htab.c

这个linux源代码是很全面的~基本完整了~使用c编译的~由于时间问题我没有亲自测试~但就算用来做参考资料也是非常好的

/*
 * PowerPC64 port by Mike Corrigan and Dave Engebretsen
 *   {mikejc|engebret}@us.ibm.com
 *
 *    Copyright (c) 2000 Mike Corrigan <[e-mail]>
 *
 * SMP scalability work:
 *    Copyright (C) 2001 Anton Blanchard <[e-mail]>, IBM
 * 
 *    Module name: htab.c
 *
 *    Description:
 *      PowerPC Hashed Page Table functions
 *
 * 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/config.h>
#include <linux/spinlock.h>
#include <linux/errno.h>
#include <linux/sched.h>
#include <linux/proc_fs.h>
#include <linux/stat.h>
#include <linux/sysctl.h>
#include <linux/ctype.h>
#include <linux/cache.h>

#include <asm/ppcdebug.h>
#include <asm/processor.h>
#include <asm/pgtable.h>
#include <asm/mmu.h>
#include <asm/mmu_context.h>
#include <asm/page.h>
#include <asm/types.h>
#include <asm/uaccess.h>
#include <asm/naca.h>
#include <asm/pmc.h>
#include <asm/machdep.h>
#include <asm/lmb.h>
#include <asm/abs_addr.h>
#include <asm/io.h>
#include <asm/eeh.h>
#include <asm/hvcall.h>
#include <asm/iSeries/LparData.h>
#include <asm/iSeries/HvCallHpt.h>

/*
 * Note:  pte   --> Linux PTE
 *        HPTE  --> PowerPC Hashed Page Table Entry
 *
 * Execution context:
 *   htab_initialize is called with the MMU off (of course), but
 *   the kernel has been copied down to zero so it can directly
 *   reference global data.  At this point it is very difficult
 *   to print debug info.
 *
 */

HTAB htab_data = {NULL, 0, 0, 0, 0};

extern unsigned long _SDR1;
extern unsigned long klimit;

void make_pte(HPTE *htab, unsigned long va, unsigned long pa,
	      int mode, unsigned long hash_mask, int large);
long plpar_pte_enter(unsigned long flags,
		     unsigned long ptex,
		     unsigned long new_pteh, unsigned long new_ptel,
		     unsigned long *old_pteh_ret, unsigned long *old_ptel_ret);
static long hpte_remove(unsigned long hpte_group);
static long rpa_lpar_hpte_remove(unsigned long hpte_group);
static long iSeries_hpte_remove(unsigned long hpte_group);

static spinlock_t pSeries_tlbie_lock = SPIN_LOCK_UNLOCKED;
static spinlock_t pSeries_lpar_tlbie_lock = SPIN_LOCK_UNLOCKED;
spinlock_t hash_table_lock __cacheline_aligned_in_smp = SPIN_LOCK_UNLOCKED;

#define KB (1024)
#define MB (1024*KB)

static inline void
loop_forever(void)
{
	volatile unsigned long x = 1;
	for(;x;x|=1)
		;
}

static inline void
create_pte_mapping(unsigned long start, unsigned long end,
		   unsigned long mode, unsigned long mask, int large)
{
	unsigned long addr;
	HPTE *htab = (HPTE *)__v2a(htab_data.htab);
	unsigned int step;

	if (large)
		step = 16*MB;
	else
		step = 4*KB;

	for (addr = start; addr < end; addr += step) {
		unsigned long vsid = get_kernel_vsid(addr);
		unsigned long va = (vsid << 28) | (addr & 0xfffffff);
		make_pte(htab, va, (unsigned long)__v2a(addr), 
			 mode, mask, large);
	}
}

void
htab_initialize(void)
{
	unsigned long table, htab_size_bytes;
	unsigned long pteg_count;
	unsigned long mode_rw, mask;

#if 0
	/* Can't really do the call below since it calls the normal RTAS
	 * entry point and we're still relocate off at the moment.
	 * Temporarily diabling until it can call through the relocate off
	 * RTAS entry point.  -Peter
	 */
	ppc64_boot_msg(0x05, "htab init");
#endif
	/*
	 * Calculate the required size of the htab.  We want the number of
	 * PTEGs to equal one half the number of real pages.
	 */ 
	htab_size_bytes = 1UL << naca->pftSize;
	pteg_count = htab_size_bytes >> 7;

	/* For debug, make the HTAB 1/8 as big as it normally would be. */
	ifppcdebug(PPCDBG_HTABSIZE) {
		pteg_count >>= 3;
		htab_size_bytes = pteg_count << 7;
	}

	htab_data.htab_num_ptegs = pteg_count;
	htab_data.htab_hash_mask = pteg_count - 1;

	if(naca->platform == PLATFORM_PSERIES) {
		/* Find storage for the HPT.  Must be contiguous in
		 * the absolute address space.
		 */
		table = lmb_alloc(htab_size_bytes, htab_size_bytes);
		if ( !table ) {
			ppc64_terminate_msg(0x20, "hpt space");
			loop_forever();
		}
		htab_data.htab = (HPTE *)__a2v(table);

		/* htab absolute addr + encoded htabsize */
		_SDR1 = table + __ilog2(pteg_count) - 11;

		/* Initialize the HPT with no entries */
		memset((void *)table, 0, htab_size_bytes);
	} else {
		/* Using a hypervisor which owns the htab */
		htab_data.htab = NULL;
		_SDR1 = 0; 
	}

	mode_rw = _PAGE_ACCESSED | _PAGE_COHERENT | PP_RWXX;
	mask = pteg_count-1;

	/* XXX we currently map kernel text rw, should fix this */
	if ((naca->platform & PLATFORM_PSERIES) &&
	   cpu_has_largepage() && (naca->physicalMemorySize > 256*MB)) {
		create_pte_mapping((unsigned long)KERNELBASE, 
				   KERNELBASE + 256*MB, mode_rw, mask, 0);
		create_pte_mapping((unsigned long)KERNELBASE + 256*MB, 
				   KERNELBASE + (naca->physicalMemorySize), 
				   mode_rw, mask, 1);
	} else {
		create_pte_mapping((unsigned long)KERNELBASE, 
				   KERNELBASE+(naca->physicalMemorySize), 
				   mode_rw, mask, 0);
	}
#if 0
	/* Can't really do the call below since it calls the normal RTAS
	 * entry point and we're still relocate off at the moment.
	 * Temporarily diabling until it can call through the relocate off
	 * RTAS entry point.  -Peter
	 */
	ppc64_boot_msg(0x06, "htab done");
#endif
}
#undef KB
#undef MB

/*
 * Create a pte. Used during initialization only.
 * We assume the PTE will fit in the primary PTEG.
 */
void make_pte(HPTE *htab, unsigned long va, unsigned long pa,
	      int mode, unsigned long hash_mask, int large)
{
	HPTE *hptep, local_hpte, rhpte;
	unsigned long hash, vpn, flags, lpar_rc;
	unsigned long i, dummy1, dummy2;
	long slot;

	if (large)
		vpn = va >> LARGE_PAGE_SHIFT;
	else
		vpn = va >> PAGE_SHIFT;

	hash = hpt_hash(vpn, large);

	local_hpte.dw1.dword1 = pa | mode;
	local_hpte.dw0.dword0 = 0;
	local_hpte.dw0.dw0.avpn = va >> 23;
	local_hpte.dw0.dw0.bolted = 1;		/* bolted */
	if (large) {
		local_hpte.dw0.dw0.l = 1;	/* large page */
		local_hpte.dw0.dw0.avpn &= ~0x1UL;
	}
	local_hpte.dw0.dw0.v = 1;

	if (naca->platform == PLATFORM_PSERIES) {
		hptep  = htab + ((hash & hash_mask)*HPTES_PER_GROUP);

		for (i = 0; i < 8; ++i, ++hptep) {
			if (hptep->dw0.dw0.v == 0) {		/* !valid */
				*hptep = local_hpte;
				return;
			}
		}
	} else if (naca->platform == PLATFORM_PSERIES_LPAR) {
		slot = ((hash & hash_mask)*HPTES_PER_GROUP);
		
		/* Set CEC cookie to 0                   */
		/* Zero page = 0                         */
		/* I-cache Invalidate = 0                */
		/* I-cache synchronize = 0               */
		/* Exact = 0 - modify any entry in group */
		flags = 0;
		
		lpar_rc =  plpar_pte_enter(flags, slot, local_hpte.dw0.dword0,
					   local_hpte.dw1.dword1, 
					   &dummy1, &dummy2);
		if (lpar_rc != H_Success) {
			ppc64_terminate_msg(0x21, "hpte enter");
			loop_forever();
		}
		return;
	} else if (naca->platform == PLATFORM_ISERIES_LPAR) {
		slot = HvCallHpt_findValid(&rhpte, vpn);
		if (slot < 0) {
			/* Must find space in primary group */
			panic("hash_page: hpte already exists\n");
		}
		HvCallHpt_addValidate(slot, 0, (HPTE *)&local_hpte );
		return;
	}

	/* We should _never_ get here and too early to call xmon. */
	ppc64_terminate_msg(0x22, "hpte platform");
	loop_forever();
}

/*
 * find_linux_pte returns the address of a linux pte for a given 
 * effective address and directory.  If not found, it returns zero.
 */
pte_t *find_linux_pte(pgd_t *pgdir, unsigned long ea)
{
	pgd_t *pg;
	pmd_t *pm;
	pte_t *pt = NULL;
	pte_t pte;

	pg = pgdir + pgd_index(ea);
	if (!pgd_none(*pg)) {
		pm = pmd_offset(pg, ea);
		if (!pmd_none(*pm)) { 
			pt = pte_offset(pm, ea);
			pte = *pt;
			if (!pte_present(pte))
				pt = NULL;
		}
	}

	return pt;
}

static inline unsigned long computeHptePP(unsigned long pte)
{
	return (pte & _PAGE_USER) |
		(((pte & _PAGE_USER) >> 1) &
		 ((~((pte >> 2) &	/* _PAGE_RW */
		     (pte >> 7))) &	/* _PAGE_DIRTY */
		  1));
}

/*
 * Handle a fault by adding an HPTE. If the address can't be determined
 * to be valid via Linux page tables, return 1. If handled return 0
 */
int __hash_page(unsigned long ea, unsigned long access, 
		unsigned long vsid, pte_t *ptep)
{
	unsigned long va, vpn;
	unsigned long newpp, prpn;
	unsigned long hpteflags;
	long slot;
	pte_t old_pte, new_pte;

	/* Search the Linux page table for a match with va */
	va = (vsid << 28) | (ea & 0x0fffffff);
	vpn = va >> PAGE_SHIFT;

	/* Acquire the hash table lock to guarantee that the linux
	 * pte we fetch will not change
	 */
	spin_lock( &hash_table_lock );
	
	/* 
	 * Check the user's access rights to the page.  If access should be
	 * prevented then send the problem up to do_page_fault.
	 */
	access |= _PAGE_PRESENT;
	if (unlikely(access & ~(pte_val(*ptep)))) {
		spin_unlock( &hash_table_lock );
		return 1;
	}

	/* 
	 * We have found a pte (which was present).
	 * The spinlocks prevent this status from changing
	 * The hash_table_lock prevents the _PAGE_HASHPTE status
	 * from changing (RPN, DIRTY and ACCESSED too)
	 * The page_table_lock prevents the pte from being 
	 * invalidated or modified
	 */

	/*
	 * At this point, we have a pte (old_pte) which can be used to build
	 * or update an HPTE. There are 2 cases:
	 *
	 * 1. There is a valid (present) pte with no associated HPTE (this is 
	 *	the most common case)
	 * 2. There is a valid (present) pte with an associated HPTE. The
	 *	current values of the pp bits in the HPTE prevent access
	 *	because we are doing software DIRTY bit management and the
	 *	page is currently not DIRTY. 
	 */

	old_pte = *ptep;
	new_pte = old_pte;

	/* If the attempted access was a store */
	if (access & _PAGE_RW)
		pte_val(new_pte) |= _PAGE_ACCESSED | _PAGE_DIRTY;
	else
		pte_val(new_pte) |= _PAGE_ACCESSED;

	newpp = computeHptePP(pte_val(new_pte));
	
	/* Check if pte already has an hpte (case 2) */
	if (unlikely(pte_val(old_pte) & _PAGE_HASHPTE)) {
		/* There MIGHT be an HPTE for this pte */
		unsigned long hash, slot, secondary;

		/* XXX fix large pte flag */
		hash = hpt_hash(vpn, 0);
		secondary = (pte_val(old_pte) & _PAGE_SECONDARY) >> 15;
		if (secondary)
			hash = ~hash;
		slot = (hash & htab_data.htab_hash_mask) * HPTES_PER_GROUP;
		slot += (pte_val(old_pte) & _PAGE_GROUP_IX) >> 12;

		/* XXX fix large pte flag */
		if (ppc_md.hpte_updatepp(slot, secondary, 
					 newpp, va, 0) == -1) {
			pte_val(old_pte) &= ~_PAGE_HPTEFLAGS;
		} else {
			if (!pte_same(old_pte, new_pte)) {
				*ptep = new_pte;
			}
		}
	}

	if (likely(!(pte_val(old_pte) & _PAGE_HASHPTE))) {
		/* Update the linux pte with the HPTE slot */
		pte_val(new_pte) &= ~_PAGE_HPTEFLAGS;
		pte_val(new_pte) |= _PAGE_HASHPTE;
		prpn = pte_val(old_pte) >> PTE_SHIFT;

		/* copy appropriate flags from linux pte */
		hpteflags = (pte_val(new_pte) & 0x1f8) | newpp;

		slot = ppc_md.hpte_insert(vpn, prpn, hpteflags, 0, 0);

		pte_val(new_pte) |= ((slot<<12) & 
				     (_PAGE_GROUP_IX | _PAGE_SECONDARY));

		*ptep = new_pte;
	}

	spin_unlock(&hash_table_lock);

	return 0;
}

/*
 * Handle a fault by adding an HPTE. If the address can't be determined
 * to be valid via Linux page tables, return 1. If handled return 0
 */
int hash_page(unsigned long ea, unsigned long access)
{
	void *pgdir;
	unsigned long vsid;
	struct mm_struct *mm;
	pte_t *ptep;
	int ret;

	/* Check for invalid addresses. */
	if (!IS_VALID_EA(ea)) return 1;

 	switch (REGION_ID(ea)) {
	case USER_REGION_ID:
		mm = current->mm;
		if (mm == NULL) return 1;
		vsid = get_vsid(mm->context, ea);
		break;
	case IO_REGION_ID:
		mm = &ioremap_mm;
		vsid = get_kernel_vsid(ea);
		break;
	case VMALLOC_REGION_ID:
		mm = &init_mm;
		vsid = get_kernel_vsid(ea);
		break;
	case IO_UNMAPPED_REGION_ID:
		udbg_printf("EEH Error ea = 0x%lx\n", ea);
		PPCDBG_ENTER_DEBUGGER();
		panic("EEH Error ea = 0x%lx\n", ea);
		break;
	case KERNEL_REGION_ID:
		/*
		 * As htab_initialize is now, we shouldn't ever get here since
		 * we're bolting the entire 0xC0... region.
		 */
		udbg_printf("Little faulted on kernel address 0x%lx\n", ea);
		PPCDBG_ENTER_DEBUGGER();
		panic("Little faulted on kernel address 0x%lx\n", ea);
		break;
	default:
		/* Not a valid range, send the problem up to do_page_fault */
		return 1;
		break;
	}