Fixed MTP to work with TWRP

arch/sparc/mm/Makefile

@@ -0,0 +1,19 @@
+# Makefile for the linux Sparc-specific parts of the memory manager.
+#
+
+asflags-y := -ansi
+ccflags-y := -Werror
+
+obj-$(CONFIG_SPARC64)   += ultra.o tlb.o tsb.o gup.o
+obj-y                   += fault_$(BITS).o
+obj-y                   += init_$(BITS).o
+obj-$(CONFIG_SPARC32)   += extable.o srmmu.o iommu.o io-unit.o
+obj-$(CONFIG_SPARC32)   += srmmu_access.o
+obj-$(CONFIG_SPARC32)   += hypersparc.o viking.o tsunami.o swift.o
+obj-$(CONFIG_SPARC32)   += leon_mm.o
+
+# Only used by sparc64
+obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpage.o
+
+# Only used by sparc32
+obj-$(CONFIG_HIGHMEM)   += highmem.o

arch/sparc/mm/extable.c

@@ -0,0 +1,105 @@
+/*
+ * linux/arch/sparc/mm/extable.c
+ */
+
+#include <linux/module.h>
+#include <asm/uaccess.h>
+
+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
+	 *
+	 * Deleted entries are encoded as:
+	 *	word 1: unused
+	 *	word 2: -1
+	 *
+	 * 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;
+		}
+
+		/* A deleted entry; see trim_init_extable */
+		if (walk->fixup == -1)
+			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;
+}
+
+#ifdef CONFIG_MODULES
+/* We could memmove them around; easier to mark the trimmed ones. */
+void trim_init_extable(struct module *m)
+{
+	unsigned int i;
+	bool range;
+
+	for (i = 0; i < m->num_exentries; i += range ? 2 : 1) {
+		range = m->extable[i].fixup == 0;
+
+		if (within_module_init(m->extable[i].insn, m)) {
+			m->extable[i].fixup = -1;
+			if (range)
+				m->extable[i+1].fixup = -1;
+		}
+		if (range)
+			i++;
+	}
+}
+#endif /* CONFIG_MODULES */
+
+/* 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;
+}

arch/sparc/mm/fault_32.c

@@ -0,0 +1,468 @@
+/*
+ * fault.c:  Page fault handlers for the Sparc.
+ *
+ * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
+ * Copyright (C) 1996 Eddie C. Dost (ecd@skynet.be)
+ * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.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/threads.h>
+#include <linux/kernel.h>
+#include <linux/signal.h>
+#include <linux/mm.h>
+#include <linux/smp.h>
+#include <linux/perf_event.h>
+#include <linux/interrupt.h>
+#include <linux/kdebug.h>
+
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/openprom.h>
+#include <asm/oplib.h>
+#include <asm/setup.h>
+#include <asm/smp.h>
+#include <asm/traps.h>
+#include <asm/uaccess.h>
+
+#include "mm_32.h"
+
+int show_unhandled_signals = 1;
+
+static void __noreturn 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 %08lx\n",
+		       address);
+	}
+	printk(KERN_ALERT "tsk->{mm,active_mm}->context = %08lx\n",
+		(tsk->mm ? tsk->mm->context : tsk->active_mm->context));
+	printk(KERN_ALERT "tsk->{mm,active_mm}->pgd = %08lx\n",
+		(tsk->mm ? (unsigned long) tsk->mm->pgd :
+			(unsigned long) tsk->active_mm->pgd));
+	die_if_kernel("Oops", regs);
+}
+
+asmlinkage int lookup_fault(unsigned long pc, unsigned long ret_pc,
+			    unsigned long address)
+{
+	struct pt_regs regs;
+	unsigned long g2;
+	unsigned int insn;
+	int i;
+
+	i = search_extables_range(ret_pc, &g2);
+	switch (i) {
+	case 3:
+		/* load & store will be handled by fixup */
+		return 3;
+
+	case 1:
+		/* store will be handled by fixup, load will bump out */
+		/* for _to_ macros */
+		insn = *((unsigned int *) pc);
+		if ((insn >> 21) & 1)
+			return 1;
+		break;
+
+	case 2:
+		/* load will be handled by fixup, store will bump out */
+		/* for _from_ macros */
+		insn = *((unsigned int *) pc);
+		if (!((insn >> 21) & 1) || ((insn>>19)&0x3f) == 15)
+			return 2;
+		break;
+
+	default:
+		break;
+	}
+
+	memset(&regs, 0, sizeof(regs));
+	regs.pc = pc;
+	regs.npc = pc + 4;
+	__asm__ __volatile__(
+		"rd %%psr, %0\n\t"
+		"nop\n\t"
+		"nop\n\t"
+		"nop\n" : "=r" (regs.psr));
+	unhandled_fault(address, current, &regs);
+
+	/* Not reached */
+	return 0;
+}
+
+static inline void
+show_signal_msg(struct pt_regs *regs, int sig, int code,
+		unsigned long address, struct task_struct *tsk)
+{
+	if (!unhandled_signal(tsk, sig))
+		return;
+
+	if (!printk_ratelimit())
+		return;
+
+	printk("%s%s[%d]: segfault at %lx ip %p (rpc %p) sp %p error %x",
+	       task_pid_nr(tsk) > 1 ? KERN_INFO : KERN_EMERG,
+	       tsk->comm, task_pid_nr(tsk), address,
+	       (void *)regs->pc, (void *)regs->u_regs[UREG_I7],
+	       (void *)regs->u_regs[UREG_FP], code);
+
+	print_vma_addr(KERN_CONT " in ", regs->pc);
+
+	printk(KERN_CONT "\n");
+}
+
+static void __do_fault_siginfo(int code, int sig, struct pt_regs *regs,
+			       unsigned long addr)
+{
+	siginfo_t info;
+
+	info.si_signo = sig;
+	info.si_code = code;
+	info.si_errno = 0;
+	info.si_addr = (void __user *) addr;
+	info.si_trapno = 0;
+
+	if (unlikely(show_unhandled_signals))
+		show_signal_msg(regs, sig, info.si_code,
+				addr, current);
+
+	force_sig_info (sig, &info, current);
+}
+
+static unsigned long compute_si_addr(struct pt_regs *regs, int text_fault)
+{
+	unsigned int insn;
+
+	if (text_fault)
+		return regs->pc;
+
+	if (regs->psr & PSR_PS)
+		insn = *(unsigned int *) regs->pc;
+	else
+		__get_user(insn, (unsigned int *) regs->pc);
+
+	return safe_compute_effective_address(regs, insn);
+}
+
+static noinline void do_fault_siginfo(int code, int sig, struct pt_regs *regs,
+				      int text_fault)
+{
+	unsigned long addr = compute_si_addr(regs, text_fault);
+
+	__do_fault_siginfo(code, sig, regs, addr);
+}
+
+asmlinkage void do_sparc_fault(struct pt_regs *regs, int text_fault, int write,
+			       unsigned long address)
+{
+	struct vm_area_struct *vma;
+	struct task_struct *tsk = current;
+	struct mm_struct *mm = tsk->mm;
+	unsigned int fixup;
+	unsigned long g2;
+	int from_user = !(regs->psr & PSR_PS);
+	int fault, code;
+	unsigned int flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE;
+
+	if (text_fault)
+		address = regs->pc;
+
+	/*
+	 * We fault-in kernel-space virtual memory on-demand. The
+	 * 'reference' page table is init_mm.pgd.
+	 *
+	 * NOTE! We MUST NOT take any locks for this case. We may
+	 * be in an interrupt or a critical region, and should
+	 * only copy the information from the master page table,
+	 * nothing more.
+	 */
+	code = SEGV_MAPERR;
+	if (address >= TASK_SIZE)
+		goto vmalloc_fault;
+
+	/*
+	 * If we're in an interrupt or have no user
+	 * context, we must not take the fault..
+	 */
+	if (in_atomic() || !mm)
+		goto no_context;
+
+	perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS, 1, regs, address);
+
+retry:
+	down_read(&mm->mmap_sem);
+
+	if (!from_user && address >= PAGE_OFFSET)
+		goto bad_area;
+
+	vma = find_vma(mm, address);
+	if (!vma)
+		goto bad_area;
+	if (vma->vm_start <= address)
+		goto good_area;
+	if (!(vma->vm_flags & VM_GROWSDOWN))
+		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:
+	code = SEGV_ACCERR;
+	if (write) {
+		if (!(vma->vm_flags & VM_WRITE))
+			goto bad_area;
+	} else {
+		/* Allow reads even for write-only mappings */
+		if (!(vma->vm_flags & (VM_READ | VM_EXEC)))
+			goto bad_area;
+	}
+
+	if (from_user)
+		flags |= FAULT_FLAG_USER;
+	if (write)
+		flags |= FAULT_FLAG_WRITE;
+
+	/*
+	 * If for any reason at all we couldn't handle the fault,
+	 * make sure we exit gracefully rather than endlessly redo
+	 * the fault.
+	 */
+	fault = handle_mm_fault(mm, vma, address, flags);
+
+	if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current))
+		return;
+
+	if (unlikely(fault & VM_FAULT_ERROR)) {
+		if (fault & VM_FAULT_OOM)
+			goto out_of_memory;
+		else if (fault & VM_FAULT_SIGSEGV)
+			goto bad_area;
+		else if (fault & VM_FAULT_SIGBUS)
+			goto do_sigbus;
+		BUG();
+	}
+
+	if (flags & FAULT_FLAG_ALLOW_RETRY) {
+		if (fault & VM_FAULT_MAJOR) {
+			current->maj_flt++;
+			perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MAJ,
+				      1, regs, address);
+		} else {
+			current->min_flt++;
+			perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MIN,
+				      1, regs, address);
+		}
+		if (fault & VM_FAULT_RETRY) {
+			flags &= ~FAULT_FLAG_ALLOW_RETRY;
+			flags |= FAULT_FLAG_TRIED;
+
+			/* No need to up_read(&mm->mmap_sem) as we would
+			 * have already released it in __lock_page_or_retry
+			 * in mm/filemap.c.
+			 */
+
+			goto retry;
+		}
+	}
+
+	up_read(&mm->mmap_sem);
+	return;
+
+	/*
+	 * 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:
+	up_read(&mm->mmap_sem);
+
+bad_area_nosemaphore:
+	/* User mode accesses just cause a SIGSEGV */
+	if (from_user) {
+		do_fault_siginfo(code, SIGSEGV, regs, text_fault);
+		return;
+	}
+
+	/* Is this in ex_table? */
+no_context:
+	g2 = regs->u_regs[UREG_G2];
+	if (!from_user) {
+		fixup = search_extables_range(regs->pc, &g2);
+		/* Values below 10 are reserved for other things */
+		if (fixup > 10) {
+			extern const unsigned __memset_start[];
+			extern const unsigned __memset_end[];
+			extern const unsigned __csum_partial_copy_start[];
+			extern const unsigned __csum_partial_copy_end[];
+
+#ifdef DEBUG_EXCEPTIONS
+			printk("Exception: PC<%08lx> faddr<%08lx>\n",
+			       regs->pc, address);
+			printk("EX_TABLE: insn<%08lx> fixup<%08x> g2<%08lx>\n",
+				regs->pc, fixup, g2);
+#endif
+			if ((regs->pc >= (unsigned long)__memset_start &&
+			     regs->pc < (unsigned long)__memset_end) ||
+			    (regs->pc >= (unsigned long)__csum_partial_copy_start &&
+			     regs->pc < (unsigned long)__csum_partial_copy_end)) {
+				regs->u_regs[UREG_I4] = address;
+				regs->u_regs[UREG_I5] = regs->pc;
+			}
+			regs->u_regs[UREG_G2] = g2;
+			regs->pc = fixup;
+			regs->npc = regs->pc + 4;
+			return;
+		}
+	}
+
+	unhandled_fault(address, tsk, regs);
+	do_exit(SIGKILL);
+
+/*
+ * 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:
+	up_read(&mm->mmap_sem);
+	if (from_user) {
+		pagefault_out_of_memory();
+		return;
+	}
+	goto no_context;
+
+do_sigbus:
+	up_read(&mm->mmap_sem);
+	do_fault_siginfo(BUS_ADRERR, SIGBUS, regs, text_fault);
+	if (!from_user)
+		goto no_context;
+
+vmalloc_fault:
+	{
+		/*
+		 * Synchronize this task's top level page-table
+		 * with the 'reference' page table.
+		 */
+		int offset = pgd_index(address);
+		pgd_t *pgd, *pgd_k;
+		pmd_t *pmd, *pmd_k;
+
+		pgd = tsk->active_mm->pgd + offset;
+		pgd_k = init_mm.pgd + offset;
+
+		if (!pgd_present(*pgd)) {
+			if (!pgd_present(*pgd_k))
+				goto bad_area_nosemaphore;
+			pgd_val(*pgd) = pgd_val(*pgd_k);
+			return;
+		}
+
+		pmd = pmd_offset(pgd, address);
+		pmd_k = pmd_offset(pgd_k, address);
+
+		if (pmd_present(*pmd) || !pmd_present(*pmd_k))
+			goto bad_area_nosemaphore;
+
+		*pmd = *pmd_k;
+		return;
+	}
+}
+
+/* This always deals with user addresses. */
+static void force_user_fault(unsigned long address, int write)
+{
+	struct vm_area_struct *vma;
+	struct task_struct *tsk = current;
+	struct mm_struct *mm = tsk->mm;
+	unsigned int flags = FAULT_FLAG_USER;
+	int code;
+
+	code = SEGV_MAPERR;
+
+	down_read(&mm->mmap_sem);
+	vma = find_vma(mm, address);
+	if (!vma)
+		goto bad_area;
+	if (vma->vm_start <= address)
+		goto good_area;
+	if (!(vma->vm_flags & VM_GROWSDOWN))
+		goto bad_area;
+	if (expand_stack(vma, address))
+		goto bad_area;
+good_area:
+	code = SEGV_ACCERR;
+	if (write) {
+		if (!(vma->vm_flags & VM_WRITE))
+			goto bad_area;
+		flags |= FAULT_FLAG_WRITE;
+	} else {
+		if (!(vma->vm_flags & (VM_READ | VM_EXEC)))
+			goto bad_area;
+	}
+	switch (handle_mm_fault(mm, vma, address, flags)) {
+	case VM_FAULT_SIGBUS:
+	case VM_FAULT_OOM:
+		goto do_sigbus;
+	}
+	up_read(&mm->mmap_sem);
+	return;
+bad_area:
+	up_read(&mm->mmap_sem);
+	__do_fault_siginfo(code, SIGSEGV, tsk->thread.kregs, address);
+	return;
+
+do_sigbus:
+	up_read(&mm->mmap_sem);
+	__do_fault_siginfo(BUS_ADRERR, SIGBUS, tsk->thread.kregs, address);
+}
+
+static void check_stack_aligned(unsigned long sp)
+{
+	if (sp & 0x7UL)
+		force_sig(SIGILL, current);
+}
+
+void window_overflow_fault(void)
+{
+	unsigned long sp;
+
+	sp = current_thread_info()->rwbuf_stkptrs[0];
+	if (((sp + 0x38) & PAGE_MASK) != (sp & PAGE_MASK))
+		force_user_fault(sp + 0x38, 1);
+	force_user_fault(sp, 1);
+
+	check_stack_aligned(sp);
+}
+
+void window_underflow_fault(unsigned long sp)
+{
+	if (((sp + 0x38) & PAGE_MASK) != (sp & PAGE_MASK))
+		force_user_fault(sp + 0x38, 0);
+	force_user_fault(sp, 0);
+
+	check_stack_aligned(sp);
+}
+
+void window_ret_fault(struct pt_regs *regs)
+{
+	unsigned long sp;
+
+	sp = regs->u_regs[UREG_FP];
+	if (((sp + 0x38) & PAGE_MASK) != (sp & PAGE_MASK))
+		force_user_fault(sp + 0x38, 0);
+	force_user_fault(sp, 0);
+
+	check_stack_aligned(sp);
+}

arch/sparc/mm/fault_64.c

@@ -0,0 +1,544 @@
+/*
+ * arch/sparc64/mm/fault.c: Page fault handlers for the 64-bit Sparc.
+ *
+ * Copyright (C) 1996, 2008 David S. Miller (davem@davemloft.net)
+ * 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/init.h>
+#include <linux/perf_event.h>
+#include <linux/interrupt.h>
+#include <linux/kprobes.h>
+#include <linux/kdebug.h>
+#include <linux/percpu.h>
+#include <linux/context_tracking.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/mmu_context.h>
+#include <asm/setup.h>
+
+int show_unhandled_signals = 1;
+
+static inline __kprobes int notify_page_fault(struct pt_regs *regs)
+{
+	int ret = 0;
+
+	/* kprobe_running() needs smp_processor_id() */
+	if (kprobes_built_in() && !user_mode(regs)) {
+		preempt_disable();
+		if (kprobe_running() && kprobe_fault_handler(regs, 0))
+			ret = 1;
+		preempt_enable();
+	}
+	return ret;
+}
+
+static void __kprobes 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));
+	die_if_kernel("Oops", regs);
+}
+
+static void __kprobes bad_kernel_pc(struct pt_regs *regs, unsigned long vaddr)
+{
+	printk(KERN_CRIT "OOPS: Bogus kernel PC [%016lx] in fault handler\n",
+	       regs->tpc);
+	printk(KERN_CRIT "OOPS: RPC [%016lx]\n", regs->u_regs[15]);
+	printk("OOPS: RPC <%pS>\n", (void *) regs->u_regs[15]);
+	printk(KERN_CRIT "OOPS: Fault was to vaddr[%lx]\n", vaddr);
+	dump_stack();
+	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. vmtruncate likewise. This saves us having to get pte 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;
+
+	if (pgd_none(*pgdp) || unlikely(pgd_bad(*pgdp)))
+		goto out;
+	pudp = pud_offset(pgdp, tpc);
+	if (pud_none(*pudp) || unlikely(pud_bad(*pudp)))
+		goto out;
+
+	/* This disables preemption for us as well. */
+	local_irq_disable();
+
+	pmdp = pmd_offset(pudp, tpc);
+	if (pmd_none(*pmdp) || unlikely(pmd_bad(*pmdp)))
+		goto out_irq_enable;
+
+#ifdef CONFIG_TRANSPARENT_HUGEPAGE
+	if (pmd_trans_huge(*pmdp)) {
+		if (pmd_trans_splitting(*pmdp))
+			goto out_irq_enable;
+
+		pa  = pmd_pfn(*pmdp) << PAGE_SHIFT;
+		pa += tpc & ~HPAGE_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));
+	} else
+#endif
+	{
+		ptep = pte_offset_map(pmdp, tpc);
+		pte = *ptep;
+		if (pte_present(pte)) {
+			pa  = (pte_pfn(pte) << PAGE_SHIFT);
+			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));
+		}
+		pte_unmap(ptep);
+	}
+out_irq_enable:
+	local_irq_enable();
+out:
+	return insn;
+}
+
+static inline void
+show_signal_msg(struct pt_regs *regs, int sig, int code,
+		unsigned long address, struct task_struct *tsk)
+{
+	if (!unhandled_signal(tsk, sig))
+		return;
+
+	if (!printk_ratelimit())
+		return;
+
+	printk("%s%s[%d]: segfault at %lx ip %p (rpc %p) sp %p error %x",
+	       task_pid_nr(tsk) > 1 ? KERN_INFO : KERN_EMERG,
+	       tsk->comm, task_pid_nr(tsk), address,
+	       (void *)regs->tpc, (void *)regs->u_regs[UREG_I7],
+	       (void *)regs->u_regs[UREG_FP], code);
+
+	print_vma_addr(KERN_CONT " in ", regs->tpc);
+
+	printk(KERN_CONT "\n");
+}
+
+static void do_fault_siginfo(int code, int sig, struct pt_regs *regs,
+			     unsigned long fault_addr, unsigned int insn,
+			     int fault_code)
+{
+	unsigned long addr;
+	siginfo_t info;
+
+	info.si_code = code;
+	info.si_signo = sig;
+	info.si_errno = 0;
+	if (fault_code & FAULT_CODE_ITLB) {
+		addr = regs->tpc;
+	} else {
+		/* If we were able to probe the faulting instruction, use it
+		 * to compute a precise fault address.  Otherwise use the fault
+		 * time provided address which may only have page granularity.
+		 */
+		if (insn)
+			addr = compute_effective_address(regs, insn, 0);
+		else
+			addr = fault_addr;
+	}
+	info.si_addr = (void __user *) addr;
+	info.si_trapno = 0;
+
+	if (unlikely(show_unhandled_signals))
+		show_signal_msg(regs, sig, code, addr, current);
+
+	force_sig_info(sig, &info, current);
+}
+
+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 __kprobes do_kernel_fault(struct pt_regs *regs, int si_code,
+				      int fault_code, unsigned int insn,
+				      unsigned long address)
+{
+	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;
+		}
+	}
+		
+	/* Is this in ex_table? */
+	if (regs->tstate & TSTATE_PRIV) {
+		const struct exception_table_entry *entry;
+
+		entry = search_exception_tables(regs->tpc);
+		if (entry) {
+			regs->tpc = entry->fixup;
+			regs->tnpc = regs->tpc + 4;
+			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, address, insn, fault_code);
+		return;
+	}
+
+cannot_handle:
+	unhandled_fault (address, current, regs);
+}
+
+static void noinline __kprobes bogus_32bit_fault_tpc(struct pt_regs *regs)
+{
+	static int times;
+
+	if (times++ < 10)
+		printk(KERN_ERR "FAULT[%s:%d]: 32-bit process reports "
+		       "64-bit TPC [%lx]\n",
+		       current->comm, current->pid,
+		       regs->tpc);
+	show_regs(regs);
+}
+
+asmlinkage void __kprobes do_sparc64_fault(struct pt_regs *regs)
+{
+	enum ctx_state prev_state = exception_enter();
+	struct mm_struct *mm = current->mm;
+	struct vm_area_struct *vma;
+	unsigned int insn = 0;
+	int si_code, fault_code, fault;
+	unsigned long address, mm_rss;
+	unsigned int flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE;
+
+	fault_code = get_thread_fault_code();
+
+	if (notify_page_fault(regs))
+		goto exit_exception;
+
+	si_code = SEGV_MAPERR;
+	address = current_thread_info()->fault_address;
+
+	if ((fault_code & FAULT_CODE_ITLB) &&
+	    (fault_code & FAULT_CODE_DTLB))
+		BUG();
+
+	if (test_thread_flag(TIF_32BIT)) {
+		if (!(regs->tstate & TSTATE_PRIV)) {
+			if (unlikely((regs->tpc >> 32) != 0)) {
+				bogus_32bit_fault_tpc(regs);
+				goto intr_or_no_mm;
+			}
+		}
+		if (unlikely((address >> 32) != 0))
+			goto intr_or_no_mm;
+	}
+
+	if (regs->tstate & TSTATE_PRIV) {
+		unsigned long tpc = regs->tpc;
+
+		/* Sanity check the PC. */
+		if ((tpc >= KERNBASE && tpc < (unsigned long) __init_end) ||
+		    (tpc >= MODULES_VADDR && tpc < MODULES_END)) {
+			/* Valid, no problems... */
+		} else {
+			bad_kernel_pc(regs, address);
+			goto exit_exception;
+		}
+	} else
+		flags |= FAULT_FLAG_USER;
+
+	/*
+	 * 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;
+
+	perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS, 1, regs, address);
+
+	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;
+		}
+
+retry:
+		down_read(&mm->mmap_sem);
+	}
+
+	if (fault_code & FAULT_CODE_BAD_RA)
+		goto do_sigbus;
+
+	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;
+		/* All loads, stores and atomics have bits 30 and 31 both set
+		 * in the instruction.  Bit 21 is set in all stores, but we
+		 * have to avoid prefetches which also have bit 21 set.
+		 */
+		if ((insn & 0xc0200000) == 0xc0200000 &&
+		    (insn & 0x01780000) != 0x01680000) {
+			/* 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);
+
+		flags |= FAULT_FLAG_WRITE;
+	} else {
+		/* Allow reads even for write-only mappings */
+		if (!(vma->vm_flags & (VM_READ | VM_EXEC)))
+			goto bad_area;
+	}
+
+	fault = handle_mm_fault(mm, vma, address, flags);
+
+	if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current))
+		goto exit_exception;
+
+	if (unlikely(fault & VM_FAULT_ERROR)) {
+		if (fault & VM_FAULT_OOM)
+			goto out_of_memory;
+		else if (fault & VM_FAULT_SIGSEGV)
+			goto bad_area;
+		else if (fault & VM_FAULT_SIGBUS)
+			goto do_sigbus;
+		BUG();
+	}
+
+	if (flags & FAULT_FLAG_ALLOW_RETRY) {
+		if (fault & VM_FAULT_MAJOR) {
+			current->maj_flt++;
+			perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MAJ,
+				      1, regs, address);
+		} else {
+			current->min_flt++;
+			perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MIN,
+				      1, regs, address);
+		}
+		if (fault & VM_FAULT_RETRY) {
+			flags &= ~FAULT_FLAG_ALLOW_RETRY;
+			flags |= FAULT_FLAG_TRIED;
+
+			/* No need to up_read(&mm->mmap_sem) as we would
+			 * have already released it in __lock_page_or_retry
+			 * in mm/filemap.c.
+			 */
+
+			goto retry;
+		}
+	}
+	up_read(&mm->mmap_sem);
+
+	mm_rss = get_mm_rss(mm);
+#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
+	mm_rss -= (mm->context.huge_pte_count * (HPAGE_SIZE / PAGE_SIZE));
+#endif
+	if (unlikely(mm_rss >
+		     mm->context.tsb_block[MM_TSB_BASE].tsb_rss_limit))
+		tsb_grow(mm, MM_TSB_BASE, mm_rss);
+#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
+	mm_rss = mm->context.huge_pte_count;
+	if (unlikely(mm_rss >
+		     mm->context.tsb_block[MM_TSB_HUGE].tsb_rss_limit)) {
+		if (mm->context.tsb_block[MM_TSB_HUGE].tsb)
+			tsb_grow(mm, MM_TSB_HUGE, mm_rss);
+		else
+			hugetlb_setup(regs);
+
+	}
+#endif
+exit_exception:
+	exception_exit(prev_state);
+	return;
+
+	/*
+	 * 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 exit_exception;
+
+/*
+ * 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);
+	if (!(regs->tstate & TSTATE_PRIV)) {
+		pagefault_out_of_memory();
+		goto exit_exception;
+	}
+	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, address, insn, fault_code);
+
+	/* Kernel mode? Handle exceptions or die */
+	if (regs->tstate & TSTATE_PRIV)
+		goto handle_kernel_fault;
+}

arch/sparc/mm/gup.c

@@ -0,0 +1,267 @@
+/*
+ * Lockless get_user_pages_fast for sparc, cribbed from powerpc
+ *
+ * Copyright (C) 2008 Nick Piggin
+ * Copyright (C) 2008 Novell Inc.
+ */
+
+#include <linux/sched.h>
+#include <linux/mm.h>
+#include <linux/vmstat.h>
+#include <linux/pagemap.h>
+#include <linux/rwsem.h>
+#include <asm/pgtable.h>
+
+/*
+ * The performance critical leaf functions are made noinline otherwise gcc
+ * inlines everything into a single function which results in too much
+ * register pressure.
+ */
+static noinline int gup_pte_range(pmd_t pmd, unsigned long addr,
+		unsigned long end, int write, struct page **pages, int *nr)
+{
+	unsigned long mask, result;
+	pte_t *ptep;
+
+	if (tlb_type == hypervisor) {
+		result = _PAGE_PRESENT_4V|_PAGE_P_4V;
+		if (write)
+			result |= _PAGE_WRITE_4V;
+	} else {
+		result = _PAGE_PRESENT_4U|_PAGE_P_4U;
+		if (write)
+			result |= _PAGE_WRITE_4U;
+	}
+	mask = result | _PAGE_SPECIAL;
+
+	ptep = pte_offset_kernel(&pmd, addr);
+	do {
+		struct page *page, *head;
+		pte_t pte = *ptep;
+
+		if ((pte_val(pte) & mask) != result)
+			return 0;
+		VM_BUG_ON(!pfn_valid(pte_pfn(pte)));
+
+		/* The hugepage case is simplified on sparc64 because
+		 * we encode the sub-page pfn offsets into the
+		 * hugepage PTEs.  We could optimize this in the future
+		 * use page_cache_add_speculative() for the hugepage case.
+		 */
+		page = pte_page(pte);
+		head = compound_head(page);
+		if (!page_cache_get_speculative(head))
+			return 0;
+		if (unlikely(pte_val(pte) != pte_val(*ptep))) {
+			put_page(head);
+			return 0;
+		}
+		if (head != page)
+			get_huge_page_tail(page);
+
+		pages[*nr] = page;
+		(*nr)++;
+	} while (ptep++, addr += PAGE_SIZE, addr != end);
+
+	return 1;
+}
+
+static int gup_huge_pmd(pmd_t *pmdp, pmd_t pmd, unsigned long addr,
+			unsigned long end, int write, struct page **pages,
+			int *nr)
+{
+	struct page *head, *page, *tail;
+	int refs;
+
+	if (!(pmd_val(pmd) & _PAGE_VALID))
+		return 0;
+
+	if (write && !pmd_write(pmd))
+		return 0;
+
+	refs = 0;
+	head = pmd_page(pmd);
+	page = head + ((addr & ~PMD_MASK) >> PAGE_SHIFT);
+	tail = page;
+	do {
+		VM_BUG_ON(compound_head(page) != head);
+		pages[*nr] = page;
+		(*nr)++;
+		page++;
+		refs++;
+	} while (addr += PAGE_SIZE, addr != end);
+
+	if (!page_cache_add_speculative(head, refs)) {
+		*nr -= refs;
+		return 0;
+	}
+
+	if (unlikely(pmd_val(pmd) != pmd_val(*pmdp))) {
+		*nr -= refs;
+		while (refs--)
+			put_page(head);
+		return 0;
+	}
+
+	/* Any tail page need their mapcount reference taken before we
+	 * return.
+	 */
+	while (refs--) {
+		if (PageTail(tail))
+			get_huge_page_tail(tail);
+		tail++;
+	}
+
+	return 1;
+}
+
+static int gup_pmd_range(pud_t pud, unsigned long addr, unsigned long end,
+		int write, struct page **pages, int *nr)
+{
+	unsigned long next;
+	pmd_t *pmdp;
+
+	pmdp = pmd_offset(&pud, addr);
+	do {
+		pmd_t pmd = *pmdp;
+
+		next = pmd_addr_end(addr, end);
+		if (pmd_none(pmd) || pmd_trans_splitting(pmd))
+			return 0;
+		if (unlikely(pmd_large(pmd))) {
+			if (!gup_huge_pmd(pmdp, pmd, addr, next,
+					  write, pages, nr))
+				return 0;
+		} else if (!gup_pte_range(pmd, addr, next, write,
+					  pages, nr))
+			return 0;
+	} while (pmdp++, addr = next, addr != end);
+
+	return 1;
+}
+
+static int gup_pud_range(pgd_t pgd, unsigned long addr, unsigned long end,
+		int write, struct page **pages, int *nr)
+{
+	unsigned long next;
+	pud_t *pudp;
+
+	pudp = pud_offset(&pgd, addr);
+	do {
+		pud_t pud = *pudp;
+
+		next = pud_addr_end(addr, end);
+		if (pud_none(pud))
+			return 0;
+		if (!gup_pmd_range(pud, addr, next, write, pages, nr))
+			return 0;
+	} while (pudp++, addr = next, addr != end);
+
+	return 1;
+}
+
+int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
+			  struct page **pages)
+{
+	struct mm_struct *mm = current->mm;
+	unsigned long addr, len, end;
+	unsigned long next, flags;
+	pgd_t *pgdp;
+	int nr = 0;
+
+	start &= PAGE_MASK;
+	addr = start;
+	len = (unsigned long) nr_pages << PAGE_SHIFT;
+	end = start + len;
+
+	local_irq_save(flags);
+	pgdp = pgd_offset(mm, addr);
+	do {
+		pgd_t pgd = *pgdp;
+
+		next = pgd_addr_end(addr, end);
+		if (pgd_none(pgd))
+			break;
+		if (!gup_pud_range(pgd, addr, next, write, pages, &nr))
+			break;
+	} while (pgdp++, addr = next, addr != end);
+	local_irq_restore(flags);
+
+	return nr;
+}
+
+int get_user_pages_fast(unsigned long start, int nr_pages, int write,
+			struct page **pages)
+{
+	struct mm_struct *mm = current->mm;
+	unsigned long addr, len, end;
+	unsigned long next;
+	pgd_t *pgdp;
+	int nr = 0;
+
+	start &= PAGE_MASK;
+	addr = start;
+	len = (unsigned long) nr_pages << PAGE_SHIFT;
+	end = start + len;
+
+	/*
+	 * XXX: batch / limit 'nr', to avoid large irq off latency
+	 * needs some instrumenting to determine the common sizes used by
+	 * important workloads (eg. DB2), and whether limiting the batch size
+	 * will decrease performance.
+	 *
+	 * It seems like we're in the clear for the moment. Direct-IO is
+	 * the main guy that batches up lots of get_user_pages, and even
+	 * they are limited to 64-at-a-time which is not so many.
+	 */
+	/*
+	 * This doesn't prevent pagetable teardown, but does prevent
+	 * the pagetables from being freed on sparc.
+	 *
+	 * So long as we atomically load page table pointers versus teardown,
+	 * we can follow the address down to the the page and take a ref on it.
+	 */
+	local_irq_disable();
+
+	pgdp = pgd_offset(mm, addr);
+	do {
+		pgd_t pgd = *pgdp;
+
+		next = pgd_addr_end(addr, end);
+		if (pgd_none(pgd))
+			goto slow;
+		if (!gup_pud_range(pgd, addr, next, write, pages, &nr))
+			goto slow;
+	} while (pgdp++, addr = next, addr != end);
+
+	local_irq_enable();
+
+	VM_BUG_ON(nr != (end - start) >> PAGE_SHIFT);
+	return nr;
+
+	{
+		int ret;
+
+slow:
+		local_irq_enable();
+
+		/* Try to get the remaining pages with get_user_pages */
+		start += nr << PAGE_SHIFT;
+		pages += nr;
+
+		down_read(&mm->mmap_sem);
+		ret = get_user_pages(current, mm, start,
+			(end - start) >> PAGE_SHIFT, write, 0, pages, NULL);
+		up_read(&mm->mmap_sem);
+
+		/* Have to be a bit careful with return values */
+		if (nr > 0) {
+			if (ret < 0)
+				ret = nr;
+			else
+				ret += nr;
+		}
+
+		return ret;
+	}
+}

arch/sparc/mm/highmem.c

@@ -0,0 +1,130 @@
+/*
+ *  highmem.c: virtual kernel memory mappings for high memory
+ *
+ *  Provides kernel-static versions of atomic kmap functions originally
+ *  found as inlines in include/asm-sparc/highmem.h.  These became
+ *  needed as kmap_atomic() and kunmap_atomic() started getting
+ *  called from within modules.
+ *  -- Tomas Szepe <szepe@pinerecords.com>, September 2002
+ *
+ *  But kmap_atomic() and kunmap_atomic() cannot be inlined in
+ *  modules because they are loaded with btfixup-ped functions.
+ */
+
+/*
+ * The use of kmap_atomic/kunmap_atomic is discouraged - kmap/kunmap
+ * gives a more generic (and caching) interface. But kmap_atomic can
+ * be used in IRQ contexts, so in some (very limited) cases we need it.
+ *
+ * XXX This is an old text. Actually, it's good to use atomic kmaps,
+ * provided you remember that they are atomic and not try to sleep
+ * with a kmap taken, much like a spinlock. Non-atomic kmaps are
+ * shared by CPUs, and so precious, and establishing them requires IPI.
+ * Atomic kmaps are lightweight and we may have NCPUS more of them.
+ */
+#include <linux/highmem.h>
+#include <linux/export.h>
+#include <linux/mm.h>
+
+#include <asm/cacheflush.h>
+#include <asm/tlbflush.h>
+#include <asm/pgalloc.h>
+#include <asm/vaddrs.h>
+
+pgprot_t kmap_prot;
+
+static pte_t *kmap_pte;
+
+void __init kmap_init(void)
+{
+	unsigned long address;
+	pmd_t *dir;
+
+	address = __fix_to_virt(FIX_KMAP_BEGIN);
+	dir = pmd_offset(pgd_offset_k(address), address);
+
+        /* cache the first kmap pte */
+        kmap_pte = pte_offset_kernel(dir, address);
+        kmap_prot = __pgprot(SRMMU_ET_PTE | SRMMU_PRIV | SRMMU_CACHE);
+}
+
+void *kmap_atomic(struct page *page)
+{
+	unsigned long vaddr;
+	long idx, type;
+
+	/* even !CONFIG_PREEMPT needs this, for in_atomic in do_page_fault */
+	pagefault_disable();
+	if (!PageHighMem(page))
+		return page_address(page);
+
+	type = kmap_atomic_idx_push();
+	idx = type + KM_TYPE_NR*smp_processor_id();
+	vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
+
+/* XXX Fix - Anton */
+#if 0
+	__flush_cache_one(vaddr);
+#else
+	flush_cache_all();
+#endif
+
+#ifdef CONFIG_DEBUG_HIGHMEM
+	BUG_ON(!pte_none(*(kmap_pte-idx)));
+#endif
+	set_pte(kmap_pte-idx, mk_pte(page, kmap_prot));
+/* XXX Fix - Anton */
+#if 0
+	__flush_tlb_one(vaddr);
+#else
+	flush_tlb_all();
+#endif
+
+	return (void*) vaddr;
+}
+EXPORT_SYMBOL(kmap_atomic);
+
+void __kunmap_atomic(void *kvaddr)
+{
+	unsigned long vaddr = (unsigned long) kvaddr & PAGE_MASK;
+	int type;
+
+	if (vaddr < FIXADDR_START) { // FIXME
+		pagefault_enable();
+		return;
+	}
+
+	type = kmap_atomic_idx();
+
+#ifdef CONFIG_DEBUG_HIGHMEM
+	{
+		unsigned long idx;
+
+		idx = type + KM_TYPE_NR * smp_processor_id();
+		BUG_ON(vaddr != __fix_to_virt(FIX_KMAP_BEGIN+idx));
+
+		/* XXX Fix - Anton */
+#if 0
+		__flush_cache_one(vaddr);
+#else
+		flush_cache_all();
+#endif
+
+		/*
+		 * force other mappings to Oops if they'll try to access
+		 * this pte without first remap it
+		 */
+		pte_clear(&init_mm, vaddr, kmap_pte-idx);
+		/* XXX Fix - Anton */
+#if 0
+		__flush_tlb_one(vaddr);
+#else
+		flush_tlb_all();
+#endif
+	}
+#endif
+
+	kmap_atomic_idx_pop();
+	pagefault_enable();
+}
+EXPORT_SYMBOL(__kunmap_atomic);

arch/sparc/mm/hugetlbpage.c

@@ -0,0 +1,226 @@
+/*
+ * SPARC64 Huge TLB page support.
+ *
+ * Copyright (C) 2002, 2003, 2006 David S. Miller (davem@davemloft.net)
+ */
+
+#include <linux/fs.h>
+#include <linux/mm.h>
+#include <linux/hugetlb.h>
+#include <linux/pagemap.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>
+
+/* Slightly simplified from the non-hugepage variant because by
+ * definition we don't have to worry about any page coloring stuff
+ */
+
+static unsigned long hugetlb_get_unmapped_area_bottomup(struct file *filp,
+							unsigned long addr,
+							unsigned long len,
+							unsigned long pgoff,
+							unsigned long flags)
+{
+	unsigned long task_size = TASK_SIZE;
+	struct vm_unmapped_area_info info;
+
+	if (test_thread_flag(TIF_32BIT))
+		task_size = STACK_TOP32;
+
+	info.flags = 0;
+	info.length = len;
+	info.low_limit = TASK_UNMAPPED_BASE;
+	info.high_limit = min(task_size, VA_EXCLUDE_START);
+	info.align_mask = PAGE_MASK & ~HPAGE_MASK;
+	info.align_offset = 0;
+	addr = vm_unmapped_area(&info);
+
+	if ((addr & ~PAGE_MASK) && task_size > VA_EXCLUDE_END) {
+		VM_BUG_ON(addr != -ENOMEM);
+		info.low_limit = VA_EXCLUDE_END;
+		info.high_limit = task_size;
+		addr = vm_unmapped_area(&info);
+	}
+
+	return addr;
+}
+
+static unsigned long
+hugetlb_get_unmapped_area_topdown(struct file *filp, const unsigned long addr0,
+				  const unsigned long len,
+				  const unsigned long pgoff,
+				  const unsigned long flags)
+{
+	struct mm_struct *mm = current->mm;
+	unsigned long addr = addr0;
+	struct vm_unmapped_area_info info;
+
+	/* This should only ever run for 32-bit processes.  */
+	BUG_ON(!test_thread_flag(TIF_32BIT));
+
+	info.flags = VM_UNMAPPED_AREA_TOPDOWN;
+	info.length = len;
+	info.low_limit = PAGE_SIZE;
+	info.high_limit = mm->mmap_base;
+	info.align_mask = PAGE_MASK & ~HPAGE_MASK;
+	info.align_offset = 0;
+	addr = vm_unmapped_area(&info);
+
+	/*
+	 * A failed mmap() very likely causes application failure,
+	 * so fall back to the bottom-up function here. This scenario
+	 * can happen with large stack limits and large mmap()
+	 * allocations.
+	 */
+	if (addr & ~PAGE_MASK) {
+		VM_BUG_ON(addr != -ENOMEM);
+		info.flags = 0;
+		info.low_limit = TASK_UNMAPPED_BASE;
+		info.high_limit = STACK_TOP32;
+		addr = vm_unmapped_area(&info);
+	}
+
+	return addr;
+}
+
+unsigned long
+hugetlb_get_unmapped_area(struct file *file, unsigned long addr,
+		unsigned long len, unsigned long pgoff, unsigned long flags)
+{
+	struct mm_struct *mm = current->mm;
+	struct vm_area_struct *vma;
+	unsigned long task_size = TASK_SIZE;
+
+	if (test_thread_flag(TIF_32BIT))
+		task_size = STACK_TOP32;
+
+	if (len & ~HPAGE_MASK)
+		return -EINVAL;
+	if (len > task_size)
+		return -ENOMEM;
+
+	if (flags & MAP_FIXED) {
+		if (prepare_hugepage_range(file, addr, len))
+			return -EINVAL;
+		return addr;
+	}
+
+	if (addr) {
+		addr = ALIGN(addr, HPAGE_SIZE);
+		vma = find_vma(mm, addr);
+		if (task_size - len >= addr &&
+		    (!vma || addr + len <= vma->vm_start))
+			return addr;
+	}
+	if (mm->get_unmapped_area == arch_get_unmapped_area)
+		return hugetlb_get_unmapped_area_bottomup(file, addr, len,
+				pgoff, flags);
+	else
+		return hugetlb_get_unmapped_area_topdown(file, addr, len,
+				pgoff, flags);
+}
+
+pte_t *huge_pte_alloc(struct mm_struct *mm,
+			unsigned long addr, unsigned long sz)
+{
+	pgd_t *pgd;
+	pud_t *pud;
+	pmd_t *pmd;
+	pte_t *pte = NULL;
+
+	/* We must align the address, because our caller will run
+	 * set_huge_pte_at() on whatever we return, which writes out
+	 * all of the sub-ptes for the hugepage range.  So we have
+	 * to give it the first such sub-pte.
+	 */
+	addr &= HPAGE_MASK;
+
+	pgd = pgd_offset(mm, addr);
+	pud = pud_alloc(mm, pgd, addr);
+	if (pud) {
+		pmd = pmd_alloc(mm, pud, addr);
+		if (pmd)
+			pte = pte_alloc_map(mm, NULL, pmd, addr);
+	}
+	return pte;
+}
+
+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;
+
+	addr &= HPAGE_MASK;
+
+	pgd = pgd_offset(mm, addr);
+	if (!pgd_none(*pgd)) {
+		pud = pud_offset(pgd, addr);
+		if (!pud_none(*pud)) {
+			pmd = pmd_offset(pud, addr);
+			if (!pmd_none(*pmd))
+				pte = pte_offset_map(pmd, addr);
+		}
+	}
+	return pte;
+}
+
+int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep)
+{
+	return 0;
+}
+
+void set_huge_pte_at(struct mm_struct *mm, unsigned long addr,
+		     pte_t *ptep, pte_t entry)
+{
+	int i;
+
+	if (!pte_present(*ptep) && pte_present(entry))
+		mm->context.huge_pte_count++;
+
+	addr &= HPAGE_MASK;
+	for (i = 0; i < (1 << HUGETLB_PAGE_ORDER); i++) {
+		set_pte_at(mm, addr, ptep, entry);
+		ptep++;
+		addr += PAGE_SIZE;
+		pte_val(entry) += PAGE_SIZE;
+	}
+}
+
+pte_t huge_ptep_get_and_clear(struct mm_struct *mm, unsigned long addr,
+			      pte_t *ptep)
+{
+	pte_t entry;
+	int i;
+
+	entry = *ptep;
+	if (pte_present(entry))
+		mm->context.huge_pte_count--;
+
+	addr &= HPAGE_MASK;
+
+	for (i = 0; i < (1 << HUGETLB_PAGE_ORDER); i++) {
+		pte_clear(mm, addr, ptep);
+		addr += PAGE_SIZE;
+		ptep++;
+	}
+
+	return entry;
+}
+
+int pmd_huge(pmd_t pmd)
+{
+	return 0;
+}
+
+int pud_huge(pud_t pud)
+{
+	return 0;
+}

arch/sparc/mm/hypersparc.S

@@ -0,0 +1,412 @@
+/*
+ * hypersparc.S: High speed Hypersparc mmu/cache operations.
+ *
+ * Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu)
+ */
+
+#include <asm/ptrace.h>
+#include <asm/psr.h>
+#include <asm/asm-offsets.h>
+#include <asm/asi.h>
+#include <asm/page.h>
+#include <asm/pgtsrmmu.h>
+#include <linux/init.h>
+
+	.text
+	.align	4
+
+	.globl	hypersparc_flush_cache_all, hypersparc_flush_cache_mm
+	.globl	hypersparc_flush_cache_range, hypersparc_flush_cache_page
+	.globl	hypersparc_flush_page_to_ram
+	.globl	hypersparc_flush_page_for_dma, hypersparc_flush_sig_insns
+	.globl	hypersparc_flush_tlb_all, hypersparc_flush_tlb_mm
+	.globl	hypersparc_flush_tlb_range, hypersparc_flush_tlb_page
+
+hypersparc_flush_cache_all:
+	WINDOW_FLUSH(%g4, %g5)
+	sethi	%hi(vac_cache_size), %g4
+	ld	[%g4 + %lo(vac_cache_size)], %g5
+	sethi	%hi(vac_line_size), %g1
+	ld	[%g1 + %lo(vac_line_size)], %g2
+1:	
+	subcc	%g5, %g2, %g5			! hyper_flush_unconditional_combined
+	bne	1b
+	 sta	%g0, [%g5] ASI_M_FLUSH_CTX
+	retl
+	 sta	%g0, [%g0] ASI_M_FLUSH_IWHOLE	! hyper_flush_whole_icache
+
+	/* We expand the window flush to get maximum performance. */
+hypersparc_flush_cache_mm:
+#ifndef CONFIG_SMP
+	ld	[%o0 + AOFF_mm_context], %g1
+	cmp	%g1, -1
+	be	hypersparc_flush_cache_mm_out
+#endif
+	WINDOW_FLUSH(%g4, %g5)
+
+	sethi	%hi(vac_line_size), %g1
+	ld	[%g1 + %lo(vac_line_size)], %o1
+	sethi	%hi(vac_cache_size), %g2
+	ld	[%g2 + %lo(vac_cache_size)], %o0
+	add	%o1, %o1, %g1
+	add	%o1, %g1, %g2
+	add	%o1, %g2, %g3
+	add	%o1, %g3, %g4
+	add	%o1, %g4, %g5
+	add	%o1, %g5, %o4
+	add	%o1, %o4, %o5
+
+	/* BLAMMO! */
+1:
+	subcc	%o0, %o5, %o0				! hyper_flush_cache_user
+	sta	%g0, [%o0 + %g0] ASI_M_FLUSH_USER
+	sta	%g0, [%o0 + %o1] ASI_M_FLUSH_USER
+	sta	%g0, [%o0 + %g1] ASI_M_FLUSH_USER
+	sta	%g0, [%o0 + %g2] ASI_M_FLUSH_USER
+	sta	%g0, [%o0 + %g3] ASI_M_FLUSH_USER
+	sta	%g0, [%o0 + %g4] ASI_M_FLUSH_USER
+	sta	%g0, [%o0 + %g5] ASI_M_FLUSH_USER
+	bne	1b
+	 sta	%g0, [%o0 + %o4] ASI_M_FLUSH_USER
+hypersparc_flush_cache_mm_out:
+	retl
+	 nop
+
+	/* The things we do for performance... */
+hypersparc_flush_cache_range:
+	ld	[%o0 + VMA_VM_MM], %o0
+#ifndef CONFIG_SMP
+	ld	[%o0 + AOFF_mm_context], %g1
+	cmp	%g1, -1
+	be	hypersparc_flush_cache_range_out
+#endif
+	WINDOW_FLUSH(%g4, %g5)
+
+	sethi	%hi(vac_line_size), %g1
+	ld	[%g1 + %lo(vac_line_size)], %o4
+	sethi	%hi(vac_cache_size), %g2
+	ld	[%g2 + %lo(vac_cache_size)], %o3
+
+	/* Here comes the fun part... */
+	add	%o2, (PAGE_SIZE - 1), %o2
+	andn	%o1, (PAGE_SIZE - 1), %o1
+	add	%o4, %o4, %o5
+	andn	%o2, (PAGE_SIZE - 1), %o2
+	add	%o4, %o5, %g1
+	sub	%o2, %o1, %g4
+	add	%o4, %g1, %g2
+	sll	%o3, 2, %g5
+	add	%o4, %g2, %g3
+	cmp	%g4, %g5
+	add	%o4, %g3, %g4
+	blu	0f
+	 add	%o4, %g4, %g5
+	add	%o4, %g5, %g7
+
+	/* Flush entire user space, believe it or not this is quicker
+	 * than page at a time flushings for range > (cache_size<<2).
+	 */
+1:
+	subcc	%o3, %g7, %o3
+	sta	%g0, [%o3 + %g0] ASI_M_FLUSH_USER
+	sta	%g0, [%o3 + %o4] ASI_M_FLUSH_USER
+	sta	%g0, [%o3 + %o5] ASI_M_FLUSH_USER
+	sta	%g0, [%o3 + %g1] ASI_M_FLUSH_USER
+	sta	%g0, [%o3 + %g2] ASI_M_FLUSH_USER
+	sta	%g0, [%o3 + %g3] ASI_M_FLUSH_USER
+	sta	%g0, [%o3 + %g4] ASI_M_FLUSH_USER
+	bne	1b
+	 sta	%g0, [%o3 + %g5] ASI_M_FLUSH_USER
+	retl
+	 nop
+
+	/* Below our threshold, flush one page at a time. */
+0:
+	ld	[%o0 + AOFF_mm_context], %o0
+	mov	SRMMU_CTX_REG, %g7
+	lda	[%g7] ASI_M_MMUREGS, %o3
+	sta	%o0, [%g7] ASI_M_MMUREGS
+	add	%o2, -PAGE_SIZE, %o0
+1:
+	or	%o0, 0x400, %g7
+	lda	[%g7] ASI_M_FLUSH_PROBE, %g7
+	orcc	%g7, 0, %g0
+	be,a	3f
+	 mov	%o0, %o2
+	add	%o4, %g5, %g7
+2:
+	sub	%o2, %g7, %o2
+	sta	%g0, [%o2 + %g0] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o2 + %o4] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o2 + %o5] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o2 + %g1] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o2 + %g2] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o2 + %g3] ASI_M_FLUSH_PAGE
+	andcc	%o2, 0xffc, %g0
+	sta	%g0, [%o2 + %g4] ASI_M_FLUSH_PAGE
+	bne	2b
+	 sta	%g0, [%o2 + %g5] ASI_M_FLUSH_PAGE
+3:
+	cmp	%o2, %o1
+	bne	1b
+	 add	%o2, -PAGE_SIZE, %o0
+	mov	SRMMU_FAULT_STATUS, %g5
+	lda	[%g5] ASI_M_MMUREGS, %g0
+	mov	SRMMU_CTX_REG, %g7
+	sta	%o3, [%g7] ASI_M_MMUREGS
+hypersparc_flush_cache_range_out:
+	retl
+	 nop
+
+	/* HyperSparc requires a valid mapping where we are about to flush
+	 * in order to check for a physical tag match during the flush.
+	 */
+	/* Verified, my ass... */
+hypersparc_flush_cache_page:
+	ld	[%o0 + VMA_VM_MM], %o0
+	ld	[%o0 + AOFF_mm_context], %g2
+#ifndef CONFIG_SMP
+	cmp	%g2, -1
+	be	hypersparc_flush_cache_page_out
+#endif
+	WINDOW_FLUSH(%g4, %g5)
+
+	sethi	%hi(vac_line_size), %g1
+	ld	[%g1 + %lo(vac_line_size)], %o4
+	mov	SRMMU_CTX_REG, %o3
+	andn	%o1, (PAGE_SIZE - 1), %o1
+	lda	[%o3] ASI_M_MMUREGS, %o2
+	sta	%g2, [%o3] ASI_M_MMUREGS
+	or	%o1, 0x400, %o5
+	lda	[%o5] ASI_M_FLUSH_PROBE, %g1
+	orcc	%g0, %g1, %g0
+	be	2f
+	 add	%o4, %o4, %o5
+	sub	%o1, -PAGE_SIZE, %o1
+	add	%o4, %o5, %g1
+	add	%o4, %g1, %g2
+	add	%o4, %g2, %g3
+	add	%o4, %g3, %g4
+	add	%o4, %g4, %g5
+	add	%o4, %g5, %g7
+
+	/* BLAMMO! */
+1:
+	sub	%o1, %g7, %o1
+	sta	%g0, [%o1 + %g0] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %o4] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %o5] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %g1] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %g2] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %g3] ASI_M_FLUSH_PAGE
+	andcc	%o1, 0xffc, %g0
+	sta	%g0, [%o1 + %g4] ASI_M_FLUSH_PAGE
+	bne	1b
+	 sta	%g0, [%o1 + %g5] ASI_M_FLUSH_PAGE
+2:
+	mov	SRMMU_FAULT_STATUS, %g7
+	mov	SRMMU_CTX_REG, %g4
+	lda	[%g7] ASI_M_MMUREGS, %g0
+	sta	%o2, [%g4] ASI_M_MMUREGS
+hypersparc_flush_cache_page_out:
+	retl
+	 nop
+
+hypersparc_flush_sig_insns:
+	flush	%o1
+	retl
+	 flush	%o1 + 4
+
+	/* HyperSparc is copy-back. */
+hypersparc_flush_page_to_ram:
+	sethi	%hi(vac_line_size), %g1
+	ld	[%g1 + %lo(vac_line_size)], %o4
+	andn	%o0, (PAGE_SIZE - 1), %o0
+	add	%o4, %o4, %o5
+	or	%o0, 0x400, %g7
+	lda	[%g7] ASI_M_FLUSH_PROBE, %g5
+	add	%o4, %o5, %g1
+	orcc	%g5, 0, %g0
+	be	2f
+	 add	%o4, %g1, %g2
+	add	%o4, %g2, %g3
+	sub	%o0, -PAGE_SIZE, %o0
+	add	%o4, %g3, %g4
+	add	%o4, %g4, %g5
+	add	%o4, %g5, %g7
+
+	/* BLAMMO! */
+1:
+	sub	%o0, %g7, %o0
+	sta	%g0, [%o0 + %g0] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o0 + %o4] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o0 + %o5] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o0 + %g1] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o0 + %g2] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o0 + %g3] ASI_M_FLUSH_PAGE
+	andcc	%o0, 0xffc, %g0
+	sta	%g0, [%o0 + %g4] ASI_M_FLUSH_PAGE
+	bne	1b
+	 sta	%g0, [%o0 + %g5] ASI_M_FLUSH_PAGE
+2:
+	mov	SRMMU_FAULT_STATUS, %g1
+	retl
+	 lda	[%g1] ASI_M_MMUREGS, %g0
+
+	/* HyperSparc is IO cache coherent. */
+hypersparc_flush_page_for_dma:
+	retl
+	 nop
+
+	/* It was noted that at boot time a TLB flush all in a delay slot
+	 * can deliver an illegal instruction to the processor if the timing
+	 * is just right...
+	 */
+hypersparc_flush_tlb_all:
+	mov	0x400, %g1
+	sta	%g0, [%g1] ASI_M_FLUSH_PROBE
+	retl
+	 nop
+
+hypersparc_flush_tlb_mm:
+	mov	SRMMU_CTX_REG, %g1
+	ld	[%o0 + AOFF_mm_context], %o1
+	lda	[%g1] ASI_M_MMUREGS, %g5
+#ifndef CONFIG_SMP
+	cmp	%o1, -1
+	be	hypersparc_flush_tlb_mm_out
+#endif
+	 mov	0x300, %g2
+	sta	%o1, [%g1] ASI_M_MMUREGS
+	sta	%g0, [%g2] ASI_M_FLUSH_PROBE
+hypersparc_flush_tlb_mm_out:
+	retl
+	 sta	%g5, [%g1] ASI_M_MMUREGS
+
+hypersparc_flush_tlb_range:
+	ld	[%o0 + VMA_VM_MM], %o0
+	mov	SRMMU_CTX_REG, %g1
+	ld	[%o0 + AOFF_mm_context], %o3
+	lda	[%g1] ASI_M_MMUREGS, %g5
+#ifndef CONFIG_SMP
+	cmp	%o3, -1
+	be	hypersparc_flush_tlb_range_out
+#endif
+	 sethi	%hi(~((1 << SRMMU_PGDIR_SHIFT) - 1)), %o4
+	sta	%o3, [%g1] ASI_M_MMUREGS
+	and	%o1, %o4, %o1
+	add	%o1, 0x200, %o1
+	sta	%g0, [%o1] ASI_M_FLUSH_PROBE
+1:
+	sub	%o1, %o4, %o1
+	cmp	%o1, %o2
+	blu,a	1b
+	 sta	%g0, [%o1] ASI_M_FLUSH_PROBE
+hypersparc_flush_tlb_range_out:
+	retl
+	 sta	%g5, [%g1] ASI_M_MMUREGS
+
+hypersparc_flush_tlb_page:
+	ld	[%o0 + VMA_VM_MM], %o0
+	mov	SRMMU_CTX_REG, %g1
+	ld	[%o0 + AOFF_mm_context], %o3
+	andn	%o1, (PAGE_SIZE - 1), %o1
+#ifndef CONFIG_SMP
+	cmp	%o3, -1
+	be	hypersparc_flush_tlb_page_out
+#endif
+	 lda	[%g1] ASI_M_MMUREGS, %g5
+	sta	%o3, [%g1] ASI_M_MMUREGS
+	sta	%g0, [%o1] ASI_M_FLUSH_PROBE
+hypersparc_flush_tlb_page_out:
+	retl
+	 sta	%g5, [%g1] ASI_M_MMUREGS
+
+	__INIT
+	
+	/* High speed page clear/copy. */
+hypersparc_bzero_1page:
+/* NOTE: This routine has to be shorter than 40insns --jj */
+	clr	%g1
+	mov	32, %g2
+	mov	64, %g3
+	mov	96, %g4
+	mov	128, %g5
+	mov	160, %g7
+	mov	192, %o2
+	mov	224, %o3
+	mov	16, %o1
+1:
+	stda	%g0, [%o0 + %g0] ASI_M_BFILL
+	stda	%g0, [%o0 + %g2] ASI_M_BFILL
+	stda	%g0, [%o0 + %g3] ASI_M_BFILL
+	stda	%g0, [%o0 + %g4] ASI_M_BFILL
+	stda	%g0, [%o0 + %g5] ASI_M_BFILL
+	stda	%g0, [%o0 + %g7] ASI_M_BFILL
+	stda	%g0, [%o0 + %o2] ASI_M_BFILL
+	stda	%g0, [%o0 + %o3] ASI_M_BFILL
+	subcc	%o1, 1, %o1
+	bne	1b
+	 add	%o0, 256, %o0
+
+	retl
+	 nop
+
+hypersparc_copy_1page:
+/* NOTE: This routine has to be shorter than 70insns --jj */
+	sub	%o1, %o0, %o2		! difference
+	mov	16, %g1
+1:
+	sta	%o0, [%o0 + %o2] ASI_M_BCOPY
+	add	%o0, 32, %o0
+	sta	%o0, [%o0 + %o2] ASI_M_BCOPY
+	add	%o0, 32, %o0
+	sta	%o0, [%o0 + %o2] ASI_M_BCOPY
+	add	%o0, 32, %o0
+	sta	%o0, [%o0 + %o2] ASI_M_BCOPY
+	add	%o0, 32, %o0
+	sta	%o0, [%o0 + %o2] ASI_M_BCOPY
+	add	%o0, 32, %o0
+	sta	%o0, [%o0 + %o2] ASI_M_BCOPY
+	add	%o0, 32, %o0
+	sta	%o0, [%o0 + %o2] ASI_M_BCOPY
+	add	%o0, 32, %o0
+	sta	%o0, [%o0 + %o2] ASI_M_BCOPY
+	subcc	%g1, 1, %g1
+	bne	1b
+	 add	%o0, 32, %o0
+
+	retl
+	 nop
+
+	.globl	hypersparc_setup_blockops
+hypersparc_setup_blockops:
+	sethi	%hi(bzero_1page), %o0
+	or	%o0, %lo(bzero_1page), %o0
+	sethi	%hi(hypersparc_bzero_1page), %o1
+	or	%o1, %lo(hypersparc_bzero_1page), %o1
+	sethi	%hi(hypersparc_copy_1page), %o2
+	or	%o2, %lo(hypersparc_copy_1page), %o2
+	ld	[%o1], %o4
+1:
+	add	%o1, 4, %o1
+	st	%o4, [%o0]
+	add	%o0, 4, %o0
+	cmp	%o1, %o2
+	bne	1b
+	 ld	[%o1], %o4
+	sethi	%hi(__copy_1page), %o0
+	or	%o0, %lo(__copy_1page), %o0
+	sethi	%hi(hypersparc_setup_blockops), %o2
+	or	%o2, %lo(hypersparc_setup_blockops), %o2
+	ld	[%o1], %o4
+1:
+	add	%o1, 4, %o1
+	st	%o4, [%o0]
+	add	%o0, 4, %o0
+	cmp	%o1, %o2
+	bne	1b
+	 ld	[%o1], %o4
+	sta	%g0, [%g0] ASI_M_FLUSH_IWHOLE
+	retl
+	 nop

arch/sparc/mm/init_32.c

@@ -0,0 +1,359 @@
+/*
+ *  linux/arch/sparc/mm/init.c
+ *
+ *  Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
+ *  Copyright (C) 1995 Eddie C. Dost (ecd@skynet.be)
+ *  Copyright (C) 1998 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ *  Copyright (C) 2000 Anton Blanchard (anton@samba.org)
+ */
+
+#include <linux/module.h>
+#include <linux/signal.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/string.h>
+#include <linux/types.h>
+#include <linux/ptrace.h>
+#include <linux/mman.h>
+#include <linux/mm.h>
+#include <linux/swap.h>
+#include <linux/initrd.h>
+#include <linux/init.h>
+#include <linux/highmem.h>
+#include <linux/bootmem.h>
+#include <linux/pagemap.h>
+#include <linux/poison.h>
+#include <linux/gfp.h>
+
+#include <asm/sections.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/vaddrs.h>
+#include <asm/pgalloc.h>	/* bug in asm-generic/tlb.h: check_pgt_cache */
+#include <asm/setup.h>
+#include <asm/tlb.h>
+#include <asm/prom.h>
+#include <asm/leon.h>
+
+#include "mm_32.h"
+
+unsigned long *sparc_valid_addr_bitmap;
+EXPORT_SYMBOL(sparc_valid_addr_bitmap);
+
+unsigned long phys_base;
+EXPORT_SYMBOL(phys_base);
+
+unsigned long pfn_base;
+EXPORT_SYMBOL(pfn_base);
+
+struct sparc_phys_banks sp_banks[SPARC_PHYS_BANKS+1];
+
+/* Initial ramdisk setup */
+extern unsigned int sparc_ramdisk_image;
+extern unsigned int sparc_ramdisk_size;
+
+unsigned long highstart_pfn, highend_pfn;
+
+void show_mem(unsigned int filter)
+{
+	printk("Mem-info:\n");
+	show_free_areas(filter);
+	printk("Free swap:       %6ldkB\n",
+	       get_nr_swap_pages() << (PAGE_SHIFT-10));
+	printk("%ld pages of RAM\n", totalram_pages);
+	printk("%ld free pages\n", nr_free_pages());
+}
+
+
+unsigned long last_valid_pfn;
+
+unsigned long calc_highpages(void)
+{
+	int i;
+	int nr = 0;
+
+	for (i = 0; sp_banks[i].num_bytes != 0; i++) {
+		unsigned long start_pfn = sp_banks[i].base_addr >> PAGE_SHIFT;
+		unsigned long end_pfn = (sp_banks[i].base_addr + sp_banks[i].num_bytes) >> PAGE_SHIFT;
+
+		if (end_pfn <= max_low_pfn)
+			continue;
+
+		if (start_pfn < max_low_pfn)
+			start_pfn = max_low_pfn;
+
+		nr += end_pfn - start_pfn;
+	}
+
+	return nr;
+}
+
+static unsigned long calc_max_low_pfn(void)
+{
+	int i;
+	unsigned long tmp = pfn_base + (SRMMU_MAXMEM >> PAGE_SHIFT);
+	unsigned long curr_pfn, last_pfn;
+
+	last_pfn = (sp_banks[0].base_addr + sp_banks[0].num_bytes) >> PAGE_SHIFT;
+	for (i = 1; sp_banks[i].num_bytes != 0; i++) {
+		curr_pfn = sp_banks[i].base_addr >> PAGE_SHIFT;
+
+		if (curr_pfn >= tmp) {
+			if (last_pfn < tmp)
+				tmp = last_pfn;
+			break;
+		}
+
+		last_pfn = (sp_banks[i].base_addr + sp_banks[i].num_bytes) >> PAGE_SHIFT;
+	}
+
+	return tmp;
+}
+
+unsigned long __init bootmem_init(unsigned long *pages_avail)
+{
+	unsigned long bootmap_size, start_pfn;
+	unsigned long end_of_phys_memory = 0UL;
+	unsigned long bootmap_pfn, bytes_avail, size;
+	int i;
+
+	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;
+			}
+		}
+	}
+
+	/* Start with page aligned address of last symbol in kernel
+	 * image.  
+	 */
+	start_pfn  = (unsigned long)__pa(PAGE_ALIGN((unsigned long) &_end));
+
+	/* Now shift down to get the real physical page frame number. */
+	start_pfn >>= PAGE_SHIFT;
+
+	bootmap_pfn = start_pfn;
+
+	max_pfn = end_of_phys_memory >> PAGE_SHIFT;
+
+	max_low_pfn = max_pfn;
+	highstart_pfn = highend_pfn = max_pfn;
+
+	if (max_low_pfn > pfn_base + (SRMMU_MAXMEM >> PAGE_SHIFT)) {
+		highstart_pfn = pfn_base + (SRMMU_MAXMEM >> PAGE_SHIFT);
+		max_low_pfn = calc_max_low_pfn();
+		printk(KERN_NOTICE "%ldMB HIGHMEM available.\n",
+		    calc_highpages() >> (20 - PAGE_SHIFT));
+	}
+
+#ifdef CONFIG_BLK_DEV_INITRD
+	/* Now have to check initial ramdisk, so that bootmap does not overwrite it */
+	if (sparc_ramdisk_image) {
+		if (sparc_ramdisk_image >= (unsigned long)&_end - 2 * PAGE_SIZE)
+			sparc_ramdisk_image -= KERNBASE;
+		initrd_start = sparc_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. */
+	bootmap_size = init_bootmem_node(NODE_DATA(0), bootmap_pfn, pfn_base,
+					 max_low_pfn);
+
+	/* Now register the available physical memory with the
+	 * allocator.
+	 */
+	*pages_avail = 0;
+	for (i = 0; sp_banks[i].num_bytes != 0; i++) {
+		unsigned long curr_pfn, last_pfn;
+
+		curr_pfn = sp_banks[i].base_addr >> PAGE_SHIFT;
+		if (curr_pfn >= max_low_pfn)
+			break;
+
+		last_pfn = (sp_banks[i].base_addr + sp_banks[i].num_bytes) >> PAGE_SHIFT;
+		if (last_pfn > max_low_pfn)
+			last_pfn = max_low_pfn;
+
+		/*
+		 * .. finally, did all the rounding and playing
+		 * around just make the area go away?
+		 */
+		if (last_pfn <= curr_pfn)
+			continue;
+
+		size = (last_pfn - curr_pfn) << PAGE_SHIFT;
+		*pages_avail += last_pfn - curr_pfn;
+
+		free_bootmem(sp_banks[i].base_addr, size);
+	}
+
+#ifdef CONFIG_BLK_DEV_INITRD
+	if (initrd_start) {
+		/* Reserve the initrd image area. */
+		size = initrd_end - initrd_start;
+		reserve_bootmem(initrd_start, size, BOOTMEM_DEFAULT);
+		*pages_avail -= PAGE_ALIGN(size) >> PAGE_SHIFT;
+
+		initrd_start = (initrd_start - phys_base) + PAGE_OFFSET;
+		initrd_end = (initrd_end - phys_base) + PAGE_OFFSET;		
+	}
+#endif
+	/* Reserve the kernel text/data/bss. */
+	size = (start_pfn << PAGE_SHIFT) - phys_base;
+	reserve_bootmem(phys_base, size, BOOTMEM_DEFAULT);
+	*pages_avail -= PAGE_ALIGN(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;
+	reserve_bootmem((bootmap_pfn << PAGE_SHIFT), size, BOOTMEM_DEFAULT);
+	*pages_avail -= PAGE_ALIGN(size) >> PAGE_SHIFT;
+
+	return max_pfn;
+}
+
+/*
+ * paging_init() sets up the page tables: We call the MMU specific
+ * init routine based upon the Sun model type on the Sparc.
+ *
+ */
+void __init paging_init(void)
+{
+	srmmu_paging_init();
+	prom_build_devicetree();
+	of_fill_in_cpu_data();
+	device_scan();
+}
+
+static void __init taint_real_pages(void)
+{
+	int i;
+
+	for (i = 0; sp_banks[i].num_bytes; i++) {
+		unsigned long start, end;
+
+		start = sp_banks[i].base_addr;
+		end = start + sp_banks[i].num_bytes;
+
+		while (start < end) {
+			set_bit(start >> 20, sparc_valid_addr_bitmap);
+			start += PAGE_SIZE;
+		}
+	}
+}
+
+static void map_high_region(unsigned long start_pfn, unsigned long end_pfn)
+{
+	unsigned long tmp;
+
+#ifdef CONFIG_DEBUG_HIGHMEM
+	printk("mapping high region %08lx - %08lx\n", start_pfn, end_pfn);
+#endif
+
+	for (tmp = start_pfn; tmp < end_pfn; tmp++)
+		free_highmem_page(pfn_to_page(tmp));
+}
+
+void __init mem_init(void)
+{
+	int i;
+
+	if (PKMAP_BASE+LAST_PKMAP*PAGE_SIZE >= FIXADDR_START) {
+		prom_printf("BUG: fixmap and pkmap areas overlap\n");
+		prom_printf("pkbase: 0x%lx pkend: 0x%lx fixstart 0x%lx\n",
+		       PKMAP_BASE,
+		       (unsigned long)PKMAP_BASE+LAST_PKMAP*PAGE_SIZE,
+		       FIXADDR_START);
+		prom_printf("Please mail sparclinux@vger.kernel.org.\n");
+		prom_halt();
+	}
+
+
+	/* Saves us work later. */
+	memset((void *)&empty_zero_page, 0, PAGE_SIZE);
+
+	i = last_valid_pfn >> ((20 - PAGE_SHIFT) + 5);
+	i += 1;
+	sparc_valid_addr_bitmap = (unsigned long *)
+		__alloc_bootmem(i << 2, SMP_CACHE_BYTES, 0UL);
+
+	if (sparc_valid_addr_bitmap == NULL) {
+		prom_printf("mem_init: Cannot alloc valid_addr_bitmap.\n");
+		prom_halt();
+	}
+	memset(sparc_valid_addr_bitmap, 0, i << 2);
+
+	taint_real_pages();
+
+	max_mapnr = last_valid_pfn - pfn_base;
+	high_memory = __va(max_low_pfn << PAGE_SHIFT);
+	free_all_bootmem();
+
+	for (i = 0; sp_banks[i].num_bytes != 0; i++) {
+		unsigned long start_pfn = sp_banks[i].base_addr >> PAGE_SHIFT;
+		unsigned long end_pfn = (sp_banks[i].base_addr + sp_banks[i].num_bytes) >> PAGE_SHIFT;
+
+		if (end_pfn <= highstart_pfn)
+			continue;
+
+		if (start_pfn < highstart_pfn)
+			start_pfn = highstart_pfn;
+
+		map_high_region(start_pfn, end_pfn);
+	}
+	
+	mem_init_print_info(NULL);
+}
+
+void free_initmem (void)
+{
+	free_initmem_default(POISON_FREE_INITMEM);
+}
+
+#ifdef CONFIG_BLK_DEV_INITRD
+void free_initrd_mem(unsigned long start, unsigned long end)
+{
+	free_reserved_area((void *)start, (void *)end, POISON_FREE_INITMEM,
+			   "initrd");
+}
+#endif
+
+void sparc_flush_page_to_ram(struct page *page)
+{
+	unsigned long vaddr = (unsigned long)page_address(page);
+
+	if (vaddr)
+		__flush_page_to_ram(vaddr);
+}
+EXPORT_SYMBOL(sparc_flush_page_to_ram);

arch/sparc/mm/init_64.h

@@ -0,0 +1,34 @@
+#ifndef _SPARC64_MM_INIT_H
+#define _SPARC64_MM_INIT_H
+
+#include <asm/page.h>
+
+/* Most of the symbols in this file are defined in init.c and
+ * marked non-static so that assembler code can get at them.
+ */
+
+#define MAX_PHYS_ADDRESS	(1UL << MAX_PHYS_ADDRESS_BITS)
+
+extern unsigned long kern_linear_pte_xor[4];
+extern unsigned int sparc64_highest_unlocked_tlb_ent;
+extern unsigned long sparc64_kern_pri_context;
+extern unsigned long sparc64_kern_pri_nuc_bits;
+extern unsigned long sparc64_kern_sec_context;
+void mmu_info(struct seq_file *m);
+
+struct linux_prom_translation {
+	unsigned long virt;
+	unsigned long size;
+	unsigned long data;
+};
+
+/* Exported for kernel TLB miss handling in ktlb.S */
+extern struct linux_prom_translation prom_trans[512];
+extern unsigned int prom_trans_ents;
+
+/* Exported for SMP bootup purposes. */
+extern unsigned long kern_locked_tte_data;
+
+void prom_world(int enter);
+
+#endif /* _SPARC64_MM_INIT_H */

arch/sparc/mm/io-unit.c

@@ -0,0 +1,262 @@
+/*
+ * io-unit.c:  IO-UNIT specific routines for memory management.
+ *
+ * Copyright (C) 1997,1998 Jakub Jelinek    (jj@sunsite.mff.cuni.cz)
+ */
+ 
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+#include <linux/spinlock.h>
+#include <linux/mm.h>
+#include <linux/highmem.h>	/* pte_offset_map => kmap_atomic */
+#include <linux/bitops.h>
+#include <linux/scatterlist.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+
+#include <asm/pgalloc.h>
+#include <asm/pgtable.h>
+#include <asm/io.h>
+#include <asm/io-unit.h>
+#include <asm/mxcc.h>
+#include <asm/cacheflush.h>
+#include <asm/tlbflush.h>
+#include <asm/dma.h>
+#include <asm/oplib.h>
+
+#include "mm_32.h"
+
+/* #define IOUNIT_DEBUG */
+#ifdef IOUNIT_DEBUG
+#define IOD(x) printk(x)
+#else
+#define IOD(x) do { } while (0)
+#endif
+
+#define IOPERM        (IOUPTE_CACHE | IOUPTE_WRITE | IOUPTE_VALID)
+#define MKIOPTE(phys) __iopte((((phys)>>4) & IOUPTE_PAGE) | IOPERM)
+
+static void __init iounit_iommu_init(struct platform_device *op)
+{
+	struct iounit_struct *iounit;
+	iopte_t __iomem *xpt;
+	iopte_t __iomem *xptend;
+
+	iounit = kzalloc(sizeof(struct iounit_struct), GFP_ATOMIC);
+	if (!iounit) {
+		prom_printf("SUN4D: Cannot alloc iounit, halting.\n");
+		prom_halt();
+	}
+
+	iounit->limit[0] = IOUNIT_BMAP1_START;
+	iounit->limit[1] = IOUNIT_BMAP2_START;
+	iounit->limit[2] = IOUNIT_BMAPM_START;
+	iounit->limit[3] = IOUNIT_BMAPM_END;
+	iounit->rotor[1] = IOUNIT_BMAP2_START;
+	iounit->rotor[2] = IOUNIT_BMAPM_START;
+
+	xpt = of_ioremap(&op->resource[2], 0, PAGE_SIZE * 16, "XPT");
+	if (!xpt) {
+		prom_printf("SUN4D: Cannot map External Page Table.");
+		prom_halt();
+	}
+	
+	op->dev.archdata.iommu = iounit;
+	iounit->page_table = xpt;
+	spin_lock_init(&iounit->lock);
+
+	xptend = iounit->page_table + (16 * PAGE_SIZE) / sizeof(iopte_t);
+	for (; xpt < xptend; xpt++)
+		sbus_writel(0, xpt);
+}
+
+static int __init iounit_init(void)
+{
+	extern void sun4d_init_sbi_irq(void);
+	struct device_node *dp;
+
+	for_each_node_by_name(dp, "sbi") {
+		struct platform_device *op = of_find_device_by_node(dp);
+
+		iounit_iommu_init(op);
+		of_propagate_archdata(op);
+	}
+
+	sun4d_init_sbi_irq();
+
+	return 0;
+}
+
+subsys_initcall(iounit_init);
+
+/* One has to hold iounit->lock to call this */
+static unsigned long iounit_get_area(struct iounit_struct *iounit, unsigned long vaddr, int size)
+{
+	int i, j, k, npages;
+	unsigned long rotor, scan, limit;
+	iopte_t iopte;
+
+        npages = ((vaddr & ~PAGE_MASK) + size + (PAGE_SIZE-1)) >> PAGE_SHIFT;
+
+	/* A tiny bit of magic ingredience :) */
+	switch (npages) {
+	case 1: i = 0x0231; break;
+	case 2: i = 0x0132; break;
+	default: i = 0x0213; break;
+	}
+	
+	IOD(("iounit_get_area(%08lx,%d[%d])=", vaddr, size, npages));
+	
+next:	j = (i & 15);
+	rotor = iounit->rotor[j - 1];
+	limit = iounit->limit[j];
+	scan = rotor;
+nexti:	scan = find_next_zero_bit(iounit->bmap, limit, scan);
+	if (scan + npages > limit) {
+		if (limit != rotor) {
+			limit = rotor;
+			scan = iounit->limit[j - 1];
+			goto nexti;
+		}
+		i >>= 4;
+		if (!(i & 15))
+			panic("iounit_get_area: Couldn't find free iopte slots for (%08lx,%d)\n", vaddr, size);
+		goto next;
+	}
+	for (k = 1, scan++; k < npages; k++)
+		if (test_bit(scan++, iounit->bmap))
+			goto nexti;
+	iounit->rotor[j - 1] = (scan < limit) ? scan : iounit->limit[j - 1];
+	scan -= npages;
+	iopte = MKIOPTE(__pa(vaddr & PAGE_MASK));
+	vaddr = IOUNIT_DMA_BASE + (scan << PAGE_SHIFT) + (vaddr & ~PAGE_MASK);
+	for (k = 0; k < npages; k++, iopte = __iopte(iopte_val(iopte) + 0x100), scan++) {
+		set_bit(scan, iounit->bmap);
+		sbus_writel(iopte, &iounit->page_table[scan]);
+	}
+	IOD(("%08lx\n", vaddr));
+	return vaddr;
+}
+
+static __u32 iounit_get_scsi_one(struct device *dev, char *vaddr, unsigned long len)
+{
+	struct iounit_struct *iounit = dev->archdata.iommu;
+	unsigned long ret, flags;
+	
+	spin_lock_irqsave(&iounit->lock, flags);
+	ret = iounit_get_area(iounit, (unsigned long)vaddr, len);
+	spin_unlock_irqrestore(&iounit->lock, flags);
+	return ret;
+}
+
+static void iounit_get_scsi_sgl(struct device *dev, struct scatterlist *sg, int sz)
+{
+	struct iounit_struct *iounit = dev->archdata.iommu;
+	unsigned long flags;
+
+	/* FIXME: Cache some resolved pages - often several sg entries are to the same page */
+	spin_lock_irqsave(&iounit->lock, flags);
+	while (sz != 0) {
+		--sz;
+		sg->dma_address = iounit_get_area(iounit, (unsigned long) sg_virt(sg), sg->length);
+		sg->dma_length = sg->length;
+		sg = sg_next(sg);
+	}
+	spin_unlock_irqrestore(&iounit->lock, flags);
+}
+
+static void iounit_release_scsi_one(struct device *dev, __u32 vaddr, unsigned long len)
+{
+	struct iounit_struct *iounit = dev->archdata.iommu;
+	unsigned long flags;
+	
+	spin_lock_irqsave(&iounit->lock, flags);
+	len = ((vaddr & ~PAGE_MASK) + len + (PAGE_SIZE-1)) >> PAGE_SHIFT;
+	vaddr = (vaddr - IOUNIT_DMA_BASE) >> PAGE_SHIFT;
+	IOD(("iounit_release %08lx-%08lx\n", (long)vaddr, (long)len+vaddr));
+	for (len += vaddr; vaddr < len; vaddr++)
+		clear_bit(vaddr, iounit->bmap);
+	spin_unlock_irqrestore(&iounit->lock, flags);
+}
+
+static void iounit_release_scsi_sgl(struct device *dev, struct scatterlist *sg, int sz)
+{
+	struct iounit_struct *iounit = dev->archdata.iommu;
+	unsigned long flags;
+	unsigned long vaddr, len;
+
+	spin_lock_irqsave(&iounit->lock, flags);
+	while (sz != 0) {
+		--sz;
+		len = ((sg->dma_address & ~PAGE_MASK) + sg->length + (PAGE_SIZE-1)) >> PAGE_SHIFT;
+		vaddr = (sg->dma_address - IOUNIT_DMA_BASE) >> PAGE_SHIFT;
+		IOD(("iounit_release %08lx-%08lx\n", (long)vaddr, (long)len+vaddr));
+		for (len += vaddr; vaddr < len; vaddr++)
+			clear_bit(vaddr, iounit->bmap);
+		sg = sg_next(sg);
+	}
+	spin_unlock_irqrestore(&iounit->lock, flags);
+}
+
+#ifdef CONFIG_SBUS
+static int iounit_map_dma_area(struct device *dev, dma_addr_t *pba, unsigned long va, unsigned long addr, int len)
+{
+	struct iounit_struct *iounit = dev->archdata.iommu;
+	unsigned long page, end;
+	pgprot_t dvma_prot;
+	iopte_t __iomem *iopte;
+
+	*pba = addr;
+
+	dvma_prot = __pgprot(SRMMU_CACHE | SRMMU_ET_PTE | SRMMU_PRIV);
+	end = PAGE_ALIGN((addr + len));
+	while(addr < end) {
+		page = va;
+		{
+			pgd_t *pgdp;
+			pmd_t *pmdp;
+			pte_t *ptep;
+			long i;
+
+			pgdp = pgd_offset(&init_mm, addr);
+			pmdp = pmd_offset(pgdp, addr);
+			ptep = pte_offset_map(pmdp, addr);
+
+			set_pte(ptep, mk_pte(virt_to_page(page), dvma_prot));
+			
+			i = ((addr - IOUNIT_DMA_BASE) >> PAGE_SHIFT);
+
+			iopte = iounit->page_table + i;
+			sbus_writel(MKIOPTE(__pa(page)), iopte);
+		}
+		addr += PAGE_SIZE;
+		va += PAGE_SIZE;
+	}
+	flush_cache_all();
+	flush_tlb_all();
+
+	return 0;
+}
+
+static void iounit_unmap_dma_area(struct device *dev, unsigned long addr, int len)
+{
+	/* XXX Somebody please fill this in */
+}
+#endif
+
+static const struct sparc32_dma_ops iounit_dma_ops = {
+	.get_scsi_one		= iounit_get_scsi_one,
+	.get_scsi_sgl		= iounit_get_scsi_sgl,
+	.release_scsi_one	= iounit_release_scsi_one,
+	.release_scsi_sgl	= iounit_release_scsi_sgl,
+#ifdef CONFIG_SBUS
+	.map_dma_area		= iounit_map_dma_area,
+	.unmap_dma_area		= iounit_unmap_dma_area,
+#endif
+};
+
+void __init ld_mmu_iounit(void)
+{
+	sparc32_dma_ops = &iounit_dma_ops;
+}

arch/sparc/mm/iommu.c

@@ -0,0 +1,451 @@
+/*
+ * iommu.c:  IOMMU specific routines for memory management.
+ *
+ * Copyright (C) 1995 David S. Miller  (davem@caip.rutgers.edu)
+ * Copyright (C) 1995,2002 Pete Zaitcev     (zaitcev@yahoo.com)
+ * Copyright (C) 1996 Eddie C. Dost    (ecd@skynet.be)
+ * Copyright (C) 1997,1998 Jakub Jelinek    (jj@sunsite.mff.cuni.cz)
+ */
+ 
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/mm.h>
+#include <linux/slab.h>
+#include <linux/highmem.h>	/* pte_offset_map => kmap_atomic */
+#include <linux/scatterlist.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+
+#include <asm/pgalloc.h>
+#include <asm/pgtable.h>
+#include <asm/io.h>
+#include <asm/mxcc.h>
+#include <asm/mbus.h>
+#include <asm/cacheflush.h>
+#include <asm/tlbflush.h>
+#include <asm/bitext.h>
+#include <asm/iommu.h>
+#include <asm/dma.h>
+
+#include "mm_32.h"
+
+/*
+ * This can be sized dynamically, but we will do this
+ * only when we have a guidance about actual I/O pressures.
+ */
+#define IOMMU_RNGE	IOMMU_RNGE_256MB
+#define IOMMU_START	0xF0000000
+#define IOMMU_WINSIZE	(256*1024*1024U)
+#define IOMMU_NPTES	(IOMMU_WINSIZE/PAGE_SIZE)	/* 64K PTEs, 256KB */
+#define IOMMU_ORDER	6				/* 4096 * (1<<6) */
+
+static int viking_flush;
+/* viking.S */
+extern void viking_flush_page(unsigned long page);
+extern void viking_mxcc_flush_page(unsigned long page);
+
+/*
+ * Values precomputed according to CPU type.
+ */
+static unsigned int ioperm_noc;		/* Consistent mapping iopte flags */
+static pgprot_t dvma_prot;		/* Consistent mapping pte flags */
+
+#define IOPERM        (IOPTE_CACHE | IOPTE_WRITE | IOPTE_VALID)
+#define MKIOPTE(pfn, perm) (((((pfn)<<8) & IOPTE_PAGE) | (perm)) & ~IOPTE_WAZ)
+
+static void __init sbus_iommu_init(struct platform_device *op)
+{
+	struct iommu_struct *iommu;
+	unsigned int impl, vers;
+	unsigned long *bitmap;
+	unsigned long control;
+	unsigned long base;
+	unsigned long tmp;
+
+	iommu = kmalloc(sizeof(struct iommu_struct), GFP_KERNEL);
+	if (!iommu) {
+		prom_printf("Unable to allocate iommu structure\n");
+		prom_halt();
+	}
+
+	iommu->regs = of_ioremap(&op->resource[0], 0, PAGE_SIZE * 3,
+				 "iommu_regs");
+	if (!iommu->regs) {
+		prom_printf("Cannot map IOMMU registers\n");
+		prom_halt();
+	}
+
+	control = sbus_readl(&iommu->regs->control);
+	impl = (control & IOMMU_CTRL_IMPL) >> 28;
+	vers = (control & IOMMU_CTRL_VERS) >> 24;
+	control &= ~(IOMMU_CTRL_RNGE);
+	control |= (IOMMU_RNGE_256MB | IOMMU_CTRL_ENAB);
+	sbus_writel(control, &iommu->regs->control);
+
+	iommu_invalidate(iommu->regs);
+	iommu->start = IOMMU_START;
+	iommu->end = 0xffffffff;
+
+	/* Allocate IOMMU page table */
+	/* Stupid alignment constraints give me a headache. 
+	   We need 256K or 512K or 1M or 2M area aligned to
+           its size and current gfp will fortunately give
+           it to us. */
+        tmp = __get_free_pages(GFP_KERNEL, IOMMU_ORDER);
+	if (!tmp) {
+		prom_printf("Unable to allocate iommu table [0x%lx]\n",
+			    IOMMU_NPTES * sizeof(iopte_t));
+		prom_halt();
+	}
+	iommu->page_table = (iopte_t *)tmp;
+
+	/* Initialize new table. */
+	memset(iommu->page_table, 0, IOMMU_NPTES*sizeof(iopte_t));
+	flush_cache_all();
+	flush_tlb_all();
+
+	base = __pa((unsigned long)iommu->page_table) >> 4;
+	sbus_writel(base, &iommu->regs->base);
+	iommu_invalidate(iommu->regs);
+
+	bitmap = kmalloc(IOMMU_NPTES>>3, GFP_KERNEL);
+	if (!bitmap) {
+		prom_printf("Unable to allocate iommu bitmap [%d]\n",
+			    (int)(IOMMU_NPTES>>3));
+		prom_halt();
+	}
+	bit_map_init(&iommu->usemap, bitmap, IOMMU_NPTES);
+	/* To be coherent on HyperSparc, the page color of DVMA
+	 * and physical addresses must match.
+	 */
+	if (srmmu_modtype == HyperSparc)
+		iommu->usemap.num_colors = vac_cache_size >> PAGE_SHIFT;
+	else
+		iommu->usemap.num_colors = 1;
+
+	printk(KERN_INFO "IOMMU: impl %d vers %d table 0x%p[%d B] map [%d b]\n",
+	       impl, vers, iommu->page_table,
+	       (int)(IOMMU_NPTES*sizeof(iopte_t)), (int)IOMMU_NPTES);
+
+	op->dev.archdata.iommu = iommu;
+}
+
+static int __init iommu_init(void)
+{
+	struct device_node *dp;
+
+	for_each_node_by_name(dp, "iommu") {
+		struct platform_device *op = of_find_device_by_node(dp);
+
+		sbus_iommu_init(op);
+		of_propagate_archdata(op);
+	}
+
+	return 0;
+}
+
+subsys_initcall(iommu_init);
+
+/* Flush the iotlb entries to ram. */
+/* This could be better if we didn't have to flush whole pages. */
+static void iommu_flush_iotlb(iopte_t *iopte, unsigned int niopte)
+{
+	unsigned long start;
+	unsigned long end;
+
+	start = (unsigned long)iopte;
+	end = PAGE_ALIGN(start + niopte*sizeof(iopte_t));
+	start &= PAGE_MASK;
+	if (viking_mxcc_present) {
+		while(start < end) {
+			viking_mxcc_flush_page(start);
+			start += PAGE_SIZE;
+		}
+	} else if (viking_flush) {
+		while(start < end) {
+			viking_flush_page(start);
+			start += PAGE_SIZE;
+		}
+	} else {
+		while(start < end) {
+			__flush_page_to_ram(start);
+			start += PAGE_SIZE;
+		}
+	}
+}
+
+static u32 iommu_get_one(struct device *dev, struct page *page, int npages)
+{
+	struct iommu_struct *iommu = dev->archdata.iommu;
+	int ioptex;
+	iopte_t *iopte, *iopte0;
+	unsigned int busa, busa0;
+	int i;
+
+	/* page color = pfn of page */
+	ioptex = bit_map_string_get(&iommu->usemap, npages, page_to_pfn(page));
+	if (ioptex < 0)
+		panic("iommu out");
+	busa0 = iommu->start + (ioptex << PAGE_SHIFT);
+	iopte0 = &iommu->page_table[ioptex];
+
+	busa = busa0;
+	iopte = iopte0;
+	for (i = 0; i < npages; i++) {
+		iopte_val(*iopte) = MKIOPTE(page_to_pfn(page), IOPERM);
+		iommu_invalidate_page(iommu->regs, busa);
+		busa += PAGE_SIZE;
+		iopte++;
+		page++;
+	}
+
+	iommu_flush_iotlb(iopte0, npages);
+
+	return busa0;
+}
+
+static u32 iommu_get_scsi_one(struct device *dev, char *vaddr, unsigned int len)
+{
+	unsigned long off;
+	int npages;
+	struct page *page;
+	u32 busa;
+
+	off = (unsigned long)vaddr & ~PAGE_MASK;
+	npages = (off + len + PAGE_SIZE-1) >> PAGE_SHIFT;
+	page = virt_to_page((unsigned long)vaddr & PAGE_MASK);
+	busa = iommu_get_one(dev, page, npages);
+	return busa + off;
+}
+
+static __u32 iommu_get_scsi_one_gflush(struct device *dev, char *vaddr, unsigned long len)
+{
+	flush_page_for_dma(0);
+	return iommu_get_scsi_one(dev, vaddr, len);
+}
+
+static __u32 iommu_get_scsi_one_pflush(struct device *dev, char *vaddr, unsigned long len)
+{
+	unsigned long page = ((unsigned long) vaddr) & PAGE_MASK;
+
+	while(page < ((unsigned long)(vaddr + len))) {
+		flush_page_for_dma(page);
+		page += PAGE_SIZE;
+	}
+	return iommu_get_scsi_one(dev, vaddr, len);
+}
+
+static void iommu_get_scsi_sgl_gflush(struct device *dev, struct scatterlist *sg, int sz)
+{
+	int n;
+
+	flush_page_for_dma(0);
+	while (sz != 0) {
+		--sz;
+		n = (sg->length + sg->offset + PAGE_SIZE-1) >> PAGE_SHIFT;
+		sg->dma_address = iommu_get_one(dev, sg_page(sg), n) + sg->offset;
+		sg->dma_length = sg->length;
+		sg = sg_next(sg);
+	}
+}
+
+static void iommu_get_scsi_sgl_pflush(struct device *dev, struct scatterlist *sg, int sz)
+{
+	unsigned long page, oldpage = 0;
+	int n, i;
+
+	while(sz != 0) {
+		--sz;
+
+		n = (sg->length + sg->offset + PAGE_SIZE-1) >> PAGE_SHIFT;
+
+		/*
+		 * We expect unmapped highmem pages to be not in the cache.
+		 * XXX Is this a good assumption?
+		 * XXX What if someone else unmaps it here and races us?
+		 */
+		if ((page = (unsigned long) page_address(sg_page(sg))) != 0) {
+			for (i = 0; i < n; i++) {
+				if (page != oldpage) {	/* Already flushed? */
+					flush_page_for_dma(page);
+					oldpage = page;
+				}
+				page += PAGE_SIZE;
+			}
+		}
+
+		sg->dma_address = iommu_get_one(dev, sg_page(sg), n) + sg->offset;
+		sg->dma_length = sg->length;
+		sg = sg_next(sg);
+	}
+}
+
+static void iommu_release_one(struct device *dev, u32 busa, int npages)
+{
+	struct iommu_struct *iommu = dev->archdata.iommu;
+	int ioptex;
+	int i;
+
+	BUG_ON(busa < iommu->start);
+	ioptex = (busa - iommu->start) >> PAGE_SHIFT;
+	for (i = 0; i < npages; i++) {
+		iopte_val(iommu->page_table[ioptex + i]) = 0;
+		iommu_invalidate_page(iommu->regs, busa);
+		busa += PAGE_SIZE;
+	}
+	bit_map_clear(&iommu->usemap, ioptex, npages);
+}
+
+static void iommu_release_scsi_one(struct device *dev, __u32 vaddr, unsigned long len)
+{
+	unsigned long off;
+	int npages;
+
+	off = vaddr & ~PAGE_MASK;
+	npages = (off + len + PAGE_SIZE-1) >> PAGE_SHIFT;
+	iommu_release_one(dev, vaddr & PAGE_MASK, npages);
+}
+
+static void iommu_release_scsi_sgl(struct device *dev, struct scatterlist *sg, int sz)
+{
+	int n;
+
+	while(sz != 0) {
+		--sz;
+
+		n = (sg->length + sg->offset + PAGE_SIZE-1) >> PAGE_SHIFT;
+		iommu_release_one(dev, sg->dma_address & PAGE_MASK, n);
+		sg->dma_address = 0x21212121;
+		sg = sg_next(sg);
+	}
+}
+
+#ifdef CONFIG_SBUS
+static int iommu_map_dma_area(struct device *dev, dma_addr_t *pba, unsigned long va,
+			      unsigned long addr, int len)
+{
+	struct iommu_struct *iommu = dev->archdata.iommu;
+	unsigned long page, end;
+	iopte_t *iopte = iommu->page_table;
+	iopte_t *first;
+	int ioptex;
+
+	BUG_ON((va & ~PAGE_MASK) != 0);
+	BUG_ON((addr & ~PAGE_MASK) != 0);
+	BUG_ON((len & ~PAGE_MASK) != 0);
+
+	/* page color = physical address */
+	ioptex = bit_map_string_get(&iommu->usemap, len >> PAGE_SHIFT,
+		addr >> PAGE_SHIFT);
+	if (ioptex < 0)
+		panic("iommu out");
+
+	iopte += ioptex;
+	first = iopte;
+	end = addr + len;
+	while(addr < end) {
+		page = va;
+		{
+			pgd_t *pgdp;
+			pmd_t *pmdp;
+			pte_t *ptep;
+
+			if (viking_mxcc_present)
+				viking_mxcc_flush_page(page);
+			else if (viking_flush)
+				viking_flush_page(page);
+			else
+				__flush_page_to_ram(page);
+
+			pgdp = pgd_offset(&init_mm, addr);
+			pmdp = pmd_offset(pgdp, addr);
+			ptep = pte_offset_map(pmdp, addr);
+
+			set_pte(ptep, mk_pte(virt_to_page(page), dvma_prot));
+		}
+		iopte_val(*iopte++) =
+		    MKIOPTE(page_to_pfn(virt_to_page(page)), ioperm_noc);
+		addr += PAGE_SIZE;
+		va += PAGE_SIZE;
+	}
+	/* P3: why do we need this?
+	 *
+	 * DAVEM: Because there are several aspects, none of which
+	 *        are handled by a single interface.  Some cpus are
+	 *        completely not I/O DMA coherent, and some have
+	 *        virtually indexed caches.  The driver DMA flushing
+	 *        methods handle the former case, but here during
+	 *        IOMMU page table modifications, and usage of non-cacheable
+	 *        cpu mappings of pages potentially in the cpu caches, we have
+	 *        to handle the latter case as well.
+	 */
+	flush_cache_all();
+	iommu_flush_iotlb(first, len >> PAGE_SHIFT);
+	flush_tlb_all();
+	iommu_invalidate(iommu->regs);
+
+	*pba = iommu->start + (ioptex << PAGE_SHIFT);
+	return 0;
+}
+
+static void iommu_unmap_dma_area(struct device *dev, unsigned long busa, int len)
+{
+	struct iommu_struct *iommu = dev->archdata.iommu;
+	iopte_t *iopte = iommu->page_table;
+	unsigned long end;
+	int ioptex = (busa - iommu->start) >> PAGE_SHIFT;
+
+	BUG_ON((busa & ~PAGE_MASK) != 0);
+	BUG_ON((len & ~PAGE_MASK) != 0);
+
+	iopte += ioptex;
+	end = busa + len;
+	while (busa < end) {
+		iopte_val(*iopte++) = 0;
+		busa += PAGE_SIZE;
+	}
+	flush_tlb_all();
+	iommu_invalidate(iommu->regs);
+	bit_map_clear(&iommu->usemap, ioptex, len >> PAGE_SHIFT);
+}
+#endif
+
+static const struct sparc32_dma_ops iommu_dma_gflush_ops = {
+	.get_scsi_one		= iommu_get_scsi_one_gflush,
+	.get_scsi_sgl		= iommu_get_scsi_sgl_gflush,
+	.release_scsi_one	= iommu_release_scsi_one,
+	.release_scsi_sgl	= iommu_release_scsi_sgl,
+#ifdef CONFIG_SBUS
+	.map_dma_area		= iommu_map_dma_area,
+	.unmap_dma_area		= iommu_unmap_dma_area,
+#endif
+};
+
+static const struct sparc32_dma_ops iommu_dma_pflush_ops = {
+	.get_scsi_one		= iommu_get_scsi_one_pflush,
+	.get_scsi_sgl		= iommu_get_scsi_sgl_pflush,
+	.release_scsi_one	= iommu_release_scsi_one,
+	.release_scsi_sgl	= iommu_release_scsi_sgl,
+#ifdef CONFIG_SBUS
+	.map_dma_area		= iommu_map_dma_area,
+	.unmap_dma_area		= iommu_unmap_dma_area,
+#endif
+};
+
+void __init ld_mmu_iommu(void)
+{
+	if (flush_page_for_dma_global) {
+		/* flush_page_for_dma flushes everything, no matter of what page is it */
+		sparc32_dma_ops = &iommu_dma_gflush_ops;
+	} else {
+		sparc32_dma_ops = &iommu_dma_pflush_ops;
+	}
+
+	if (viking_mxcc_present || srmmu_modtype == HyperSparc) {
+		dvma_prot = __pgprot(SRMMU_CACHE | SRMMU_ET_PTE | SRMMU_PRIV);
+		ioperm_noc = IOPTE_CACHE | IOPTE_WRITE | IOPTE_VALID;
+	} else {
+		dvma_prot = __pgprot(SRMMU_ET_PTE | SRMMU_PRIV);
+		ioperm_noc = IOPTE_WRITE | IOPTE_VALID;
+	}
+}

arch/sparc/mm/leon_mm.c

@@ -0,0 +1,351 @@
+/*
+ *  linux/arch/sparc/mm/leon_m.c
+ *
+ * Copyright (C) 2004 Konrad Eisele (eiselekd@web.de, konrad@gaisler.com) Gaisler Research
+ * Copyright (C) 2009 Daniel Hellstrom (daniel@gaisler.com) Aeroflex Gaisler AB
+ * Copyright (C) 2009 Konrad Eisele (konrad@gaisler.com) Aeroflex Gaisler AB
+ *
+ * do srmmu probe in software
+ *
+ */
+
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <asm/asi.h>
+#include <asm/leon.h>
+#include <asm/tlbflush.h>
+
+#include "mm_32.h"
+
+int leon_flush_during_switch = 1;
+static int srmmu_swprobe_trace;
+
+static inline unsigned long leon_get_ctable_ptr(void)
+{
+	unsigned int retval;
+
+	__asm__ __volatile__("lda [%1] %2, %0\n\t" :
+			     "=r" (retval) :
+			     "r" (SRMMU_CTXTBL_PTR),
+			     "i" (ASI_LEON_MMUREGS));
+	return (retval & SRMMU_CTX_PMASK) << 4;
+}
+
+
+unsigned long leon_swprobe(unsigned long vaddr, unsigned long *paddr)
+{
+
+	unsigned int ctxtbl;
+	unsigned int pgd, pmd, ped;
+	unsigned int ptr;
+	unsigned int lvl, pte, paddrbase;
+	unsigned int ctx;
+	unsigned int paddr_calc;
+
+	paddrbase = 0;
+
+	if (srmmu_swprobe_trace)
+		printk(KERN_INFO "swprobe: trace on\n");
+
+	ctxtbl = leon_get_ctable_ptr();
+	if (!(ctxtbl)) {
+		if (srmmu_swprobe_trace)
+			printk(KERN_INFO "swprobe: leon_get_ctable_ptr returned 0=>0\n");
+		return 0;
+	}
+	if (!_pfn_valid(PFN(ctxtbl))) {
+		if (srmmu_swprobe_trace)
+			printk(KERN_INFO
+			       "swprobe: !_pfn_valid(%x)=>0\n",
+			       PFN(ctxtbl));
+		return 0;
+	}
+
+	ctx = srmmu_get_context();
+	if (srmmu_swprobe_trace)
+		printk(KERN_INFO "swprobe:  --- ctx (%x) ---\n", ctx);
+
+	pgd = LEON_BYPASS_LOAD_PA(ctxtbl + (ctx * 4));
+
+	if (((pgd & SRMMU_ET_MASK) == SRMMU_ET_PTE)) {
+		if (srmmu_swprobe_trace)
+			printk(KERN_INFO "swprobe: pgd is entry level 3\n");
+		lvl = 3;
+		pte = pgd;
+		paddrbase = pgd & _SRMMU_PTE_PMASK_LEON;
+		goto ready;
+	}
+	if (((pgd & SRMMU_ET_MASK) != SRMMU_ET_PTD)) {
+		if (srmmu_swprobe_trace)
+			printk(KERN_INFO "swprobe: pgd is invalid => 0\n");
+		return 0;
+	}
+
+	if (srmmu_swprobe_trace)
+		printk(KERN_INFO "swprobe:  --- pgd (%x) ---\n", pgd);
+
+	ptr = (pgd & SRMMU_PTD_PMASK) << 4;
+	ptr += ((((vaddr) >> LEON_PGD_SH) & LEON_PGD_M) * 4);
+	if (!_pfn_valid(PFN(ptr)))
+		return 0;
+
+	pmd = LEON_BYPASS_LOAD_PA(ptr);
+	if (((pmd & SRMMU_ET_MASK) == SRMMU_ET_PTE)) {
+		if (srmmu_swprobe_trace)
+			printk(KERN_INFO "swprobe: pmd is entry level 2\n");
+		lvl = 2;
+		pte = pmd;
+		paddrbase = pmd & _SRMMU_PTE_PMASK_LEON;
+		goto ready;
+	}
+	if (((pmd & SRMMU_ET_MASK) != SRMMU_ET_PTD)) {
+		if (srmmu_swprobe_trace)
+			printk(KERN_INFO "swprobe: pmd is invalid => 0\n");
+		return 0;
+	}
+
+	if (srmmu_swprobe_trace)
+		printk(KERN_INFO "swprobe:  --- pmd (%x) ---\n", pmd);
+
+	ptr = (pmd & SRMMU_PTD_PMASK) << 4;
+	ptr += (((vaddr >> LEON_PMD_SH) & LEON_PMD_M) * 4);
+	if (!_pfn_valid(PFN(ptr))) {
+		if (srmmu_swprobe_trace)
+			printk(KERN_INFO "swprobe: !_pfn_valid(%x)=>0\n",
+			       PFN(ptr));
+		return 0;
+	}
+
+	ped = LEON_BYPASS_LOAD_PA(ptr);
+
+	if (((ped & SRMMU_ET_MASK) == SRMMU_ET_PTE)) {
+		if (srmmu_swprobe_trace)
+			printk(KERN_INFO "swprobe: ped is entry level 1\n");
+		lvl = 1;
+		pte = ped;
+		paddrbase = ped & _SRMMU_PTE_PMASK_LEON;
+		goto ready;
+	}
+	if (((ped & SRMMU_ET_MASK) != SRMMU_ET_PTD)) {
+		if (srmmu_swprobe_trace)
+			printk(KERN_INFO "swprobe: ped is invalid => 0\n");
+		return 0;
+	}
+
+	if (srmmu_swprobe_trace)
+		printk(KERN_INFO "swprobe:  --- ped (%x) ---\n", ped);
+
+	ptr = (ped & SRMMU_PTD_PMASK) << 4;
+	ptr += (((vaddr >> LEON_PTE_SH) & LEON_PTE_M) * 4);
+	if (!_pfn_valid(PFN(ptr)))
+		return 0;
+
+	ptr = LEON_BYPASS_LOAD_PA(ptr);
+	if (((ptr & SRMMU_ET_MASK) == SRMMU_ET_PTE)) {
+		if (srmmu_swprobe_trace)
+			printk(KERN_INFO "swprobe: ptr is entry level 0\n");
+		lvl = 0;
+		pte = ptr;
+		paddrbase = ptr & _SRMMU_PTE_PMASK_LEON;
+		goto ready;
+	}
+	if (srmmu_swprobe_trace)
+		printk(KERN_INFO "swprobe: ptr is invalid => 0\n");
+	return 0;
+
+ready:
+	switch (lvl) {
+	case 0:
+		paddr_calc =
+		    (vaddr & ~(-1 << LEON_PTE_SH)) | ((pte & ~0xff) << 4);
+		break;
+	case 1:
+		paddr_calc =
+		    (vaddr & ~(-1 << LEON_PMD_SH)) | ((pte & ~0xff) << 4);
+		break;
+	case 2:
+		paddr_calc =
+		    (vaddr & ~(-1 << LEON_PGD_SH)) | ((pte & ~0xff) << 4);
+		break;
+	default:
+	case 3:
+		paddr_calc = vaddr;
+		break;
+	}
+	if (srmmu_swprobe_trace)
+		printk(KERN_INFO "swprobe: padde %x\n", paddr_calc);
+	if (paddr)
+		*paddr = paddr_calc;
+	return pte;
+}
+
+void leon_flush_icache_all(void)
+{
+	__asm__ __volatile__(" flush ");	/*iflush*/
+}
+
+void leon_flush_dcache_all(void)
+{
+	__asm__ __volatile__("sta %%g0, [%%g0] %0\n\t" : :
+			     "i"(ASI_LEON_DFLUSH) : "memory");
+}
+
+void leon_flush_pcache_all(struct vm_area_struct *vma, unsigned long page)
+{
+	if (vma->vm_flags & VM_EXEC)
+		leon_flush_icache_all();
+	leon_flush_dcache_all();
+}
+
+void leon_flush_cache_all(void)
+{
+	__asm__ __volatile__(" flush ");	/*iflush*/
+	__asm__ __volatile__("sta %%g0, [%%g0] %0\n\t" : :
+			     "i"(ASI_LEON_DFLUSH) : "memory");
+}
+
+void leon_flush_tlb_all(void)
+{
+	leon_flush_cache_all();
+	__asm__ __volatile__("sta %%g0, [%0] %1\n\t" : : "r"(0x400),
+			     "i"(ASI_LEON_MMUFLUSH) : "memory");
+}
+
+/* get all cache regs */
+void leon3_getCacheRegs(struct leon3_cacheregs *regs)
+{
+	unsigned long ccr, iccr, dccr;
+
+	if (!regs)
+		return;
+	/* Get Cache regs from "Cache ASI" address 0x0, 0x8 and 0xC */
+	__asm__ __volatile__("lda [%%g0] %3, %0\n\t"
+			     "mov 0x08, %%g1\n\t"
+			     "lda [%%g1] %3, %1\n\t"
+			     "mov 0x0c, %%g1\n\t"
+			     "lda [%%g1] %3, %2\n\t"
+			     : "=r"(ccr), "=r"(iccr), "=r"(dccr)
+			       /* output */
+			     : "i"(ASI_LEON_CACHEREGS)	/* input */
+			     : "g1"	/* clobber list */
+	    );
+	regs->ccr = ccr;
+	regs->iccr = iccr;
+	regs->dccr = dccr;
+}
+
+/* Due to virtual cache we need to check cache configuration if
+ * it is possible to skip flushing in some cases.
+ *
+ * Leon2 and Leon3 differ in their way of telling cache information
+ *
+ */
+int __init leon_flush_needed(void)
+{
+	int flush_needed = -1;
+	unsigned int ssize, sets;
+	char *setStr[4] =
+	    { "direct mapped", "2-way associative", "3-way associative",
+		"4-way associative"
+	};
+	/* leon 3 */
+	struct leon3_cacheregs cregs;
+	leon3_getCacheRegs(&cregs);
+	sets = (cregs.dccr & LEON3_XCCR_SETS_MASK) >> 24;
+	/* (ssize=>realsize) 0=>1k, 1=>2k, 2=>4k, 3=>8k ... */
+	ssize = 1 << ((cregs.dccr & LEON3_XCCR_SSIZE_MASK) >> 20);
+
+	printk(KERN_INFO "CACHE: %s cache, set size %dk\n",
+	       sets > 3 ? "unknown" : setStr[sets], ssize);
+	if ((ssize <= (PAGE_SIZE / 1024)) && (sets == 0)) {
+		/* Set Size <= Page size  ==>
+		   flush on every context switch not needed. */
+		flush_needed = 0;
+		printk(KERN_INFO "CACHE: not flushing on every context switch\n");
+	}
+	return flush_needed;
+}
+
+void leon_switch_mm(void)
+{
+	flush_tlb_mm((void *)0);
+	if (leon_flush_during_switch)
+		leon_flush_cache_all();
+}
+
+static void leon_flush_cache_mm(struct mm_struct *mm)
+{
+	leon_flush_cache_all();
+}
+
+static void leon_flush_cache_page(struct vm_area_struct *vma, unsigned long page)
+{
+	leon_flush_pcache_all(vma, page);
+}
+
+static void leon_flush_cache_range(struct vm_area_struct *vma,
+				   unsigned long start,
+				   unsigned long end)
+{
+	leon_flush_cache_all();
+}
+
+static void leon_flush_tlb_mm(struct mm_struct *mm)
+{
+	leon_flush_tlb_all();
+}
+
+static void leon_flush_tlb_page(struct vm_area_struct *vma,
+				unsigned long page)
+{
+	leon_flush_tlb_all();
+}
+
+static void leon_flush_tlb_range(struct vm_area_struct *vma,
+				 unsigned long start,
+				 unsigned long end)
+{
+	leon_flush_tlb_all();
+}
+
+static void leon_flush_page_to_ram(unsigned long page)
+{
+	leon_flush_cache_all();
+}
+
+static void leon_flush_sig_insns(struct mm_struct *mm, unsigned long page)
+{
+	leon_flush_cache_all();
+}
+
+static void leon_flush_page_for_dma(unsigned long page)
+{
+	leon_flush_dcache_all();
+}
+
+void __init poke_leonsparc(void)
+{
+}
+
+static const struct sparc32_cachetlb_ops leon_ops = {
+	.cache_all	= leon_flush_cache_all,
+	.cache_mm	= leon_flush_cache_mm,
+	.cache_page	= leon_flush_cache_page,
+	.cache_range	= leon_flush_cache_range,
+	.tlb_all	= leon_flush_tlb_all,
+	.tlb_mm		= leon_flush_tlb_mm,
+	.tlb_page	= leon_flush_tlb_page,
+	.tlb_range	= leon_flush_tlb_range,
+	.page_to_ram	= leon_flush_page_to_ram,
+	.sig_insns	= leon_flush_sig_insns,
+	.page_for_dma	= leon_flush_page_for_dma,
+};
+
+void __init init_leon(void)
+{
+	srmmu_name = "LEON";
+	sparc32_cachetlb_ops = &leon_ops;
+	poke_srmmu = poke_leonsparc;
+
+	leon_flush_during_switch = leon_flush_needed();
+}

arch/sparc/mm/mm_32.h

@@ -0,0 +1,24 @@
+/* fault_32.c - visible as they are called from assembler */
+asmlinkage int lookup_fault(unsigned long pc, unsigned long ret_pc,
+                            unsigned long address);
+asmlinkage void do_sparc_fault(struct pt_regs *regs, int text_fault, int write,
+                               unsigned long address);
+
+void window_overflow_fault(void);
+void window_underflow_fault(unsigned long sp);
+void window_ret_fault(struct pt_regs *regs);
+
+/* srmmu.c */
+extern char *srmmu_name;
+extern int viking_mxcc_present;
+extern int flush_page_for_dma_global;
+
+extern void (*poke_srmmu)(void);
+
+void __init srmmu_paging_init(void);
+
+/* iommu.c */
+void ld_mmu_iommu(void);
+
+/* io-unit.c */
+void ld_mmu_iounit(void);

arch/sparc/mm/srmmu_access.S

@@ -0,0 +1,82 @@
+/* Assembler variants of srmmu access functions.
+ * Implemented in assembler to allow run-time patching.
+ * LEON uses a different ASI for MMUREGS than SUN.
+ *
+ * The leon_1insn_patch infrastructure is used
+ * for the run-time patching.
+ */
+
+#include <linux/linkage.h>
+
+#include <asm/asmmacro.h>
+#include <asm/pgtsrmmu.h>
+#include <asm/asi.h>
+
+/* unsigned int srmmu_get_mmureg(void) */
+ENTRY(srmmu_get_mmureg)
+LEON_PI(lda	[%g0] ASI_LEON_MMUREGS, %o0)
+SUN_PI_(lda	[%g0] ASI_M_MMUREGS, %o0)
+	retl
+	 nop
+ENDPROC(srmmu_get_mmureg)
+
+/* void srmmu_set_mmureg(unsigned long regval) */
+ENTRY(srmmu_set_mmureg)
+LEON_PI(sta	%o0, [%g0] ASI_LEON_MMUREGS)
+SUN_PI_(sta	%o0, [%g0] ASI_M_MMUREGS)
+	retl
+	 nop
+ENDPROC(srmmu_set_mmureg)
+
+/* void srmmu_set_ctable_ptr(unsigned long paddr) */
+ENTRY(srmmu_set_ctable_ptr)
+	/* paddr = ((paddr >> 4) & SRMMU_CTX_PMASK); */
+	srl	%o0, 4, %g1
+	and	%g1, SRMMU_CTX_PMASK, %g1
+
+	mov	SRMMU_CTXTBL_PTR, %g2
+LEON_PI(sta	%g1, [%g2] ASI_LEON_MMUREGS)
+SUN_PI_(sta	%g1, [%g2] ASI_M_MMUREGS)
+	retl
+	 nop
+ENDPROC(srmmu_set_ctable_ptr)
+
+
+/* void srmmu_set_context(int context) */
+ENTRY(srmmu_set_context)
+	mov	SRMMU_CTX_REG, %g1
+LEON_PI(sta	%o0, [%g1] ASI_LEON_MMUREGS)
+SUN_PI_(sta	%o0, [%g1] ASI_M_MMUREGS)
+	retl
+	 nop
+ENDPROC(srmmu_set_context)
+
+
+/* int srmmu_get_context(void) */
+ENTRY(srmmu_get_context)
+	mov	SRMMU_CTX_REG, %o0
+LEON_PI(lda     [%o0] ASI_LEON_MMUREGS, %o0)
+SUN_PI_(lda	[%o0] ASI_M_MMUREGS, %o0)
+	retl
+	 nop
+ENDPROC(srmmu_get_context)
+
+
+/* unsigned int srmmu_get_fstatus(void) */
+ENTRY(srmmu_get_fstatus)
+	mov	SRMMU_FAULT_STATUS, %o0
+LEON_PI(lda     [%o0] ASI_LEON_MMUREGS, %o0)
+SUN_PI_(lda	[%o0] ASI_M_MMUREGS, %o0)
+	retl
+	 nop
+ENDPROC(srmmu_get_fstatus)
+
+
+/* unsigned int srmmu_get_faddr(void) */
+ENTRY(srmmu_get_faddr)
+	mov	SRMMU_FAULT_ADDR, %o0
+LEON_PI(lda     [%o0] ASI_LEON_MMUREGS, %o0)
+SUN_PI_(lda	[%o0] ASI_M_MMUREGS, %o0)
+	retl
+	 nop
+ENDPROC(srmmu_get_faddr)

arch/sparc/mm/swift.S

@@ -0,0 +1,255 @@
+/*
+ * swift.S: MicroSparc-II mmu/cache operations.
+ *
+ * Copyright (C) 1999 David S. Miller (davem@redhat.com)
+ */
+
+#include <asm/psr.h>
+#include <asm/asi.h>
+#include <asm/page.h>
+#include <asm/pgtsrmmu.h>
+#include <asm/asm-offsets.h>
+
+	.text
+	.align	4
+
+#if 1	/* XXX screw this, I can't get the VAC flushes working
+	 * XXX reliably... -DaveM
+	 */
+	.globl	swift_flush_cache_all, swift_flush_cache_mm
+	.globl	swift_flush_cache_range, swift_flush_cache_page
+	.globl	swift_flush_page_for_dma
+	.globl	swift_flush_page_to_ram
+
+swift_flush_cache_all:
+swift_flush_cache_mm:
+swift_flush_cache_range:
+swift_flush_cache_page:
+swift_flush_page_for_dma:
+swift_flush_page_to_ram:
+	sethi	%hi(0x2000), %o0
+1:	subcc	%o0, 0x10, %o0
+	add	%o0, %o0, %o1
+	sta	%g0, [%o0] ASI_M_DATAC_TAG
+	bne	1b
+	 sta	%g0, [%o1] ASI_M_TXTC_TAG
+	retl
+	 nop
+#else
+
+	.globl	swift_flush_cache_all
+swift_flush_cache_all:
+	WINDOW_FLUSH(%g4, %g5)
+
+	/* Just clear out all the tags. */
+	sethi	%hi(16 * 1024), %o0
+1:	subcc	%o0, 16, %o0
+	sta	%g0, [%o0] ASI_M_TXTC_TAG
+	bne	1b
+	 sta	%g0, [%o0] ASI_M_DATAC_TAG
+	retl
+	 nop
+
+	.globl	swift_flush_cache_mm
+swift_flush_cache_mm:
+	ld	[%o0 + AOFF_mm_context], %g2
+	cmp	%g2, -1
+	be	swift_flush_cache_mm_out
+	WINDOW_FLUSH(%g4, %g5)
+	rd	%psr, %g1
+	andn	%g1, PSR_ET, %g3
+	wr	%g3, 0x0, %psr
+	nop
+	nop
+	mov	SRMMU_CTX_REG, %g7
+	lda	[%g7] ASI_M_MMUREGS, %g5
+	sta	%g2, [%g7] ASI_M_MMUREGS
+
+#if 1
+	sethi	%hi(0x2000), %o0
+1:	subcc	%o0, 0x10, %o0
+	sta	%g0, [%o0] ASI_M_FLUSH_CTX
+	bne	1b
+	 nop
+#else
+	clr	%o0
+	or	%g0, 2048, %g7
+	or	%g0, 2048, %o1
+	add	%o1, 2048, %o2
+	add	%o2, 2048, %o3
+	mov	16, %o4
+	add	%o4, 2048, %o5
+	add	%o5, 2048, %g2
+	add	%g2, 2048, %g3
+1:	sta	%g0, [%o0      ] ASI_M_FLUSH_CTX
+	sta	%g0, [%o0 + %o1] ASI_M_FLUSH_CTX
+	sta	%g0, [%o0 + %o2] ASI_M_FLUSH_CTX
+	sta	%g0, [%o0 + %o3] ASI_M_FLUSH_CTX
+	sta	%g0, [%o0 + %o4] ASI_M_FLUSH_CTX
+	sta	%g0, [%o0 + %o5] ASI_M_FLUSH_CTX
+	sta	%g0, [%o0 + %g2] ASI_M_FLUSH_CTX
+	sta	%g0, [%o0 + %g3] ASI_M_FLUSH_CTX
+	subcc	%g7, 32, %g7
+	bne	1b
+	 add	%o0, 32, %o0
+#endif
+
+	mov	SRMMU_CTX_REG, %g7
+	sta	%g5, [%g7] ASI_M_MMUREGS
+	wr	%g1, 0x0, %psr
+	nop
+	nop
+swift_flush_cache_mm_out:
+	retl
+	 nop
+
+	.globl	swift_flush_cache_range
+swift_flush_cache_range:
+	ld	[%o0 + VMA_VM_MM], %o0
+	sub	%o2, %o1, %o2
+	sethi	%hi(4096), %o3
+	cmp	%o2, %o3
+	bgu	swift_flush_cache_mm
+	 nop
+	b	70f
+	 nop
+
+	.globl	swift_flush_cache_page
+swift_flush_cache_page:
+	ld	[%o0 + VMA_VM_MM], %o0
+70:
+	ld	[%o0 + AOFF_mm_context], %g2
+	cmp	%g2, -1
+	be	swift_flush_cache_page_out
+	WINDOW_FLUSH(%g4, %g5)
+	rd	%psr, %g1
+	andn	%g1, PSR_ET, %g3
+	wr	%g3, 0x0, %psr
+	nop
+	nop
+	mov	SRMMU_CTX_REG, %g7
+	lda	[%g7] ASI_M_MMUREGS, %g5
+	sta	%g2, [%g7] ASI_M_MMUREGS
+
+	andn	%o1, (PAGE_SIZE - 1), %o1
+#if 1
+	sethi	%hi(0x1000), %o0
+1:	subcc	%o0, 0x10, %o0
+	sta	%g0, [%o1 + %o0] ASI_M_FLUSH_PAGE
+	bne	1b
+	 nop
+#else
+	or	%g0, 512, %g7
+	or	%g0, 512, %o0
+	add	%o0, 512, %o2
+	add	%o2, 512, %o3
+	add	%o3, 512, %o4
+	add	%o4, 512, %o5
+	add	%o5, 512, %g3
+	add	%g3, 512, %g4
+1:	sta	%g0, [%o1      ] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %o0] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %o2] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %o3] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %o4] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %o5] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %g3] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %g4] ASI_M_FLUSH_PAGE
+	subcc	%g7, 16, %g7
+	bne	1b
+	 add	%o1, 16, %o1
+#endif
+
+	mov	SRMMU_CTX_REG, %g7
+	sta	%g5, [%g7] ASI_M_MMUREGS
+	wr	%g1, 0x0, %psr
+	nop
+	nop
+swift_flush_cache_page_out:
+	retl
+	 nop
+
+	/* Swift is write-thru, however it is not
+	 * I/O nor TLB-walk coherent.  Also it has
+	 * caches which are virtually indexed and tagged.
+	 */
+	.globl	swift_flush_page_for_dma
+	.globl	swift_flush_page_to_ram
+swift_flush_page_for_dma:
+swift_flush_page_to_ram:
+	andn	%o0, (PAGE_SIZE - 1), %o1
+#if 1
+	sethi	%hi(0x1000), %o0
+1:	subcc	%o0, 0x10, %o0
+	sta	%g0, [%o1 + %o0] ASI_M_FLUSH_PAGE
+	bne	1b
+	 nop
+#else
+	or	%g0, 512, %g7
+	or	%g0, 512, %o0
+	add	%o0, 512, %o2
+	add	%o2, 512, %o3
+	add	%o3, 512, %o4
+	add	%o4, 512, %o5
+	add	%o5, 512, %g3
+	add	%g3, 512, %g4
+1:	sta	%g0, [%o1      ] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %o0] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %o2] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %o3] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %o4] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %o5] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %g3] ASI_M_FLUSH_PAGE
+	sta	%g0, [%o1 + %g4] ASI_M_FLUSH_PAGE
+	subcc	%g7, 16, %g7
+	bne	1b
+	 add	%o1, 16, %o1
+#endif
+	retl
+	 nop
+#endif
+
+	.globl	swift_flush_sig_insns
+swift_flush_sig_insns:
+	flush	%o1
+	retl
+	 flush	%o1 + 4
+
+	.globl	swift_flush_tlb_mm
+	.globl	swift_flush_tlb_range
+	.globl	swift_flush_tlb_all
+swift_flush_tlb_range:
+	ld	[%o0 + VMA_VM_MM], %o0
+swift_flush_tlb_mm:
+	ld	[%o0 + AOFF_mm_context], %g2
+	cmp	%g2, -1
+	be	swift_flush_tlb_all_out
+swift_flush_tlb_all:
+	mov	0x400, %o1
+	sta	%g0, [%o1] ASI_M_FLUSH_PROBE
+swift_flush_tlb_all_out:
+	retl
+	 nop
+
+	.globl	swift_flush_tlb_page
+swift_flush_tlb_page:
+	ld	[%o0 + VMA_VM_MM], %o0
+	mov	SRMMU_CTX_REG, %g1
+	ld	[%o0 + AOFF_mm_context], %o3
+	andn	%o1, (PAGE_SIZE - 1), %o1
+	cmp	%o3, -1
+	be	swift_flush_tlb_page_out
+	 nop
+#if 1
+	mov	0x400, %o1
+	sta	%g0, [%o1] ASI_M_FLUSH_PROBE	
+#else
+	lda	[%g1] ASI_M_MMUREGS, %g5
+	sta	%o3, [%g1] ASI_M_MMUREGS
+	sta	%g0, [%o1] ASI_M_FLUSH_PAGE	/* rem. virt. cache. prot. */
+	sta	%g0, [%o1] ASI_M_FLUSH_PROBE
+	sta	%g5, [%g1] ASI_M_MMUREGS
+#endif
+swift_flush_tlb_page_out:
+	retl
+	 nop

arch/sparc/mm/tlb.c

@@ -0,0 +1,243 @@
+/* arch/sparc64/mm/tlb.c
+ *
+ * Copyright (C) 2004 David S. Miller <davem@redhat.com>
+ */
+
+#include <linux/kernel.h>
+#include <linux/percpu.h>
+#include <linux/mm.h>
+#include <linux/swap.h>
+#include <linux/preempt.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.  */
+
+static DEFINE_PER_CPU(struct tlb_batch, tlb_batch);
+
+void flush_tlb_pending(void)
+{
+	struct tlb_batch *tb = &get_cpu_var(tlb_batch);
+	struct mm_struct *mm = tb->mm;
+
+	if (!tb->tlb_nr)
+		goto out;
+
+	flush_tsb_user(tb);
+
+	if (CTX_VALID(mm->context)) {
+		if (tb->tlb_nr == 1) {
+			global_flush_tlb_page(mm, tb->vaddrs[0]);
+		} else {
+#ifdef CONFIG_SMP
+			smp_flush_tlb_pending(tb->mm, tb->tlb_nr,
+					      &tb->vaddrs[0]);
+#else
+			__flush_tlb_pending(CTX_HWBITS(tb->mm->context),
+					    tb->tlb_nr, &tb->vaddrs[0]);
+#endif
+		}
+	}
+
+	tb->tlb_nr = 0;
+
+out:
+	put_cpu_var(tlb_batch);
+}
+
+void arch_enter_lazy_mmu_mode(void)
+{
+	struct tlb_batch *tb = this_cpu_ptr(&tlb_batch);
+
+	tb->active = 1;
+}
+
+void arch_leave_lazy_mmu_mode(void)
+{
+	struct tlb_batch *tb = this_cpu_ptr(&tlb_batch);
+
+	if (tb->tlb_nr)
+		flush_tlb_pending();
+	tb->active = 0;
+}
+
+static void tlb_batch_add_one(struct mm_struct *mm, unsigned long vaddr,
+			      bool exec)
+{
+	struct tlb_batch *tb = &get_cpu_var(tlb_batch);
+	unsigned long nr;
+
+	vaddr &= PAGE_MASK;
+	if (exec)
+		vaddr |= 0x1UL;
+
+	nr = tb->tlb_nr;
+
+	if (unlikely(nr != 0 && mm != tb->mm)) {
+		flush_tlb_pending();
+		nr = 0;
+	}
+
+	if (!tb->active) {
+		flush_tsb_user_page(mm, vaddr);
+		global_flush_tlb_page(mm, vaddr);
+		goto out;
+	}
+
+	if (nr == 0)
+		tb->mm = mm;
+
+	tb->vaddrs[nr] = vaddr;
+	tb->tlb_nr = ++nr;
+	if (nr >= TLB_BATCH_NR)
+		flush_tlb_pending();
+
+out:
+	put_cpu_var(tlb_batch);
+}
+
+void tlb_batch_add(struct mm_struct *mm, unsigned long vaddr,
+		   pte_t *ptep, pte_t orig, int fullmm)
+{
+	if (tlb_type != hypervisor &&
+	    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 (!fullmm)
+		tlb_batch_add_one(mm, vaddr, pte_exec(orig));
+}
+
+#ifdef CONFIG_TRANSPARENT_HUGEPAGE
+static void tlb_batch_pmd_scan(struct mm_struct *mm, unsigned long vaddr,
+			       pmd_t pmd)
+{
+	unsigned long end;
+	pte_t *pte;
+
+	pte = pte_offset_map(&pmd, vaddr);
+	end = vaddr + HPAGE_SIZE;
+	while (vaddr < end) {
+		if (pte_val(*pte) & _PAGE_VALID) {
+			bool exec = pte_exec(*pte);
+
+			tlb_batch_add_one(mm, vaddr, exec);
+		}
+		pte++;
+		vaddr += PAGE_SIZE;
+	}
+	pte_unmap(pte);
+}
+
+void set_pmd_at(struct mm_struct *mm, unsigned long addr,
+		pmd_t *pmdp, pmd_t pmd)
+{
+	pmd_t orig = *pmdp;
+
+	*pmdp = pmd;
+
+	if (mm == &init_mm)
+		return;
+
+	if ((pmd_val(pmd) ^ pmd_val(orig)) & _PAGE_PMD_HUGE) {
+		if (pmd_val(pmd) & _PAGE_PMD_HUGE)
+			mm->context.huge_pte_count++;
+		else
+			mm->context.huge_pte_count--;
+
+		/* Do not try to allocate the TSB hash table if we
+		 * don't have one already.  We have various locks held
+		 * and thus we'll end up doing a GFP_KERNEL allocation
+		 * in an atomic context.
+		 *
+		 * Instead, we let the first TLB miss on a hugepage
+		 * take care of this.
+		 */
+	}
+
+	if (!pmd_none(orig)) {
+		addr &= HPAGE_MASK;
+		if (pmd_trans_huge(orig)) {
+			pte_t orig_pte = __pte(pmd_val(orig));
+			bool exec = pte_exec(orig_pte);
+
+			tlb_batch_add_one(mm, addr, exec);
+			tlb_batch_add_one(mm, addr + REAL_HPAGE_SIZE, exec);
+		} else {
+			tlb_batch_pmd_scan(mm, addr, orig);
+		}
+	}
+}
+
+void pmdp_invalidate(struct vm_area_struct *vma, unsigned long address,
+		     pmd_t *pmdp)
+{
+	pmd_t entry = *pmdp;
+
+	pmd_val(entry) &= ~_PAGE_VALID;
+
+	set_pmd_at(vma->vm_mm, address, pmdp, entry);
+	flush_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
+}
+
+void pgtable_trans_huge_deposit(struct mm_struct *mm, pmd_t *pmdp,
+				pgtable_t pgtable)
+{
+	struct list_head *lh = (struct list_head *) pgtable;
+
+	assert_spin_locked(&mm->page_table_lock);
+
+	/* FIFO */
+	if (!pmd_huge_pte(mm, pmdp))
+		INIT_LIST_HEAD(lh);
+	else
+		list_add(lh, (struct list_head *) pmd_huge_pte(mm, pmdp));
+	pmd_huge_pte(mm, pmdp) = pgtable;
+}
+
+pgtable_t pgtable_trans_huge_withdraw(struct mm_struct *mm, pmd_t *pmdp)
+{
+	struct list_head *lh;
+	pgtable_t pgtable;
+
+	assert_spin_locked(&mm->page_table_lock);
+
+	/* FIFO */
+	pgtable = pmd_huge_pte(mm, pmdp);
+	lh = (struct list_head *) pgtable;
+	if (list_empty(lh))
+		pmd_huge_pte(mm, pmdp) = NULL;
+	else {
+		pmd_huge_pte(mm, pmdp) = (pgtable_t) lh->next;
+		list_del(lh);
+	}
+	pte_val(pgtable[0]) = 0;
+	pte_val(pgtable[1]) = 0;
+
+	return pgtable;
+}
+#endif /* CONFIG_TRANSPARENT_HUGEPAGE */

arch/sparc/mm/tsb.c

@@ -0,0 +1,538 @@
+/* arch/sparc64/mm/tsb.c
+ *
+ * Copyright (C) 2006, 2008 David S. Miller <davem@davemloft.net>
+ */
+
+#include <linux/kernel.h>
+#include <linux/preempt.h>
+#include <linux/slab.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/mmu_context.h>
+#include <asm/setup.h>
+#include <asm/tsb.h>
+#include <asm/tlb.h>
+#include <asm/oplib.h>
+
+extern struct tsb swapper_tsb[KERNEL_TSB_NENTRIES];
+
+static inline unsigned long tsb_hash(unsigned long vaddr, unsigned long hash_shift, unsigned long nentries)
+{
+	vaddr >>= hash_shift;
+	return vaddr & (nentries - 1);
+}
+
+static inline int tag_compare(unsigned long tag, unsigned long vaddr)
+{
+	return (tag == (vaddr >> 22));
+}
+
+/* TSB flushes need only occur on the processor initiating the address
+ * space modification, not on each cpu the address space has run on.
+ * Only the TLB flush needs that treatment.
+ */
+
+void flush_tsb_kernel_range(unsigned long start, unsigned long end)
+{
+	unsigned long v;
+
+	for (v = start; v < end; v += PAGE_SIZE) {
+		unsigned long hash = tsb_hash(v, PAGE_SHIFT,
+					      KERNEL_TSB_NENTRIES);
+		struct tsb *ent = &swapper_tsb[hash];
+
+		if (tag_compare(ent->tag, v))
+			ent->tag = (1UL << TSB_TAG_INVALID_BIT);
+	}
+}
+
+static void __flush_tsb_one_entry(unsigned long tsb, unsigned long v,
+				  unsigned long hash_shift,
+				  unsigned long nentries)
+{
+	unsigned long tag, ent, hash;
+
+	v &= ~0x1UL;
+	hash = tsb_hash(v, hash_shift, nentries);
+	ent = tsb + (hash * sizeof(struct tsb));
+	tag = (v >> 22UL);
+
+	tsb_flush(ent, tag);
+}
+
+static void __flush_tsb_one(struct tlb_batch *tb, unsigned long hash_shift,
+			    unsigned long tsb, unsigned long nentries)
+{
+	unsigned long i;
+
+	for (i = 0; i < tb->tlb_nr; i++)
+		__flush_tsb_one_entry(tsb, tb->vaddrs[i], hash_shift, nentries);
+}
+
+void flush_tsb_user(struct tlb_batch *tb)
+{
+	struct mm_struct *mm = tb->mm;
+	unsigned long nentries, base, flags;
+
+	spin_lock_irqsave(&mm->context.lock, flags);
+
+	base = (unsigned long) mm->context.tsb_block[MM_TSB_BASE].tsb;
+	nentries = mm->context.tsb_block[MM_TSB_BASE].tsb_nentries;
+	if (tlb_type == cheetah_plus || tlb_type == hypervisor)
+		base = __pa(base);
+	__flush_tsb_one(tb, PAGE_SHIFT, base, nentries);
+
+#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
+	if (mm->context.tsb_block[MM_TSB_HUGE].tsb) {
+		base = (unsigned long) mm->context.tsb_block[MM_TSB_HUGE].tsb;
+		nentries = mm->context.tsb_block[MM_TSB_HUGE].tsb_nentries;
+		if (tlb_type == cheetah_plus || tlb_type == hypervisor)
+			base = __pa(base);
+		__flush_tsb_one(tb, REAL_HPAGE_SHIFT, base, nentries);
+	}
+#endif
+	spin_unlock_irqrestore(&mm->context.lock, flags);
+}
+
+void flush_tsb_user_page(struct mm_struct *mm, unsigned long vaddr)
+{
+	unsigned long nentries, base, flags;
+
+	spin_lock_irqsave(&mm->context.lock, flags);
+
+	base = (unsigned long) mm->context.tsb_block[MM_TSB_BASE].tsb;
+	nentries = mm->context.tsb_block[MM_TSB_BASE].tsb_nentries;
+	if (tlb_type == cheetah_plus || tlb_type == hypervisor)
+		base = __pa(base);
+	__flush_tsb_one_entry(base, vaddr, PAGE_SHIFT, nentries);
+
+#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
+	if (mm->context.tsb_block[MM_TSB_HUGE].tsb) {
+		base = (unsigned long) mm->context.tsb_block[MM_TSB_HUGE].tsb;
+		nentries = mm->context.tsb_block[MM_TSB_HUGE].tsb_nentries;
+		if (tlb_type == cheetah_plus || tlb_type == hypervisor)
+			base = __pa(base);
+		__flush_tsb_one_entry(base, vaddr, REAL_HPAGE_SHIFT, nentries);
+	}
+#endif
+	spin_unlock_irqrestore(&mm->context.lock, flags);
+}
+
+#define HV_PGSZ_IDX_BASE	HV_PGSZ_IDX_8K
+#define HV_PGSZ_MASK_BASE	HV_PGSZ_MASK_8K
+
+#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
+#define HV_PGSZ_IDX_HUGE	HV_PGSZ_IDX_4MB
+#define HV_PGSZ_MASK_HUGE	HV_PGSZ_MASK_4MB
+#endif
+
+static void setup_tsb_params(struct mm_struct *mm, unsigned long tsb_idx, unsigned long tsb_bytes)
+{
+	unsigned long tsb_reg, base, tsb_paddr;
+	unsigned long page_sz, tte;
+
+	mm->context.tsb_block[tsb_idx].tsb_nentries =
+		tsb_bytes / sizeof(struct tsb);
+
+	switch (tsb_idx) {
+	case MM_TSB_BASE:
+		base = TSBMAP_8K_BASE;
+		break;
+#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
+	case MM_TSB_HUGE:
+		base = TSBMAP_4M_BASE;
+		break;
+#endif
+	default:
+		BUG();
+	}
+
+	tte = pgprot_val(PAGE_KERNEL_LOCKED);
+	tsb_paddr = __pa(mm->context.tsb_block[tsb_idx].tsb);
+	BUG_ON(tsb_paddr & (tsb_bytes - 1UL));
+
+	/* Use the smallest page size that can map the whole TSB
+	 * in one TLB entry.
+	 */
+	switch (tsb_bytes) {
+	case 8192 << 0:
+		tsb_reg = 0x0UL;
+#ifdef DCACHE_ALIASING_POSSIBLE
+		base += (tsb_paddr & 8192);
+#endif
+		page_sz = 8192;
+		break;
+
+	case 8192 << 1:
+		tsb_reg = 0x1UL;
+		page_sz = 64 * 1024;
+		break;
+
+	case 8192 << 2:
+		tsb_reg = 0x2UL;
+		page_sz = 64 * 1024;
+		break;
+
+	case 8192 << 3:
+		tsb_reg = 0x3UL;
+		page_sz = 64 * 1024;
+		break;
+
+	case 8192 << 4:
+		tsb_reg = 0x4UL;
+		page_sz = 512 * 1024;
+		break;
+
+	case 8192 << 5:
+		tsb_reg = 0x5UL;
+		page_sz = 512 * 1024;
+		break;
+
+	case 8192 << 6:
+		tsb_reg = 0x6UL;
+		page_sz = 512 * 1024;
+		break;
+
+	case 8192 << 7:
+		tsb_reg = 0x7UL;
+		page_sz = 4 * 1024 * 1024;
+		break;
+
+	default:
+		printk(KERN_ERR "TSB[%s:%d]: Impossible TSB size %lu, killing process.\n",
+		       current->comm, current->pid, tsb_bytes);
+		do_exit(SIGSEGV);
+	}
+	tte |= pte_sz_bits(page_sz);
+
+	if (tlb_type == cheetah_plus || tlb_type == hypervisor) {
+		/* Physical mapping, no locked TLB entry for TSB.  */
+		tsb_reg |= tsb_paddr;
+
+		mm->context.tsb_block[tsb_idx].tsb_reg_val = tsb_reg;
+		mm->context.tsb_block[tsb_idx].tsb_map_vaddr = 0;
+		mm->context.tsb_block[tsb_idx].tsb_map_pte = 0;
+	} else {
+		tsb_reg |= base;
+		tsb_reg |= (tsb_paddr & (page_sz - 1UL));
+		tte |= (tsb_paddr & ~(page_sz - 1UL));
+
+		mm->context.tsb_block[tsb_idx].tsb_reg_val = tsb_reg;
+		mm->context.tsb_block[tsb_idx].tsb_map_vaddr = base;
+		mm->context.tsb_block[tsb_idx].tsb_map_pte = tte;
+	}
+
+	/* Setup the Hypervisor TSB descriptor.  */
+	if (tlb_type == hypervisor) {
+		struct hv_tsb_descr *hp = &mm->context.tsb_descr[tsb_idx];
+
+		switch (tsb_idx) {
+		case MM_TSB_BASE:
+			hp->pgsz_idx = HV_PGSZ_IDX_BASE;
+			break;
+#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
+		case MM_TSB_HUGE:
+			hp->pgsz_idx = HV_PGSZ_IDX_HUGE;
+			break;
+#endif
+		default:
+			BUG();
+		}
+		hp->assoc = 1;
+		hp->num_ttes = tsb_bytes / 16;
+		hp->ctx_idx = 0;
+		switch (tsb_idx) {
+		case MM_TSB_BASE:
+			hp->pgsz_mask = HV_PGSZ_MASK_BASE;
+			break;
+#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
+		case MM_TSB_HUGE:
+			hp->pgsz_mask = HV_PGSZ_MASK_HUGE;
+			break;
+#endif
+		default:
+			BUG();
+		}
+		hp->tsb_base = tsb_paddr;
+		hp->resv = 0;
+	}
+}
+
+struct kmem_cache *pgtable_cache __read_mostly;
+
+static struct kmem_cache *tsb_caches[8] __read_mostly;
+
+static const char *tsb_cache_names[8] = {
+	"tsb_8KB",
+	"tsb_16KB",
+	"tsb_32KB",
+	"tsb_64KB",
+	"tsb_128KB",
+	"tsb_256KB",
+	"tsb_512KB",
+	"tsb_1MB",
+};
+
+void __init pgtable_cache_init(void)
+{
+	unsigned long i;
+
+	pgtable_cache = kmem_cache_create("pgtable_cache",
+					  PAGE_SIZE, PAGE_SIZE,
+					  0,
+					  _clear_page);
+	if (!pgtable_cache) {
+		prom_printf("pgtable_cache_init(): Could not create!\n");
+		prom_halt();
+	}
+
+	for (i = 0; i < ARRAY_SIZE(tsb_cache_names); i++) {
+		unsigned long size = 8192 << i;
+		const char *name = tsb_cache_names[i];
+
+		tsb_caches[i] = kmem_cache_create(name,
+						  size, size,
+						  0, NULL);
+		if (!tsb_caches[i]) {
+			prom_printf("Could not create %s cache\n", name);
+			prom_halt();
+		}
+	}
+}
+
+int sysctl_tsb_ratio = -2;
+
+static unsigned long tsb_size_to_rss_limit(unsigned long new_size)
+{
+	unsigned long num_ents = (new_size / sizeof(struct tsb));
+
+	if (sysctl_tsb_ratio < 0)
+		return num_ents - (num_ents >> -sysctl_tsb_ratio);
+	else
+		return num_ents + (num_ents >> sysctl_tsb_ratio);
+}
+
+/* When the RSS of an address space exceeds tsb_rss_limit for a TSB,
+ * do_sparc64_fault() invokes this routine to try and grow it.
+ *
+ * When we reach the maximum TSB size supported, we stick ~0UL into
+ * tsb_rss_limit for that TSB so the grow checks in do_sparc64_fault()
+ * will not trigger any longer.
+ *
+ * The TSB can be anywhere from 8K to 1MB in size, in increasing powers
+ * of two.  The TSB must be aligned to it's size, so f.e. a 512K TSB
+ * must be 512K aligned.  It also must be physically contiguous, so we
+ * cannot use vmalloc().
+ *
+ * The idea here is to grow the TSB when the RSS of the process approaches
+ * the number of entries that the current TSB can hold at once.  Currently,
+ * we trigger when the RSS hits 3/4 of the TSB capacity.
+ */
+void tsb_grow(struct mm_struct *mm, unsigned long tsb_index, unsigned long rss)
+{
+	unsigned long max_tsb_size = 1 * 1024 * 1024;
+	unsigned long new_size, old_size, flags;
+	struct tsb *old_tsb, *new_tsb;
+	unsigned long new_cache_index, old_cache_index;
+	unsigned long new_rss_limit;
+	gfp_t gfp_flags;
+
+	if (max_tsb_size > (PAGE_SIZE << MAX_ORDER))
+		max_tsb_size = (PAGE_SIZE << MAX_ORDER);
+
+	new_cache_index = 0;
+	for (new_size = 8192; new_size < max_tsb_size; new_size <<= 1UL) {
+		new_rss_limit = tsb_size_to_rss_limit(new_size);
+		if (new_rss_limit > rss)
+			break;
+		new_cache_index++;
+	}
+
+	if (new_size == max_tsb_size)
+		new_rss_limit = ~0UL;
+
+retry_tsb_alloc:
+	gfp_flags = GFP_KERNEL;
+	if (new_size > (PAGE_SIZE * 2))
+		gfp_flags |= __GFP_NOWARN | __GFP_NORETRY;
+
+	new_tsb = kmem_cache_alloc_node(tsb_caches[new_cache_index],
+					gfp_flags, numa_node_id());
+	if (unlikely(!new_tsb)) {
+		/* Not being able to fork due to a high-order TSB
+		 * allocation failure is very bad behavior.  Just back
+		 * down to a 0-order allocation and force no TSB
+		 * growing for this address space.
+		 */
+		if (mm->context.tsb_block[tsb_index].tsb == NULL &&
+		    new_cache_index > 0) {
+			new_cache_index = 0;
+			new_size = 8192;
+			new_rss_limit = ~0UL;
+			goto retry_tsb_alloc;
+		}
+
+		/* If we failed on a TSB grow, we are under serious
+		 * memory pressure so don't try to grow any more.
+		 */
+		if (mm->context.tsb_block[tsb_index].tsb != NULL)
+			mm->context.tsb_block[tsb_index].tsb_rss_limit = ~0UL;
+		return;
+	}
+
+	/* Mark all tags as invalid.  */
+	tsb_init(new_tsb, new_size);
+
+	/* Ok, we are about to commit the changes.  If we are
+	 * growing an existing TSB the locking is very tricky,
+	 * so WATCH OUT!
+	 *
+	 * We have to hold mm->context.lock while committing to the
+	 * new TSB, this synchronizes us with processors in
+	 * flush_tsb_user() and switch_mm() for this address space.
+	 *
+	 * But even with that lock held, processors run asynchronously
+	 * accessing the old TSB via TLB miss handling.  This is OK
+	 * because those actions are just propagating state from the
+	 * Linux page tables into the TSB, page table mappings are not
+	 * being changed.  If a real fault occurs, the processor will
+	 * synchronize with us when it hits flush_tsb_user(), this is
+	 * also true for the case where vmscan is modifying the page
+	 * tables.  The only thing we need to be careful with is to
+	 * skip any locked TSB entries during copy_tsb().
+	 *
+	 * When we finish committing to the new TSB, we have to drop
+	 * the lock and ask all other cpus running this address space
+	 * to run tsb_context_switch() to see the new TSB table.
+	 */
+	spin_lock_irqsave(&mm->context.lock, flags);
+
+	old_tsb = mm->context.tsb_block[tsb_index].tsb;
+	old_cache_index =
+		(mm->context.tsb_block[tsb_index].tsb_reg_val & 0x7UL);
+	old_size = (mm->context.tsb_block[tsb_index].tsb_nentries *
+		    sizeof(struct tsb));
+
+
+	/* Handle multiple threads trying to grow the TSB at the same time.
+	 * One will get in here first, and bump the size and the RSS limit.
+	 * The others will get in here next and hit this check.
+	 */
+	if (unlikely(old_tsb &&
+		     (rss < mm->context.tsb_block[tsb_index].tsb_rss_limit))) {
+		spin_unlock_irqrestore(&mm->context.lock, flags);
+
+		kmem_cache_free(tsb_caches[new_cache_index], new_tsb);
+		return;
+	}
+
+	mm->context.tsb_block[tsb_index].tsb_rss_limit = new_rss_limit;
+
+	if (old_tsb) {
+		extern void copy_tsb(unsigned long old_tsb_base,
+				     unsigned long old_tsb_size,
+				     unsigned long new_tsb_base,
+				     unsigned long new_tsb_size);
+		unsigned long old_tsb_base = (unsigned long) old_tsb;
+		unsigned long new_tsb_base = (unsigned long) new_tsb;
+
+		if (tlb_type == cheetah_plus || tlb_type == hypervisor) {
+			old_tsb_base = __pa(old_tsb_base);
+			new_tsb_base = __pa(new_tsb_base);
+		}
+		copy_tsb(old_tsb_base, old_size, new_tsb_base, new_size);
+	}
+
+	mm->context.tsb_block[tsb_index].tsb = new_tsb;
+	setup_tsb_params(mm, tsb_index, new_size);
+
+	spin_unlock_irqrestore(&mm->context.lock, flags);
+
+	/* If old_tsb is NULL, we're being invoked for the first time
+	 * from init_new_context().
+	 */
+	if (old_tsb) {
+		/* Reload it on the local cpu.  */
+		tsb_context_switch(mm);
+
+		/* Now force other processors to do the same.  */
+		preempt_disable();
+		smp_tsb_sync(mm);
+		preempt_enable();
+
+		/* Now it is safe to free the old tsb.  */
+		kmem_cache_free(tsb_caches[old_cache_index], old_tsb);
+	}
+}
+
+int init_new_context(struct task_struct *tsk, struct mm_struct *mm)
+{
+#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
+	unsigned long huge_pte_count;
+#endif
+	unsigned int i;
+
+	spin_lock_init(&mm->context.lock);
+
+	mm->context.sparc64_ctx_val = 0UL;
+
+#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
+	/* We reset it to zero because the fork() page copying
+	 * will re-increment the counters as the parent PTEs are
+	 * copied into the child address space.
+	 */
+	huge_pte_count = mm->context.huge_pte_count;
+	mm->context.huge_pte_count = 0;
+#endif
+
+	/* copy_mm() copies over the parent's mm_struct before calling
+	 * us, so we need to zero out the TSB pointer or else tsb_grow()
+	 * will be confused and think there is an older TSB to free up.
+	 */
+	for (i = 0; i < MM_NUM_TSBS; i++)
+		mm->context.tsb_block[i].tsb = NULL;
+
+	/* If this is fork, inherit the parent's TSB size.  We would
+	 * grow it to that size on the first page fault anyways.
+	 */
+	tsb_grow(mm, MM_TSB_BASE, get_mm_rss(mm));
+
+#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
+	if (unlikely(huge_pte_count))
+		tsb_grow(mm, MM_TSB_HUGE, huge_pte_count);
+#endif
+
+	if (unlikely(!mm->context.tsb_block[MM_TSB_BASE].tsb))
+		return -ENOMEM;
+
+	return 0;
+}
+
+static void tsb_destroy_one(struct tsb_config *tp)
+{
+	unsigned long cache_index;
+
+	if (!tp->tsb)
+		return;
+	cache_index = tp->tsb_reg_val & 0x7UL;
+	kmem_cache_free(tsb_caches[cache_index], tp->tsb);
+	tp->tsb = NULL;
+	tp->tsb_reg_val = 0UL;
+}
+
+void destroy_context(struct mm_struct *mm)
+{
+	unsigned long flags, i;
+
+	for (i = 0; i < MM_NUM_TSBS; i++)
+		tsb_destroy_one(&mm->context.tsb_block[i]);
+
+	spin_lock_irqsave(&ctx_alloc_lock, flags);
+
+	if (CTX_VALID(mm->context)) {
+		unsigned long nr = CTX_NRBITS(mm->context);
+		mmu_context_bmap[nr>>6] &= ~(1UL << (nr & 63));
+	}
+
+	spin_unlock_irqrestore(&ctx_alloc_lock, flags);
+}

arch/sparc/mm/tsunami.S

@@ -0,0 +1,131 @@
+/*
+ * tsunami.S: High speed MicroSparc-I mmu/cache operations.
+ *
+ * Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu)
+ */
+
+#include <asm/ptrace.h>
+#include <asm/asm-offsets.h>
+#include <asm/psr.h>
+#include <asm/asi.h>
+#include <asm/page.h>
+#include <asm/pgtsrmmu.h>
+
+	.text
+	.align	4
+
+	.globl	tsunami_flush_cache_all, tsunami_flush_cache_mm
+	.globl	tsunami_flush_cache_range, tsunami_flush_cache_page
+	.globl	tsunami_flush_page_to_ram, tsunami_flush_page_for_dma
+	.globl	tsunami_flush_sig_insns
+	.globl	tsunami_flush_tlb_all, tsunami_flush_tlb_mm
+	.globl	tsunami_flush_tlb_range, tsunami_flush_tlb_page
+
+	/* Sliiick... */
+tsunami_flush_cache_page:
+tsunami_flush_cache_range:
+	ld	[%o0 + VMA_VM_MM], %o0
+tsunami_flush_cache_mm:
+	ld	[%o0 + AOFF_mm_context], %g2
+	cmp	%g2, -1
+	be	tsunami_flush_cache_out
+tsunami_flush_cache_all:
+	WINDOW_FLUSH(%g4, %g5)
+tsunami_flush_page_for_dma:
+	sta	%g0, [%g0] ASI_M_IC_FLCLEAR
+	sta	%g0, [%g0] ASI_M_DC_FLCLEAR
+tsunami_flush_cache_out:
+tsunami_flush_page_to_ram:
+	retl
+	 nop
+
+tsunami_flush_sig_insns:
+	flush	%o1
+	retl
+	 flush	%o1 + 4
+
+	/* More slick stuff... */
+tsunami_flush_tlb_range:
+	ld	[%o0 + VMA_VM_MM], %o0
+tsunami_flush_tlb_mm:
+	ld	[%o0 + AOFF_mm_context], %g2
+	cmp	%g2, -1
+	be	tsunami_flush_tlb_out
+tsunami_flush_tlb_all:
+	 mov	0x400, %o1
+	sta	%g0, [%o1] ASI_M_FLUSH_PROBE
+	nop
+	nop
+	nop
+	nop
+	nop
+tsunami_flush_tlb_out:
+	retl
+	 nop
+
+	/* This one can be done in a fine grained manner... */
+tsunami_flush_tlb_page:
+	ld	[%o0 + VMA_VM_MM], %o0
+	mov	SRMMU_CTX_REG, %g1
+	ld	[%o0 + AOFF_mm_context], %o3
+	andn	%o1, (PAGE_SIZE - 1), %o1
+	cmp	%o3, -1
+	be	tsunami_flush_tlb_page_out
+	 lda	[%g1] ASI_M_MMUREGS, %g5
+	sta	%o3, [%g1] ASI_M_MMUREGS
+	sta	%g0, [%o1] ASI_M_FLUSH_PROBE
+	nop
+	nop
+	nop
+	nop
+	nop
+tsunami_flush_tlb_page_out:
+	retl
+	 sta	%g5, [%g1] ASI_M_MMUREGS
+
+#define MIRROR_BLOCK(dst, src, offset, t0, t1, t2, t3) \
+	ldd	[src + offset + 0x18], t0; \
+	std	t0, [dst + offset + 0x18]; \
+	ldd	[src + offset + 0x10], t2; \
+	std	t2, [dst + offset + 0x10]; \
+	ldd	[src + offset + 0x08], t0; \
+	std	t0, [dst + offset + 0x08]; \
+	ldd	[src + offset + 0x00], t2; \
+	std	t2, [dst + offset + 0x00];
+
+tsunami_copy_1page:
+/* NOTE: This routine has to be shorter than 70insns --jj */
+	or	%g0, (PAGE_SIZE >> 8), %g1
+1:
+	MIRROR_BLOCK(%o0, %o1, 0x00, %o2, %o3, %o4, %o5)
+	MIRROR_BLOCK(%o0, %o1, 0x20, %o2, %o3, %o4, %o5)
+	MIRROR_BLOCK(%o0, %o1, 0x40, %o2, %o3, %o4, %o5)
+	MIRROR_BLOCK(%o0, %o1, 0x60, %o2, %o3, %o4, %o5)
+	MIRROR_BLOCK(%o0, %o1, 0x80, %o2, %o3, %o4, %o5)
+	MIRROR_BLOCK(%o0, %o1, 0xa0, %o2, %o3, %o4, %o5)
+	MIRROR_BLOCK(%o0, %o1, 0xc0, %o2, %o3, %o4, %o5)
+	MIRROR_BLOCK(%o0, %o1, 0xe0, %o2, %o3, %o4, %o5)
+	subcc	%g1, 1, %g1
+	add	%o0, 0x100, %o0
+	bne	1b
+	 add	%o1, 0x100, %o1
+
+	.globl	tsunami_setup_blockops
+tsunami_setup_blockops:
+	sethi	%hi(__copy_1page), %o0
+	or	%o0, %lo(__copy_1page), %o0
+	sethi	%hi(tsunami_copy_1page), %o1
+	or	%o1, %lo(tsunami_copy_1page), %o1
+	sethi	%hi(tsunami_setup_blockops), %o2
+	or	%o2, %lo(tsunami_setup_blockops), %o2
+	ld	[%o1], %o4
+1:	add	%o1, 4, %o1
+	st	%o4, [%o0]
+	add	%o0, 4, %o0
+	cmp	%o1, %o2
+	bne	1b
+	ld	[%o1], %o4
+	sta	%g0, [%g0] ASI_M_IC_FLCLEAR
+	sta	%g0, [%g0] ASI_M_DC_FLCLEAR
+	retl
+	 nop

arch/sparc/mm/ultra.S

@@ -0,0 +1,878 @@
+/*
+ * ultra.S: Don't expand these all over the place...
+ *
+ * Copyright (C) 1997, 2000, 2008 David S. Miller (davem@davemloft.net)
+ */
+
+#include <asm/asi.h>
+#include <asm/pgtable.h>
+#include <asm/page.h>
+#include <asm/spitfire.h>
+#include <asm/mmu_context.h>
+#include <asm/mmu.h>
+#include <asm/pil.h>
+#include <asm/head.h>
+#include <asm/thread_info.h>
+#include <asm/cacheflush.h>
+#include <asm/hypervisor.h>
+#include <asm/cpudata.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:		/* 18 insns */
+	/* %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
+	sethi		%hi(KERNBASE), %g3
+	flush		%g3
+	retl
+	 nop
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+
+	.align		32
+	.globl		__flush_tlb_page
+__flush_tlb_page:	/* 22 insns */
+	/* %o0 = context, %o1 = vaddr */
+	rdpr		%pstate, %g7
+	andn		%g7, PSTATE_IE, %g2
+	wrpr		%g2, %pstate
+	mov		SECONDARY_CONTEXT, %o4
+	ldxa		[%o4] ASI_DMMU, %g2
+	stxa		%o0, [%o4] ASI_DMMU
+	andcc		%o1, 1, %g0
+	andn		%o1, 1, %o3
+	be,pn		%icc, 1f
+	 or		%o3, 0x10, %o3
+	stxa		%g0, [%o3] ASI_IMMU_DEMAP
+1:	stxa		%g0, [%o3] ASI_DMMU_DEMAP
+	membar		#Sync
+	stxa		%g2, [%o4] ASI_DMMU
+	sethi		%hi(KERNBASE), %o4
+	flush		%o4
+	retl
+	 wrpr		%g7, 0x0, %pstate
+	nop
+	nop
+	nop
+	nop
+
+	.align		32
+	.globl		__flush_tlb_pending
+__flush_tlb_pending:	/* 26 insns */
+	/* %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
+	sethi		%hi(KERNBASE), %o4
+	flush		%o4
+	retl
+	 wrpr		%g7, 0x0, %pstate
+	nop
+	nop
+	nop
+	nop
+
+	.align		32
+	.globl		__flush_tlb_kernel_range
+__flush_tlb_kernel_range:	/* 16 insns */
+	/* %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:	sethi		%hi(KERNBASE), %o3
+	flush		%o3
+	retl
+	 nop
+	nop
+
+__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
+	sethi		%hi(KERNBASE), %o1
+	flush		%o1
+	retl
+	 wrpr		%g1, 0, %pstate
+
+/*
+ * The following code flushes one page_size worth.
+ */
+	.section .kprobes.text, "ax"
+	.align		32
+	.globl		__flush_icache_page
+__flush_icache_page:	/* %o0 = phys_page */
+	srlx		%o0, PAGE_SHIFT, %o0
+	sethi		%hi(PAGE_OFFSET), %g1
+	sllx		%o0, PAGE_SHIFT, %o0
+	sethi		%hi(PAGE_SIZE), %g2
+	ldx		[%g1 + %lo(PAGE_OFFSET)], %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
+
+	/* This routine is Spitfire specific so the hardcoded
+	 * D-cache size and line-size are OK.
+	 */
+	.align		64
+	.globl		__flush_dcache_page
+__flush_dcache_page:	/* %o0=kaddr, %o1=flush_icache */
+	sethi		%hi(PAGE_OFFSET), %g1
+	ldx		[%g1 + %lo(PAGE_OFFSET)], %g1
+	sub		%o0, %g1, %o0			! physical address
+	srlx		%o0, 11, %o0			! make D-cache TAG
+	sethi		%hi(1 << 14), %o2		! D-cache size
+	sub		%o2, (1 << 5), %o2		! D-cache line size
+1:	ldxa		[%o2] ASI_DCACHE_TAG, %o3	! load D-cache TAG
+	andcc		%o3, DTAG_MASK, %g0		! Valid?
+	be,pn		%xcc, 2f			! Nope, branch
+	 andn		%o3, DTAG_MASK, %o3		! Clear valid bits
+	cmp		%o3, %o0			! TAG match?
+	bne,pt		%xcc, 2f			! Nope, branch
+	 nop
+	stxa		%g0, [%o2] ASI_DCACHE_TAG	! Invalidate TAG
+	membar		#Sync
+2:	brnz,pt		%o2, 1b
+	 sub		%o2, (1 << 5), %o2		! D-cache line size
+
+	/* 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
+
+#endif /* DCACHE_ALIASING_POSSIBLE */
+
+	.previous
+
+	/* Cheetah specific versions, patched at boot time. */
+__cheetah_flush_tlb_mm: /* 19 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
+	srlx		%g2, CTX_PGSZ1_NUC_SHIFT, %o1
+	sllx		%o1, CTX_PGSZ1_NUC_SHIFT, %o1
+	or		%o0, %o1, %o0	/* Preserve nucleus page size fields */
+	stxa		%o0, [%o2] ASI_DMMU
+	stxa		%g0, [%g3] ASI_DMMU_DEMAP
+	stxa		%g0, [%g3] ASI_IMMU_DEMAP
+	stxa		%g2, [%o2] ASI_DMMU
+	sethi		%hi(KERNBASE), %o2
+	flush		%o2
+	wrpr		%g0, 0, %tl
+	retl
+	 wrpr		%g7, 0x0, %pstate
+
+__cheetah_flush_tlb_page:	/* 22 insns */
+	/* %o0 = context, %o1 = vaddr */
+	rdpr		%pstate, %g7
+	andn		%g7, PSTATE_IE, %g2
+	wrpr		%g2, 0x0, %pstate
+	wrpr		%g0, 1, %tl
+	mov		PRIMARY_CONTEXT, %o4
+	ldxa		[%o4] ASI_DMMU, %g2
+	srlx		%g2, CTX_PGSZ1_NUC_SHIFT, %o3
+	sllx		%o3, CTX_PGSZ1_NUC_SHIFT, %o3
+	or		%o0, %o3, %o0	/* Preserve nucleus page size fields */
+	stxa		%o0, [%o4] ASI_DMMU
+	andcc		%o1, 1, %g0
+	be,pn		%icc, 1f
+	 andn		%o1, 1, %o3
+	stxa		%g0, [%o3] ASI_IMMU_DEMAP
+1:	stxa		%g0, [%o3] ASI_DMMU_DEMAP	
+	membar		#Sync
+	stxa		%g2, [%o4] ASI_DMMU
+	sethi		%hi(KERNBASE), %o4
+	flush		%o4
+	wrpr		%g0, 0, %tl
+	retl
+	 wrpr		%g7, 0x0, %pstate
+
+__cheetah_flush_tlb_pending:	/* 27 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
+	srlx		%g2, CTX_PGSZ1_NUC_SHIFT, %o3
+	sllx		%o3, CTX_PGSZ1_NUC_SHIFT, %o3
+	or		%o0, %o3, %o0	/* Preserve nucleus page size fields */
+	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	
+	membar		#Sync
+	brnz,pt		%o1, 1b
+	 nop
+	stxa		%g2, [%o4] ASI_DMMU
+	sethi		%hi(KERNBASE), %o4
+	flush		%o4
+	wrpr		%g0, 0, %tl
+	retl
+	 wrpr		%g7, 0x0, %pstate
+
+#ifdef DCACHE_ALIASING_POSSIBLE
+__cheetah_flush_dcache_page: /* 11 insns */
+	sethi		%hi(PAGE_OFFSET), %g1
+	ldx		[%g1 + %lo(PAGE_OFFSET)], %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 */
+
+	/* Hypervisor specific versions, patched at boot time.  */
+__hypervisor_tlb_tl0_error:
+	save		%sp, -192, %sp
+	mov		%i0, %o0
+	call		hypervisor_tlbop_error
+	 mov		%i1, %o1
+	ret
+	 restore
+
+__hypervisor_flush_tlb_mm: /* 10 insns */
+	mov		%o0, %o2	/* ARG2: mmu context */
+	mov		0, %o0		/* ARG0: CPU lists unimplemented */
+	mov		0, %o1		/* ARG1: CPU lists unimplemented */
+	mov		HV_MMU_ALL, %o3	/* ARG3: flags */
+	mov		HV_FAST_MMU_DEMAP_CTX, %o5
+	ta		HV_FAST_TRAP
+	brnz,pn		%o0, __hypervisor_tlb_tl0_error
+	 mov		HV_FAST_MMU_DEMAP_CTX, %o1
+	retl
+	 nop
+
+__hypervisor_flush_tlb_page: /* 11 insns */
+	/* %o0 = context, %o1 = vaddr */
+	mov		%o0, %g2
+	mov		%o1, %o0              /* ARG0: vaddr + IMMU-bit */
+	mov		%g2, %o1	      /* ARG1: mmu context */
+	mov		HV_MMU_ALL, %o2	      /* ARG2: flags */
+	srlx		%o0, PAGE_SHIFT, %o0
+	sllx		%o0, PAGE_SHIFT, %o0
+	ta		HV_MMU_UNMAP_ADDR_TRAP
+	brnz,pn		%o0, __hypervisor_tlb_tl0_error
+	 mov		HV_MMU_UNMAP_ADDR_TRAP, %o1
+	retl
+	 nop
+
+__hypervisor_flush_tlb_pending: /* 16 insns */
+	/* %o0 = context, %o1 = nr, %o2 = vaddrs[] */
+	sllx		%o1, 3, %g1
+	mov		%o2, %g2
+	mov		%o0, %g3
+1:	sub		%g1, (1 << 3), %g1
+	ldx		[%g2 + %g1], %o0      /* ARG0: vaddr + IMMU-bit */
+	mov		%g3, %o1	      /* ARG1: mmu context */
+	mov		HV_MMU_ALL, %o2	      /* ARG2: flags */
+	srlx		%o0, PAGE_SHIFT, %o0
+	sllx		%o0, PAGE_SHIFT, %o0
+	ta		HV_MMU_UNMAP_ADDR_TRAP
+	brnz,pn		%o0, __hypervisor_tlb_tl0_error
+	 mov		HV_MMU_UNMAP_ADDR_TRAP, %o1
+	brnz,pt		%g1, 1b
+	 nop
+	retl
+	 nop
+
+__hypervisor_flush_tlb_kernel_range: /* 16 insns */
+	/* %o0=start, %o1=end */
+	cmp		%o0, %o1
+	be,pn		%xcc, 2f
+	 sethi		%hi(PAGE_SIZE), %g3
+	mov		%o0, %g1
+	sub		%o1, %g1, %g2
+	sub		%g2, %g3, %g2
+1:	add		%g1, %g2, %o0	/* ARG0: virtual address */
+	mov		0, %o1		/* ARG1: mmu context */
+	mov		HV_MMU_ALL, %o2	/* ARG2: flags */
+	ta		HV_MMU_UNMAP_ADDR_TRAP
+	brnz,pn		%o0, __hypervisor_tlb_tl0_error
+	 mov		HV_MMU_UNMAP_ADDR_TRAP, %o1
+	brnz,pt		%g2, 1b
+	 sub		%g2, %g3, %g2
+2:	retl
+	 nop
+
+#ifdef DCACHE_ALIASING_POSSIBLE
+	/* XXX Niagara and friends have an 8K cache, so no aliasing is
+	 * XXX possible, but nothing explicit in the Hypervisor API
+	 * XXX guarantees this.
+	 */
+__hypervisor_flush_dcache_page:	/* 2 insns */
+	retl
+	 nop
+#endif
+
+tlb_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		tlb_patch_one
+	 mov		19, %o2
+
+	sethi		%hi(__flush_tlb_page), %o0
+	or		%o0, %lo(__flush_tlb_page), %o0
+	sethi		%hi(__cheetah_flush_tlb_page), %o1
+	or		%o1, %lo(__cheetah_flush_tlb_page), %o1
+	call		tlb_patch_one
+	 mov		22, %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		tlb_patch_one
+	 mov		27, %o2
+
+#ifdef DCACHE_ALIASING_POSSIBLE
+	sethi		%hi(__flush_dcache_page), %o0
+	or		%o0, %lo(__flush_dcache_page), %o0
+	sethi		%hi(__cheetah_flush_dcache_page), %o1
+	or		%o1, %lo(__cheetah_flush_dcache_page), %o1
+	call		tlb_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	scratch 1
+	 *   %g2	scratch 2
+	 *   %g3	scratch 3
+	 *   %g4	scratch 4
+	 */
+	.align		32
+	.globl		xcall_flush_tlb_mm
+xcall_flush_tlb_mm:	/* 21 insns */
+	mov		PRIMARY_CONTEXT, %g2
+	ldxa		[%g2] ASI_DMMU, %g3
+	srlx		%g3, CTX_PGSZ1_NUC_SHIFT, %g4
+	sllx		%g4, CTX_PGSZ1_NUC_SHIFT, %g4
+	or		%g5, %g4, %g5	/* Preserve nucleus page size fields */
+	stxa		%g5, [%g2] ASI_DMMU
+	mov		0x40, %g4
+	stxa		%g0, [%g4] ASI_DMMU_DEMAP
+	stxa		%g0, [%g4] ASI_IMMU_DEMAP
+	stxa		%g3, [%g2] ASI_DMMU
+	retry
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+
+	.globl		xcall_flush_tlb_page
+xcall_flush_tlb_page:	/* 17 insns */
+	/* %g5=context, %g1=vaddr */
+	mov		PRIMARY_CONTEXT, %g4
+	ldxa		[%g4] ASI_DMMU, %g2
+	srlx		%g2, CTX_PGSZ1_NUC_SHIFT, %g4
+	sllx		%g4, CTX_PGSZ1_NUC_SHIFT, %g4
+	or		%g5, %g4, %g5
+	mov		PRIMARY_CONTEXT, %g4
+	stxa		%g5, [%g4] ASI_DMMU
+	andcc		%g1, 0x1, %g0
+	be,pn		%icc, 2f
+	 andn		%g1, 0x1, %g5
+	stxa		%g0, [%g5] ASI_IMMU_DEMAP
+2:	stxa		%g0, [%g5] ASI_DMMU_DEMAP
+	membar		#Sync
+	stxa		%g2, [%g4] ASI_DMMU
+	retry
+	nop
+	nop
+
+	.globl		xcall_flush_tlb_kernel_range
+xcall_flush_tlb_kernel_range:	/* 25 insns */
+	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
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+	nop
+	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:
+
+661:	rdpr		%pstate, %g2
+	wrpr		%g2, PSTATE_IG | PSTATE_AG, %pstate
+	.section	.sun4v_2insn_patch, "ax"
+	.word		661b
+	nop
+	nop
+	.previous
+
+	rdpr		%pil, %g2
+	wrpr		%g0, PIL_NORMAL_MAX, %pil
+	sethi		%hi(109f), %g7
+	b,pt		%xcc, etrap_irq
+109:	 or		%g7, %lo(109b), %g7
+#ifdef CONFIG_TRACE_IRQFLAGS
+	call		trace_hardirqs_off
+	 nop
+#endif
+	call		smp_synchronize_tick_client
+	 nop
+	b		rtrap_xcall
+	 ldx		[%sp + PTREGS_OFF + PT_V9_TSTATE], %l1
+
+	.globl		xcall_fetch_glob_regs
+xcall_fetch_glob_regs:
+	sethi		%hi(global_cpu_snapshot), %g1
+	or		%g1, %lo(global_cpu_snapshot), %g1
+	__GET_CPUID(%g2)
+	sllx		%g2, 6, %g3
+	add		%g1, %g3, %g1
+	rdpr		%tstate, %g7
+	stx		%g7, [%g1 + GR_SNAP_TSTATE]
+	rdpr		%tpc, %g7
+	stx		%g7, [%g1 + GR_SNAP_TPC]
+	rdpr		%tnpc, %g7
+	stx		%g7, [%g1 + GR_SNAP_TNPC]
+	stx		%o7, [%g1 + GR_SNAP_O7]
+	stx		%i7, [%g1 + GR_SNAP_I7]
+	/* Don't try this at home kids... */
+	rdpr		%cwp, %g3
+	sub		%g3, 1, %g7
+	wrpr		%g7, %cwp
+	mov		%i7, %g7
+	wrpr		%g3, %cwp
+	stx		%g7, [%g1 + GR_SNAP_RPC]
+	sethi		%hi(trap_block), %g7
+	or		%g7, %lo(trap_block), %g7
+	sllx		%g2, TRAP_BLOCK_SZ_SHIFT, %g2
+	add		%g7, %g2, %g7
+	ldx		[%g7 + TRAP_PER_CPU_THREAD], %g3
+	stx		%g3, [%g1 + GR_SNAP_THREAD]
+	retry
+
+	.globl		xcall_fetch_glob_pmu
+xcall_fetch_glob_pmu:
+	sethi		%hi(global_cpu_snapshot), %g1
+	or		%g1, %lo(global_cpu_snapshot), %g1
+	__GET_CPUID(%g2)
+	sllx		%g2, 6, %g3
+	add		%g1, %g3, %g1
+	rd		%pic, %g7
+	stx		%g7, [%g1 + (4 * 8)]
+	rd		%pcr, %g7
+	stx		%g7, [%g1 + (0 * 8)]
+	retry
+
+	.globl		xcall_fetch_glob_pmu_n4
+xcall_fetch_glob_pmu_n4:
+	sethi		%hi(global_cpu_snapshot), %g1
+	or		%g1, %lo(global_cpu_snapshot), %g1
+	__GET_CPUID(%g2)
+	sllx		%g2, 6, %g3
+	add		%g1, %g3, %g1
+
+	ldxa		[%g0] ASI_PIC, %g7
+	stx		%g7, [%g1 + (4 * 8)]
+	mov		0x08, %g3
+	ldxa		[%g3] ASI_PIC, %g7
+	stx		%g7, [%g1 + (5 * 8)]
+	mov		0x10, %g3
+	ldxa		[%g3] ASI_PIC, %g7
+	stx		%g7, [%g1 + (6 * 8)]
+	mov		0x18, %g3
+	ldxa		[%g3] ASI_PIC, %g7
+	stx		%g7, [%g1 + (7 * 8)]
+
+	mov		%o0, %g2
+	mov		%o1, %g3
+	mov		%o5, %g7
+
+	mov		HV_FAST_VT_GET_PERFREG, %o5
+	mov		3, %o0
+	ta		HV_FAST_TRAP
+	stx		%o1, [%g1 + (3 * 8)]
+	mov		HV_FAST_VT_GET_PERFREG, %o5
+	mov		2, %o0
+	ta		HV_FAST_TRAP
+	stx		%o1, [%g1 + (2 * 8)]
+	mov		HV_FAST_VT_GET_PERFREG, %o5
+	mov		1, %o0
+	ta		HV_FAST_TRAP
+	stx		%o1, [%g1 + (1 * 8)]
+	mov		HV_FAST_VT_GET_PERFREG, %o5
+	mov		0, %o0
+	ta		HV_FAST_TRAP
+	stx		%o1, [%g1 + (0 * 8)]
+
+	mov		%g2, %o0
+	mov		%g3, %o1
+	mov		%g7, %o5
+
+	retry
+
+#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
+
+	/* %g5:	error
+	 * %g6:	tlb op
+	 */
+__hypervisor_tlb_xcall_error:
+	mov	%g5, %g4
+	mov	%g6, %g5
+	ba,pt	%xcc, etrap
+	 rd	%pc, %g7
+	mov	%l4, %o0
+	call	hypervisor_tlbop_error_xcall
+	 mov	%l5, %o1
+	ba,a,pt	%xcc, rtrap
+
+	.globl		__hypervisor_xcall_flush_tlb_mm
+__hypervisor_xcall_flush_tlb_mm: /* 21 insns */
+	/* %g5=ctx, g1,g2,g3,g4,g7=scratch, %g6=unusable */
+	mov		%o0, %g2
+	mov		%o1, %g3
+	mov		%o2, %g4
+	mov		%o3, %g1
+	mov		%o5, %g7
+	clr		%o0		/* ARG0: CPU lists unimplemented */
+	clr		%o1		/* ARG1: CPU lists unimplemented */
+	mov		%g5, %o2	/* ARG2: mmu context */
+	mov		HV_MMU_ALL, %o3	/* ARG3: flags */
+	mov		HV_FAST_MMU_DEMAP_CTX, %o5
+	ta		HV_FAST_TRAP
+	mov		HV_FAST_MMU_DEMAP_CTX, %g6
+	brnz,pn		%o0, __hypervisor_tlb_xcall_error
+	 mov		%o0, %g5
+	mov		%g2, %o0
+	mov		%g3, %o1
+	mov		%g4, %o2
+	mov		%g1, %o3
+	mov		%g7, %o5
+	membar		#Sync
+	retry
+
+	.globl		__hypervisor_xcall_flush_tlb_page
+__hypervisor_xcall_flush_tlb_page: /* 17 insns */
+	/* %g5=ctx, %g1=vaddr */
+	mov		%o0, %g2
+	mov		%o1, %g3
+	mov		%o2, %g4
+	mov		%g1, %o0	        /* ARG0: virtual address */
+	mov		%g5, %o1		/* ARG1: mmu context */
+	mov		HV_MMU_ALL, %o2		/* ARG2: flags */
+	srlx		%o0, PAGE_SHIFT, %o0
+	sllx		%o0, PAGE_SHIFT, %o0
+	ta		HV_MMU_UNMAP_ADDR_TRAP
+	mov		HV_MMU_UNMAP_ADDR_TRAP, %g6
+	brnz,a,pn	%o0, __hypervisor_tlb_xcall_error
+	 mov		%o0, %g5
+	mov		%g2, %o0
+	mov		%g3, %o1
+	mov		%g4, %o2
+	membar		#Sync
+	retry
+
+	.globl		__hypervisor_xcall_flush_tlb_kernel_range
+__hypervisor_xcall_flush_tlb_kernel_range: /* 25 insns */
+	/* %g1=start, %g7=end, g2,g3,g4,g5,g6=scratch */
+	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
+	mov		%o0, %g2
+	mov		%o1, %g4
+	mov		%o2, %g7
+1:	add		%g1, %g3, %o0	/* ARG0: virtual address */
+	mov		0, %o1		/* ARG1: mmu context */
+	mov		HV_MMU_ALL, %o2	/* ARG2: flags */
+	ta		HV_MMU_UNMAP_ADDR_TRAP
+	mov		HV_MMU_UNMAP_ADDR_TRAP, %g6
+	brnz,pn		%o0, __hypervisor_tlb_xcall_error
+	 mov		%o0, %g5
+	sethi		%hi(PAGE_SIZE), %o2
+	brnz,pt		%g3, 1b
+	 sub		%g3, %o2, %g3
+	mov		%g2, %o0
+	mov		%g4, %o1
+	mov		%g7, %o2
+	membar		#Sync
+	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_call_function_single
+xcall_call_function_single:
+	wr		%g0, (1 << PIL_SMP_CALL_FUNC_SNGL), %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
+
+	.globl		xcall_new_mmu_context_version
+xcall_new_mmu_context_version:
+	wr		%g0, (1 << PIL_SMP_CTX_NEW_VERSION), %set_softint
+	retry
+
+#ifdef CONFIG_KGDB
+	.globl		xcall_kgdb_capture
+xcall_kgdb_capture:
+	wr		%g0, (1 << PIL_KGDB_CAPTURE), %set_softint
+	retry
+#endif
+
+#endif /* CONFIG_SMP */
+
+
+	.globl		hypervisor_patch_cachetlbops
+hypervisor_patch_cachetlbops:
+	save		%sp, -128, %sp
+
+	sethi		%hi(__flush_tlb_mm), %o0
+	or		%o0, %lo(__flush_tlb_mm), %o0
+	sethi		%hi(__hypervisor_flush_tlb_mm), %o1
+	or		%o1, %lo(__hypervisor_flush_tlb_mm), %o1
+	call		tlb_patch_one
+	 mov		10, %o2
+
+	sethi		%hi(__flush_tlb_page), %o0
+	or		%o0, %lo(__flush_tlb_page), %o0
+	sethi		%hi(__hypervisor_flush_tlb_page), %o1
+	or		%o1, %lo(__hypervisor_flush_tlb_page), %o1
+	call		tlb_patch_one
+	 mov		11, %o2
+
+	sethi		%hi(__flush_tlb_pending), %o0
+	or		%o0, %lo(__flush_tlb_pending), %o0
+	sethi		%hi(__hypervisor_flush_tlb_pending), %o1
+	or		%o1, %lo(__hypervisor_flush_tlb_pending), %o1
+	call		tlb_patch_one
+	 mov		16, %o2
+
+	sethi		%hi(__flush_tlb_kernel_range), %o0
+	or		%o0, %lo(__flush_tlb_kernel_range), %o0
+	sethi		%hi(__hypervisor_flush_tlb_kernel_range), %o1
+	or		%o1, %lo(__hypervisor_flush_tlb_kernel_range), %o1
+	call		tlb_patch_one
+	 mov		16, %o2
+
+#ifdef DCACHE_ALIASING_POSSIBLE
+	sethi		%hi(__flush_dcache_page), %o0
+	or		%o0, %lo(__flush_dcache_page), %o0
+	sethi		%hi(__hypervisor_flush_dcache_page), %o1
+	or		%o1, %lo(__hypervisor_flush_dcache_page), %o1
+	call		tlb_patch_one
+	 mov		2, %o2
+#endif /* DCACHE_ALIASING_POSSIBLE */
+
+#ifdef CONFIG_SMP
+	sethi		%hi(xcall_flush_tlb_mm), %o0
+	or		%o0, %lo(xcall_flush_tlb_mm), %o0
+	sethi		%hi(__hypervisor_xcall_flush_tlb_mm), %o1
+	or		%o1, %lo(__hypervisor_xcall_flush_tlb_mm), %o1
+	call		tlb_patch_one
+	 mov		21, %o2
+
+	sethi		%hi(xcall_flush_tlb_page), %o0
+	or		%o0, %lo(xcall_flush_tlb_page), %o0
+	sethi		%hi(__hypervisor_xcall_flush_tlb_page), %o1
+	or		%o1, %lo(__hypervisor_xcall_flush_tlb_page), %o1
+	call		tlb_patch_one
+	 mov		17, %o2
+
+	sethi		%hi(xcall_flush_tlb_kernel_range), %o0
+	or		%o0, %lo(xcall_flush_tlb_kernel_range), %o0
+	sethi		%hi(__hypervisor_xcall_flush_tlb_kernel_range), %o1
+	or		%o1, %lo(__hypervisor_xcall_flush_tlb_kernel_range), %o1
+	call		tlb_patch_one
+	 mov		25, %o2
+#endif /* CONFIG_SMP */
+
+	ret
+	 restore

arch/sparc/mm/viking.S

@@ -0,0 +1,282 @@
+/*
+ * viking.S: High speed Viking cache/mmu operations
+ *
+ * Copyright (C) 1997  Eddie C. Dost  (ecd@skynet.be)
+ * Copyright (C) 1997,1998,1999  Jakub Jelinek  (jj@ultra.linux.cz)
+ * Copyright (C) 1999  Pavel Semerad  (semerad@ss1000.ms.mff.cuni.cz)
+ */
+
+#include <asm/ptrace.h>
+#include <asm/psr.h>
+#include <asm/asm-offsets.h>
+#include <asm/asi.h>
+#include <asm/mxcc.h>
+#include <asm/page.h>
+#include <asm/pgtsrmmu.h>
+#include <asm/viking.h>
+
+#ifdef CONFIG_SMP
+	.data
+	.align	4
+sun4dsmp_flush_tlb_spin:
+	.word	0
+#endif
+
+	.text
+	.align	4
+
+	.globl	viking_flush_cache_all, viking_flush_cache_mm
+	.globl	viking_flush_cache_range, viking_flush_cache_page
+	.globl	viking_flush_page, viking_mxcc_flush_page
+	.globl	viking_flush_page_for_dma, viking_flush_page_to_ram
+	.globl	viking_flush_sig_insns
+	.globl	viking_flush_tlb_all, viking_flush_tlb_mm
+	.globl	viking_flush_tlb_range, viking_flush_tlb_page
+
+viking_flush_page:
+	sethi	%hi(PAGE_OFFSET), %g2
+	sub	%o0, %g2, %g3
+	srl	%g3, 12, %g1		! ppage >> 12
+
+	clr	%o1			! set counter, 0 - 127
+	sethi	%hi(PAGE_OFFSET + PAGE_SIZE - 0x80000000), %o3
+	sethi	%hi(0x80000000), %o4
+	sethi	%hi(VIKING_PTAG_VALID), %o5
+	sethi	%hi(2*PAGE_SIZE), %o0
+	sethi	%hi(PAGE_SIZE), %g7
+	clr	%o2			! block counter, 0 - 3
+5:
+	sll	%o1, 5, %g4
+	or	%g4, %o4, %g4		! 0x80000000 | (set << 5)
+
+	sll	%o2, 26, %g5		! block << 26
+6:
+	or	%g5, %g4, %g5
+	ldda	[%g5] ASI_M_DATAC_TAG, %g2
+	cmp	%g3, %g1		! ptag == ppage?
+	bne	7f
+	 inc	%o2
+
+	andcc	%g2, %o5, %g0		! ptag VALID?
+	be	7f
+	 add	%g4, %o3, %g2		! (PAGE_OFFSET + PAGE_SIZE) | (set << 5)
+	ld	[%g2], %g3
+	ld	[%g2 + %g7], %g3
+	add	%g2, %o0, %g2
+	ld	[%g2], %g3
+	ld	[%g2 + %g7], %g3
+	add	%g2, %o0, %g2
+	ld	[%g2], %g3
+	ld	[%g2 + %g7], %g3
+	add	%g2, %o0, %g2
+	ld	[%g2], %g3
+	b	8f
+	 ld	[%g2 + %g7], %g3
+
+7:
+	cmp	%o2, 3
+	ble	6b
+	 sll	%o2, 26, %g5			! block << 26
+
+8:	inc	%o1
+	cmp	%o1, 0x7f
+	ble	5b
+	 clr	%o2
+
+9:	retl
+	 nop
+
+viking_mxcc_flush_page:
+	sethi	%hi(PAGE_OFFSET), %g2
+	sub	%o0, %g2, %g3
+	sub	%g3, -PAGE_SIZE, %g3		! ppage + PAGE_SIZE
+	sethi	%hi(MXCC_SRCSTREAM), %o3	! assume %hi(MXCC_SRCSTREAM) == %hi(MXCC_DESTSTREAM)
+	mov	0x10, %g2			! set cacheable bit
+	or	%o3, %lo(MXCC_SRCSTREAM), %o2
+	or	%o3, %lo(MXCC_DESSTREAM), %o3
+	sub	%g3, MXCC_STREAM_SIZE, %g3
+6:
+	stda	%g2, [%o2] ASI_M_MXCC
+	stda	%g2, [%o3] ASI_M_MXCC
+	andncc	%g3, PAGE_MASK, %g0
+	bne	6b
+	 sub	%g3, MXCC_STREAM_SIZE, %g3
+
+9:	retl
+	 nop
+
+viking_flush_cache_page:
+viking_flush_cache_range:
+#ifndef CONFIG_SMP
+	ld	[%o0 + VMA_VM_MM], %o0
+#endif
+viking_flush_cache_mm:
+#ifndef CONFIG_SMP
+	ld	[%o0 + AOFF_mm_context], %g1
+	cmp	%g1, -1
+	bne	viking_flush_cache_all
+	 nop
+	b,a	viking_flush_cache_out
+#endif
+viking_flush_cache_all:
+	WINDOW_FLUSH(%g4, %g5)
+viking_flush_cache_out:
+	retl
+	 nop
+
+viking_flush_tlb_all:
+	mov	0x400, %g1
+	retl
+	 sta	%g0, [%g1] ASI_M_FLUSH_PROBE
+
+viking_flush_tlb_mm:
+	mov	SRMMU_CTX_REG, %g1
+	ld	[%o0 + AOFF_mm_context], %o1
+	lda	[%g1] ASI_M_MMUREGS, %g5
+#ifndef CONFIG_SMP
+	cmp	%o1, -1
+	be	1f
+#endif
+	mov	0x300, %g2
+	sta	%o1, [%g1] ASI_M_MMUREGS
+	sta	%g0, [%g2] ASI_M_FLUSH_PROBE
+	retl
+	 sta	%g5, [%g1] ASI_M_MMUREGS
+#ifndef CONFIG_SMP
+1:	retl
+	 nop
+#endif
+
+viking_flush_tlb_range:
+	ld	[%o0 + VMA_VM_MM], %o0
+	mov	SRMMU_CTX_REG, %g1
+	ld	[%o0 + AOFF_mm_context], %o3
+	lda	[%g1] ASI_M_MMUREGS, %g5
+#ifndef CONFIG_SMP
+	cmp	%o3, -1
+	be	2f
+#endif
+	sethi	%hi(~((1 << SRMMU_PGDIR_SHIFT) - 1)), %o4
+	sta	%o3, [%g1] ASI_M_MMUREGS
+	and	%o1, %o4, %o1
+	add	%o1, 0x200, %o1
+	sta	%g0, [%o1] ASI_M_FLUSH_PROBE
+1:	sub	%o1, %o4, %o1
+	cmp	%o1, %o2
+	blu,a	1b
+	 sta	%g0, [%o1] ASI_M_FLUSH_PROBE
+	retl
+	 sta	%g5, [%g1] ASI_M_MMUREGS
+#ifndef CONFIG_SMP
+2:	retl
+	 nop
+#endif
+
+viking_flush_tlb_page:
+	ld	[%o0 + VMA_VM_MM], %o0
+	mov	SRMMU_CTX_REG, %g1
+	ld	[%o0 + AOFF_mm_context], %o3
+	lda	[%g1] ASI_M_MMUREGS, %g5
+#ifndef CONFIG_SMP
+	cmp	%o3, -1
+	be	1f
+#endif
+	and	%o1, PAGE_MASK, %o1
+	sta	%o3, [%g1] ASI_M_MMUREGS
+	sta	%g0, [%o1] ASI_M_FLUSH_PROBE
+	retl
+	 sta	%g5, [%g1] ASI_M_MMUREGS
+#ifndef CONFIG_SMP
+1:	retl
+	 nop
+#endif
+
+viking_flush_page_to_ram:
+viking_flush_page_for_dma:
+viking_flush_sig_insns:
+	retl
+	 nop
+
+#ifdef CONFIG_SMP
+	.globl	sun4dsmp_flush_tlb_all, sun4dsmp_flush_tlb_mm
+	.globl	sun4dsmp_flush_tlb_range, sun4dsmp_flush_tlb_page
+sun4dsmp_flush_tlb_all:
+	sethi	%hi(sun4dsmp_flush_tlb_spin), %g3
+1:	ldstub	[%g3 + %lo(sun4dsmp_flush_tlb_spin)], %g5
+	tst	%g5
+	bne	2f
+	 mov	0x400, %g1
+	sta	%g0, [%g1] ASI_M_FLUSH_PROBE
+	retl
+	 stb	%g0, [%g3 + %lo(sun4dsmp_flush_tlb_spin)]
+2:	tst	%g5
+	bne,a	2b
+	 ldub	[%g3 + %lo(sun4dsmp_flush_tlb_spin)], %g5
+	b,a	1b
+
+sun4dsmp_flush_tlb_mm:
+	sethi	%hi(sun4dsmp_flush_tlb_spin), %g3
+1:	ldstub	[%g3 + %lo(sun4dsmp_flush_tlb_spin)], %g5
+	tst	%g5
+	bne	2f
+	 mov	SRMMU_CTX_REG, %g1
+	ld	[%o0 + AOFF_mm_context], %o1
+	lda	[%g1] ASI_M_MMUREGS, %g5
+	mov	0x300, %g2
+	sta	%o1, [%g1] ASI_M_MMUREGS
+	sta	%g0, [%g2] ASI_M_FLUSH_PROBE
+	sta	%g5, [%g1] ASI_M_MMUREGS
+	retl
+	 stb	%g0, [%g3 + %lo(sun4dsmp_flush_tlb_spin)]
+2:	tst	%g5
+	bne,a	2b
+	 ldub	[%g3 + %lo(sun4dsmp_flush_tlb_spin)], %g5
+	b,a	1b
+
+sun4dsmp_flush_tlb_range:
+	sethi	%hi(sun4dsmp_flush_tlb_spin), %g3
+1:	ldstub	[%g3 + %lo(sun4dsmp_flush_tlb_spin)], %g5
+	tst	%g5
+	bne	3f
+	 mov	SRMMU_CTX_REG, %g1
+	ld	[%o0 + VMA_VM_MM], %o0
+	ld	[%o0 + AOFF_mm_context], %o3
+	lda	[%g1] ASI_M_MMUREGS, %g5
+	sethi	%hi(~((1 << SRMMU_PGDIR_SHIFT) - 1)), %o4
+	sta	%o3, [%g1] ASI_M_MMUREGS
+	and	%o1, %o4, %o1
+	add	%o1, 0x200, %o1
+	sta	%g0, [%o1] ASI_M_FLUSH_PROBE
+2:	sub	%o1, %o4, %o1
+	cmp	%o1, %o2
+	blu,a	2b
+	 sta	%g0, [%o1] ASI_M_FLUSH_PROBE
+	sta	%g5, [%g1] ASI_M_MMUREGS
+	retl
+	 stb	%g0, [%g3 + %lo(sun4dsmp_flush_tlb_spin)]
+3:	tst	%g5
+	bne,a	3b
+	 ldub	[%g3 + %lo(sun4dsmp_flush_tlb_spin)], %g5
+	b,a	1b
+
+sun4dsmp_flush_tlb_page:
+	sethi	%hi(sun4dsmp_flush_tlb_spin), %g3
+1:	ldstub	[%g3 + %lo(sun4dsmp_flush_tlb_spin)], %g5
+	tst	%g5
+	bne	2f
+	 mov	SRMMU_CTX_REG, %g1
+	ld	[%o0 + VMA_VM_MM], %o0
+	ld	[%o0 + AOFF_mm_context], %o3
+	lda	[%g1] ASI_M_MMUREGS, %g5
+	and	%o1, PAGE_MASK, %o1
+	sta	%o3, [%g1] ASI_M_MMUREGS
+	sta	%g0, [%o1] ASI_M_FLUSH_PROBE
+	sta	%g5, [%g1] ASI_M_MMUREGS
+	retl
+	 stb	%g0, [%g3 + %lo(sun4dsmp_flush_tlb_spin)]
+2:	tst	%g5
+	bne,a	2b
+	 ldub	[%g3 + %lo(sun4dsmp_flush_tlb_spin)], %g5
+	b,a	1b
+	 nop
+#endif