1 files changed, 58 insertions, 42 deletions

diff --git a/arch/arm/mm/fault-armv.c b/arch/arm/mm/fault-armv.c
index d0d17b6a370..0d414c28eb2 100644
--- a/arch/arm/mm/fault-armv.c
+++ b/arch/arm/mm/fault-armv.c
@@ -16,6 +16,7 @@
 #include <linux/vmalloc.h>
 #include <linux/init.h>
 #include <linux/pagemap.h>
+#include <linux/gfp.h>
 
 #include <asm/bugs.h>
 #include <asm/cacheflush.h>
@@ -23,6 +24,8 @@
 #include <asm/pgtable.h>
 #include <asm/tlbflush.h>
 
+#include "mm.h"
+
 static unsigned long shared_pte_mask = L_PTE_MT_BUFFERABLE;
 
 /*
@@ -34,28 +37,12 @@ static unsigned long shared_pte_mask = L_PTE_MT_BUFFERABLE;
  * Therefore those configurations which might call adjust_pte (those
  * without CONFIG_CPU_CACHE_VIPT) cannot support split page_table_lock.
  */
-static int adjust_pte(struct vm_area_struct *vma, unsigned long address)
+static int do_adjust_pte(struct vm_area_struct *vma, unsigned long address,
+	unsigned long pfn, pte_t *ptep)
 {
-	pgd_t *pgd;
-	pmd_t *pmd;
-	pte_t *pte, entry;
+	pte_t entry = *ptep;
 	int ret;
 
-	pgd = pgd_offset(vma->vm_mm, address);
-	if (pgd_none(*pgd))
-		goto no_pgd;
-	if (pgd_bad(*pgd))
-		goto bad_pgd;
-
-	pmd = pmd_offset(pgd, address);
-	if (pmd_none(*pmd))
-		goto no_pmd;
-	if (pmd_bad(*pmd))
-		goto bad_pmd;
-
-	pte = pte_offset_map(pmd, address);
-	entry = *pte;
-
 	/*
 	 * If this page is present, it's actually being shared.
 	 */
@@ -66,33 +53,55 @@ static int adjust_pte(struct vm_area_struct *vma, unsigned long address)
 	 * fault (ie, is old), we can safely ignore any issues.
 	 */
 	if (ret && (pte_val(entry) & L_PTE_MT_MASK) != shared_pte_mask) {
-		unsigned long pfn = pte_pfn(entry);
 		flush_cache_page(vma, address, pfn);
 		outer_flush_range((pfn << PAGE_SHIFT),
 				  (pfn << PAGE_SHIFT) + PAGE_SIZE);
 		pte_val(entry) &= ~L_PTE_MT_MASK;
 		pte_val(entry) |= shared_pte_mask;
-		set_pte_at(vma->vm_mm, address, pte, entry);
+		set_pte_at(vma->vm_mm, address, ptep, entry);
 		flush_tlb_page(vma, address);
 	}
-	pte_unmap(pte);
+
 	return ret;
+}
+
+static int adjust_pte(struct vm_area_struct *vma, unsigned long address,
+	unsigned long pfn)
+{
+	spinlock_t *ptl;
+	pgd_t *pgd;
+	pmd_t *pmd;
+	pte_t *pte;
+	int ret;
 
-bad_pgd:
-	pgd_ERROR(*pgd);
-	pgd_clear(pgd);
-no_pgd:
-	return 0;
-
-bad_pmd:
-	pmd_ERROR(*pmd);
-	pmd_clear(pmd);
-no_pmd:
-	return 0;
+	pgd = pgd_offset(vma->vm_mm, address);
+	if (pgd_none_or_clear_bad(pgd))
+		return 0;
+
+	pmd = pmd_offset(pgd, address);
+	if (pmd_none_or_clear_bad(pmd))
+		return 0;
+
+	/*
+	 * This is called while another page table is mapped, so we
+	 * must use the nested version.  This also means we need to
+	 * open-code the spin-locking.
+	 */
+	ptl = pte_lockptr(vma->vm_mm, pmd);
+	pte = pte_offset_map_nested(pmd, address);
+	spin_lock(ptl);
+
+	ret = do_adjust_pte(vma, address, pfn, pte);
+
+	spin_unlock(ptl);
+	pte_unmap_nested(pte);
+
+	return ret;
 }
 
 static void
-make_coherent(struct address_space *mapping, struct vm_area_struct *vma, unsigned long addr, unsigned long pfn)
+make_coherent(struct address_space *mapping, struct vm_area_struct *vma,
+	unsigned long addr, pte_t *ptep, unsigned long pfn)
 {
 	struct mm_struct *mm = vma->vm_mm;
 	struct vm_area_struct *mpnt;
@@ -120,11 +129,11 @@ make_coherent(struct address_space *mapping, struct vm_area_struct *vma, unsigne
 		if (!(mpnt->vm_flags & VM_MAYSHARE))
 			continue;
 		offset = (pgoff - mpnt->vm_pgoff) << PAGE_SHIFT;
-		aliases += adjust_pte(mpnt, mpnt->vm_start + offset);
+		aliases += adjust_pte(mpnt, mpnt->vm_start + offset, pfn);
 	}
 	flush_dcache_mmap_unlock(mapping);
 	if (aliases)
-		adjust_pte(vma, addr);
+		do_adjust_pte(vma, addr, pfn, ptep);
 	else
 		flush_cache_page(vma, addr, pfn);
 }
@@ -142,16 +151,24 @@ make_coherent(struct address_space *mapping, struct vm_area_struct *vma, unsigne
  *
  * Note that the pte lock will be held.
  */
-void update_mmu_cache(struct vm_area_struct *vma, unsigned long addr, pte_t pte)
+void update_mmu_cache(struct vm_area_struct *vma, unsigned long addr,
+	pte_t *ptep)
 {
-	unsigned long pfn = pte_pfn(pte);
+	unsigned long pfn = pte_pfn(*ptep);
 	struct address_space *mapping;
 	struct page *page;
 
 	if (!pfn_valid(pfn))
 		return;
 
+	/*
+	 * The zero page is never written to, so never has any dirty
+	 * cache lines, and therefore never needs to be flushed.
+	 */
 	page = pfn_to_page(pfn);
+	if (page == ZERO_PAGE(0))
+		return;
+
 	mapping = page_mapping(page);
 #ifndef CONFIG_SMP
 	if (test_and_clear_bit(PG_dcache_dirty, &page->flags))
@@ -159,7 +176,7 @@ void update_mmu_cache(struct vm_area_struct *vma, unsigned long addr, pte_t pte)
 #endif
 	if (mapping) {
 		if (cache_is_vivt())
-			make_coherent(mapping, vma, addr, pfn);
+			make_coherent(mapping, vma, addr, ptep, pfn);
 		else if (vma->vm_flags & VM_EXEC)
 			__flush_icache_all();
 	}
@@ -198,9 +215,8 @@ void __init check_writebuffer_bugs(void)
 	page = alloc_page(GFP_KERNEL);
 	if (page) {
 		unsigned long *p1, *p2;
-		pgprot_t prot = __pgprot(L_PTE_PRESENT|L_PTE_YOUNG|
-					 L_PTE_DIRTY|L_PTE_WRITE|
-					 L_PTE_MT_BUFFERABLE);
+		pgprot_t prot = __pgprot_modify(PAGE_KERNEL,
+					L_PTE_MT_MASK, L_PTE_MT_BUFFERABLE);
 
 		p1 = vmap(&page, 1, VM_IOREMAP, prot);
 		p2 = vmap(&page, 1, VM_IOREMAP, prot);