned long ea,
		  unsigned long access, unsigned long trap)
{
	pmd_t *pmd;

	if (Hash == 0)
		return;
	pmd = pmd_offset(pud_offset(pgd_offset(mm, ea), ea), ea);
	if (!pmd_none(*pmd))
		add_hash_page(mm->context.id, ea, pmd_val(*pmd));
}

/*
 * Initialize the hash table and patch the instructions in hashtable.S.
 */
void __init MMU_init_hw(void)
{
	unsigned int hmask, mb, mb2;
	unsigned int n_hpteg, lg_n_hpteg;

	extern unsigned int hash_page_patch_A[];
	extern unsigned int hash_page_patch_B[], hash_page_patch_C[];
	extern unsigned int hash_page[];
	extern unsigned int flush_hash_patch_A[], flush_hash_patch_B[];

	if (!mmu_has_feature(MMU_FTR_HPTE_TABLE)) {
		/*
		 * Put a blr (procedure return) instruction at the
		 * start of hash_page, since we can still get DSI
		 * exceptions on a 603.
		 */
		hash_page[0] = 0x4e800020;
		flush_icache_range((unsigned long) &hash_page[0],
				   (unsigned long) &hash_page[1]);
		return;
	}

	if ( ppc_md.progress ) ppc_md.progress("hash:enter", 0x105);

#define LG_HPTEG_SIZE	6		/* 64 bytes per HPTEG */
#define SDR1_LOW_BITS	((n_hpteg - 1) >> 10)
#define MIN_N_HPTEG	1024		/* min 64kB hash table */

	/*
	 * Allow 1 HPTE (1/8 HPTEG) for each page of memory.
	 * This is less than the recommended amount, but then
	 * Linux ain't AIX.
	 */
	n_hpteg = total_memory / (PAGE_SIZE * 8);
	if (n_hpteg < MIN_N_HPTEG)
		n_hpteg = MIN_N_HPTEG;
	lg_n_hpteg = __ilog2(n_hpteg);
	if (n_hpteg & (n_hpteg - 1)) {
		++lg_n_hpteg;		/* round up if not power of 2 */
		n_hpteg = 1 << lg_n_hpteg;
	}
	Hash_size = n_hpteg << LG_HPTEG_SIZE;

	/*
	 * Find some memory for the hash table.
	 */
	if ( ppc_md.progress ) ppc_md.progress("hash:find piece", 0x322);
	Hash = __va(memblock_alloc_base(Hash_size, Hash_size,
				   __initial_memory_limit_addr));
	cacheable_memzero(Hash, Hash_size);
	_SDR1 = __pa(Hash) | SDR1_LOW_BITS;

	Hash_end = (struct hash_pte *) ((unsigned long)Hash + Hash_size);

	printk("Total memory = %lldMB; using %ldkB for hash table (at %p)\n",
	       (unsigned long long)(total_memory >> 20), Hash_size >> 10, Hash);


	/*
	 * Patch up the instructions in hashtable.S:create_hpte
	 */
	if ( ppc_md.progress ) ppc_md.progress("hash:patch", 0x345);
	Hash_mask = n_hpteg - 1;
	hmask = Hash_mask >> (16 - LG_HPTEG_SIZE);
	mb2 = mb = 32 - LG_HPTEG_SIZE - lg_n_hpteg;
	if (lg_n_hpteg > 16)
		mb2 = 16 - LG_HPTEG_SIZE;

	hash_page_patch_A[0] = (hash_page_patch_A[0] & ~0xffff)
		| ((unsigned int)(Hash) >> 16);
	hash_page_patch_A[1] = (hash_page_patch_A[1] & ~0x7c0) | (mb << 6);
	hash_page_patch_A[2] = (hash_page_patch_A[2] & ~0x7c0) | (mb2 << 6);
	hash_page_patch_B[0] = (hash_page_patch_B[0] & ~0xffff) | hmask;
	hash_page_patch_C[0] = (hash_page_patch_C[0] & ~0xffff) | hmask;

	/*
	 * Ensure that the locations we've patched have been written
	 * out from the data cache and invalidated in the instruction
	 * cache, on those machines with split caches.
	 */
	flush_icache_range((unsigned long) &hash_page_patch_A[0],
			   (unsigned long) &hash_page_patch_C[1]);

	/*
	 * Patch up the instructions in hashtable.S:flush_hash_page
	 */
	flush_hash_patch_A[0] = (flush_hash_patch_A[0] & ~0xffff)
		| ((unsigned int)(Hash) >> 16);
	flush_hash_patch_A[1] = (flush_hash_patch_A[1] & ~0x7c0) | (mb << 6);
	flush_hash_patch_A[2] = (flush_hash_patch_A[2] & ~0x7c0) | (mb2 << 6);
	flush_hash_patch_B[0] = (flush_hash_patch_B[0] & ~0xffff) | hmask;
	flush_icache_range((unsigned long) &flush_hash_patch_A[0],
			   (unsigned long) &flush_hash_patch_B[1]);

	if ( ppc_md.progress ) ppc_md.progress("hash:done", 0x205);
}
Commit	Line	Data
14cf11af PM	1	/*
	2	* This file contains the routines for handling the MMU on those
	3	* PowerPC implementations where the MMU substantially follows the
	4	* architecture specification. This includes the 6xx, 7xx, 7xxx,
	5	* 8260, and POWER3 implementations but excludes the 8xx and 4xx.
	6	* -- paulus
	7	*
	8	* Derived from arch/ppc/mm/init.c:
	9	* Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
	10	*
	11	* Modifications by Paul Mackerras (PowerMac) (paulus@cs.anu.edu.au)
	12	* and Cort Dougan (PReP) (cort@cs.nmt.edu)
	13	* Copyright (C) 1996 Paul Mackerras
14cf11af PM	14	*
	15	* Derived from "arch/i386/mm/init.c"
	16	* Copyright (C) 1991, 1992, 1993, 1994 Linus Torvalds
	17	*
	18	* This program is free software; you can redistribute it and/or
	19	* modify it under the terms of the GNU General Public License
	20	* as published by the Free Software Foundation; either version
	21	* 2 of the License, or (at your option) any later version.
	22	*
	23	*/
	24
14cf11af PM	25	#include <linux/kernel.h>
	26	#include <linux/mm.h>
	27	#include <linux/init.h>
	28	#include <linux/highmem.h>
95f72d1e	29	#include <linux/memblock.h>
14cf11af PM	30
	31	#include <asm/prom.h>
	32	#include <asm/mmu.h>
	33	#include <asm/machdep.h>
	34
	35	#include "mmu_decl.h"
14cf11af	36
8e561e7e	37	struct hash_pte Hash, Hash_end;
14cf11af PM	38	unsigned long Hash_size, Hash_mask;
	39	unsigned long _SDR1;
	40
316a4058	41	struct ppc_bat BATS[8][2]; /* 8 pairs of IBAT, DBAT */
14cf11af PM	42
	43	struct batrange { /* stores address ranges mapped by BATs */
	44	unsigned long start;
	45	unsigned long limit;
7c5c4325	46	phys_addr_t phys;
ee0339f2	47	} bat_addrs[8];
14cf11af PM	48
	49	/*
	50	* Return PA for this VA if it is mapped by a BAT, or 0
	51	*/
7c5c4325	52	phys_addr_t v_mapped_by_bats(unsigned long va)
14cf11af PM	53	{
	54	int b;
	55	for (b = 0; b < 4; ++b)
	56	if (va >= bat_addrs[b].start && va < bat_addrs[b].limit)
	57	return bat_addrs[b].phys + (va - bat_addrs[b].start);
	58	return 0;
	59	}
	60
	61	/*
	62	* Return VA for a given PA or 0 if not mapped
	63	*/
7c5c4325	64	unsigned long p_mapped_by_bats(phys_addr_t pa)
14cf11af PM	65	{
	66	int b;
	67	for (b = 0; b < 4; ++b)
	68	if (pa >= bat_addrs[b].phys
	69	&& pa < (bat_addrs[b].limit-bat_addrs[b].start)
	70	+bat_addrs[b].phys)
	71	return bat_addrs[b].start+(pa-bat_addrs[b].phys);
	72	return 0;
	73	}
	74
de32400d	75	unsigned long __init mmu_mapin_ram(unsigned long top)
14cf11af	76	{
14cf11af PM	77	unsigned long tot, bl, done;
14cf11af PM	78	unsigned long max_size = (256<<20);
14cf11af	79
88df6e90 BH	80	if (__map_without_bats) {
88df6e90 BH	81	printk(KERN_DEBUG "RAM mapped without BATs\n");
14cf11af	82	return 0;
88df6e90	83	}
14cf11af PM	84
	85	/* Set up BAT2 and if necessary BAT3 to cover RAM. */
	86
	87	/* Make sure we don't map a block larger than the
	88	smallest alignment of the physical address. */
de32400d	89	tot = top;
14cf11af PM	90	for (bl = 128<<10; bl < max_size; bl <<= 1) {
	91	if (bl * 2 > tot)
	92	break;
	93	}
	94
8d1cf34e	95	setbat(2, PAGE_OFFSET, 0, bl, PAGE_KERNEL_X);
ccdcef72	96	done = (unsigned long)bat_addrs[2].limit - PAGE_OFFSET + 1;
14cf11af PM	97	if ((done < tot) && !bat_addrs[3].limit) {
	98	/* use BAT3 to cover a bit more */
	99	tot -= done;
	100	for (bl = 128<<10; bl < max_size; bl <<= 1)
	101	if (bl * 2 > tot)
	102	break;
8d1cf34e	103	setbat(3, PAGE_OFFSET+done, done, bl, PAGE_KERNEL_X);
ccdcef72	104	done = (unsigned long)bat_addrs[3].limit - PAGE_OFFSET + 1;
14cf11af PM	105	}
	106
	107	return done;
14cf11af PM	108	}
	109
	110	/*
	111	* Set up one of the I/D BAT (block address translation) register pairs.
	112	* The parameters are not checked; in particular size must be a power
	113	* of 2 between 128k and 256M.
	114	*/
7c5c4325	115	void __init setbat(int index, unsigned long virt, phys_addr_t phys,
14cf11af PM	116	unsigned int size, int flags)
	117	{
	118	unsigned int bl;
	119	int wimgxpp;
316a4058	120	struct ppc_bat *bat = BATS[index];
14cf11af	121
4c456a67 GP	122	if ((flags & _PAGE_NO_CACHE) \|\|
	123	(cpu_has_feature(CPU_FTR_NEED_COHERENT) == 0))
	124	flags &= ~_PAGE_COHERENT;
14cf11af PM	125
	126	bl = (size >> 17) - 1;
	127	if (PVR_VER(mfspr(SPRN_PVR)) != 1) {
	128	/* 603, 604, etc. */
	129	/* Do DBAT first */
	130	wimgxpp = flags & (_PAGE_WRITETHRU \| _PAGE_NO_CACHE
	131	\| _PAGE_COHERENT \| _PAGE_GUARDED);
	132	wimgxpp \|= (flags & _PAGE_RW)? BPP_RW: BPP_RX;
316a4058 BB	133	bat[1].batu = virt \| (bl << 2) \| 2; /* Vs=1, Vp=0 */
316a4058 BB	134	bat[1].batl = BAT_PHYS_ADDR(phys) \| wimgxpp;
14cf11af	135	if (flags & _PAGE_USER)
316a4058	136	bat[1].batu \|= 1; /* Vp = 1 */
14cf11af PM	137	if (flags & _PAGE_GUARDED) {
14cf11af PM	138	/* G bit must be zero in IBATs */
316a4058	139	bat[0].batu = bat[0].batl = 0;
14cf11af PM	140	} else {
	141	/* make IBAT same as DBAT */
	142	bat[0] = bat[1];
	143	}
	144	} else {
	145	/* 601 cpu */
	146	if (bl > BL_8M)
	147	bl = BL_8M;
	148	wimgxpp = flags & (_PAGE_WRITETHRU \| _PAGE_NO_CACHE
	149	\| _PAGE_COHERENT);
	150	wimgxpp \|= (flags & _PAGE_RW)?
	151	((flags & _PAGE_USER)? PP_RWRW: PP_RWXX): PP_RXRX;
316a4058 BB	152	bat->batu = virt \| wimgxpp \| 4; /* Ks=0, Ku=1 */
316a4058 BB	153	bat->batl = phys \| bl \| 0x40; /* V=1 */
14cf11af PM	154	}
	155
	156	bat_addrs[index].start = virt;
	157	bat_addrs[index].limit = virt + ((bl + 1) << 17) - 1;
	158	bat_addrs[index].phys = phys;
	159	}
	160
3c726f8d BH	161	/*
	162	* Preload a translation in the hash table
	163	*/
	164	void hash_preload(struct mm_struct *mm, unsigned long ea,
	165	unsigned long access, unsigned long trap)
	166	{
	167	pmd_t *pmd;
	168
	169	if (Hash == 0)
	170	return;
f1a1eb29	171	pmd = pmd_offset(pud_offset(pgd_offset(mm, ea), ea), ea);
3c726f8d	172	if (!pmd_none(*pmd))
6218a761	173	add_hash_page(mm->context.id, ea, pmd_val(*pmd));
3c726f8d BH	174	}
3c726f8d BH	175
14cf11af PM	176	/*
	177	* Initialize the hash table and patch the instructions in hashtable.S.
	178	*/
	179	void __init MMU_init_hw(void)
	180	{
	181	unsigned int hmask, mb, mb2;
	182	unsigned int n_hpteg, lg_n_hpteg;
	183
	184	extern unsigned int hash_page_patch_A[];
	185	extern unsigned int hash_page_patch_B[], hash_page_patch_C[];
	186	extern unsigned int hash_page[];
	187	extern unsigned int flush_hash_patch_A[], flush_hash_patch_B[];
	188
7c03d653	189	if (!mmu_has_feature(MMU_FTR_HPTE_TABLE)) {
14cf11af PM	190	/*
	191	* Put a blr (procedure return) instruction at the
	192	* start of hash_page, since we can still get DSI
	193	* exceptions on a 603.
	194	*/
	195	hash_page[0] = 0x4e800020;
	196	flush_icache_range((unsigned long) &hash_page[0],
	197	(unsigned long) &hash_page[1]);
	198	return;
	199	}
	200
	201	if ( ppc_md.progress ) ppc_md.progress("hash:enter", 0x105);
	202
14cf11af PM	203	#define LG_HPTEG_SIZE 6 /* 64 bytes per HPTEG */
	204	#define SDR1_LOW_BITS ((n_hpteg - 1) >> 10)
	205	#define MIN_N_HPTEG 1024 /* min 64kB hash table */
14cf11af	206
14cf11af PM	207	/*
	208	* Allow 1 HPTE (1/8 HPTEG) for each page of memory.
	209	* This is less than the recommended amount, but then
	210	* Linux ain't AIX.
	211	*/
	212	n_hpteg = total_memory / (PAGE_SIZE * 8);
	213	if (n_hpteg < MIN_N_HPTEG)
	214	n_hpteg = MIN_N_HPTEG;
	215	lg_n_hpteg = __ilog2(n_hpteg);
	216	if (n_hpteg & (n_hpteg - 1)) {
	217	++lg_n_hpteg; /* round up if not power of 2 */
	218	n_hpteg = 1 << lg_n_hpteg;
	219	}
	220	Hash_size = n_hpteg << LG_HPTEG_SIZE;
	221
	222	/*
	223	* Find some memory for the hash table.
	224	*/
	225	if ( ppc_md.progress ) ppc_md.progress("hash:find piece", 0x322);
95f72d1e	226	Hash = __va(memblock_alloc_base(Hash_size, Hash_size,
09b5e63f	227	__initial_memory_limit_addr));
14cf11af PM	228	cacheable_memzero(Hash, Hash_size);
14cf11af PM	229	_SDR1 = __pa(Hash) \| SDR1_LOW_BITS;
14cf11af	230
8e561e7e	231	Hash_end = (struct hash_pte *) ((unsigned long)Hash + Hash_size);
14cf11af	232
c7c8eede TB	233	printk("Total memory = %lldMB; using %ldkB for hash table (at %p)\n",
c7c8eede TB	234	(unsigned long long)(total_memory >> 20), Hash_size >> 10, Hash);
14cf11af PM	235
	236
	237	/*
	238	* Patch up the instructions in hashtable.S:create_hpte
	239	*/
	240	if ( ppc_md.progress ) ppc_md.progress("hash:patch", 0x345);
	241	Hash_mask = n_hpteg - 1;
	242	hmask = Hash_mask >> (16 - LG_HPTEG_SIZE);
	243	mb2 = mb = 32 - LG_HPTEG_SIZE - lg_n_hpteg;
	244	if (lg_n_hpteg > 16)
	245	mb2 = 16 - LG_HPTEG_SIZE;
	246
	247	hash_page_patch_A[0] = (hash_page_patch_A[0] & ~0xffff)
	248	\| ((unsigned int)(Hash) >> 16);
	249	hash_page_patch_A[1] = (hash_page_patch_A[1] & ~0x7c0) \| (mb << 6);
	250	hash_page_patch_A[2] = (hash_page_patch_A[2] & ~0x7c0) \| (mb2 << 6);
	251	hash_page_patch_B[0] = (hash_page_patch_B[0] & ~0xffff) \| hmask;
	252	hash_page_patch_C[0] = (hash_page_patch_C[0] & ~0xffff) \| hmask;
	253
	254	/*
	255	* Ensure that the locations we've patched have been written
	256	* out from the data cache and invalidated in the instruction
	257	* cache, on those machines with split caches.
	258	*/
	259	flush_icache_range((unsigned long) &hash_page_patch_A[0],
	260	(unsigned long) &hash_page_patch_C[1]);
	261
	262	/*
	263	* Patch up the instructions in hashtable.S:flush_hash_page
	264	*/
	265	flush_hash_patch_A[0] = (flush_hash_patch_A[0] & ~0xffff)
	266	\| ((unsigned int)(Hash) >> 16);
	267	flush_hash_patch_A[1] = (flush_hash_patch_A[1] & ~0x7c0) \| (mb << 6);
	268	flush_hash_patch_A[2] = (flush_hash_patch_A[2] & ~0x7c0) \| (mb2 << 6);
	269	flush_hash_patch_B[0] = (flush_hash_patch_B[0] & ~0xffff) \| hmask;
	270	flush_icache_range((unsigned long) &flush_hash_patch_A[0],
	271	(unsigned long) &flush_hash_patch_B[1]);
	272
	273	if ( ppc_md.progress ) ppc_md.progress("hash:done", 0x205);
	274	}