diff --git a/arch/sparc64/mm/Makefile b/arch/sparc64/mm/Makefile
new file mode 100644
index 0000000..cda8733
--- /dev/null
+++ b/arch/sparc64/mm/Makefile
@@ -0,0 +1,10 @@
+# $Id: Makefile,v 1.8 2000/12/14 22:57:25 davem Exp $
+# Makefile for the linux Sparc64-specific parts of the memory manager.
+#
+
+EXTRA_AFLAGS := -ansi
+EXTRA_CFLAGS := -Werror
+
+obj-y    := ultra.o tlb.o fault.o init.o generic.o extable.o
+
+obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpage.o
diff --git a/arch/sparc64/mm/extable.c b/arch/sparc64/mm/extable.c
new file mode 100644
index 0000000..ec33429
--- /dev/null
+++ b/arch/sparc64/mm/extable.c
@@ -0,0 +1,80 @@
+/*
+ * linux/arch/sparc64/mm/extable.c
+ */
+
+#include <linux/config.h>
+#include <linux/module.h>
+#include <asm/uaccess.h>
+
+extern const struct exception_table_entry __start___ex_table[];
+extern const struct exception_table_entry __stop___ex_table[];
+
+void sort_extable(struct exception_table_entry *start,
+		  struct exception_table_entry *finish)
+{
+}
+
+/* Caller knows they are in a range if ret->fixup == 0 */
+const struct exception_table_entry *
+search_extable(const struct exception_table_entry *start,
+	       const struct exception_table_entry *last,
+	       unsigned long value)
+{
+	const struct exception_table_entry *walk;
+
+	/* Single insn entries are encoded as:
+	 *	word 1:	insn address
+	 *	word 2:	fixup code address
+	 *
+	 * Range entries are encoded as:
+	 *	word 1: first insn address
+	 *	word 2: 0
+	 *	word 3: last insn address + 4 bytes
+	 *	word 4: fixup code address
+	 *
+	 * See asm/uaccess.h for more details.
+	 */
+
+	/* 1. Try to find an exact match. */
+	for (walk = start; walk <= last; walk++) {
+		if (walk->fixup == 0) {
+			/* A range entry, skip both parts. */
+			walk++;
+			continue;
+		}
+
+		if (walk->insn == value)
+			return walk;
+	}
+
+	/* 2. Try to find a range match. */
+	for (walk = start; walk <= (last - 1); walk++) {
+		if (walk->fixup)
+			continue;
+
+		if (walk[0].insn <= value && walk[1].insn > value)
+			return walk;
+
+		walk++;
+	}
+
+        return NULL;
+}
+
+/* Special extable search, which handles ranges.  Returns fixup */
+unsigned long search_extables_range(unsigned long addr, unsigned long *g2)
+{
+	const struct exception_table_entry *entry;
+
+	entry = search_exception_tables(addr);
+	if (!entry)
+		return 0;
+
+	/* Inside range?  Fix g2 and return correct fixup */
+	if (!entry->fixup) {
+		*g2 = (addr - entry->insn) / 4;
+		return (entry + 1)->fixup;
+	}
+
+	return entry->fixup;
+}
diff --git a/arch/sparc64/mm/fault.c b/arch/sparc64/mm/fault.c
new file mode 100644
index 0000000..3ffee7b
--- /dev/null
+++ b/arch/sparc64/mm/fault.c
@@ -0,0 +1,527 @@
+/* $Id: fault.c,v 1.59 2002/02/09 19:49:31 davem Exp $
+ * arch/sparc64/mm/fault.c: Page fault handlers for the 64-bit Sparc.
+ *
+ * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
+ * Copyright (C) 1997, 1999 Jakub Jelinek (jj@ultra.linux.cz)
+ */
+
+#include <asm/head.h>
+
+#include <linux/string.h>
+#include <linux/types.h>
+#include <linux/sched.h>
+#include <linux/ptrace.h>
+#include <linux/mman.h>
+#include <linux/signal.h>
+#include <linux/mm.h>
+#include <linux/module.h>
+#include <linux/smp_lock.h>
+#include <linux/init.h>
+#include <linux/interrupt.h>
+
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/openprom.h>
+#include <asm/oplib.h>
+#include <asm/uaccess.h>
+#include <asm/asi.h>
+#include <asm/lsu.h>
+#include <asm/sections.h>
+#include <asm/kdebug.h>
+
+#define ELEMENTS(arr) (sizeof (arr)/sizeof (arr[0]))
+
+extern struct sparc_phys_banks sp_banks[SPARC_PHYS_BANKS];
+
+/*
+ * To debug kernel during syscall entry.
+ */
+void syscall_trace_entry(struct pt_regs *regs)
+{
+	printk("scall entry: %s[%d]/cpu%d: %d\n", current->comm, current->pid, smp_processor_id(), (int) regs->u_regs[UREG_G1]);
+}
+
+/*
+ * To debug kernel during syscall exit.
+ */
+void syscall_trace_exit(struct pt_regs *regs)
+{
+	printk("scall exit: %s[%d]/cpu%d: %d\n", current->comm, current->pid, smp_processor_id(), (int) regs->u_regs[UREG_G1]);
+}
+
+/*
+ * To debug kernel to catch accesses to certain virtual/physical addresses.
+ * Mode = 0 selects physical watchpoints, mode = 1 selects virtual watchpoints.
+ * flags = VM_READ watches memread accesses, flags = VM_WRITE watches memwrite accesses.
+ * Caller passes in a 64bit aligned addr, with mask set to the bytes that need to be
+ * watched. This is only useful on a single cpu machine for now. After the watchpoint
+ * is detected, the process causing it will be killed, thus preventing an infinite loop.
+ */
+void set_brkpt(unsigned long addr, unsigned char mask, int flags, int mode)
+{
+	unsigned long lsubits;
+
+	__asm__ __volatile__("ldxa [%%g0] %1, %0"
+			     : "=r" (lsubits)
+			     : "i" (ASI_LSU_CONTROL));
+	lsubits &= ~(LSU_CONTROL_PM | LSU_CONTROL_VM |
+		     LSU_CONTROL_PR | LSU_CONTROL_VR |
+		     LSU_CONTROL_PW | LSU_CONTROL_VW);
+
+	__asm__ __volatile__("stxa	%0, [%1] %2\n\t"
+			     "membar	#Sync"
+			     : /* no outputs */
+			     : "r" (addr), "r" (mode ? VIRT_WATCHPOINT : PHYS_WATCHPOINT),
+			       "i" (ASI_DMMU));
+
+	lsubits |= ((unsigned long)mask << (mode ? 25 : 33));
+	if (flags & VM_READ)
+		lsubits |= (mode ? LSU_CONTROL_VR : LSU_CONTROL_PR);
+	if (flags & VM_WRITE)
+		lsubits |= (mode ? LSU_CONTROL_VW : LSU_CONTROL_PW);
+	__asm__ __volatile__("stxa %0, [%%g0] %1\n\t"
+			     "membar #Sync"
+			     : /* no outputs */
+			     : "r" (lsubits), "i" (ASI_LSU_CONTROL)
+			     : "memory");
+}
+
+/* Nice, simple, prom library does all the sweating for us. ;) */
+unsigned long __init prom_probe_memory (void)
+{
+	register struct linux_mlist_p1275 *mlist;
+	register unsigned long bytes, base_paddr, tally;
+	register int i;
+
+	i = 0;
+	mlist = *prom_meminfo()->p1275_available;
+	bytes = tally = mlist->num_bytes;
+	base_paddr = mlist->start_adr;
+  
+	sp_banks[0].base_addr = base_paddr;
+	sp_banks[0].num_bytes = bytes;
+
+	while (mlist->theres_more != (void *) 0) {
+		i++;
+		mlist = mlist->theres_more;
+		bytes = mlist->num_bytes;
+		tally += bytes;
+		if (i >= SPARC_PHYS_BANKS-1) {
+			printk ("The machine has more banks than "
+				"this kernel can support\n"
+				"Increase the SPARC_PHYS_BANKS "
+				"setting (currently %d)\n",
+				SPARC_PHYS_BANKS);
+			i = SPARC_PHYS_BANKS-1;
+			break;
+		}
+    
+		sp_banks[i].base_addr = mlist->start_adr;
+		sp_banks[i].num_bytes = mlist->num_bytes;
+	}
+
+	i++;
+	sp_banks[i].base_addr = 0xdeadbeefbeefdeadUL;
+	sp_banks[i].num_bytes = 0;
+
+	/* Now mask all bank sizes on a page boundary, it is all we can
+	 * use anyways.
+	 */
+	for (i = 0; sp_banks[i].num_bytes != 0; i++)
+		sp_banks[i].num_bytes &= PAGE_MASK;
+
+	return tally;
+}
+
+static void unhandled_fault(unsigned long address, struct task_struct *tsk,
+			    struct pt_regs *regs)
+{
+	if ((unsigned long) address < PAGE_SIZE) {
+		printk(KERN_ALERT "Unable to handle kernel NULL "
+		       "pointer dereference\n");
+	} else {
+		printk(KERN_ALERT "Unable to handle kernel paging request "
+		       "at virtual address %016lx\n", (unsigned long)address);
+	}
+	printk(KERN_ALERT "tsk->{mm,active_mm}->context = %016lx\n",
+	       (tsk->mm ?
+		CTX_HWBITS(tsk->mm->context) :
+		CTX_HWBITS(tsk->active_mm->context)));
+	printk(KERN_ALERT "tsk->{mm,active_mm}->pgd = %016lx\n",
+	       (tsk->mm ? (unsigned long) tsk->mm->pgd :
+		          (unsigned long) tsk->active_mm->pgd));
+	if (notify_die(DIE_GPF, "general protection fault", regs,
+		       0, 0, SIGSEGV) == NOTIFY_STOP)
+		return;
+	die_if_kernel("Oops", regs);
+}
+
+static void bad_kernel_pc(struct pt_regs *regs)
+{
+	unsigned long *ksp;
+
+	printk(KERN_CRIT "OOPS: Bogus kernel PC [%016lx] in fault handler\n",
+	       regs->tpc);
+	__asm__("mov %%sp, %0" : "=r" (ksp));
+	show_stack(current, ksp);
+	unhandled_fault(regs->tpc, current, regs);
+}
+
+/*
+ * We now make sure that mmap_sem is held in all paths that call 
+ * this. Additionally, to prevent kswapd from ripping ptes from
+ * under us, raise interrupts around the time that we look at the
+ * pte, kswapd will have to wait to get his smp ipi response from
+ * us. This saves us having to get page_table_lock.
+ */
+static unsigned int get_user_insn(unsigned long tpc)
+{
+	pgd_t *pgdp = pgd_offset(current->mm, tpc);
+	pud_t *pudp;
+	pmd_t *pmdp;
+	pte_t *ptep, pte;
+	unsigned long pa;
+	u32 insn = 0;
+	unsigned long pstate;
+
+	if (pgd_none(*pgdp))
+		goto outret;
+	pudp = pud_offset(pgdp, tpc);
+	if (pud_none(*pudp))
+		goto outret;
+	pmdp = pmd_offset(pudp, tpc);
+	if (pmd_none(*pmdp))
+		goto outret;
+
+	/* This disables preemption for us as well. */
+	__asm__ __volatile__("rdpr %%pstate, %0" : "=r" (pstate));
+	__asm__ __volatile__("wrpr %0, %1, %%pstate"
+				: : "r" (pstate), "i" (PSTATE_IE));
+	ptep = pte_offset_map(pmdp, tpc);
+	pte = *ptep;
+	if (!pte_present(pte))
+		goto out;
+
+	pa  = (pte_val(pte) & _PAGE_PADDR);
+	pa += (tpc & ~PAGE_MASK);
+
+	/* Use phys bypass so we don't pollute dtlb/dcache. */
+	__asm__ __volatile__("lduwa [%1] %2, %0"
+			     : "=r" (insn)
+			     : "r" (pa), "i" (ASI_PHYS_USE_EC));
+
+out:
+	pte_unmap(ptep);
+	__asm__ __volatile__("wrpr %0, 0x0, %%pstate" : : "r" (pstate));
+outret:
+	return insn;
+}
+
+extern unsigned long compute_effective_address(struct pt_regs *, unsigned int, unsigned int);
+
+static void do_fault_siginfo(int code, int sig, struct pt_regs *regs,
+			     unsigned int insn, int fault_code)
+{
+	siginfo_t info;
+
+	info.si_code = code;
+	info.si_signo = sig;
+	info.si_errno = 0;
+	if (fault_code & FAULT_CODE_ITLB)
+		info.si_addr = (void __user *) regs->tpc;
+	else
+		info.si_addr = (void __user *)
+			compute_effective_address(regs, insn, 0);
+	info.si_trapno = 0;
+	force_sig_info(sig, &info, current);
+}
+
+extern int handle_ldf_stq(u32, struct pt_regs *);
+extern int handle_ld_nf(u32, struct pt_regs *);
+
+static unsigned int get_fault_insn(struct pt_regs *regs, unsigned int insn)
+{
+	if (!insn) {
+		if (!regs->tpc || (regs->tpc & 0x3))
+			return 0;
+		if (regs->tstate & TSTATE_PRIV) {
+			insn = *(unsigned int *) regs->tpc;
+		} else {
+			insn = get_user_insn(regs->tpc);
+		}
+	}
+	return insn;
+}
+
+static void do_kernel_fault(struct pt_regs *regs, int si_code, int fault_code,
+			    unsigned int insn, unsigned long address)
+{
+	unsigned long g2;
+	unsigned char asi = ASI_P;
+ 
+	if ((!insn) && (regs->tstate & TSTATE_PRIV))
+		goto cannot_handle;
+
+	/* If user insn could be read (thus insn is zero), that
+	 * is fine.  We will just gun down the process with a signal
+	 * in that case.
+	 */
+
+	if (!(fault_code & (FAULT_CODE_WRITE|FAULT_CODE_ITLB)) &&
+	    (insn & 0xc0800000) == 0xc0800000) {
+		if (insn & 0x2000)
+			asi = (regs->tstate >> 24);
+		else
+			asi = (insn >> 5);
+		if ((asi & 0xf2) == 0x82) {
+			if (insn & 0x1000000) {
+				handle_ldf_stq(insn, regs);
+			} else {
+				/* This was a non-faulting load. Just clear the
+				 * destination register(s) and continue with the next
+				 * instruction. -jj
+				 */
+				handle_ld_nf(insn, regs);
+			}
+			return;
+		}
+	}
+		
+	g2 = regs->u_regs[UREG_G2];
+
+	/* Is this in ex_table? */
+	if (regs->tstate & TSTATE_PRIV) {
+		unsigned long fixup;
+
+		if (asi == ASI_P && (insn & 0xc0800000) == 0xc0800000) {
+			if (insn & 0x2000)
+				asi = (regs->tstate >> 24);
+			else
+				asi = (insn >> 5);
+		}
+	
+		/* Look in asi.h: All _S asis have LS bit set */
+		if ((asi & 0x1) &&
+		    (fixup = search_extables_range(regs->tpc, &g2))) {
+			regs->tpc = fixup;
+			regs->tnpc = regs->tpc + 4;
+			regs->u_regs[UREG_G2] = g2;
+			return;
+		}
+	} else {
+		/* The si_code was set to make clear whether
+		 * this was a SEGV_MAPERR or SEGV_ACCERR fault.
+		 */
+		do_fault_siginfo(si_code, SIGSEGV, regs, insn, fault_code);
+		return;
+	}
+
+cannot_handle:
+	unhandled_fault (address, current, regs);
+}
+
+asmlinkage void do_sparc64_fault(struct pt_regs *regs)
+{
+	struct mm_struct *mm = current->mm;
+	struct vm_area_struct *vma;
+	unsigned int insn = 0;
+	int si_code, fault_code;
+	unsigned long address;
+
+	fault_code = get_thread_fault_code();
+
+	if (notify_die(DIE_PAGE_FAULT, "page_fault", regs,
+		       fault_code, 0, SIGSEGV) == NOTIFY_STOP)
+		return;
+
+	si_code = SEGV_MAPERR;
+	address = current_thread_info()->fault_address;
+
+	if ((fault_code & FAULT_CODE_ITLB) &&
+	    (fault_code & FAULT_CODE_DTLB))
+		BUG();
+
+	if (regs->tstate & TSTATE_PRIV) {
+		unsigned long tpc = regs->tpc;
+
+		/* Sanity check the PC. */
+		if ((tpc >= KERNBASE && tpc < (unsigned long) _etext) ||
+		    (tpc >= MODULES_VADDR && tpc < MODULES_END)) {
+			/* Valid, no problems... */
+		} else {
+			bad_kernel_pc(regs);
+			return;
+		}
+	}
+
+	/*
+	 * If we're in an interrupt or have no user
+	 * context, we must not take the fault..
+	 */
+	if (in_atomic() || !mm)
+		goto intr_or_no_mm;
+
+	if (test_thread_flag(TIF_32BIT)) {
+		if (!(regs->tstate & TSTATE_PRIV))
+			regs->tpc &= 0xffffffff;
+		address &= 0xffffffff;
+	}
+
+	if (!down_read_trylock(&mm->mmap_sem)) {
+		if ((regs->tstate & TSTATE_PRIV) &&
+		    !search_exception_tables(regs->tpc)) {
+			insn = get_fault_insn(regs, insn);
+			goto handle_kernel_fault;
+		}
+		down_read(&mm->mmap_sem);
+	}
+
+	vma = find_vma(mm, address);
+	if (!vma)
+		goto bad_area;
+
+	/* Pure DTLB misses do not tell us whether the fault causing
+	 * load/store/atomic was a write or not, it only says that there
+	 * was no match.  So in such a case we (carefully) read the
+	 * instruction to try and figure this out.  It's an optimization
+	 * so it's ok if we can't do this.
+	 *
+	 * Special hack, window spill/fill knows the exact fault type.
+	 */
+	if (((fault_code &
+	      (FAULT_CODE_DTLB | FAULT_CODE_WRITE | FAULT_CODE_WINFIXUP)) == FAULT_CODE_DTLB) &&
+	    (vma->vm_flags & VM_WRITE) != 0) {
+		insn = get_fault_insn(regs, 0);
+		if (!insn)
+			goto continue_fault;
+		if ((insn & 0xc0200000) == 0xc0200000 &&
+		    (insn & 0x1780000) != 0x1680000) {
+			/* Don't bother updating thread struct value,
+			 * because update_mmu_cache only cares which tlb
+			 * the access came from.
+			 */
+			fault_code |= FAULT_CODE_WRITE;
+		}
+	}
+continue_fault:
+
+	if (vma->vm_start <= address)
+		goto good_area;
+	if (!(vma->vm_flags & VM_GROWSDOWN))
+		goto bad_area;
+	if (!(fault_code & FAULT_CODE_WRITE)) {
+		/* Non-faulting loads shouldn't expand stack. */
+		insn = get_fault_insn(regs, insn);
+		if ((insn & 0xc0800000) == 0xc0800000) {
+			unsigned char asi;
+
+			if (insn & 0x2000)
+				asi = (regs->tstate >> 24);
+			else
+				asi = (insn >> 5);
+			if ((asi & 0xf2) == 0x82)
+				goto bad_area;
+		}
+	}
+	if (expand_stack(vma, address))
+		goto bad_area;
+	/*
+	 * Ok, we have a good vm_area for this memory access, so
+	 * we can handle it..
+	 */
+good_area:
+	si_code = SEGV_ACCERR;
+
+	/* If we took a ITLB miss on a non-executable page, catch
+	 * that here.
+	 */
+	if ((fault_code & FAULT_CODE_ITLB) && !(vma->vm_flags & VM_EXEC)) {
+		BUG_ON(address != regs->tpc);
+		BUG_ON(regs->tstate & TSTATE_PRIV);
+		goto bad_area;
+	}
+
+	if (fault_code & FAULT_CODE_WRITE) {
+		if (!(vma->vm_flags & VM_WRITE))
+			goto bad_area;
+
+		/* Spitfire has an icache which does not snoop
+		 * processor stores.  Later processors do...
+		 */
+		if (tlb_type == spitfire &&
+		    (vma->vm_flags & VM_EXEC) != 0 &&
+		    vma->vm_file != NULL)
+			set_thread_fault_code(fault_code |
+					      FAULT_CODE_BLKCOMMIT);
+	} else {
+		/* Allow reads even for write-only mappings */
+		if (!(vma->vm_flags & (VM_READ | VM_EXEC)))
+			goto bad_area;
+	}
+
+	switch (handle_mm_fault(mm, vma, address, (fault_code & FAULT_CODE_WRITE))) {
+	case VM_FAULT_MINOR:
+		current->min_flt++;
+		break;
+	case VM_FAULT_MAJOR:
+		current->maj_flt++;
+		break;
+	case VM_FAULT_SIGBUS:
+		goto do_sigbus;
+	case VM_FAULT_OOM:
+		goto out_of_memory;
+	default:
+		BUG();
+	}
+
+	up_read(&mm->mmap_sem);
+	goto fault_done;
+
+	/*
+	 * Something tried to access memory that isn't in our memory map..
+	 * Fix it, but check if it's kernel or user first..
+	 */
+bad_area:
+	insn = get_fault_insn(regs, insn);
+	up_read(&mm->mmap_sem);
+
+handle_kernel_fault:
+	do_kernel_fault(regs, si_code, fault_code, insn, address);
+
+	goto fault_done;
+
+/*
+ * We ran out of memory, or some other thing happened to us that made
+ * us unable to handle the page fault gracefully.
+ */
+out_of_memory:
+	insn = get_fault_insn(regs, insn);
+	up_read(&mm->mmap_sem);
+	printk("VM: killing process %s\n", current->comm);
+	if (!(regs->tstate & TSTATE_PRIV))
+		do_exit(SIGKILL);
+	goto handle_kernel_fault;
+
+intr_or_no_mm:
+	insn = get_fault_insn(regs, 0);
+	goto handle_kernel_fault;
+
+do_sigbus:
+	insn = get_fault_insn(regs, insn);
+	up_read(&mm->mmap_sem);
+
+	/*
+	 * Send a sigbus, regardless of whether we were in kernel
+	 * or user mode.
+	 */
+	do_fault_siginfo(BUS_ADRERR, SIGBUS, regs, insn, fault_code);
+
+	/* Kernel mode? Handle exceptions or die */
+	if (regs->tstate & TSTATE_PRIV)
+		goto handle_kernel_fault;
+
+fault_done:
+	/* These values are no longer needed, clear them. */
+	set_thread_fault_code(0);
+	current_thread_info()->fault_address = 0;
+}
diff --git a/arch/sparc64/mm/generic.c b/arch/sparc64/mm/generic.c
new file mode 100644
index 0000000..6b31f61
--- /dev/null
+++ b/arch/sparc64/mm/generic.c
@@ -0,0 +1,182 @@
+/* $Id: generic.c,v 1.18 2001/12/21 04:56:15 davem Exp $
+ * generic.c: Generic Sparc mm routines that are not dependent upon
+ *            MMU type but are Sparc specific.
+ *
+ * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
+ */
+
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/swap.h>
+#include <linux/pagemap.h>
+
+#include <asm/pgalloc.h>
+#include <asm/pgtable.h>
+#include <asm/page.h>
+#include <asm/tlbflush.h>
+
+/* Remap IO memory, the same way as remap_pfn_range(), but use
+ * the obio memory space.
+ *
+ * They use a pgprot that sets PAGE_IO and does not check the
+ * mem_map table as this is independent of normal memory.
+ */
+static inline void io_remap_pte_range(struct mm_struct *mm, pte_t * pte,
+				      unsigned long address,
+				      unsigned long size,
+				      unsigned long offset, pgprot_t prot,
+				      int space)
+{
+	unsigned long end;
+
+	/* clear hack bit that was used as a write_combine side-effect flag */
+	offset &= ~0x1UL;
+	address &= ~PMD_MASK;
+	end = address + size;
+	if (end > PMD_SIZE)
+		end = PMD_SIZE;
+	do {
+		pte_t entry;
+		unsigned long curend = address + PAGE_SIZE;
+		
+		entry = mk_pte_io(offset, prot, space);
+		if (!(address & 0xffff)) {
+			if (!(address & 0x3fffff) && !(offset & 0x3ffffe) && end >= address + 0x400000) {
+				entry = mk_pte_io(offset,
+						  __pgprot(pgprot_val (prot) | _PAGE_SZ4MB),
+						  space);
+				curend = address + 0x400000;
+				offset += 0x400000;
+			} else if (!(address & 0x7ffff) && !(offset & 0x7fffe) && end >= address + 0x80000) {
+				entry = mk_pte_io(offset,
+						  __pgprot(pgprot_val (prot) | _PAGE_SZ512K),
+						  space);
+				curend = address + 0x80000;
+				offset += 0x80000;
+			} else if (!(offset & 0xfffe) && end >= address + 0x10000) {
+				entry = mk_pte_io(offset,
+						  __pgprot(pgprot_val (prot) | _PAGE_SZ64K),
+						  space);
+				curend = address + 0x10000;
+				offset += 0x10000;
+			} else
+				offset += PAGE_SIZE;
+		} else
+			offset += PAGE_SIZE;
+
+		do {
+			BUG_ON(!pte_none(*pte));
+			set_pte_at(mm, address, pte, entry);
+			address += PAGE_SIZE;
+			pte++;
+		} while (address < curend);
+	} while (address < end);
+}
+
+static inline int io_remap_pmd_range(struct mm_struct *mm, pmd_t * pmd, unsigned long address, unsigned long size,
+	unsigned long offset, pgprot_t prot, int space)
+{
+	unsigned long end;
+
+	address &= ~PGDIR_MASK;
+	end = address + size;
+	if (end > PGDIR_SIZE)
+		end = PGDIR_SIZE;
+	offset -= address;
+	do {
+		pte_t * pte = pte_alloc_map(mm, pmd, address);
+		if (!pte)
+			return -ENOMEM;
+		io_remap_pte_range(mm, pte, address, end - address, address + offset, prot, space);
+		pte_unmap(pte);
+		address = (address + PMD_SIZE) & PMD_MASK;
+		pmd++;
+	} while (address < end);
+	return 0;
+}
+
+static inline int io_remap_pud_range(struct mm_struct *mm, pud_t * pud, unsigned long address, unsigned long size,
+	unsigned long offset, pgprot_t prot, int space)
+{
+	unsigned long end;
+
+	address &= ~PUD_MASK;
+	end = address + size;
+	if (end > PUD_SIZE)
+		end = PUD_SIZE;
+	offset -= address;
+	do {
+		pmd_t *pmd = pmd_alloc(mm, pud, address);
+		if (!pud)
+			return -ENOMEM;
+		io_remap_pmd_range(mm, pmd, address, end - address, address + offset, prot, space);
+		address = (address + PUD_SIZE) & PUD_MASK;
+		pud++;
+	} while (address < end);
+	return 0;
+}
+
+int io_remap_page_range(struct vm_area_struct *vma, unsigned long from, unsigned long offset, unsigned long size, pgprot_t prot, int space)
+{
+	int error = 0;
+	pgd_t * dir;
+	unsigned long beg = from;
+	unsigned long end = from + size;
+	struct mm_struct *mm = vma->vm_mm;
+
+	prot = __pgprot(pg_iobits);
+	offset -= from;
+	dir = pgd_offset(mm, from);
+	flush_cache_range(vma, beg, end);
+
+	spin_lock(&mm->page_table_lock);
+	while (from < end) {
+		pud_t *pud = pud_alloc(mm, dir, from);
+		error = -ENOMEM;
+		if (!pud)
+			break;
+		error = io_remap_pud_range(mm, pud, from, end - from, offset + from, prot, space);
+		if (error)
+			break;
+		from = (from + PGDIR_SIZE) & PGDIR_MASK;
+		dir++;
+	}
+	flush_tlb_range(vma, beg, end);
+	spin_unlock(&mm->page_table_lock);
+
+	return error;
+}
+
+int io_remap_pfn_range(struct vm_area_struct *vma, unsigned long from,
+		unsigned long pfn, unsigned long size, pgprot_t prot)
+{
+	int error = 0;
+	pgd_t * dir;
+	unsigned long beg = from;
+	unsigned long end = from + size;
+	struct mm_struct *mm = vma->vm_mm;
+	int space = GET_IOSPACE(pfn);
+	unsigned long offset = GET_PFN(pfn) << PAGE_SHIFT;
+
+	prot = __pgprot(pg_iobits);
+	offset -= from;
+	dir = pgd_offset(mm, from);
+	flush_cache_range(vma, beg, end);
+
+	spin_lock(&mm->page_table_lock);
+	while (from < end) {
+		pud_t *pud = pud_alloc(current->mm, dir, from);
+		error = -ENOMEM;
+		if (!pud)
+			break;
+		error = io_remap_pud_range(mm, pud, from, end - from, offset + from, prot, space);
+		if (error)
+			break;
+		from = (from + PGDIR_SIZE) & PGDIR_MASK;
+		dir++;
+	}
+	flush_tlb_range(vma, beg, end);
+	spin_unlock(&mm->page_table_lock);
+
+	return error;
+}
diff --git a/arch/sparc64/mm/hugetlbpage.c b/arch/sparc64/mm/hugetlbpage.c
new file mode 100644
index 0000000..5a1f831
--- /dev/null
+++ b/arch/sparc64/mm/hugetlbpage.c
@@ -0,0 +1,310 @@
+/*
+ * SPARC64 Huge TLB page support.
+ *
+ * Copyright (C) 2002, 2003 David S. Miller (davem@redhat.com)
+ */
+
+#include <linux/config.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/fs.h>
+#include <linux/mm.h>
+#include <linux/hugetlb.h>
+#include <linux/pagemap.h>
+#include <linux/smp_lock.h>
+#include <linux/slab.h>
+#include <linux/sysctl.h>
+
+#include <asm/mman.h>
+#include <asm/pgalloc.h>
+#include <asm/tlb.h>
+#include <asm/tlbflush.h>
+#include <asm/cacheflush.h>
+#include <asm/mmu_context.h>
+
+static pte_t *huge_pte_alloc(struct mm_struct *mm, unsigned long addr)
+{
+	pgd_t *pgd;
+	pud_t *pud;
+	pmd_t *pmd;
+	pte_t *pte = NULL;
+
+	pgd = pgd_offset(mm, addr);
+	if (pgd) {
+		pud = pud_offset(pgd, addr);
+		if (pud) {
+			pmd = pmd_alloc(mm, pud, addr);
+			if (pmd)
+				pte = pte_alloc_map(mm, pmd, addr);
+		}
+	}
+	return pte;
+}
+
+static pte_t *huge_pte_offset(struct mm_struct *mm, unsigned long addr)
+{
+	pgd_t *pgd;
+	pud_t *pud;
+	pmd_t *pmd;
+	pte_t *pte = NULL;
+
+	pgd = pgd_offset(mm, addr);
+	if (pgd) {
+		pud = pud_offset(pgd, addr);
+		if (pud) {
+			pmd = pmd_offset(pud, addr);
+			if (pmd)
+				pte = pte_offset_map(pmd, addr);
+		}
+	}
+	return pte;
+}
+
+#define mk_pte_huge(entry) do { pte_val(entry) |= _PAGE_SZHUGE; } while (0)
+
+static void set_huge_pte(struct mm_struct *mm, struct vm_area_struct *vma,
+			 unsigned long addr,
+			 struct page *page, pte_t * page_table, int write_access)
+{
+	unsigned long i;
+	pte_t entry;
+
+	add_mm_counter(mm, rss, HPAGE_SIZE / PAGE_SIZE);
+
+	if (write_access)
+		entry = pte_mkwrite(pte_mkdirty(mk_pte(page,
+						       vma->vm_page_prot)));
+	else
+		entry = pte_wrprotect(mk_pte(page, vma->vm_page_prot));
+	entry = pte_mkyoung(entry);
+	mk_pte_huge(entry);
+
+	for (i = 0; i < (1 << HUGETLB_PAGE_ORDER); i++) {
+		set_pte_at(mm, addr, page_table, entry);
+		page_table++;
+		addr += PAGE_SIZE;
+
+		pte_val(entry) += PAGE_SIZE;
+	}
+}
+
+/*
+ * This function checks for proper alignment of input addr and len parameters.
+ */
+int is_aligned_hugepage_range(unsigned long addr, unsigned long len)
+{
+	if (len & ~HPAGE_MASK)
+		return -EINVAL;
+	if (addr & ~HPAGE_MASK)
+		return -EINVAL;
+	return 0;
+}
+
+int copy_hugetlb_page_range(struct mm_struct *dst, struct mm_struct *src,
+			    struct vm_area_struct *vma)
+{
+	pte_t *src_pte, *dst_pte, entry;
+	struct page *ptepage;
+	unsigned long addr = vma->vm_start;
+	unsigned long end = vma->vm_end;
+	int i;
+
+	while (addr < end) {
+		dst_pte = huge_pte_alloc(dst, addr);
+		if (!dst_pte)
+			goto nomem;
+		src_pte = huge_pte_offset(src, addr);
+		BUG_ON(!src_pte || pte_none(*src_pte));
+		entry = *src_pte;
+		ptepage = pte_page(entry);
+		get_page(ptepage);
+		for (i = 0; i < (1 << HUGETLB_PAGE_ORDER); i++) {
+			set_pte_at(dst, addr, dst_pte, entry);
+			pte_val(entry) += PAGE_SIZE;
+			dst_pte++;
+			addr += PAGE_SIZE;
+		}
+		add_mm_counter(dst, rss, HPAGE_SIZE / PAGE_SIZE);
+	}
+	return 0;
+
+nomem:
+	return -ENOMEM;
+}
+
+int follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct *vma,
+			struct page **pages, struct vm_area_struct **vmas,
+			unsigned long *position, int *length, int i)
+{
+	unsigned long vaddr = *position;
+	int remainder = *length;
+
+	WARN_ON(!is_vm_hugetlb_page(vma));
+
+	while (vaddr < vma->vm_end && remainder) {
+		if (pages) {
+			pte_t *pte;
+			struct page *page;
+
+			pte = huge_pte_offset(mm, vaddr);
+
+			/* hugetlb should be locked, and hence, prefaulted */
+			BUG_ON(!pte || pte_none(*pte));
+
+			page = pte_page(*pte);
+
+			WARN_ON(!PageCompound(page));
+
+			get_page(page);
+			pages[i] = page;
+		}
+
+		if (vmas)
+			vmas[i] = vma;
+
+		vaddr += PAGE_SIZE;
+		--remainder;
+		++i;
+	}
+
+	*length = remainder;
+	*position = vaddr;
+
+	return i;
+}
+
+struct page *follow_huge_addr(struct mm_struct *mm,
+			      unsigned long address, int write)
+{
+	return ERR_PTR(-EINVAL);
+}
+
+int pmd_huge(pmd_t pmd)
+{
+	return 0;
+}
+
+struct page *follow_huge_pmd(struct mm_struct *mm, unsigned long address,
+			     pmd_t *pmd, int write)
+{
+	return NULL;
+}
+
+void unmap_hugepage_range(struct vm_area_struct *vma,
+			  unsigned long start, unsigned long end)
+{
+	struct mm_struct *mm = vma->vm_mm;
+	unsigned long address;
+	pte_t *pte;
+	struct page *page;
+	int i;
+
+	BUG_ON(start & (HPAGE_SIZE - 1));
+	BUG_ON(end & (HPAGE_SIZE - 1));
+
+	for (address = start; address < end; address += HPAGE_SIZE) {
+		pte = huge_pte_offset(mm, address);
+		BUG_ON(!pte);
+		if (pte_none(*pte))
+			continue;
+		page = pte_page(*pte);
+		put_page(page);
+		for (i = 0; i < (1 << HUGETLB_PAGE_ORDER); i++) {
+			pte_clear(mm, address+(i*PAGE_SIZE), pte);
+			pte++;
+		}
+	}
+	add_mm_counter(mm, rss, -((end - start) >> PAGE_SHIFT));
+	flush_tlb_range(vma, start, end);
+}
+
+static void context_reload(void *__data)
+{
+	struct mm_struct *mm = __data;
+
+	if (mm == current->mm)
+		load_secondary_context(mm);
+}
+
+int hugetlb_prefault(struct address_space *mapping, struct vm_area_struct *vma)
+{
+	struct mm_struct *mm = current->mm;
+	unsigned long addr;
+	int ret = 0;
+
+	/* On UltraSPARC-III+ and later, configure the second half of
+	 * the Data-TLB for huge pages.
+	 */
+	if (tlb_type == cheetah_plus) {
+		unsigned long ctx;
+
+		spin_lock(&ctx_alloc_lock);
+		ctx = mm->context.sparc64_ctx_val;
+		ctx &= ~CTX_PGSZ_MASK;
+		ctx |= CTX_PGSZ_BASE << CTX_PGSZ0_SHIFT;
+		ctx |= CTX_PGSZ_HUGE << CTX_PGSZ1_SHIFT;
+
+		if (ctx != mm->context.sparc64_ctx_val) {
+			/* When changing the page size fields, we
+			 * must perform a context flush so that no
+			 * stale entries match.  This flush must
+			 * occur with the original context register
+			 * settings.
+			 */
+			do_flush_tlb_mm(mm);
+
+			/* Reload the context register of all processors
+			 * also executing in this address space.
+			 */
+			mm->context.sparc64_ctx_val = ctx;
+			on_each_cpu(context_reload, mm, 0, 0);
+		}
+		spin_unlock(&ctx_alloc_lock);
+	}
+
+	BUG_ON(vma->vm_start & ~HPAGE_MASK);
+	BUG_ON(vma->vm_end & ~HPAGE_MASK);
+
+	spin_lock(&mm->page_table_lock);
+	for (addr = vma->vm_start; addr < vma->vm_end; addr += HPAGE_SIZE) {
+		unsigned long idx;
+		pte_t *pte = huge_pte_alloc(mm, addr);
+		struct page *page;
+
+		if (!pte) {
+			ret = -ENOMEM;
+			goto out;
+		}
+		if (!pte_none(*pte))
+			continue;
+
+		idx = ((addr - vma->vm_start) >> HPAGE_SHIFT)
+			+ (vma->vm_pgoff >> (HPAGE_SHIFT - PAGE_SHIFT));
+		page = find_get_page(mapping, idx);
+		if (!page) {
+			/* charge the fs quota first */
+			if (hugetlb_get_quota(mapping)) {
+				ret = -ENOMEM;
+				goto out;
+			}
+			page = alloc_huge_page();
+			if (!page) {
+				hugetlb_put_quota(mapping);
+				ret = -ENOMEM;
+				goto out;
+			}
+			ret = add_to_page_cache(page, mapping, idx, GFP_ATOMIC);
+			if (! ret) {
+				unlock_page(page);
+			} else {
+				hugetlb_put_quota(mapping);
+				free_huge_page(page);
+				goto out;
+			}
+		}
+		set_huge_pte(mm, vma, addr, page, pte, vma->vm_flags & VM_WRITE);
+	}
+out:
+	spin_unlock(&mm->page_table_lock);
+	return ret;
+}
diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c
new file mode 100644
index 0000000..89022cc
--- /dev/null
+++ b/arch/sparc64/mm/init.c
@@ -0,0 +1,1769 @@
+/*  $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 (davem@caip.rutgers.edu)
+ *  Copyright (C) 1997-1999 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ */
+ 
+#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".  */
+DEFINE_SPINLOCK(ctx_alloc_lock);
+unsigned long tlb_context_cache = CTX_FIRST_VERSION - 1;
+#define CTX_BMAP_SLOTS (1UL << (CTX_NR_BITS - 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 {
+			if (pgd_quicklist)
+				free_pgd_slow(get_pgd_fast());
+			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);
+	}
+	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
+
+#ifdef DCACHE_ALIASING_POSSIBLE
+	__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, %%g1\n\t"
+			     "or	%%g1, %0, %%g1\n\t"
+			     "casx	[%2], %%g7, %%g1\n\t"
+			     "cmp	%%g7, %%g1\n\t"
+			     "bne,pn	%%xcc, 1b\n\t"
+			     " membar	#StoreLoad | #StoreStore"
+			     : /* no outputs */
+			     : "r" (mask), "r" (non_cpu_bits), "r" (&page->flags)
+			     : "g1", "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, %%g1\n\t"
+			     "and	%%g1, %3, %%g1\n\t"
+			     "cmp	%%g1, %0\n\t"
+			     "bne,pn	%%icc, 2f\n\t"
+			     " andn	%%g7, %1, %%g1\n\t"
+			     "casx	[%2], %%g7, %%g1\n\t"
+			     "cmp	%%g7, %%g1\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)
+			     : "g1", "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(CTX_NRBITS(vma->vm_mm->context),
+				   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);
+}
+
+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_at(&init_mm, vaddr,
+					   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);
+	flushi((long)&obp_daddr_patch[0]);
+
+	/* Now fixup OBP's idea about where we really are mapped. */
+	prom_printf("Remapping the kernel... ");
+
+	/* Spitfire Errata #32 workaround */
+	/* NOTE: Using plain zero for the context value is
+	 *       correct here, we are not using the Linux trap
+	 *       tables yet so we should not use the special
+	 *       UltraSPARC-III+ page size encodings yet.
+	 */
+	__asm__ __volatile__("stxa	%0, [%1] %2\n\t"
+			     "flush	%%g6"
+			     : /* No outputs */
+			     : "r" (0), "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU));
+
+	switch (tlb_type) {
+	default:
+	case spitfire:
+		phys_page = spitfire_get_dtlb_data(sparc64_highest_locked_tlbent());
+		break;
+
+	case cheetah:
+	case cheetah_plus:
+		phys_page = cheetah_get_litlb_data(sparc64_highest_locked_tlbent());
+		break;
+	};
+
+	phys_page &= _PAGE_PADDR;
+	phys_page += ((unsigned long)&prom_boot_page -
+		      (unsigned long)KERNBASE);
+
+	if (tlb_type == spitfire) {
+		/* Lock this into i/d tlb entry 59 */
+		__asm__ __volatile__(
+			"stxa	%%g0, [%2] %3\n\t"
+			"stxa	%0, [%1] %4\n\t"
+			"membar	#Sync\n\t"
+			"flush	%%g6\n\t"
+			"stxa	%%g0, [%2] %5\n\t"
+			"stxa	%0, [%1] %6\n\t"
+			"membar	#Sync\n\t"
+			"flush	%%g6"
+			: : "r" (phys_page | _PAGE_VALID | _PAGE_SZ8K | _PAGE_CP |
+				 _PAGE_CV | _PAGE_P | _PAGE_L | _PAGE_W),
+			"r" (59 << 3), "r" (TLB_TAG_ACCESS),
+			"i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS),
+			"i" (ASI_IMMU), "i" (ASI_ITLB_DATA_ACCESS)
+			: "memory");
+	} else if (tlb_type == cheetah || tlb_type == cheetah_plus) {
+		/* Lock this into i/d tlb-0 entry 11 */
+		__asm__ __volatile__(
+			"stxa	%%g0, [%2] %3\n\t"
+			"stxa	%0, [%1] %4\n\t"
+			"membar	#Sync\n\t"
+			"flush	%%g6\n\t"
+			"stxa	%%g0, [%2] %5\n\t"
+			"stxa	%0, [%1] %6\n\t"
+			"membar	#Sync\n\t"
+			"flush	%%g6"
+			: : "r" (phys_page | _PAGE_VALID | _PAGE_SZ8K | _PAGE_CP |
+				 _PAGE_CV | _PAGE_P | _PAGE_L | _PAGE_W),
+			"r" ((0 << 16) | (11 << 3)), "r" (TLB_TAG_ACCESS),
+			"i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS),
+			"i" (ASI_IMMU), "i" (ASI_ITLB_DATA_ACCESS)
+			: "memory");
+	} else {
+		/* Implement me :-) */
+		BUG();
+	}
+
+	tte_vaddr = (unsigned long) KERNBASE;
+
+	/* Spitfire Errata #32 workaround */
+	/* NOTE: Using plain zero for the context value is
+	 *       correct here, we are not using the Linux trap
+	 *       tables yet so we should not use the special
+	 *       UltraSPARC-III+ page size encodings yet.
+	 */
+	__asm__ __volatile__("stxa	%0, [%1] %2\n\t"
+			     "flush	%%g6"
+			     : /* No outputs */
+			     : "r" (0),
+			     "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU));
+
+	if (tlb_type == spitfire)
+		tte_data = spitfire_get_dtlb_data(sparc64_highest_locked_tlbent());
+	else
+		tte_data = cheetah_get_ldtlb_data(sparc64_highest_locked_tlbent());
+
+	kern_locked_tte_data = tte_data;
+
+	remap_func = (void *)  ((unsigned long) &prom_remap -
+				(unsigned long) &prom_boot_page);
+
+
+	/* Spitfire Errata #32 workaround */
+	/* NOTE: Using plain zero for the context value is
+	 *       correct here, we are not using the Linux trap
+	 *       tables yet so we should not use the special
+	 *       UltraSPARC-III+ page size encodings yet.
+	 */
+	__asm__ __volatile__("stxa	%0, [%1] %2\n\t"
+			     "flush	%%g6"
+			     : /* No outputs */
+			     : "r" (0),
+			     "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU));
+
+	remap_func((tlb_type == spitfire ?
+		    (spitfire_get_dtlb_data(sparc64_highest_locked_tlbent()) & _PAGE_PADDR) :
+		    (cheetah_get_litlb_data(sparc64_highest_locked_tlbent()) & _PAGE_PADDR)),
+		   (unsigned long) KERNBASE,
+		   prom_get_mmu_ihandle());
+
+	if (bigkernel)
+		remap_func(((tte_data + 0x400000) & _PAGE_PADDR),
+			(unsigned long) KERNBASE + 0x400000, prom_get_mmu_ihandle());
+
+	/* Flush out that temporary mapping. */
+	spitfire_flush_dtlb_nucleus_page(0x0);
+	spitfire_flush_itlb_nucleus_page(0x0);
+
+	/* Now lock us back into the TLBs via OBP. */
+	prom_dtlb_load(sparc64_highest_locked_tlbent(), tte_data, tte_vaddr);
+	prom_itlb_load(sparc64_highest_locked_tlbent(), tte_data, tte_vaddr);
+	if (bigkernel) {
+		prom_dtlb_load(sparc64_highest_locked_tlbent()-1, tte_data + 0x400000, 
+								tte_vaddr + 0x400000);
+		prom_itlb_load(sparc64_highest_locked_tlbent()-1, tte_data + 0x400000, 
+								tte_vaddr + 0x400000);
+	}
+
+	/* Re-read translations property. */
+	if ((n = prom_getproperty(node, "translations", (char *)trans, tsz)) == -1) {
+		prom_printf("Couldn't get translation property\n");
+		prom_halt();
+	}
+	n = n / sizeof(*trans);
+
+	for (i = 0; i < n; i++) {
+		unsigned long vaddr = trans[i].virt;
+		unsigned long size = trans[i].size;
+
+		if (vaddr < 0xf0000000UL) {
+			unsigned long avoid_start = (unsigned long) KERNBASE;
+			unsigned long avoid_end = avoid_start + (4 * 1024 * 1024);
+
+			if (bigkernel)
+				avoid_end += (4 * 1024 * 1024);
+			if (vaddr < avoid_start) {
+				unsigned long top = vaddr + size;
+
+				if (top > avoid_start)
+					top = avoid_start;
+				prom_unmap(top - vaddr, vaddr);
+			}
+			if ((vaddr + size) > avoid_end) {
+				unsigned long bottom = vaddr;
+
+				if (bottom < avoid_end)
+					bottom = avoid_end;
+				prom_unmap((vaddr + size) - bottom, bottom);
+			}
+		}
+	}
+
+	prom_printf("done.\n");
+
+	register_prom_callbacks();
+}
+
+/* The OBP specifications for sun4u mark 0xfffffffc00000000 and
+ * upwards as reserved for use by the firmware (I wonder if this
+ * will be the same on Cheetah...).  We use this virtual address
+ * range for the VPTE table mappings of the nucleus so we need
+ * to zap them when we enter the PROM.  -DaveM
+ */
+static void __flush_nucleus_vptes(void)
+{
+	unsigned long prom_reserved_base = 0xfffffffc00000000UL;
+	int i;
+
+	/* Only DTLB must be checked for VPTE entries. */
+	if (tlb_type == spitfire) {
+		for (i = 0; i < 63; i++) {
+			unsigned long tag;
+
+			/* Spitfire Errata #32 workaround */
+			/* NOTE: Always runs on spitfire, so no cheetah+
+			 *       page size encodings.
+			 */
+			__asm__ __volatile__("stxa	%0, [%1] %2\n\t"
+					     "flush	%%g6"
+					     : /* No outputs */
+					     : "r" (0),
+					     "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU));
+
+			tag = spitfire_get_dtlb_tag(i);
+			if (((tag & ~(PAGE_MASK)) == 0) &&
+			    ((tag &  (PAGE_MASK)) >= prom_reserved_base)) {
+				__asm__ __volatile__("stxa %%g0, [%0] %1\n\t"
+						     "membar #Sync"
+						     : /* no outputs */
+						     : "r" (TLB_TAG_ACCESS), "i" (ASI_DMMU));
+				spitfire_put_dtlb_data(i, 0x0UL);
+			}
+		}
+	} else if (tlb_type == cheetah || tlb_type == cheetah_plus) {
+		for (i = 0; i < 512; i++) {
+			unsigned long tag = cheetah_get_dtlb_tag(i, 2);
+
+			if ((tag & ~PAGE_MASK) == 0 &&
+			    (tag & PAGE_MASK) >= prom_reserved_base) {
+				__asm__ __volatile__("stxa %%g0, [%0] %1\n\t"
+						     "membar #Sync"
+						     : /* no outputs */
+						     : "r" (TLB_TAG_ACCESS), "i" (ASI_DMMU));
+				cheetah_put_dtlb_data(i, 0x0UL, 2);
+			}
+
+			if (tlb_type != cheetah_plus)
+				continue;
+
+			tag = cheetah_get_dtlb_tag(i, 3);
+
+			if ((tag & ~PAGE_MASK) == 0 &&
+			    (tag & PAGE_MASK) >= prom_reserved_base) {
+				__asm__ __volatile__("stxa %%g0, [%0] %1\n\t"
+						     "membar #Sync"
+						     : /* no outputs */
+						     : "r" (TLB_TAG_ACCESS), "i" (ASI_DMMU));
+				cheetah_put_dtlb_data(i, 0x0UL, 3);
+			}
+		}
+	} else {
+		/* Implement me :-) */
+		BUG();
+	}
+}
+
+static int prom_ditlb_set;
+struct prom_tlb_entry {
+	int		tlb_ent;
+	unsigned long	tlb_tag;
+	unsigned long	tlb_data;
+};
+struct prom_tlb_entry prom_itlb[16], prom_dtlb[16];
+
+void prom_world(int enter)
+{
+	unsigned long pstate;
+	int i;
+
+	if (!enter)
+		set_fs((mm_segment_t) { get_thread_current_ds() });
+
+	if (!prom_ditlb_set)
+		return;
+
+	/* Make sure the following runs atomically. */
+	__asm__ __volatile__("flushw\n\t"
+			     "rdpr	%%pstate, %0\n\t"
+			     "wrpr	%0, %1, %%pstate"
+			     : "=r" (pstate)
+			     : "i" (PSTATE_IE));
+
+	if (enter) {
+		/* Kick out nucleus VPTEs. */
+		__flush_nucleus_vptes();
+
+		/* Install PROM world. */
+		for (i = 0; i < 16; i++) {
+			if (prom_dtlb[i].tlb_ent != -1) {
+				__asm__ __volatile__("stxa %0, [%1] %2\n\t"
+						     "membar #Sync"
+					: : "r" (prom_dtlb[i].tlb_tag), "r" (TLB_TAG_ACCESS),
+					"i" (ASI_DMMU));
+				if (tlb_type == spitfire)
+					spitfire_put_dtlb_data(prom_dtlb[i].tlb_ent,
+							       prom_dtlb[i].tlb_data);
+				else if (tlb_type == cheetah || tlb_type == cheetah_plus)
+					cheetah_put_ldtlb_data(prom_dtlb[i].tlb_ent,
+							       prom_dtlb[i].tlb_data);
+			}
+			if (prom_itlb[i].tlb_ent != -1) {
+				__asm__ __volatile__("stxa %0, [%1] %2\n\t"
+						     "membar #Sync"
+						     : : "r" (prom_itlb[i].tlb_tag),
+						     "r" (TLB_TAG_ACCESS),
+						     "i" (ASI_IMMU));
+				if (tlb_type == spitfire)
+					spitfire_put_itlb_data(prom_itlb[i].tlb_ent,
+							       prom_itlb[i].tlb_data);
+				else if (tlb_type == cheetah || tlb_type == cheetah_plus)
+					cheetah_put_litlb_data(prom_itlb[i].tlb_ent,
+							       prom_itlb[i].tlb_data);
+			}
+		}
+	} else {
+		for (i = 0; i < 16; i++) {
+			if (prom_dtlb[i].tlb_ent != -1) {
+				__asm__ __volatile__("stxa %%g0, [%0] %1\n\t"
+						     "membar #Sync"
+					: : "r" (TLB_TAG_ACCESS), "i" (ASI_DMMU));
+				if (tlb_type == spitfire)
+					spitfire_put_dtlb_data(prom_dtlb[i].tlb_ent, 0x0UL);
+				else
+					cheetah_put_ldtlb_data(prom_dtlb[i].tlb_ent, 0x0UL);
+			}
+			if (prom_itlb[i].tlb_ent != -1) {
+				__asm__ __volatile__("stxa %%g0, [%0] %1\n\t"
+						     "membar #Sync"
+						     : : "r" (TLB_TAG_ACCESS),
+						     "i" (ASI_IMMU));
+				if (tlb_type == spitfire)
+					spitfire_put_itlb_data(prom_itlb[i].tlb_ent, 0x0UL);
+				else
+					cheetah_put_litlb_data(prom_itlb[i].tlb_ent, 0x0UL);
+			}
+		}
+	}
+	__asm__ __volatile__("wrpr	%0, 0, %%pstate"
+			     : : "r" (pstate));
+}
+
+void inherit_locked_prom_mappings(int save_p)
+{
+	int i;
+	int dtlb_seen = 0;
+	int itlb_seen = 0;
+
+	/* Fucking losing PROM has more mappings in the TLB, but
+	 * it (conveniently) fails to mention any of these in the
+	 * translations property.  The only ones that matter are
+	 * the locked PROM tlb entries, so we impose the following
+	 * irrecovable rule on the PROM, it is allowed 8 locked
+	 * entries in the ITLB and 8 in the DTLB.
+	 *
+	 * Supposedly the upper 16GB of the address space is
+	 * reserved for OBP, BUT I WISH THIS WAS DOCUMENTED
+	 * SOMEWHERE!!!!!!!!!!!!!!!!!  Furthermore the entire interface
+	 * used between the client program and the firmware on sun5
+	 * systems to coordinate mmu mappings is also COMPLETELY
+	 * UNDOCUMENTED!!!!!! Thanks S(t)un!
+	 */
+	if (save_p) {
+		for (i = 0; i < 16; i++) {
+			prom_itlb[i].tlb_ent = -1;
+			prom_dtlb[i].tlb_ent = -1;
+		}
+	}
+	if (tlb_type == spitfire) {
+		int high = SPITFIRE_HIGHEST_LOCKED_TLBENT - bigkernel;
+		for (i = 0; i < high; i++) {
+			unsigned long data;
+
+			/* Spitfire Errata #32 workaround */
+			/* NOTE: Always runs on spitfire, so no cheetah+
+			 *       page size encodings.
+			 */
+			__asm__ __volatile__("stxa	%0, [%1] %2\n\t"
+					     "flush	%%g6"
+					     : /* No outputs */
+					     : "r" (0),
+					     "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU));
+
+			data = spitfire_get_dtlb_data(i);
+			if ((data & (_PAGE_L|_PAGE_VALID)) == (_PAGE_L|_PAGE_VALID)) {
+				unsigned long tag;
+
+				/* Spitfire Errata #32 workaround */
+				/* NOTE: Always runs on spitfire, so no
+				 *       cheetah+ page size encodings.
+				 */
+				__asm__ __volatile__("stxa	%0, [%1] %2\n\t"
+						     "flush	%%g6"
+						     : /* No outputs */
+						     : "r" (0),
+						     "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU));
+
+				tag = spitfire_get_dtlb_tag(i);
+				if (save_p) {
+					prom_dtlb[dtlb_seen].tlb_ent = i;
+					prom_dtlb[dtlb_seen].tlb_tag = tag;
+					prom_dtlb[dtlb_seen].tlb_data = data;
+				}
+				__asm__ __volatile__("stxa %%g0, [%0] %1\n\t"
+						     "membar #Sync"
+						     : : "r" (TLB_TAG_ACCESS), "i" (ASI_DMMU));
+				spitfire_put_dtlb_data(i, 0x0UL);
+
+				dtlb_seen++;
+				if (dtlb_seen > 15)
+					break;
+			}
+		}
+
+		for (i = 0; i < high; i++) {
+			unsigned long data;
+
+			/* Spitfire Errata #32 workaround */
+			/* NOTE: Always runs on spitfire, so no
+			 *       cheetah+ page size encodings.
+			 */
+			__asm__ __volatile__("stxa	%0, [%1] %2\n\t"
+					     "flush	%%g6"
+					     : /* No outputs */
+					     : "r" (0),
+					     "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU));
+
+			data = spitfire_get_itlb_data(i);
+			if ((data & (_PAGE_L|_PAGE_VALID)) == (_PAGE_L|_PAGE_VALID)) {
+				unsigned long tag;
+
+				/* Spitfire Errata #32 workaround */
+				/* NOTE: Always runs on spitfire, so no
+				 *       cheetah+ page size encodings.
+				 */
+				__asm__ __volatile__("stxa	%0, [%1] %2\n\t"
+						     "flush	%%g6"
+						     : /* No outputs */
+						     : "r" (0),
+						     "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU));
+
+				tag = spitfire_get_itlb_tag(i);
+				if (save_p) {
+					prom_itlb[itlb_seen].tlb_ent = i;
+					prom_itlb[itlb_seen].tlb_tag = tag;
+					prom_itlb[itlb_seen].tlb_data = data;
+				}
+				__asm__ __volatile__("stxa %%g0, [%0] %1\n\t"
+						     "membar #Sync"
+						     : : "r" (TLB_TAG_ACCESS), "i" (ASI_IMMU));
+				spitfire_put_itlb_data(i, 0x0UL);
+
+				itlb_seen++;
+				if (itlb_seen > 15)
+					break;
+			}
+		}
+	} else if (tlb_type == cheetah || tlb_type == cheetah_plus) {
+		int high = CHEETAH_HIGHEST_LOCKED_TLBENT - bigkernel;
+
+		for (i = 0; i < high; i++) {
+			unsigned long data;
+
+			data = cheetah_get_ldtlb_data(i);
+			if ((data & (_PAGE_L|_PAGE_VALID)) == (_PAGE_L|_PAGE_VALID)) {
+				unsigned long tag;
+
+				tag = cheetah_get_ldtlb_tag(i);
+				if (save_p) {
+					prom_dtlb[dtlb_seen].tlb_ent = i;
+					prom_dtlb[dtlb_seen].tlb_tag = tag;
+					prom_dtlb[dtlb_seen].tlb_data = data;
+				}
+				__asm__ __volatile__("stxa %%g0, [%0] %1\n\t"
+						     "membar #Sync"
+						     : : "r" (TLB_TAG_ACCESS), "i" (ASI_DMMU));
+				cheetah_put_ldtlb_data(i, 0x0UL);
+
+				dtlb_seen++;
+				if (dtlb_seen > 15)
+					break;
+			}
+		}
+
+		for (i = 0; i < high; i++) {
+			unsigned long data;
+
+			data = cheetah_get_litlb_data(i);
+			if ((data & (_PAGE_L|_PAGE_VALID)) == (_PAGE_L|_PAGE_VALID)) {
+				unsigned long tag;
+
+				tag = cheetah_get_litlb_tag(i);
+				if (save_p) {
+					prom_itlb[itlb_seen].tlb_ent = i;
+					prom_itlb[itlb_seen].tlb_tag = tag;
+					prom_itlb[itlb_seen].tlb_data = data;
+				}
+				__asm__ __volatile__("stxa %%g0, [%0] %1\n\t"
+						     "membar #Sync"
+						     : : "r" (TLB_TAG_ACCESS), "i" (ASI_IMMU));
+				cheetah_put_litlb_data(i, 0x0UL);
+
+				itlb_seen++;
+				if (itlb_seen > 15)
+					break;
+			}
+		}
+	} else {
+		/* Implement me :-) */
+		BUG();
+	}
+	if (save_p)
+		prom_ditlb_set = 1;
+}
+
+/* Give PROM back his world, done during reboots... */
+void prom_reload_locked(void)
+{
+	int i;
+
+	for (i = 0; i < 16; i++) {
+		if (prom_dtlb[i].tlb_ent != -1) {
+			__asm__ __volatile__("stxa %0, [%1] %2\n\t"
+					     "membar #Sync"
+				: : "r" (prom_dtlb[i].tlb_tag), "r" (TLB_TAG_ACCESS),
+				"i" (ASI_DMMU));
+			if (tlb_type == spitfire)
+				spitfire_put_dtlb_data(prom_dtlb[i].tlb_ent,
+						       prom_dtlb[i].tlb_data);
+			else if (tlb_type == cheetah || tlb_type == cheetah_plus)
+				cheetah_put_ldtlb_data(prom_dtlb[i].tlb_ent,
+						      prom_dtlb[i].tlb_data);
+		}
+
+		if (prom_itlb[i].tlb_ent != -1) {
+			__asm__ __volatile__("stxa %0, [%1] %2\n\t"
+					     "membar #Sync"
+					     : : "r" (prom_itlb[i].tlb_tag),
+					     "r" (TLB_TAG_ACCESS),
+					     "i" (ASI_IMMU));
+			if (tlb_type == spitfire)
+				spitfire_put_itlb_data(prom_itlb[i].tlb_ent,
+						       prom_itlb[i].tlb_data);
+			else
+				cheetah_put_litlb_data(prom_itlb[i].tlb_ent,
+						       prom_itlb[i].tlb_data);
+		}
+	}
+}
+
+#ifdef DCACHE_ALIASING_POSSIBLE
+void __flush_dcache_range(unsigned long start, unsigned long end)
+{
+	unsigned long va;
+
+	if (tlb_type == spitfire) {
+		int n = 0;
+
+		for (va = start; va < end; va += 32) {
+			spitfire_put_dcache_tag(va & 0x3fe0, 0x0);
+			if (++n >= 512)
+				break;
+		}
+	} else {
+		start = __pa(start);
+		end = __pa(end);
+		for (va = start; va < end; va += 32)
+			__asm__ __volatile__("stxa %%g0, [%0] %1\n\t"
+					     "membar #Sync"
+					     : /* no outputs */
+					     : "r" (va),
+					       "i" (ASI_DCACHE_INVALIDATE));
+	}
+}
+#endif /* DCACHE_ALIASING_POSSIBLE */
+
+/* If not locked, zap it. */
+void __flush_tlb_all(void)
+{
+	unsigned long pstate;
+	int i;
+
+	__asm__ __volatile__("flushw\n\t"
+			     "rdpr	%%pstate, %0\n\t"
+			     "wrpr	%0, %1, %%pstate"
+			     : "=r" (pstate)
+			     : "i" (PSTATE_IE));
+	if (tlb_type == spitfire) {
+		for (i = 0; i < 64; i++) {
+			/* Spitfire Errata #32 workaround */
+			/* NOTE: Always runs on spitfire, so no
+			 *       cheetah+ page size encodings.
+			 */
+			__asm__ __volatile__("stxa	%0, [%1] %2\n\t"
+					     "flush	%%g6"
+					     : /* No outputs */
+					     : "r" (0),
+					     "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU));
+
+			if (!(spitfire_get_dtlb_data(i) & _PAGE_L)) {
+				__asm__ __volatile__("stxa %%g0, [%0] %1\n\t"
+						     "membar #Sync"
+						     : /* no outputs */
+						     : "r" (TLB_TAG_ACCESS), "i" (ASI_DMMU));
+				spitfire_put_dtlb_data(i, 0x0UL);
+			}
+
+			/* Spitfire Errata #32 workaround */
+			/* NOTE: Always runs on spitfire, so no
+			 *       cheetah+ page size encodings.
+			 */
+			__asm__ __volatile__("stxa	%0, [%1] %2\n\t"
+					     "flush	%%g6"
+					     : /* No outputs */
+					     : "r" (0),
+					     "r" (PRIMARY_CONTEXT), "i" (ASI_DMMU));
+
+			if (!(spitfire_get_itlb_data(i) & _PAGE_L)) {
+				__asm__ __volatile__("stxa %%g0, [%0] %1\n\t"
+						     "membar #Sync"
+						     : /* no outputs */
+						     : "r" (TLB_TAG_ACCESS), "i" (ASI_IMMU));
+				spitfire_put_itlb_data(i, 0x0UL);
+			}
+		}
+	} else if (tlb_type == cheetah || tlb_type == cheetah_plus) {
+		cheetah_flush_dtlb_all();
+		cheetah_flush_itlb_all();
+	}
+	__asm__ __volatile__("wrpr	%0, 0, %%pstate"
+			     : : "r" (pstate));
+}
+
+/* Caller does TLB context flushing on local CPU if necessary.
+ * The caller also ensures that CTX_VALID(mm->context) is false.
+ *
+ * We must be careful about boundary cases so that we never
+ * let the user have CTX 0 (nucleus) or we ever use a CTX
+ * version of zero (and thus NO_CONTEXT would not be caught
+ * by version mis-match tests in mmu_context.h).
+ */
+void get_new_mmu_context(struct mm_struct *mm)
+{
+	unsigned long ctx, new_ctx;
+	unsigned long orig_pgsz_bits;
+	
+
+	spin_lock(&ctx_alloc_lock);
+	orig_pgsz_bits = (mm->context.sparc64_ctx_val & CTX_PGSZ_MASK);
+	ctx = (tlb_context_cache + 1) & CTX_NR_MASK;
+	new_ctx = find_next_zero_bit(mmu_context_bmap, 1 << CTX_NR_BITS, ctx);
+	if (new_ctx >= (1 << CTX_NR_BITS)) {
+		new_ctx = find_next_zero_bit(mmu_context_bmap, ctx, 1);
+		if (new_ctx >= ctx) {
+			int i;
+			new_ctx = (tlb_context_cache & CTX_VERSION_MASK) +
+				CTX_FIRST_VERSION;
+			if (new_ctx == 1)
+				new_ctx = CTX_FIRST_VERSION;
+
+			/* Don't call memset, for 16 entries that's just
+			 * plain silly...
+			 */
+			mmu_context_bmap[0] = 3;
+			mmu_context_bmap[1] = 0;
+			mmu_context_bmap[2] = 0;
+			mmu_context_bmap[3] = 0;
+			for (i = 4; i < CTX_BMAP_SLOTS; i += 4) {
+				mmu_context_bmap[i + 0] = 0;
+				mmu_context_bmap[i + 1] = 0;
+				mmu_context_bmap[i + 2] = 0;
+				mmu_context_bmap[i + 3] = 0;
+			}
+			goto out;
+		}
+	}
+	mmu_context_bmap[new_ctx>>6] |= (1UL << (new_ctx & 63));
+	new_ctx |= (tlb_context_cache & CTX_VERSION_MASK);
+out:
+	tlb_context_cache = new_ctx;
+	mm->context.sparc64_ctx_val = new_ctx | orig_pgsz_bits;
+	spin_unlock(&ctx_alloc_lock);
+}
+
+#ifndef CONFIG_SMP
+struct pgtable_cache_struct pgt_quicklists;
+#endif
+
+/* OK, we have to color these pages. The page tables are accessed
+ * by non-Dcache enabled mapping in the VPTE area by the dtlb_backend.S
+ * code, as well as by PAGE_OFFSET range direct-mapped addresses by 
+ * other parts of the kernel. By coloring, we make sure that the tlbmiss 
+ * fast handlers do not get data from old/garbage dcache lines that 
+ * correspond to an old/stale virtual address (user/kernel) that 
+ * previously mapped the pagetable page while accessing vpte range 
+ * addresses. The idea is that if the vpte color and PAGE_OFFSET range 
+ * color is the same, then when the kernel initializes the pagetable 
+ * using the later address range, accesses with the first address
+ * range will see the newly initialized data rather than the garbage.
+ */
+#ifdef DCACHE_ALIASING_POSSIBLE
+#define DC_ALIAS_SHIFT	1
+#else
+#define DC_ALIAS_SHIFT	0
+#endif
+pte_t *__pte_alloc_one_kernel(struct mm_struct *mm, unsigned long address)
+{
+	struct page *page;
+	unsigned long color;
+
+	{
+		pte_t *ptep = pte_alloc_one_fast(mm, address);
+
+		if (ptep)
+			return ptep;
+	}
+
+	color = VPTE_COLOR(address);
+	page = alloc_pages(GFP_KERNEL|__GFP_REPEAT, DC_ALIAS_SHIFT);
+	if (page) {
+		unsigned long *to_free;
+		unsigned long paddr;
+		pte_t *pte;
+
+#ifdef DCACHE_ALIASING_POSSIBLE
+		set_page_count(page, 1);
+		ClearPageCompound(page);
+
+		set_page_count((page + 1), 1);
+		ClearPageCompound(page + 1);
+#endif
+		paddr = (unsigned long) page_address(page);
+		memset((char *)paddr, 0, (PAGE_SIZE << DC_ALIAS_SHIFT));
+
+		if (!color) {
+			pte = (pte_t *) paddr;
+			to_free = (unsigned long *) (paddr + PAGE_SIZE);
+		} else {
+			pte = (pte_t *) (paddr + PAGE_SIZE);
+			to_free = (unsigned long *) paddr;
+		}
+
+#ifdef DCACHE_ALIASING_POSSIBLE
+		/* Now free the other one up, adjust cache size. */
+		preempt_disable();
+		*to_free = (unsigned long) pte_quicklist[color ^ 0x1];
+		pte_quicklist[color ^ 0x1] = to_free;
+		pgtable_cache_size++;
+		preempt_enable();
+#endif
+
+		return pte;
+	}
+	return NULL;
+}
+
+void sparc_ultra_dump_itlb(void)
+{
+        int slot;
+
+	if (tlb_type == spitfire) {
+		printk ("Contents of itlb: ");
+		for (slot = 0; slot < 14; slot++) printk ("    ");
+		printk ("%2x:%016lx,%016lx\n",
+			0,
+			spitfire_get_itlb_tag(0), spitfire_get_itlb_data(0));
+		for (slot = 1; slot < 64; slot+=3) {
+			printk ("%2x:%016lx,%016lx %2x:%016lx,%016lx %2x:%016lx,%016lx\n", 
+				slot,
+				spitfire_get_itlb_tag(slot), spitfire_get_itlb_data(slot),
+				slot+1,
+				spitfire_get_itlb_tag(slot+1), spitfire_get_itlb_data(slot+1),
+				slot+2,
+				spitfire_get_itlb_tag(slot+2), spitfire_get_itlb_data(slot+2));
+		}
+	} else if (tlb_type == cheetah || tlb_type == cheetah_plus) {
+		printk ("Contents of itlb0:\n");
+		for (slot = 0; slot < 16; slot+=2) {
+			printk ("%2x:%016lx,%016lx %2x:%016lx,%016lx\n",
+				slot,
+				cheetah_get_litlb_tag(slot), cheetah_get_litlb_data(slot),
+				slot+1,
+				cheetah_get_litlb_tag(slot+1), cheetah_get_litlb_data(slot+1));
+		}
+		printk ("Contents of itlb2:\n");
+		for (slot = 0; slot < 128; slot+=2) {
+			printk ("%2x:%016lx,%016lx %2x:%016lx,%016lx\n",
+				slot,
+				cheetah_get_itlb_tag(slot), cheetah_get_itlb_data(slot),
+				slot+1,
+				cheetah_get_itlb_tag(slot+1), cheetah_get_itlb_data(slot+1));
+		}
+	}
+}
+
+void sparc_ultra_dump_dtlb(void)
+{
+        int slot;
+
+	if (tlb_type == spitfire) {
+		printk ("Contents of dtlb: ");
+		for (slot = 0; slot < 14; slot++) printk ("    ");
+		printk ("%2x:%016lx,%016lx\n", 0,
+			spitfire_get_dtlb_tag(0), spitfire_get_dtlb_data(0));
+		for (slot = 1; slot < 64; slot+=3) {
+			printk ("%2x:%016lx,%016lx %2x:%016lx,%016lx %2x:%016lx,%016lx\n", 
+				slot,
+				spitfire_get_dtlb_tag(slot), spitfire_get_dtlb_data(slot),
+				slot+1,
+				spitfire_get_dtlb_tag(slot+1), spitfire_get_dtlb_data(slot+1),
+				slot+2,
+				spitfire_get_dtlb_tag(slot+2), spitfire_get_dtlb_data(slot+2));
+		}
+	} else if (tlb_type == cheetah || tlb_type == cheetah_plus) {
+		printk ("Contents of dtlb0:\n");
+		for (slot = 0; slot < 16; slot+=2) {
+			printk ("%2x:%016lx,%016lx %2x:%016lx,%016lx\n",
+				slot,
+				cheetah_get_ldtlb_tag(slot), cheetah_get_ldtlb_data(slot),
+				slot+1,
+				cheetah_get_ldtlb_tag(slot+1), cheetah_get_ldtlb_data(slot+1));
+		}
+		printk ("Contents of dtlb2:\n");
+		for (slot = 0; slot < 512; slot+=2) {
+			printk ("%2x:%016lx,%016lx %2x:%016lx,%016lx\n",
+				slot,
+				cheetah_get_dtlb_tag(slot, 2), cheetah_get_dtlb_data(slot, 2),
+				slot+1,
+				cheetah_get_dtlb_tag(slot+1, 2), cheetah_get_dtlb_data(slot+1, 2));
+		}
+		if (tlb_type == cheetah_plus) {
+			printk ("Contents of dtlb3:\n");
+			for (slot = 0; slot < 512; slot+=2) {
+				printk ("%2x:%016lx,%016lx %2x:%016lx,%016lx\n",
+					slot,
+					cheetah_get_dtlb_tag(slot, 3), cheetah_get_dtlb_data(slot, 3),
+					slot+1,
+					cheetah_get_dtlb_tag(slot+1, 3), cheetah_get_dtlb_data(slot+1, 3));
+			}
+		}
+	}
+}
+
+extern unsigned long cmdline_memory_size;
+
+unsigned long __init bootmem_init(unsigned long *pages_avail)
+{
+	unsigned long bootmap_size, start_pfn, end_pfn;
+	unsigned long end_of_phys_memory = 0UL;
+	unsigned long bootmap_pfn, bytes_avail, size;
+	int i;
+
+#ifdef CONFIG_DEBUG_BOOTMEM
+	prom_printf("bootmem_init: Scan sp_banks, ");
+#endif
+
+	bytes_avail = 0UL;
+	for (i = 0; sp_banks[i].num_bytes != 0; i++) {
+		end_of_phys_memory = sp_banks[i].base_addr +
+			sp_banks[i].num_bytes;
+		bytes_avail += sp_banks[i].num_bytes;
+		if (cmdline_memory_size) {
+			if (bytes_avail > cmdline_memory_size) {
+				unsigned long slack = bytes_avail - cmdline_memory_size;
+
+				bytes_avail -= slack;
+				end_of_phys_memory -= slack;
+
+				sp_banks[i].num_bytes -= slack;
+				if (sp_banks[i].num_bytes == 0) {
+					sp_banks[i].base_addr = 0xdeadbeef;
+				} else {
+					sp_banks[i+1].num_bytes = 0;
+					sp_banks[i+1].base_addr = 0xdeadbeef;
+				}
+				break;
+			}
+		}
+	}
+
+	*pages_avail = bytes_avail >> PAGE_SHIFT;
+
+	/* Start with page aligned address of last symbol in kernel
+	 * image.  The kernel is hard mapped below PAGE_OFFSET in a
+	 * 4MB locked TLB translation.
+	 */
+	start_pfn = PAGE_ALIGN(kern_base + kern_size) >> PAGE_SHIFT;
+
+	bootmap_pfn = start_pfn;
+
+	end_pfn = end_of_phys_memory >> PAGE_SHIFT;
+
+#ifdef CONFIG_BLK_DEV_INITRD
+	/* Now have to check initial ramdisk, so that bootmap does not overwrite it */
+	if (sparc_ramdisk_image || sparc_ramdisk_image64) {
+		unsigned long ramdisk_image = sparc_ramdisk_image ?
+			sparc_ramdisk_image : sparc_ramdisk_image64;
+		if (ramdisk_image >= (unsigned long)_end - 2 * PAGE_SIZE)
+			ramdisk_image -= KERNBASE;
+		initrd_start = ramdisk_image + phys_base;
+		initrd_end = initrd_start + sparc_ramdisk_size;
+		if (initrd_end > end_of_phys_memory) {
+			printk(KERN_CRIT "initrd extends beyond end of memory "
+		                 	 "(0x%016lx > 0x%016lx)\ndisabling initrd\n",
+			       initrd_end, end_of_phys_memory);
+			initrd_start = 0;
+		}
+		if (initrd_start) {
+			if (initrd_start >= (start_pfn << PAGE_SHIFT) &&
+			    initrd_start < (start_pfn << PAGE_SHIFT) + 2 * PAGE_SIZE)
+				bootmap_pfn = PAGE_ALIGN (initrd_end) >> PAGE_SHIFT;
+		}
+	}
+#endif	
+	/* Initialize the boot-time allocator. */
+	max_pfn = max_low_pfn = end_pfn;
+	min_low_pfn = pfn_base;
+
+#ifdef CONFIG_DEBUG_BOOTMEM
+	prom_printf("init_bootmem(min[%lx], bootmap[%lx], max[%lx])\n",
+		    min_low_pfn, bootmap_pfn, max_low_pfn);
+#endif
+	bootmap_size = init_bootmem_node(NODE_DATA(0), bootmap_pfn, pfn_base, end_pfn);
+
+	bootmap_base = bootmap_pfn << PAGE_SHIFT;
+
+	/* Now register the available physical memory with the
+	 * allocator.
+	 */
+	for (i = 0; sp_banks[i].num_bytes != 0; i++) {
+#ifdef CONFIG_DEBUG_BOOTMEM
+		prom_printf("free_bootmem(sp_banks:%d): base[%lx] size[%lx]\n",
+			    i, sp_banks[i].base_addr, sp_banks[i].num_bytes);
+#endif
+		free_bootmem(sp_banks[i].base_addr, sp_banks[i].num_bytes);
+	}
+
+#ifdef CONFIG_BLK_DEV_INITRD
+	if (initrd_start) {
+		size = initrd_end - initrd_start;
+
+		/* Resert the initrd image area. */
+#ifdef CONFIG_DEBUG_BOOTMEM
+		prom_printf("reserve_bootmem(initrd): base[%llx] size[%lx]\n",
+			initrd_start, initrd_end);
+#endif
+		reserve_bootmem(initrd_start, size);
+		*pages_avail -= PAGE_ALIGN(size) >> PAGE_SHIFT;
+
+		initrd_start += PAGE_OFFSET;
+		initrd_end += PAGE_OFFSET;
+	}
+#endif
+	/* Reserve the kernel text/data/bss. */
+#ifdef CONFIG_DEBUG_BOOTMEM
+	prom_printf("reserve_bootmem(kernel): base[%lx] size[%lx]\n", kern_base, kern_size);
+#endif
+	reserve_bootmem(kern_base, kern_size);
+	*pages_avail -= PAGE_ALIGN(kern_size) >> PAGE_SHIFT;
+
+	/* Reserve the bootmem map.   We do not account for it
+	 * in pages_avail because we will release that memory
+	 * in free_all_bootmem.
+	 */
+	size = bootmap_size;
+#ifdef CONFIG_DEBUG_BOOTMEM
+	prom_printf("reserve_bootmem(bootmap): base[%lx] size[%lx]\n",
+		    (bootmap_pfn << PAGE_SHIFT), size);
+#endif
+	reserve_bootmem((bootmap_pfn << PAGE_SHIFT), size);
+	*pages_avail -= PAGE_ALIGN(size) >> PAGE_SHIFT;
+
+	return end_pfn;
+}
+
+/* paging_init() sets up the page tables */
+
+extern void cheetah_ecache_flush_init(void);
+
+static unsigned long last_valid_pfn;
+
+void __init paging_init(void)
+{
+	extern pmd_t swapper_pmd_dir[1024];
+	extern unsigned int sparc64_vpte_patchme1[1];
+	extern unsigned int sparc64_vpte_patchme2[1];
+	unsigned long alias_base = kern_base + PAGE_OFFSET;
+	unsigned long second_alias_page = 0;
+	unsigned long pt, flags, end_pfn, pages_avail;
+	unsigned long shift = alias_base - ((unsigned long)KERNBASE);
+	unsigned long real_end;
+
+	set_bit(0, mmu_context_bmap);
+
+	real_end = (unsigned long)_end;
+	if ((real_end > ((unsigned long)KERNBASE + 0x400000)))
+		bigkernel = 1;
+#ifdef CONFIG_BLK_DEV_INITRD
+	if (sparc_ramdisk_image || sparc_ramdisk_image64)
+		real_end = (PAGE_ALIGN(real_end) + PAGE_ALIGN(sparc_ramdisk_size));
+#endif
+
+	/* We assume physical memory starts at some 4mb multiple,
+	 * if this were not true we wouldn't boot up to this point
+	 * anyways.
+	 */
+	pt  = kern_base | _PAGE_VALID | _PAGE_SZ4MB;
+	pt |= _PAGE_CP | _PAGE_CV | _PAGE_P | _PAGE_L | _PAGE_W;
+	local_irq_save(flags);
+	if (tlb_type == spitfire) {
+		__asm__ __volatile__(
+	"	stxa	%1, [%0] %3\n"
+	"	stxa	%2, [%5] %4\n"
+	"	membar	#Sync\n"
+	"	flush	%%g6\n"
+	"	nop\n"
+	"	nop\n"
+	"	nop\n"
+		: /* No outputs */
+		: "r" (TLB_TAG_ACCESS), "r" (alias_base), "r" (pt),
+		  "i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS), "r" (61 << 3)
+		: "memory");
+		if (real_end >= KERNBASE + 0x340000) {
+			second_alias_page = alias_base + 0x400000;
+			__asm__ __volatile__(
+		"	stxa	%1, [%0] %3\n"
+		"	stxa	%2, [%5] %4\n"
+		"	membar	#Sync\n"
+		"	flush	%%g6\n"
+		"	nop\n"
+		"	nop\n"
+		"	nop\n"
+			: /* No outputs */
+			: "r" (TLB_TAG_ACCESS), "r" (second_alias_page), "r" (pt + 0x400000),
+			  "i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS), "r" (60 << 3)
+			: "memory");
+		}
+	} else if (tlb_type == cheetah || tlb_type == cheetah_plus) {
+		__asm__ __volatile__(
+	"	stxa	%1, [%0] %3\n"
+	"	stxa	%2, [%5] %4\n"
+	"	membar	#Sync\n"
+	"	flush	%%g6\n"
+	"	nop\n"
+	"	nop\n"
+	"	nop\n"
+		: /* No outputs */
+		: "r" (TLB_TAG_ACCESS), "r" (alias_base), "r" (pt),
+		  "i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS), "r" ((0<<16) | (13<<3))
+		: "memory");
+		if (real_end >= KERNBASE + 0x340000) {
+			second_alias_page = alias_base + 0x400000;
+			__asm__ __volatile__(
+		"	stxa	%1, [%0] %3\n"
+		"	stxa	%2, [%5] %4\n"
+		"	membar	#Sync\n"
+		"	flush	%%g6\n"
+		"	nop\n"
+		"	nop\n"
+		"	nop\n"
+			: /* No outputs */
+			: "r" (TLB_TAG_ACCESS), "r" (second_alias_page), "r" (pt + 0x400000),
+			  "i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS), "r" ((0<<16) | (12<<3))
+			: "memory");
+		}
+	}
+	local_irq_restore(flags);
+	
+	/* Now set kernel pgd to upper alias so physical page computations
+	 * work.
+	 */
+	init_mm.pgd += ((shift) / (sizeof(pgd_t)));
+	
+	memset(swapper_pmd_dir, 0, sizeof(swapper_pmd_dir));
+
+	/* Now can init the kernel/bad page tables. */
+	pud_set(pud_offset(&swapper_pg_dir[0], 0),
+		swapper_pmd_dir + (shift / sizeof(pgd_t)));
+	
+	sparc64_vpte_patchme1[0] |=
+		(((unsigned long)pgd_val(init_mm.pgd[0])) >> 10);
+	sparc64_vpte_patchme2[0] |=
+		(((unsigned long)pgd_val(init_mm.pgd[0])) & 0x3ff);
+	flushi((long)&sparc64_vpte_patchme1[0]);
+	
+	/* Setup bootmem... */
+	pages_avail = 0;
+	last_valid_pfn = end_pfn = bootmem_init(&pages_avail);
+
+	/* Inherit non-locked OBP mappings. */
+	inherit_prom_mappings();
+	
+	/* Ok, we can use our TLB miss and window trap handlers safely.
+	 * We need to do a quick peek here to see if we are on StarFire
+	 * or not, so setup_tba can setup the IRQ globals correctly (it
+	 * needs to get the hard smp processor id correctly).
+	 */
+	{
+		extern void setup_tba(int);
+		setup_tba(this_is_starfire);
+	}
+
+	inherit_locked_prom_mappings(1);
+
+	/* We only created DTLB mapping of this stuff. */
+	spitfire_flush_dtlb_nucleus_page(alias_base);
+	if (second_alias_page)
+		spitfire_flush_dtlb_nucleus_page(second_alias_page);
+
+	__flush_tlb_all();
+
+	{
+		unsigned long zones_size[MAX_NR_ZONES];
+		unsigned long zholes_size[MAX_NR_ZONES];
+		unsigned long npages;
+		int znum;
+
+		for (znum = 0; znum < MAX_NR_ZONES; znum++)
+			zones_size[znum] = zholes_size[znum] = 0;
+
+		npages = end_pfn - pfn_base;
+		zones_size[ZONE_DMA] = npages;
+		zholes_size[ZONE_DMA] = npages - pages_avail;
+
+		free_area_init_node(0, &contig_page_data, zones_size,
+				    phys_base >> PAGE_SHIFT, zholes_size);
+	}
+
+	device_scan();
+}
+
+/* Ok, it seems that the prom can allocate some more memory chunks
+ * as a side effect of some prom calls we perform during the
+ * boot sequence.  My most likely theory is that it is from the
+ * prom_set_traptable() call, and OBP is allocating a scratchpad
+ * for saving client program register state etc.
+ */
+static void __init sort_memlist(struct linux_mlist_p1275 *thislist)
+{
+	int swapi = 0;
+	int i, mitr;
+	unsigned long tmpaddr, tmpsize;
+	unsigned long lowest;
+
+	for (i = 0; thislist[i].theres_more != 0; i++) {
+		lowest = thislist[i].start_adr;
+		for (mitr = i+1; thislist[mitr-1].theres_more != 0; mitr++)
+			if (thislist[mitr].start_adr < lowest) {
+				lowest = thislist[mitr].start_adr;
+				swapi = mitr;
+			}
+		if (lowest == thislist[i].start_adr)
+			continue;
+		tmpaddr = thislist[swapi].start_adr;
+		tmpsize = thislist[swapi].num_bytes;
+		for (mitr = swapi; mitr > i; mitr--) {
+			thislist[mitr].start_adr = thislist[mitr-1].start_adr;
+			thislist[mitr].num_bytes = thislist[mitr-1].num_bytes;
+		}
+		thislist[i].start_adr = tmpaddr;
+		thislist[i].num_bytes = tmpsize;
+	}
+}
+
+void __init rescan_sp_banks(void)
+{
+	struct linux_prom64_registers memlist[64];
+	struct linux_mlist_p1275 avail[64], *mlist;
+	unsigned long bytes, base_paddr;
+	int num_regs, node = prom_finddevice("/memory");
+	int i;
+
+	num_regs = prom_getproperty(node, "available",
+				    (char *) memlist, sizeof(memlist));
+	num_regs = (num_regs / sizeof(struct linux_prom64_registers));
+	for (i = 0; i < num_regs; i++) {
+		avail[i].start_adr = memlist[i].phys_addr;
+		avail[i].num_bytes = memlist[i].reg_size;
+		avail[i].theres_more = &avail[i + 1];
+	}
+	avail[i - 1].theres_more = NULL;
+	sort_memlist(avail);
+
+	mlist = &avail[0];
+	i = 0;
+	bytes = mlist->num_bytes;
+	base_paddr = mlist->start_adr;
+  
+	sp_banks[0].base_addr = base_paddr;
+	sp_banks[0].num_bytes = bytes;
+
+	while (mlist->theres_more != NULL){
+		i++;
+		mlist = mlist->theres_more;
+		bytes = mlist->num_bytes;
+		if (i >= SPARC_PHYS_BANKS-1) {
+			printk ("The machine has more banks than "
+				"this kernel can support\n"
+				"Increase the SPARC_PHYS_BANKS "
+				"setting (currently %d)\n",
+				SPARC_PHYS_BANKS);
+			i = SPARC_PHYS_BANKS-1;
+			break;
+		}
+    
+		sp_banks[i].base_addr = mlist->start_adr;
+		sp_banks[i].num_bytes = mlist->num_bytes;
+	}
+
+	i++;
+	sp_banks[i].base_addr = 0xdeadbeefbeefdeadUL;
+	sp_banks[i].num_bytes = 0;
+
+	for (i = 0; sp_banks[i].num_bytes != 0; i++)
+		sp_banks[i].num_bytes &= PAGE_MASK;
+}
+
+static void __init taint_real_pages(void)
+{
+	struct sparc_phys_banks saved_sp_banks[SPARC_PHYS_BANKS];
+	int i;
+
+	for (i = 0; i < SPARC_PHYS_BANKS; i++) {
+		saved_sp_banks[i].base_addr =
+			sp_banks[i].base_addr;
+		saved_sp_banks[i].num_bytes =
+			sp_banks[i].num_bytes;
+	}
+
+	rescan_sp_banks();
+
+	/* Find changes discovered in the sp_bank rescan and
+	 * reserve the lost portions in the bootmem maps.
+	 */
+	for (i = 0; saved_sp_banks[i].num_bytes; i++) {
+		unsigned long old_start, old_end;
+
+		old_start = saved_sp_banks[i].base_addr;
+		old_end = old_start +
+			saved_sp_banks[i].num_bytes;
+		while (old_start < old_end) {
+			int n;
+
+			for (n = 0; sp_banks[n].num_bytes; n++) {
+				unsigned long new_start, new_end;
+
+				new_start = sp_banks[n].base_addr;
+				new_end = new_start + sp_banks[n].num_bytes;
+
+				if (new_start <= old_start &&
+				    new_end >= (old_start + PAGE_SIZE)) {
+					set_bit (old_start >> 22,
+						 sparc64_valid_addr_bitmap);
+					goto do_next_page;
+				}
+			}
+			reserve_bootmem(old_start, PAGE_SIZE);
+
+		do_next_page:
+			old_start += PAGE_SIZE;
+		}
+	}
+}
+
+void __init mem_init(void)
+{
+	unsigned long codepages, datapages, initpages;
+	unsigned long addr, last;
+	int i;
+
+	i = last_valid_pfn >> ((22 - PAGE_SHIFT) + 6);
+	i += 1;
+	sparc64_valid_addr_bitmap = (unsigned long *)
+		__alloc_bootmem(i << 3, SMP_CACHE_BYTES, bootmap_base);
+	if (sparc64_valid_addr_bitmap == NULL) {
+		prom_printf("mem_init: Cannot alloc valid_addr_bitmap.\n");
+		prom_halt();
+	}
+	memset(sparc64_valid_addr_bitmap, 0, i << 3);
+
+	addr = PAGE_OFFSET + kern_base;
+	last = PAGE_ALIGN(kern_size) + addr;
+	while (addr < last) {
+		set_bit(__pa(addr) >> 22, sparc64_valid_addr_bitmap);
+		addr += PAGE_SIZE;
+	}
+
+	taint_real_pages();
+
+	max_mapnr = last_valid_pfn - pfn_base;
+	high_memory = __va(last_valid_pfn << PAGE_SHIFT);
+
+#ifdef CONFIG_DEBUG_BOOTMEM
+	prom_printf("mem_init: Calling free_all_bootmem().\n");
+#endif
+	totalram_pages = num_physpages = free_all_bootmem() - 1;
+
+	/*
+	 * Set up the zero page, mark it reserved, so that page count
+	 * is not manipulated when freeing the page from user ptes.
+	 */
+	mem_map_zero = alloc_pages(GFP_KERNEL|__GFP_ZERO, 0);
+	if (mem_map_zero == NULL) {
+		prom_printf("paging_init: Cannot alloc zero page.\n");
+		prom_halt();
+	}
+	SetPageReserved(mem_map_zero);
+
+	codepages = (((unsigned long) _etext) - ((unsigned long) _start));
+	codepages = PAGE_ALIGN(codepages) >> PAGE_SHIFT;
+	datapages = (((unsigned long) _edata) - ((unsigned long) _etext));
+	datapages = PAGE_ALIGN(datapages) >> PAGE_SHIFT;
+	initpages = (((unsigned long) __init_end) - ((unsigned long) __init_begin));
+	initpages = PAGE_ALIGN(initpages) >> PAGE_SHIFT;
+
+	printk("Memory: %uk available (%ldk kernel code, %ldk data, %ldk init) [%016lx,%016lx]\n",
+	       nr_free_pages() << (PAGE_SHIFT-10),
+	       codepages << (PAGE_SHIFT-10),
+	       datapages << (PAGE_SHIFT-10), 
+	       initpages << (PAGE_SHIFT-10), 
+	       PAGE_OFFSET, (last_valid_pfn << PAGE_SHIFT));
+
+	if (tlb_type == cheetah || tlb_type == cheetah_plus)
+		cheetah_ecache_flush_init();
+}
+
+void free_initmem (void)
+{
+	unsigned long addr, initend;
+
+	/*
+	 * The init section is aligned to 8k in vmlinux.lds. Page align for >8k pagesizes.
+	 */
+	addr = PAGE_ALIGN((unsigned long)(__init_begin));
+	initend = (unsigned long)(__init_end) & PAGE_MASK;
+	for (; addr < initend; addr += PAGE_SIZE) {
+		unsigned long page;
+		struct page *p;
+
+		page = (addr +
+			((unsigned long) __va(kern_base)) -
+			((unsigned long) KERNBASE));
+		memset((void *)addr, 0xcc, PAGE_SIZE);
+		p = virt_to_page(page);
+
+		ClearPageReserved(p);
+		set_page_count(p, 1);
+		__free_page(p);
+		num_physpages++;
+		totalram_pages++;
+	}
+}
+
+#ifdef CONFIG_BLK_DEV_INITRD
+void free_initrd_mem(unsigned long start, unsigned long end)
+{
+	if (start < end)
+		printk ("Freeing initrd memory: %ldk freed\n", (end - start) >> 10);
+	for (; start < end; start += PAGE_SIZE) {
+		struct page *p = virt_to_page(start);
+
+		ClearPageReserved(p);
+		set_page_count(p, 1);
+		__free_page(p);
+		num_physpages++;
+		totalram_pages++;
+	}
+}
+#endif
diff --git a/arch/sparc64/mm/tlb.c b/arch/sparc64/mm/tlb.c
new file mode 100644
index 0000000..90ca99d
--- /dev/null
+++ b/arch/sparc64/mm/tlb.c
@@ -0,0 +1,151 @@
+/* arch/sparc64/mm/tlb.c
+ *
+ * Copyright (C) 2004 David S. Miller <davem@redhat.com>
+ */
+
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/percpu.h>
+#include <linux/mm.h>
+#include <linux/swap.h>
+
+#include <asm/pgtable.h>
+#include <asm/pgalloc.h>
+#include <asm/tlbflush.h>
+#include <asm/cacheflush.h>
+#include <asm/mmu_context.h>
+#include <asm/tlb.h>
+
+/* Heavily inspired by the ppc64 code.  */
+
+DEFINE_PER_CPU(struct mmu_gather, mmu_gathers) =
+	{ NULL, 0, 0, 0, 0, 0, { 0 }, { NULL }, };
+
+void flush_tlb_pending(void)
+{
+	struct mmu_gather *mp = &__get_cpu_var(mmu_gathers);
+
+	if (mp->tlb_nr) {
+		if (CTX_VALID(mp->mm->context)) {
+#ifdef CONFIG_SMP
+			smp_flush_tlb_pending(mp->mm, mp->tlb_nr,
+					      &mp->vaddrs[0]);
+#else
+			__flush_tlb_pending(CTX_HWBITS(mp->mm->context),
+					    mp->tlb_nr, &mp->vaddrs[0]);
+#endif
+		}
+		mp->tlb_nr = 0;
+	}
+}
+
+void tlb_batch_add(struct mm_struct *mm, unsigned long vaddr, pte_t *ptep, pte_t orig)
+{
+	struct mmu_gather *mp = &__get_cpu_var(mmu_gathers);
+	unsigned long nr;
+
+	vaddr &= PAGE_MASK;
+	if (pte_exec(orig))
+		vaddr |= 0x1UL;
+
+	if (pte_dirty(orig)) {
+		unsigned long paddr, pfn = pte_pfn(orig);
+		struct address_space *mapping;
+		struct page *page;
+
+		if (!pfn_valid(pfn))
+			goto no_cache_flush;
+
+		page = pfn_to_page(pfn);
+		if (PageReserved(page))
+			goto no_cache_flush;
+
+		/* A real file page? */
+		mapping = page_mapping(page);
+		if (!mapping)
+			goto no_cache_flush;
+
+		paddr = (unsigned long) page_address(page);
+		if ((paddr ^ vaddr) & (1 << 13))
+			flush_dcache_page_all(mm, page);
+	}
+
+no_cache_flush:
+
+	if (mp->tlb_frozen)
+		return;
+
+	nr = mp->tlb_nr;
+
+	if (unlikely(nr != 0 && mm != mp->mm)) {
+		flush_tlb_pending();
+		nr = 0;
+	}
+
+	if (nr == 0)
+		mp->mm = mm;
+
+	mp->vaddrs[nr] = vaddr;
+	mp->tlb_nr = ++nr;
+	if (nr >= TLB_BATCH_NR)
+		flush_tlb_pending();
+}
+
+void flush_tlb_pgtables(struct mm_struct *mm, unsigned long start, unsigned long end)
+{
+	struct mmu_gather *mp = &__get_cpu_var(mmu_gathers);
+	unsigned long nr = mp->tlb_nr;
+	long s = start, e = end, vpte_base;
+
+	if (mp->tlb_frozen)
+		return;
+
+	/* If start is greater than end, that is a real problem.  */
+	BUG_ON(start > end);
+
+	/* However, straddling the VA space hole is quite normal. */
+	s &= PMD_MASK;
+	e = (e + PMD_SIZE - 1) & PMD_MASK;
+
+	vpte_base = (tlb_type == spitfire ?
+		     VPTE_BASE_SPITFIRE :
+		     VPTE_BASE_CHEETAH);
+
+	if (unlikely(nr != 0 && mm != mp->mm)) {
+		flush_tlb_pending();
+		nr = 0;
+	}
+
+	if (nr == 0)
+		mp->mm = mm;
+
+	start = vpte_base + (s >> (PAGE_SHIFT - 3));
+	end = vpte_base + (e >> (PAGE_SHIFT - 3));
+
+	/* If the request straddles the VA space hole, we
+	 * need to swap start and end.  The reason this
+	 * occurs is that "vpte_base" is the center of
+	 * the linear page table mapping area.  Thus,
+	 * high addresses with the sign bit set map to
+	 * addresses below vpte_base and non-sign bit
+	 * addresses map to addresses above vpte_base.
+	 */
+	if (end < start) {
+		unsigned long tmp = start;
+
+		start = end;
+		end = tmp;
+	}
+
+	while (start < end) {
+		mp->vaddrs[nr] = start;
+		mp->tlb_nr = ++nr;
+		if (nr >= TLB_BATCH_NR) {
+			flush_tlb_pending();
+			nr = 0;
+		}
+		start += PAGE_SIZE;
+	}
+	if (nr)
+		flush_tlb_pending();
+}
diff --git a/arch/sparc64/mm/ultra.S b/arch/sparc64/mm/ultra.S
new file mode 100644
index 0000000..7a09343
--- /dev/null
+++ b/arch/sparc64/mm/ultra.S
@@ -0,0 +1,583 @@
+/* $Id: ultra.S,v 1.72 2002/02/09 19:49:31 davem Exp $
+ * ultra.S: Don't expand these all over the place...
+ *
+ * Copyright (C) 1997, 2000 David S. Miller (davem@redhat.com)
+ */
+
+#include <linux/config.h>
+#include <asm/asi.h>
+#include <asm/pgtable.h>
+#include <asm/page.h>
+#include <asm/spitfire.h>
+#include <asm/mmu_context.h>
+#include <asm/pil.h>
+#include <asm/head.h>
+#include <asm/thread_info.h>
+#include <asm/cacheflush.h>
+
+	/* Basically, most of the Spitfire vs. Cheetah madness
+	 * has to do with the fact that Cheetah does not support
+	 * IMMU flushes out of the secondary context.  Someone needs
+	 * to throw a south lake birthday party for the folks
+	 * in Microelectronics who refused to fix this shit.
+	 */
+
+	/* This file is meant to be read efficiently by the CPU, not humans.
+	 * Staraj sie tego nikomu nie pierdolnac...
+	 */
+	.text
+	.align		32
+	.globl		__flush_tlb_mm
+__flush_tlb_mm: /* %o0=(ctx & TAG_CONTEXT_BITS), %o1=SECONDARY_CONTEXT */
+	ldxa		[%o1] ASI_DMMU, %g2
+	cmp		%g2, %o0
+	bne,pn		%icc, __spitfire_flush_tlb_mm_slow
+	 mov		0x50, %g3
+	stxa		%g0, [%g3] ASI_DMMU_DEMAP
+	stxa		%g0, [%g3] ASI_IMMU_DEMAP
+	retl
+	 flush		%g6
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+
+	.align		32
+	.globl		__flush_tlb_pending
+__flush_tlb_pending:
+	/* %o0 = context, %o1 = nr, %o2 = vaddrs[] */
+	rdpr		%pstate, %g7
+	sllx		%o1, 3, %o1
+	andn		%g7, PSTATE_IE, %g2
+	wrpr		%g2, %pstate
+	mov		SECONDARY_CONTEXT, %o4
+	ldxa		[%o4] ASI_DMMU, %g2
+	stxa		%o0, [%o4] ASI_DMMU
+1:	sub		%o1, (1 << 3), %o1
+	ldx		[%o2 + %o1], %o3
+	andcc		%o3, 1, %g0
+	andn		%o3, 1, %o3
+	be,pn		%icc, 2f
+	 or		%o3, 0x10, %o3
+	stxa		%g0, [%o3] ASI_IMMU_DEMAP
+2:	stxa		%g0, [%o3] ASI_DMMU_DEMAP
+	membar		#Sync
+	brnz,pt		%o1, 1b
+	 nop
+	stxa		%g2, [%o4] ASI_DMMU
+	flush		%g6
+	retl
+	 wrpr		%g7, 0x0, %pstate
+
+	.align		32
+	.globl		__flush_tlb_kernel_range
+__flush_tlb_kernel_range:	/* %o0=start, %o1=end */
+	cmp		%o0, %o1
+	be,pn		%xcc, 2f
+	 sethi		%hi(PAGE_SIZE), %o4
+	sub		%o1, %o0, %o3
+	sub		%o3, %o4, %o3
+	or		%o0, 0x20, %o0		! Nucleus
+1:	stxa		%g0, [%o0 + %o3] ASI_DMMU_DEMAP
+	stxa		%g0, [%o0 + %o3] ASI_IMMU_DEMAP
+	membar		#Sync
+	brnz,pt		%o3, 1b
+	 sub		%o3, %o4, %o3
+2:	retl
+	 flush		%g6
+
+__spitfire_flush_tlb_mm_slow:
+	rdpr		%pstate, %g1
+	wrpr		%g1, PSTATE_IE, %pstate
+	stxa		%o0, [%o1] ASI_DMMU
+	stxa		%g0, [%g3] ASI_DMMU_DEMAP
+	stxa		%g0, [%g3] ASI_IMMU_DEMAP
+	flush		%g6
+	stxa		%g2, [%o1] ASI_DMMU
+	flush		%g6
+	retl
+	 wrpr		%g1, 0, %pstate
+
+/*
+ * The following code flushes one page_size worth.
+ */
+#if (PAGE_SHIFT == 13)
+#define ITAG_MASK 0xfe
+#elif (PAGE_SHIFT == 16)
+#define ITAG_MASK 0x7fe
+#else
+#error unsupported PAGE_SIZE
+#endif
+	.align		32
+	.globl		__flush_icache_page
+__flush_icache_page:	/* %o0 = phys_page */
+	membar		#StoreStore
+	srlx		%o0, PAGE_SHIFT, %o0
+	sethi		%uhi(PAGE_OFFSET), %g1
+	sllx		%o0, PAGE_SHIFT, %o0
+	sethi		%hi(PAGE_SIZE), %g2
+	sllx		%g1, 32, %g1
+	add		%o0, %g1, %o0
+1:	subcc		%g2, 32, %g2
+	bne,pt		%icc, 1b
+	 flush		%o0 + %g2
+	retl
+	 nop
+
+#ifdef DCACHE_ALIASING_POSSIBLE
+
+#if (PAGE_SHIFT != 13)
+#error only page shift of 13 is supported by dcache flush
+#endif
+
+#define DTAG_MASK 0x3
+
+	.align		64
+	.globl		__flush_dcache_page
+__flush_dcache_page:	/* %o0=kaddr, %o1=flush_icache */
+	sethi		%uhi(PAGE_OFFSET), %g1
+	sllx		%g1, 32, %g1
+	sub		%o0, %g1, %o0
+	clr		%o4
+	srlx		%o0, 11, %o0
+	sethi		%hi(1 << 14), %o2
+1:	ldxa		[%o4] ASI_DCACHE_TAG, %o3	! LSU	Group
+	add		%o4, (1 << 5), %o4		! IEU0
+	ldxa		[%o4] ASI_DCACHE_TAG, %g1	! LSU	Group
+	add		%o4, (1 << 5), %o4		! IEU0
+	ldxa		[%o4] ASI_DCACHE_TAG, %g2	! LSU	Group	o3 available
+	add		%o4, (1 << 5), %o4		! IEU0
+	andn		%o3, DTAG_MASK, %o3		! IEU1
+	ldxa		[%o4] ASI_DCACHE_TAG, %g3	! LSU	Group
+	add		%o4, (1 << 5), %o4		! IEU0
+	andn		%g1, DTAG_MASK, %g1		! IEU1
+	cmp		%o0, %o3			! IEU1	Group
+	be,a,pn		%xcc, dflush1			! CTI
+	 sub		%o4, (4 << 5), %o4		! IEU0	(Group)
+	cmp		%o0, %g1			! IEU1	Group
+	andn		%g2, DTAG_MASK, %g2		! IEU0
+	be,a,pn		%xcc, dflush2			! CTI
+	 sub		%o4, (3 << 5), %o4		! IEU0	(Group)
+	cmp		%o0, %g2			! IEU1	Group
+	andn		%g3, DTAG_MASK, %g3		! IEU0
+	be,a,pn		%xcc, dflush3			! CTI
+	 sub		%o4, (2 << 5), %o4		! IEU0	(Group)
+	cmp		%o0, %g3			! IEU1	Group
+	be,a,pn		%xcc, dflush4			! CTI
+	 sub		%o4, (1 << 5), %o4		! IEU0
+2:	cmp		%o4, %o2			! IEU1	Group
+	bne,pt		%xcc, 1b			! CTI
+	 nop						! IEU0
+
+	/* The I-cache does not snoop local stores so we
+	 * better flush that too when necessary.
+	 */
+	brnz,pt		%o1, __flush_icache_page
+	 sllx		%o0, 11, %o0
+	retl
+	 nop
+
+dflush1:stxa		%g0, [%o4] ASI_DCACHE_TAG
+	add		%o4, (1 << 5), %o4
+dflush2:stxa		%g0, [%o4] ASI_DCACHE_TAG
+	add		%o4, (1 << 5), %o4
+dflush3:stxa		%g0, [%o4] ASI_DCACHE_TAG
+	add		%o4, (1 << 5), %o4
+dflush4:stxa		%g0, [%o4] ASI_DCACHE_TAG
+	add		%o4, (1 << 5), %o4
+	membar		#Sync
+	ba,pt		%xcc, 2b
+	 nop
+#endif /* DCACHE_ALIASING_POSSIBLE */
+
+	.align		32
+__prefill_dtlb:
+	rdpr		%pstate, %g7
+	wrpr		%g7, PSTATE_IE, %pstate
+	mov		TLB_TAG_ACCESS, %g1
+	stxa		%o5, [%g1] ASI_DMMU
+	stxa		%o2, [%g0] ASI_DTLB_DATA_IN
+	flush		%g6
+	retl
+	 wrpr		%g7, %pstate
+__prefill_itlb:
+	rdpr		%pstate, %g7
+	wrpr		%g7, PSTATE_IE, %pstate
+	mov		TLB_TAG_ACCESS, %g1
+	stxa		%o5, [%g1] ASI_IMMU
+	stxa		%o2, [%g0] ASI_ITLB_DATA_IN
+	flush		%g6
+	retl
+	 wrpr		%g7, %pstate
+
+	.globl		__update_mmu_cache
+__update_mmu_cache:	/* %o0=hw_context, %o1=address, %o2=pte, %o3=fault_code */
+	srlx		%o1, PAGE_SHIFT, %o1
+	andcc		%o3, FAULT_CODE_DTLB, %g0
+	sllx		%o1, PAGE_SHIFT, %o5
+	bne,pt		%xcc, __prefill_dtlb
+	 or		%o5, %o0, %o5
+	ba,a,pt		%xcc, __prefill_itlb
+
+	/* Cheetah specific versions, patched at boot time.
+	 *
+	 * This writes of the PRIMARY_CONTEXT register in this file are
+	 * safe even on Cheetah+ and later wrt. the page size fields.
+	 * The nucleus page size fields do not matter because we make
+	 * no data references, and these instructions execute out of a
+	 * locked I-TLB entry sitting in the fully assosciative I-TLB.
+	 * This sequence should also never trap.
+	 */
+__cheetah_flush_tlb_mm: /* 15 insns */
+	rdpr		%pstate, %g7
+	andn		%g7, PSTATE_IE, %g2
+	wrpr		%g2, 0x0, %pstate
+	wrpr		%g0, 1, %tl
+	mov		PRIMARY_CONTEXT, %o2
+	mov		0x40, %g3
+	ldxa		[%o2] ASI_DMMU, %g2
+	stxa		%o0, [%o2] ASI_DMMU
+	stxa		%g0, [%g3] ASI_DMMU_DEMAP
+	stxa		%g0, [%g3] ASI_IMMU_DEMAP
+	stxa		%g2, [%o2] ASI_DMMU
+	flush		%g6
+	wrpr		%g0, 0, %tl
+	retl
+	 wrpr		%g7, 0x0, %pstate
+
+__cheetah_flush_tlb_pending:	/* 22 insns */
+	/* %o0 = context, %o1 = nr, %o2 = vaddrs[] */
+	rdpr		%pstate, %g7
+	sllx		%o1, 3, %o1
+	andn		%g7, PSTATE_IE, %g2
+	wrpr		%g2, 0x0, %pstate
+	wrpr		%g0, 1, %tl
+	mov		PRIMARY_CONTEXT, %o4
+	ldxa		[%o4] ASI_DMMU, %g2
+	stxa		%o0, [%o4] ASI_DMMU
+1:	sub		%o1, (1 << 3), %o1
+	ldx		[%o2 + %o1], %o3
+	andcc		%o3, 1, %g0
+	be,pn		%icc, 2f
+	 andn		%o3, 1, %o3
+	stxa		%g0, [%o3] ASI_IMMU_DEMAP
+2:	stxa		%g0, [%o3] ASI_DMMU_DEMAP	
+	brnz,pt		%o1, 1b
+	 membar		#Sync
+	stxa		%g2, [%o4] ASI_DMMU
+	flush		%g6
+	wrpr		%g0, 0, %tl
+	retl
+	 wrpr		%g7, 0x0, %pstate
+
+#ifdef DCACHE_ALIASING_POSSIBLE
+flush_dcpage_cheetah: /* 11 insns */
+	sethi		%uhi(PAGE_OFFSET), %g1
+	sllx		%g1, 32, %g1
+	sub		%o0, %g1, %o0
+	sethi		%hi(PAGE_SIZE), %o4
+1:	subcc		%o4, (1 << 5), %o4
+	stxa		%g0, [%o0 + %o4] ASI_DCACHE_INVALIDATE
+	membar		#Sync
+	bne,pt		%icc, 1b
+	 nop
+	retl		/* I-cache flush never needed on Cheetah, see callers. */
+	 nop
+#endif /* DCACHE_ALIASING_POSSIBLE */
+
+cheetah_patch_one:
+1:	lduw		[%o1], %g1
+	stw		%g1, [%o0]
+	flush		%o0
+	subcc		%o2, 1, %o2
+	add		%o1, 4, %o1
+	bne,pt		%icc, 1b
+	 add		%o0, 4, %o0
+	retl
+	 nop
+
+	.globl		cheetah_patch_cachetlbops
+cheetah_patch_cachetlbops:
+	save		%sp, -128, %sp
+
+	sethi		%hi(__flush_tlb_mm), %o0
+	or		%o0, %lo(__flush_tlb_mm), %o0
+	sethi		%hi(__cheetah_flush_tlb_mm), %o1
+	or		%o1, %lo(__cheetah_flush_tlb_mm), %o1
+	call		cheetah_patch_one
+	 mov		15, %o2
+
+	sethi		%hi(__flush_tlb_pending), %o0
+	or		%o0, %lo(__flush_tlb_pending), %o0
+	sethi		%hi(__cheetah_flush_tlb_pending), %o1
+	or		%o1, %lo(__cheetah_flush_tlb_pending), %o1
+	call		cheetah_patch_one
+	 mov		22, %o2
+
+#ifdef DCACHE_ALIASING_POSSIBLE
+	sethi		%hi(__flush_dcache_page), %o0
+	or		%o0, %lo(__flush_dcache_page), %o0
+	sethi		%hi(flush_dcpage_cheetah), %o1
+	or		%o1, %lo(flush_dcpage_cheetah), %o1
+	call		cheetah_patch_one
+	 mov		11, %o2
+#endif /* DCACHE_ALIASING_POSSIBLE */
+
+	ret
+	 restore
+
+#ifdef CONFIG_SMP
+	/* These are all called by the slaves of a cross call, at
+	 * trap level 1, with interrupts fully disabled.
+	 *
+	 * Register usage:
+	 *   %g5	mm->context	(all tlb flushes)
+	 *   %g1	address arg 1	(tlb page and range flushes)
+	 *   %g7	address arg 2	(tlb range flush only)
+	 *
+	 *   %g6	ivector table, don't touch
+	 *   %g2	scratch 1
+	 *   %g3	scratch 2
+	 *   %g4	scratch 3
+	 *
+	 * TODO: Make xcall TLB range flushes use the tricks above... -DaveM
+	 */
+	.align		32
+	.globl		xcall_flush_tlb_mm
+xcall_flush_tlb_mm:
+	mov		PRIMARY_CONTEXT, %g2
+	mov		0x40, %g4
+	ldxa		[%g2] ASI_DMMU, %g3
+	stxa		%g5, [%g2] ASI_DMMU
+	stxa		%g0, [%g4] ASI_DMMU_DEMAP
+	stxa		%g0, [%g4] ASI_IMMU_DEMAP
+	stxa		%g3, [%g2] ASI_DMMU
+	retry
+
+	.globl		xcall_flush_tlb_pending
+xcall_flush_tlb_pending:
+	/* %g5=context, %g1=nr, %g7=vaddrs[] */
+	sllx		%g1, 3, %g1
+	mov		PRIMARY_CONTEXT, %g4
+	ldxa		[%g4] ASI_DMMU, %g2
+	stxa		%g5, [%g4] ASI_DMMU
+1:	sub		%g1, (1 << 3), %g1
+	ldx		[%g7 + %g1], %g5
+	andcc		%g5, 0x1, %g0
+	be,pn		%icc, 2f
+
+	 andn		%g5, 0x1, %g5
+	stxa		%g0, [%g5] ASI_IMMU_DEMAP
+2:	stxa		%g0, [%g5] ASI_DMMU_DEMAP
+	membar		#Sync
+	brnz,pt		%g1, 1b
+	 nop
+	stxa		%g2, [%g4] ASI_DMMU
+	retry
+
+	.globl		xcall_flush_tlb_kernel_range
+xcall_flush_tlb_kernel_range:
+	sethi		%hi(PAGE_SIZE - 1), %g2
+	or		%g2, %lo(PAGE_SIZE - 1), %g2
+	andn		%g1, %g2, %g1
+	andn		%g7, %g2, %g7
+	sub		%g7, %g1, %g3
+	add		%g2, 1, %g2
+	sub		%g3, %g2, %g3
+	or		%g1, 0x20, %g1		! Nucleus
+1:	stxa		%g0, [%g1 + %g3] ASI_DMMU_DEMAP
+	stxa		%g0, [%g1 + %g3] ASI_IMMU_DEMAP
+	membar		#Sync
+	brnz,pt		%g3, 1b
+	 sub		%g3, %g2, %g3
+	retry
+	nop
+	nop
+
+	/* This runs in a very controlled environment, so we do
+	 * not need to worry about BH races etc.
+	 */
+	.globl		xcall_sync_tick
+xcall_sync_tick:
+	rdpr		%pstate, %g2
+	wrpr		%g2, PSTATE_IG | PSTATE_AG, %pstate
+	rdpr		%pil, %g2
+	wrpr		%g0, 15, %pil
+	sethi		%hi(109f), %g7
+	b,pt		%xcc, etrap_irq
+109:	 or		%g7, %lo(109b), %g7
+	call		smp_synchronize_tick_client
+	 nop
+	clr		%l6
+	b		rtrap_xcall
+	 ldx		[%sp + PTREGS_OFF + PT_V9_TSTATE], %l1
+
+	/* NOTE: This is SPECIAL!!  We do etrap/rtrap however
+	 *       we choose to deal with the "BH's run with
+	 *       %pil==15" problem (described in asm/pil.h)
+	 *       by just invoking rtrap directly past where
+	 *       BH's are checked for.
+	 *
+	 *       We do it like this because we do not want %pil==15
+	 *       lockups to prevent regs being reported.
+	 */
+	.globl		xcall_report_regs
+xcall_report_regs:
+	rdpr		%pstate, %g2
+	wrpr		%g2, PSTATE_IG | PSTATE_AG, %pstate
+	rdpr		%pil, %g2
+	wrpr		%g0, 15, %pil
+	sethi		%hi(109f), %g7
+	b,pt		%xcc, etrap_irq
+109:	 or		%g7, %lo(109b), %g7
+	call		__show_regs
+	 add		%sp, PTREGS_OFF, %o0
+	clr		%l6
+	/* Has to be a non-v9 branch due to the large distance. */
+	b		rtrap_xcall
+	 ldx		[%sp + PTREGS_OFF + PT_V9_TSTATE], %l1
+
+#ifdef DCACHE_ALIASING_POSSIBLE
+	.align		32
+	.globl		xcall_flush_dcache_page_cheetah
+xcall_flush_dcache_page_cheetah: /* %g1 == physical page address */
+	sethi		%hi(PAGE_SIZE), %g3
+1:	subcc		%g3, (1 << 5), %g3
+	stxa		%g0, [%g1 + %g3] ASI_DCACHE_INVALIDATE
+	membar		#Sync
+	bne,pt		%icc, 1b
+	 nop
+	retry
+	nop
+#endif /* DCACHE_ALIASING_POSSIBLE */
+
+	.globl		xcall_flush_dcache_page_spitfire
+xcall_flush_dcache_page_spitfire: /* %g1 == physical page address
+				     %g7 == kernel page virtual address
+				     %g5 == (page->mapping != NULL)  */
+#ifdef DCACHE_ALIASING_POSSIBLE
+	srlx		%g1, (13 - 2), %g1	! Form tag comparitor
+	sethi		%hi(L1DCACHE_SIZE), %g3	! D$ size == 16K
+	sub		%g3, (1 << 5), %g3	! D$ linesize == 32
+1:	ldxa		[%g3] ASI_DCACHE_TAG, %g2
+	andcc		%g2, 0x3, %g0
+	be,pn		%xcc, 2f
+	 andn		%g2, 0x3, %g2
+	cmp		%g2, %g1
+
+	bne,pt		%xcc, 2f
+	 nop
+	stxa		%g0, [%g3] ASI_DCACHE_TAG
+	membar		#Sync
+2:	cmp		%g3, 0
+	bne,pt		%xcc, 1b
+	 sub		%g3, (1 << 5), %g3
+
+	brz,pn		%g5, 2f
+#endif /* DCACHE_ALIASING_POSSIBLE */
+	 sethi		%hi(PAGE_SIZE), %g3
+
+1:	flush		%g7
+	subcc		%g3, (1 << 5), %g3
+	bne,pt		%icc, 1b
+	 add		%g7, (1 << 5), %g7
+
+2:	retry
+	nop
+	nop
+
+	.globl		xcall_promstop
+xcall_promstop:
+	rdpr		%pstate, %g2
+	wrpr		%g2, PSTATE_IG | PSTATE_AG, %pstate
+	rdpr		%pil, %g2
+	wrpr		%g0, 15, %pil
+	sethi		%hi(109f), %g7
+	b,pt		%xcc, etrap_irq
+109:	 or		%g7, %lo(109b), %g7
+	flushw
+	call		prom_stopself
+	 nop
+	/* We should not return, just spin if we do... */
+1:	b,a,pt		%xcc, 1b
+	nop
+
+	.data
+
+errata32_hwbug:
+	.xword	0
+
+	.text
+
+	/* These two are not performance critical... */
+	.globl		xcall_flush_tlb_all_spitfire
+xcall_flush_tlb_all_spitfire:
+	/* Spitfire Errata #32 workaround. */
+	sethi		%hi(errata32_hwbug), %g4
+	stx		%g0, [%g4 + %lo(errata32_hwbug)]
+
+	clr		%g2
+	clr		%g3
+1:	ldxa		[%g3] ASI_DTLB_DATA_ACCESS, %g4
+	and		%g4, _PAGE_L, %g5
+	brnz,pn		%g5, 2f
+	 mov		TLB_TAG_ACCESS, %g7
+
+	stxa		%g0, [%g7] ASI_DMMU
+	membar		#Sync
+	stxa		%g0, [%g3] ASI_DTLB_DATA_ACCESS
+	membar		#Sync
+
+	/* Spitfire Errata #32 workaround. */
+	sethi		%hi(errata32_hwbug), %g4
+	stx		%g0, [%g4 + %lo(errata32_hwbug)]
+
+2:	ldxa		[%g3] ASI_ITLB_DATA_ACCESS, %g4
+	and		%g4, _PAGE_L, %g5
+	brnz,pn		%g5, 2f
+	 mov		TLB_TAG_ACCESS, %g7
+
+	stxa		%g0, [%g7] ASI_IMMU
+	membar		#Sync
+	stxa		%g0, [%g3] ASI_ITLB_DATA_ACCESS
+	membar		#Sync
+
+	/* Spitfire Errata #32 workaround. */
+	sethi		%hi(errata32_hwbug), %g4
+	stx		%g0, [%g4 + %lo(errata32_hwbug)]
+
+2:	add		%g2, 1, %g2
+	cmp		%g2, SPITFIRE_HIGHEST_LOCKED_TLBENT
+	ble,pt		%icc, 1b
+	 sll		%g2, 3, %g3
+	flush		%g6
+	retry
+
+	.globl		xcall_flush_tlb_all_cheetah
+xcall_flush_tlb_all_cheetah:
+	mov		0x80, %g2
+	stxa		%g0, [%g2] ASI_DMMU_DEMAP
+	stxa		%g0, [%g2] ASI_IMMU_DEMAP
+	retry
+
+	/* These just get rescheduled to PIL vectors. */
+	.globl		xcall_call_function
+xcall_call_function:
+	wr		%g0, (1 << PIL_SMP_CALL_FUNC), %set_softint
+	retry
+
+	.globl		xcall_receive_signal
+xcall_receive_signal:
+	wr		%g0, (1 << PIL_SMP_RECEIVE_SIGNAL), %set_softint
+	retry
+
+	.globl		xcall_capture
+xcall_capture:
+	wr		%g0, (1 << PIL_SMP_CAPTURE), %set_softint
+	retry
+
+#endif /* CONFIG_SMP */