Merge branch 'wip-2.6.34' into old-private-masterarchived-private-master

5 files changed, 321 insertions, 86 deletions

diff --git a/arch/microblaze/mm/Makefile b/arch/microblaze/mm/Makefile
index 6c8a924d9e26..09c49ed87235 100644
--- a/arch/microblaze/mm/Makefile
+++ b/arch/microblaze/mm/Makefile
@@ -2,6 +2,6 @@
 # Makefile
 #
 
-obj-y := init.o
+obj-y := consistent.o init.o
 
 obj-$(CONFIG_MMU) += pgtable.o mmu_context.o fault.o
diff --git a/arch/microblaze/mm/consistent.c b/arch/microblaze/mm/consistent.c
new file mode 100644
index 000000000000..5a59dad62bd2
--- /dev/null
+++ b/arch/microblaze/mm/consistent.c
@@ -0,0 +1,255 @@
+/*
+ * Microblaze support for cache consistent memory.
+ * Copyright (C) 2010 Michal Simek <monstr@monstr.eu>
+ * Copyright (C) 2010 PetaLogix
+ * Copyright (C) 2005 John Williams <jwilliams@itee.uq.edu.au>
+ *
+ * Based on PowerPC version derived from arch/arm/mm/consistent.c
+ * Copyright (C) 2001 Dan Malek (dmalek@jlc.net)
+ * Copyright (C) 2000 Russell King
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#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/stddef.h>
+#include <linux/vmalloc.h>
+#include <linux/init.h>
+#include <linux/delay.h>
+#include <linux/bootmem.h>
+#include <linux/highmem.h>
+#include <linux/pci.h>
+#include <linux/interrupt.h>
+#include <linux/gfp.h>
+
+#include <asm/pgalloc.h>
+#include <linux/io.h>
+#include <linux/hardirq.h>
+#include <asm/mmu_context.h>
+#include <asm/mmu.h>
+#include <linux/uaccess.h>
+#include <asm/pgtable.h>
+#include <asm/cpuinfo.h>
+#include <asm/tlbflush.h>
+
+#ifndef CONFIG_MMU
+/* I have to use dcache values because I can't relate on ram size */
+# define UNCACHED_SHADOW_MASK (cpuinfo.dcache_high - cpuinfo.dcache_base + 1)
+#endif
+
+/*
+ * Consistent memory allocators. Used for DMA devices that want to
+ * share uncached memory with the processor core.
+ * My crufty no-MMU approach is simple. In the HW platform we can optionally
+ * mirror the DDR up above the processor cacheable region.  So, memory accessed
+ * in this mirror region will not be cached.  It's alloced from the same
+ * pool as normal memory, but the handle we return is shifted up into the
+ * uncached region.  This will no doubt cause big problems if memory allocated
+ * here is not also freed properly. -- JW
+ */
+void *consistent_alloc(int gfp, size_t size, dma_addr_t *dma_handle)
+{
+        unsigned long order, vaddr;
+        void *ret;
+        unsigned int i, err = 0;
+        struct page *page, *end;
+
+#ifdef CONFIG_MMU
+        phys_addr_t pa;
+        struct vm_struct *area;
+        unsigned long va;
+#endif
+
+        if (in_interrupt())
+                BUG();
+
+        /* Only allocate page size areas. */
+        size = PAGE_ALIGN(size);
+        order = get_order(size);
+
+        vaddr = __get_free_pages(gfp, order);
+        if (!vaddr)
+                return NULL;
+
+        /*
+         * we need to ensure that there are no cachelines in use,
+         * or worse dirty in this area.
+         */
+        flush_dcache_range(virt_to_phys((void *)vaddr),
+                                        virt_to_phys((void *)vaddr) + size);
+
+#ifndef CONFIG_MMU
+        ret = (void *)vaddr;
+        /*
+         * Here's the magic!  Note if the uncached shadow is not implemented,
+         * it's up to the calling code to also test that condition and make
+         * other arranegments, such as manually flushing the cache and so on.
+         */
+# ifdef CONFIG_XILINX_UNCACHED_SHADOW
+        ret = (void *)((unsigned) ret | UNCACHED_SHADOW_MASK);
+# endif
+        if ((unsigned int)ret > cpuinfo.dcache_base &&
+                                (unsigned int)ret < cpuinfo.dcache_high)
+                printk(KERN_WARNING
+                        "ERROR: Your cache coherent area is CACHED!!!\n");
+
+        /* dma_handle is same as physical (shadowed) address */
+        *dma_handle = (dma_addr_t)ret;
+#else
+        /* Allocate some common virtual space to map the new pages. */
+        area = get_vm_area(size, VM_ALLOC);
+        if (!area) {
+                free_pages(vaddr, order);
+                return NULL;
+        }
+        va = (unsigned long) area->addr;
+        ret = (void *)va;
+
+        /* This gives us the real physical address of the first page. */
+        *dma_handle = pa = virt_to_bus((void *)vaddr);
+#endif
+
+        /*
+         * free wasted pages.  We skip the first page since we know
+         * that it will have count = 1 and won't require freeing.
+         * We also mark the pages in use as reserved so that
+         * remap_page_range works.
+         */
+        page = virt_to_page(vaddr);
+        end = page + (1 << order);
+
+        split_page(page, order);
+
+        for (i = 0; i < size && err == 0; i += PAGE_SIZE) {
+#ifdef CONFIG_MMU
+                /* MS: This is the whole magic - use cache inhibit pages */
+                err = map_page(va + i, pa + i, _PAGE_KERNEL | _PAGE_NO_CACHE);
+#endif
+
+                SetPageReserved(page);
+                page++;
+        }
+
+        /* Free the otherwise unused pages. */
+        while (page < end) {
+                __free_page(page);
+                page++;
+        }
+
+        if (err) {
+                free_pages(vaddr, order);
+                return NULL;
+        }
+
+        return ret;
+}
+EXPORT_SYMBOL(consistent_alloc);
+
+/*
+ * free page(s) as defined by the above mapping.
+ */
+void consistent_free(size_t size, void *vaddr)
+{
+        struct page *page;
+
+        if (in_interrupt())
+                BUG();
+
+        size = PAGE_ALIGN(size);
+
+#ifndef CONFIG_MMU
+        /* Clear SHADOW_MASK bit in address, and free as per usual */
+# ifdef CONFIG_XILINX_UNCACHED_SHADOW
+        vaddr = (void *)((unsigned)vaddr & ~UNCACHED_SHADOW_MASK);
+# endif
+        page = virt_to_page(vaddr);
+
+        do {
+                ClearPageReserved(page);
+                __free_page(page);
+                page++;
+        } while (size -= PAGE_SIZE);
+#else
+        do {
+                pte_t *ptep;
+                unsigned long pfn;
+
+                ptep = pte_offset_kernel(pmd_offset(pgd_offset_k(
+                                                (unsigned int)vaddr),
+                                        (unsigned int)vaddr),
+                                (unsigned int)vaddr);
+                if (!pte_none(*ptep) && pte_present(*ptep)) {
+                        pfn = pte_pfn(*ptep);
+                        pte_clear(&init_mm, (unsigned int)vaddr, ptep);
+                        if (pfn_valid(pfn)) {
+                                page = pfn_to_page(pfn);
+
+                                ClearPageReserved(page);
+                                __free_page(page);
+                        }
+                }
+                vaddr += PAGE_SIZE;
+        } while (size -= PAGE_SIZE);
+
+        /* flush tlb */
+        flush_tlb_all();
+#endif
+}
+EXPORT_SYMBOL(consistent_free);
+
+/*
+ * make an area consistent.
+ */
+void consistent_sync(void *vaddr, size_t size, int direction)
+{
+        unsigned long start;
+        unsigned long end;
+
+        start = (unsigned long)vaddr;
+
+        /* Convert start address back down to unshadowed memory region */
+#ifdef CONFIG_XILINX_UNCACHED_SHADOW
+        start &= ~UNCACHED_SHADOW_MASK;
+#endif
+        end = start + size;
+
+        switch (direction) {
+        case PCI_DMA_NONE:
+                BUG();
+        case PCI_DMA_FROMDEVICE:        /* invalidate only */
+                invalidate_dcache_range(start, end);
+                break;
+        case PCI_DMA_TODEVICE:          /* writeback only */
+                flush_dcache_range(start, end);
+                break;
+        case PCI_DMA_BIDIRECTIONAL:     /* writeback and invalidate */
+                flush_dcache_range(start, end);
+                break;
+        }
+}
+EXPORT_SYMBOL(consistent_sync);
+
+/*
+ * consistent_sync_page makes memory consistent. identical
+ * to consistent_sync, but takes a struct page instead of a
+ * virtual address
+ */
+void consistent_sync_page(struct page *page, unsigned long offset,
+        size_t size, int direction)
+{
+        unsigned long start = (unsigned long)page_address(page) + offset;
+        consistent_sync((void *)start, size, direction);
+}
+EXPORT_SYMBOL(consistent_sync_page);
diff --git a/arch/microblaze/mm/fault.c b/arch/microblaze/mm/fault.c
index d9d249a66ff2..bab922993185 100644
--- a/arch/microblaze/mm/fault.c
+++ b/arch/microblaze/mm/fault.c
@@ -106,7 +106,7 @@ void do_page_fault(struct pt_regs *regs, unsigned long address,
         regs->esr = error_code;
 
         /* On a kernel SLB miss we can only check for a valid exception entry */
-        if (kernel_mode(regs) && (address >= TASK_SIZE)) {
+        if (unlikely(kernel_mode(regs) && (address >= TASK_SIZE))) {
                 printk(KERN_WARNING "kernel task_size exceed");
                 _exception(SIGSEGV, regs, code, address);
         }
@@ -122,7 +122,7 @@ void do_page_fault(struct pt_regs *regs, unsigned long address,
         }
 #endif /* CONFIG_KGDB */
 
-        if (in_atomic() || !mm) {
+        if (unlikely(in_atomic() || !mm)) {
                 if (kernel_mode(regs))
                         goto bad_area_nosemaphore;
 
@@ -150,7 +150,7 @@ void do_page_fault(struct pt_regs *regs, unsigned long address,
          * source.  If this is invalid we can skip the address space check,
          * thus avoiding the deadlock.
          */
-        if (!down_read_trylock(&mm->mmap_sem)) {
+        if (unlikely(!down_read_trylock(&mm->mmap_sem))) {
                 if (kernel_mode(regs) && !search_exception_tables(regs->pc))
                         goto bad_area_nosemaphore;
 
@@ -158,16 +158,16 @@ void do_page_fault(struct pt_regs *regs, unsigned long address,
         }
 
         vma = find_vma(mm, address);
-        if (!vma)
+        if (unlikely(!vma))
                 goto bad_area;
 
         if (vma->vm_start <= address)
                 goto good_area;
 
-        if (!(vma->vm_flags & VM_GROWSDOWN))
+        if (unlikely(!(vma->vm_flags & VM_GROWSDOWN)))
                 goto bad_area;
 
-        if (!is_write)
+        if (unlikely(!is_write))
                 goto bad_area;
 
         /*
@@ -179,7 +179,7 @@ void do_page_fault(struct pt_regs *regs, unsigned long address,
          * before setting the user r1.  Thus we allow the stack to
          * expand to 1MB without further checks.
          */
-        if (address + 0x100000 < vma->vm_end) {
+        if (unlikely(address + 0x100000 < vma->vm_end)) {
 
                 /* get user regs even if this fault is in kernel mode */
                 struct pt_regs *uregs = current->thread.regs;
@@ -209,15 +209,15 @@ good_area:
         code = SEGV_ACCERR;
 
         /* a write */
-        if (is_write) {
-                if (!(vma->vm_flags & VM_WRITE))
+        if (unlikely(is_write)) {
+                if (unlikely(!(vma->vm_flags & VM_WRITE)))
                         goto bad_area;
         /* a read */
         } else {
                 /* protection fault */
-                if (error_code & 0x08000000)
+                if (unlikely(error_code & 0x08000000))
                         goto bad_area;
-                if (!(vma->vm_flags & (VM_READ | VM_EXEC)))
+                if (unlikely(!(vma->vm_flags & (VM_READ | VM_EXEC))))
                         goto bad_area;
         }
 
@@ -235,7 +235,7 @@ survive:
                         goto do_sigbus;
                 BUG();
         }
-        if (fault & VM_FAULT_MAJOR)
+        if (unlikely(fault & VM_FAULT_MAJOR))
                 current->maj_flt++;
         else
                 current->min_flt++;
@@ -273,16 +273,11 @@ bad_area_nosemaphore:
  * us unable to handle the page fault gracefully.
  */
 out_of_memory:
-        if (current->pid == 1) {
-                yield();
-                down_read(&mm->mmap_sem);
-                goto survive;
-        }
         up_read(&mm->mmap_sem);
-        printk(KERN_WARNING "VM: killing process %s\n", current->comm);
-        if (user_mode(regs))
-                do_exit(SIGKILL);
-        bad_page_fault(regs, address, SIGKILL);
+        if (!user_mode(regs))
+                bad_page_fault(regs, address, SIGKILL);
+        else
+                pagefault_out_of_memory();
         return;
 
 do_sigbus:
diff --git a/arch/microblaze/mm/init.c b/arch/microblaze/mm/init.c
index a44892e7cd5b..cca3579d4268 100644
--- a/arch/microblaze/mm/init.c
+++ b/arch/microblaze/mm/init.c
@@ -15,6 +15,7 @@
 #include <linux/initrd.h>
 #include <linux/pagemap.h>
 #include <linux/pfn.h>
+#include <linux/slab.h>
 #include <linux/swap.h>
 
 #include <asm/page.h>
@@ -23,6 +24,9 @@
 #include <asm/sections.h>
 #include <asm/tlb.h>
 
+/* Use for MMU and noMMU because of PCI generic code */
+int mem_init_done;
+
 #ifndef CONFIG_MMU
 unsigned int __page_offset;
 EXPORT_SYMBOL(__page_offset);
@@ -30,7 +34,6 @@ EXPORT_SYMBOL(__page_offset);
 #else
 DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);
 
-int mem_init_done;
 static int init_bootmem_done;
 #endif /* CONFIG_MMU */
 
@@ -41,8 +44,10 @@ char *klimit = _end;
  * have available.
  */
 unsigned long memory_start;
+EXPORT_SYMBOL(memory_start);
 unsigned long memory_end; /* due to mm/nommu.c */
 unsigned long memory_size;
+EXPORT_SYMBOL(memory_size);
 
 /*
  * paging_init() sets up the page tables - in fact we've already done this.
@@ -162,7 +167,6 @@ void free_init_pages(char *what, unsigned long begin, unsigned long end)
         for (addr = begin; addr < end; addr += PAGE_SIZE) {
                 ClearPageReserved(virt_to_page(addr));
                 init_page_count(virt_to_page(addr));
-                memset((void *)addr, 0xcc, PAGE_SIZE);
                 free_page(addr);
                 totalram_pages++;
         }
@@ -192,12 +196,6 @@ void free_initmem(void)
                         (unsigned long)(&__init_end));
 }
 
-/* FIXME from arch/powerpc/mm/mem.c*/
-void show_mem(void)
-{
-        printk(KERN_NOTICE "%s\n", __func__);
-}
-
 void __init mem_init(void)
 {
         high_memory = (void *)__va(memory_end);
@@ -207,20 +205,14 @@ void __init mem_init(void)
         printk(KERN_INFO "Memory: %luk/%luk available\n",
                nr_free_pages() << (PAGE_SHIFT-10),
                num_physpages << (PAGE_SHIFT-10));
-#ifdef CONFIG_MMU
         mem_init_done = 1;
-#endif
 }
 
 #ifndef CONFIG_MMU
-/* Check against bounds of physical memory */
-int ___range_ok(unsigned long addr, unsigned long size)
+int page_is_ram(unsigned long pfn)
 {
-        return ((addr < memory_start) ||
-                ((addr + size) > memory_end));
+        return __range_ok(pfn, 0);
 }
-EXPORT_SYMBOL(___range_ok);
-
 #else
 int page_is_ram(unsigned long pfn)
 {
@@ -348,4 +340,27 @@ void __init *early_get_page(void)
         }
         return p;
 }
+
 #endif /* CONFIG_MMU */
+
+void * __init_refok alloc_maybe_bootmem(size_t size, gfp_t mask)
+{
+        if (mem_init_done)
+                return kmalloc(size, mask);
+        else
+                return alloc_bootmem(size);
+}
+
+void * __init_refok zalloc_maybe_bootmem(size_t size, gfp_t mask)
+{
+        void *p;
+
+        if (mem_init_done)
+                p = kzalloc(size, mask);
+        else {
+                p = alloc_bootmem(size);
+                if (p)
+                        memset(p, 0, size);
+        }
+        return p;
+}
diff --git a/arch/microblaze/mm/pgtable.c b/arch/microblaze/mm/pgtable.c
index 46c4ca5d15c5..59bf2335a4ce 100644
--- a/arch/microblaze/mm/pgtable.c
+++ b/arch/microblaze/mm/pgtable.c
@@ -42,6 +42,7 @@
 
 unsigned long ioremap_base;
 unsigned long ioremap_bot;
+EXPORT_SYMBOL(ioremap_bot);
 
 /* The maximum lowmem defaults to 768Mb, but this can be configured to
  * another value.
@@ -103,7 +104,7 @@ static void __iomem *__ioremap(phys_addr_t addr, unsigned long size,
                 area = get_vm_area(size, VM_IOREMAP);
                 if (area == NULL)
                         return NULL;
-                v = VMALLOC_VMADDR(area->addr);
+                v = (unsigned long) area->addr;
         } else {
                 v = (ioremap_bot -= size);
         }
@@ -144,7 +145,6 @@ int map_page(unsigned long va, phys_addr_t pa, int flags)
         pmd_t *pd;
         pte_t *pg;
         int err = -ENOMEM;
-        /* spin_lock(&init_mm.page_table_lock); */
         /* Use upper 10 bits of VA to index the first level map */
         pd = pmd_offset(pgd_offset_k(va), va);
         /* Use middle 10 bits of VA to index the second-level map */
@@ -155,39 +155,13 @@ int map_page(unsigned long va, phys_addr_t pa, int flags)
                 err = 0;
                 set_pte_at(&init_mm, va, pg, pfn_pte(pa >> PAGE_SHIFT,
                                 __pgprot(flags)));
-                if (mem_init_done)
+                if (unlikely(mem_init_done))
                         flush_HPTE(0, va, pmd_val(*pd));
                         /* flush_HPTE(0, va, pg); */
-
         }
-        /* spin_unlock(&init_mm.page_table_lock); */
         return err;
 }
 
-void __init adjust_total_lowmem(void)
-{
-/* TBD */
-#if 0
-        unsigned long max_low_mem = MAX_LOW_MEM;
-
-        if (total_lowmem > max_low_mem) {
-                total_lowmem = max_low_mem;
-#ifndef CONFIG_HIGHMEM
-                printk(KERN_INFO "Warning, memory limited to %ld Mb, use "
-                                "CONFIG_HIGHMEM to reach %ld Mb\n",
-                                max_low_mem >> 20, total_memory >> 20);
-                total_memory = total_lowmem;
-#endif /* CONFIG_HIGHMEM */
-        }
-#endif
-}
-
-static void show_tmem(unsigned long tmem)
-{
-        volatile unsigned long a;
-        a = a + tmem;
-}
-
 /*
  * Map in all of physical memory starting at CONFIG_KERNEL_START.
  */
@@ -197,7 +171,6 @@ void __init mapin_ram(void)
 
         v = CONFIG_KERNEL_START;
         p = memory_start;
-        show_tmem(memory_size);
         for (s = 0; s < memory_size; s += PAGE_SIZE) {
                 f = _PAGE_PRESENT | _PAGE_ACCESSED |
                                 _PAGE_SHARED | _PAGE_HWEXEC;
@@ -216,24 +189,6 @@ void __init mapin_ram(void)
 /* is x a power of 2? */
 #define is_power_of_2(x)        ((x) != 0 && (((x) & ((x) - 1)) == 0))
 
-/*
- * Set up a mapping for a block of I/O.
- * virt, phys, size must all be page-aligned.
- * This should only be called before ioremap is called.
- */
-void __init io_block_mapping(unsigned long virt, phys_addr_t phys,
-                             unsigned int size, int flags)
-{
-        int i;
-
-        if (virt > CONFIG_KERNEL_START && virt < ioremap_bot)
-                ioremap_bot = ioremap_base = virt;
-
-        /* Put it in the page tables. */
-        for (i = 0; i < size; i += PAGE_SIZE)
-                map_page(virt + i, phys + i, flags);
-}
-
 /* Scan the real Linux page tables and return a PTE pointer for
  * a virtual address in a context.
  * Returns true (1) if PTE was found, zero otherwise.  The pointer to
@@ -284,3 +239,18 @@ unsigned long iopa(unsigned long addr)
 
         return pa;
 }
+
+__init_refok pte_t *pte_alloc_one_kernel(struct mm_struct *mm,
+                unsigned long address)
+{
+        pte_t *pte;
+        if (mem_init_done) {
+                pte = (pte_t *)__get_free_page(GFP_KERNEL |
+                                        __GFP_REPEAT | __GFP_ZERO);
+        } else {
+                pte = (pte_t *)early_get_page();
+                if (pte)
+                        clear_page(pte);
+        }
+        return pte;
+}