init.c

来自「Linux Kernel 2.6.9 for OMAP1710」· C语言 代码 · 共 1,774 行 · 第 1/4 页

/*  $Id: init.c,v 1.209 2002/02/09 19:49:31 davem Exp $
 *  arch/sparc64/mm/init.c
 *
 *  Copyright (C) 1996-1999 David S. Miller ([e-mail])
 *  Copyright (C) 1997-1999 Jakub Jelinek ([e-mail])
 */
 
#include <linux/config.h>
#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/string.h>
#include <linux/init.h>
#include <linux/bootmem.h>
#include <linux/mm.h>
#include <linux/hugetlb.h>
#include <linux/slab.h>
#include <linux/initrd.h>
#include <linux/swap.h>
#include <linux/pagemap.h>
#include <linux/fs.h>
#include <linux/seq_file.h>

#include <asm/head.h>
#include <asm/system.h>
#include <asm/page.h>
#include <asm/pgalloc.h>
#include <asm/pgtable.h>
#include <asm/oplib.h>
#include <asm/iommu.h>
#include <asm/io.h>
#include <asm/uaccess.h>
#include <asm/mmu_context.h>
#include <asm/tlbflush.h>
#include <asm/dma.h>
#include <asm/starfire.h>
#include <asm/tlb.h>
#include <asm/spitfire.h>
#include <asm/sections.h>

extern void device_scan(void);

struct sparc_phys_banks sp_banks[SPARC_PHYS_BANKS];

unsigned long *sparc64_valid_addr_bitmap;

/* Ugly, but necessary... -DaveM */
unsigned long phys_base;
unsigned long kern_base;
unsigned long kern_size;
unsigned long pfn_base;

/* This is even uglier. We have a problem where the kernel may not be
 * located at phys_base. However, initial __alloc_bootmem() calls need to
 * be adjusted to be within the 4-8Megs that the kernel is mapped to, else
 * those page mappings wont work. Things are ok after inherit_prom_mappings
 * is called though. Dave says he'll clean this up some other time.
 * -- BenC
 */
static unsigned long bootmap_base;

/* get_new_mmu_context() uses "cache + 1".  */
spinlock_t ctx_alloc_lock = SPIN_LOCK_UNLOCKED;
unsigned long tlb_context_cache = CTX_FIRST_VERSION - 1;
#define CTX_BMAP_SLOTS (1UL << (CTX_VERSION_SHIFT - 6))
unsigned long mmu_context_bmap[CTX_BMAP_SLOTS];

/* References to special section boundaries */
extern char  _start[], _end[];

/* Initial ramdisk setup */
extern unsigned long sparc_ramdisk_image64;
extern unsigned int sparc_ramdisk_image;
extern unsigned int sparc_ramdisk_size;

struct page *mem_map_zero;

int bigkernel = 0;

/* XXX Tune this... */
#define PGT_CACHE_LOW	25
#define PGT_CACHE_HIGH	50

void check_pgt_cache(void)
{
	preempt_disable();
	if (pgtable_cache_size > PGT_CACHE_HIGH) {
		do {
#ifdef CONFIG_SMP
			if (pgd_quicklist)
				free_pgd_slow(get_pgd_fast());
#endif
			if (pte_quicklist[0])
				free_pte_slow(pte_alloc_one_fast(NULL, 0));
			if (pte_quicklist[1])
				free_pte_slow(pte_alloc_one_fast(NULL, 1 << (PAGE_SHIFT + 10)));
		} while (pgtable_cache_size > PGT_CACHE_LOW);
	}
#ifndef CONFIG_SMP
        if (pgd_cache_size > PGT_CACHE_HIGH / 4) {
		struct page *page, *page2;
                for (page2 = NULL, page = (struct page *)pgd_quicklist; page;) {
                        if ((unsigned long)page->lru.prev == 3) {
                                if (page2)
                                        page2->lru.next = page->lru.next;
                                else
                                        pgd_quicklist = (void *) page->lru.next;
                                pgd_cache_size -= 2;
                                __free_page(page);
                                if (page2)
                                        page = (struct page *)page2->lru.next;
                                else
                                        page = (struct page *)pgd_quicklist;
                                if (pgd_cache_size <= PGT_CACHE_LOW / 4)
                                        break;
                                continue;
                        }
                        page2 = page;
                        page = (struct page *)page->lru.next;
                }
        }
#endif
	preempt_enable();
}

#ifdef CONFIG_DEBUG_DCFLUSH
atomic_t dcpage_flushes = ATOMIC_INIT(0);
#ifdef CONFIG_SMP
atomic_t dcpage_flushes_xcall = ATOMIC_INIT(0);
#endif
#endif

__inline__ void flush_dcache_page_impl(struct page *page)
{
#ifdef CONFIG_DEBUG_DCFLUSH
	atomic_inc(&dcpage_flushes);
#endif

#if (L1DCACHE_SIZE > PAGE_SIZE)
	__flush_dcache_page(page_address(page),
			    ((tlb_type == spitfire) &&
			     page_mapping(page) != NULL));
#else
	if (page_mapping(page) != NULL &&
	    tlb_type == spitfire)
		__flush_icache_page(__pa(page_address(page)));
#endif
}

#define PG_dcache_dirty		PG_arch_1

#define dcache_dirty_cpu(page) \
	(((page)->flags >> 24) & (NR_CPUS - 1UL))

static __inline__ void set_dcache_dirty(struct page *page, int this_cpu)
{
	unsigned long mask = this_cpu;
	unsigned long non_cpu_bits = ~((NR_CPUS - 1UL) << 24UL);
	mask = (mask << 24) | (1UL << PG_dcache_dirty);
	__asm__ __volatile__("1:\n\t"
			     "ldx	[%2], %%g7\n\t"
			     "and	%%g7, %1, %%g5\n\t"
			     "or	%%g5, %0, %%g5\n\t"
			     "casx	[%2], %%g7, %%g5\n\t"
			     "cmp	%%g7, %%g5\n\t"
			     "bne,pn	%%xcc, 1b\n\t"
			     " membar	#StoreLoad | #StoreStore"
			     : /* no outputs */
			     : "r" (mask), "r" (non_cpu_bits), "r" (&page->flags)
			     : "g5", "g7");
}

static __inline__ void clear_dcache_dirty_cpu(struct page *page, unsigned long cpu)
{
	unsigned long mask = (1UL << PG_dcache_dirty);

	__asm__ __volatile__("! test_and_clear_dcache_dirty\n"
			     "1:\n\t"
			     "ldx	[%2], %%g7\n\t"
			     "srlx	%%g7, 24, %%g5\n\t"
			     "and	%%g5, %3, %%g5\n\t"
			     "cmp	%%g5, %0\n\t"
			     "bne,pn	%%icc, 2f\n\t"
			     " andn	%%g7, %1, %%g5\n\t"
			     "casx	[%2], %%g7, %%g5\n\t"
			     "cmp	%%g7, %%g5\n\t"
			     "bne,pn	%%xcc, 1b\n\t"
			     " membar	#StoreLoad | #StoreStore\n"
			     "2:"
			     : /* no outputs */
			     : "r" (cpu), "r" (mask), "r" (&page->flags),
			       "i" (NR_CPUS - 1UL)
			     : "g5", "g7");
}

extern void __update_mmu_cache(unsigned long mmu_context_hw, unsigned long address, pte_t pte, int code);

void update_mmu_cache(struct vm_area_struct *vma, unsigned long address, pte_t pte)
{
	struct page *page;
	unsigned long pfn;
	unsigned long pg_flags;

	pfn = pte_pfn(pte);
	if (pfn_valid(pfn) &&
	    (page = pfn_to_page(pfn), page_mapping(page)) &&
	    ((pg_flags = page->flags) & (1UL << PG_dcache_dirty))) {
		int cpu = ((pg_flags >> 24) & (NR_CPUS - 1UL));
		int this_cpu = get_cpu();

		/* This is just to optimize away some function calls
		 * in the SMP case.
		 */
		if (cpu == this_cpu)
			flush_dcache_page_impl(page);
		else
			smp_flush_dcache_page_impl(page, cpu);

		clear_dcache_dirty_cpu(page, cpu);

		put_cpu();
	}
	if (get_thread_fault_code())
		__update_mmu_cache(vma->vm_mm->context & TAG_CONTEXT_BITS,
				   address, pte, get_thread_fault_code());
}

void flush_dcache_page(struct page *page)
{
	struct address_space *mapping = page_mapping(page);
	int dirty = test_bit(PG_dcache_dirty, &page->flags);
	int dirty_cpu = dcache_dirty_cpu(page);
	int this_cpu = get_cpu();

	if (mapping && !mapping_mapped(mapping)) {
		if (dirty) {
			if (dirty_cpu == this_cpu)
				goto out;
			smp_flush_dcache_page_impl(page, dirty_cpu);
		}
		set_dcache_dirty(page, this_cpu);
	} else {
		/* We could delay the flush for the !page_mapping
		 * case too.  But that case is for exec env/arg
		 * pages and those are %99 certainly going to get
		 * faulted into the tlb (and thus flushed) anyways.
		 */
		flush_dcache_page_impl(page);
	}

out:
	put_cpu();
}

void flush_icache_range(unsigned long start, unsigned long end)
{
	/* Cheetah has coherent I-cache. */
	if (tlb_type == spitfire) {
		unsigned long kaddr;

		for (kaddr = start; kaddr < end; kaddr += PAGE_SIZE)
			__flush_icache_page(__get_phys(kaddr));
	}
}

unsigned long page_to_pfn(struct page *page)
{
	return (unsigned long) ((page - mem_map) + pfn_base);
}

struct page *pfn_to_page(unsigned long pfn)
{
	return (mem_map + (pfn - pfn_base));
}

void show_mem(void)
{
	printk("Mem-info:\n");
	show_free_areas();
	printk("Free swap:       %6ldkB\n",
	       nr_swap_pages << (PAGE_SHIFT-10));
	printk("%ld pages of RAM\n", num_physpages);
	printk("%d free pages\n", nr_free_pages());
	printk("%d pages in page table cache\n",pgtable_cache_size);
#ifndef CONFIG_SMP
	printk("%d entries in page dir cache\n",pgd_cache_size);
#endif	
}

void mmu_info(struct seq_file *m)
{
	if (tlb_type == cheetah)
		seq_printf(m, "MMU Type\t: Cheetah\n");
	else if (tlb_type == cheetah_plus)
		seq_printf(m, "MMU Type\t: Cheetah+\n");
	else if (tlb_type == spitfire)
		seq_printf(m, "MMU Type\t: Spitfire\n");
	else
		seq_printf(m, "MMU Type\t: ???\n");

#ifdef CONFIG_DEBUG_DCFLUSH
	seq_printf(m, "DCPageFlushes\t: %d\n",
		   atomic_read(&dcpage_flushes));
#ifdef CONFIG_SMP
	seq_printf(m, "DCPageFlushesXC\t: %d\n",
		   atomic_read(&dcpage_flushes_xcall));
#endif /* CONFIG_SMP */
#endif /* CONFIG_DEBUG_DCFLUSH */
}

struct linux_prom_translation {
	unsigned long virt;
	unsigned long size;
	unsigned long data;
};

extern unsigned long prom_boot_page;
extern void prom_remap(unsigned long physpage, unsigned long virtpage, int mmu_ihandle);
extern int prom_get_mmu_ihandle(void);
extern void register_prom_callbacks(void);

/* Exported for SMP bootup purposes. */
unsigned long kern_locked_tte_data;

void __init early_pgtable_allocfail(char *type)
{
	prom_printf("inherit_prom_mappings: Cannot alloc kernel %s.\n", type);
	prom_halt();
}

#define BASE_PAGE_SIZE 8192
static pmd_t *prompmd;

/*
 * Translate PROM's mapping we capture at boot time into physical address.
 * The second parameter is only set from prom_callback() invocations.
 */
unsigned long prom_virt_to_phys(unsigned long promva, int *error)
{
	pmd_t *pmdp = prompmd + ((promva >> 23) & 0x7ff);
	pte_t *ptep;
	unsigned long base;

	if (pmd_none(*pmdp)) {
		if (error)
			*error = 1;
		return(0);
	}
	ptep = (pte_t *)__pmd_page(*pmdp) + ((promva >> 13) & 0x3ff);
	if (!pte_present(*ptep)) {
		if (error)
			*error = 1;
		return(0);
	}
	if (error) {
		*error = 0;
		return(pte_val(*ptep));
	}
	base = pte_val(*ptep) & _PAGE_PADDR;
	return(base + (promva & (BASE_PAGE_SIZE - 1)));
}

static void inherit_prom_mappings(void)
{
	struct linux_prom_translation *trans;
	unsigned long phys_page, tte_vaddr, tte_data;
	void (*remap_func)(unsigned long, unsigned long, int);
	pmd_t *pmdp;
	pte_t *ptep;
	int node, n, i, tsz;
	extern unsigned int obp_iaddr_patch[2], obp_daddr_patch[2];

	node = prom_finddevice("/virtual-memory");
	n = prom_getproplen(node, "translations");
	if (n == 0 || n == -1) {
		prom_printf("Couldn't get translation property\n");
		prom_halt();
	}
	n += 5 * sizeof(struct linux_prom_translation);
	for (tsz = 1; tsz < n; tsz <<= 1)
		/* empty */;
	trans = __alloc_bootmem(tsz, SMP_CACHE_BYTES, bootmap_base);
	if (trans == NULL) {
		prom_printf("inherit_prom_mappings: Cannot alloc translations.\n");
		prom_halt();
	}
	memset(trans, 0, tsz);

	if ((n = prom_getproperty(node, "translations", (char *)trans, tsz)) == -1) {
		prom_printf("Couldn't get translation property\n");
		prom_halt();
	}
	n = n / sizeof(*trans);

	/*
	 * The obp translations are saved based on 8k pagesize, since obp can use
	 * a mixture of pagesizes. Misses to the 0xf0000000 - 0x100000000, ie obp 
	 * range, are handled in entry.S and do not use the vpte scheme (see rant
	 * in inherit_locked_prom_mappings()).
	 */
#define OBP_PMD_SIZE 2048
	prompmd = __alloc_bootmem(OBP_PMD_SIZE, OBP_PMD_SIZE, bootmap_base);
	if (prompmd == NULL)
		early_pgtable_allocfail("pmd");
	memset(prompmd, 0, OBP_PMD_SIZE);
	for (i = 0; i < n; i++) {
		unsigned long vaddr;

		if (trans[i].virt >= LOW_OBP_ADDRESS && trans[i].virt < HI_OBP_ADDRESS) {
			for (vaddr = trans[i].virt;
			     ((vaddr < trans[i].virt + trans[i].size) && 
			     (vaddr < HI_OBP_ADDRESS));
			     vaddr += BASE_PAGE_SIZE) {
				unsigned long val;

				pmdp = prompmd + ((vaddr >> 23) & 0x7ff);
				if (pmd_none(*pmdp)) {
					ptep = __alloc_bootmem(BASE_PAGE_SIZE,
							       BASE_PAGE_SIZE,
							       bootmap_base);
					if (ptep == NULL)
						early_pgtable_allocfail("pte");
					memset(ptep, 0, BASE_PAGE_SIZE);
					pmd_set(pmdp, ptep);
				}
				ptep = (pte_t *)__pmd_page(*pmdp) +
						((vaddr >> 13) & 0x3ff);

				val = trans[i].data;

				/* Clear diag TTE bits. */
				if (tlb_type == spitfire)
					val &= ~0x0003fe0000000000UL;

				set_pte (ptep, __pte(val | _PAGE_MODIFIED));
				trans[i].data += BASE_PAGE_SIZE;
			}
		}
	}
	phys_page = __pa(prompmd);
	obp_iaddr_patch[0] |= (phys_page >> 10);
	obp_iaddr_patch[1] |= (phys_page & 0x3ff);
	flushi((long)&obp_iaddr_patch[0]);
	obp_daddr_patch[0] |= (phys_page >> 10);
	obp_daddr_patch[1] |= (phys_page & 0x3ff);