mm: use vmf->page during WP faults
authorJan Kara <jack@suse.cz>
Wed, 14 Dec 2016 23:07:33 +0000 (15:07 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Thu, 15 Dec 2016 00:04:09 +0000 (16:04 -0800)
So far we set vmf->page during WP faults only when we needed to pass it
to the ->page_mkwrite handler.  Set it in all the cases now and use that
instead of passing page pointer explicitly around.

Link: http://lkml.kernel.org/r/1479460644-25076-14-git-send-email-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/memory.c

index e8a527885e8b77c22362a3594837a2bf7e534e85..ad452898e6c00424d855e2627762c39a4da7def5 100644 (file)
@@ -2102,11 +2102,12 @@ static void fault_dirty_shared_page(struct vm_area_struct *vma,
  * case, all we need to do here is to mark the page as writable and update
  * any related book-keeping.
  */
-static inline int wp_page_reuse(struct vm_fault *vmf, struct page *page,
+static inline int wp_page_reuse(struct vm_fault *vmf,
                                int page_mkwrite, int dirty_shared)
        __releases(vmf->ptl)
 {
        struct vm_area_struct *vma = vmf->vma;
+       struct page *page = vmf->page;
        pte_t entry;
        /*
         * Clear the pages cpupid information as the existing
@@ -2150,10 +2151,11 @@ static inline int wp_page_reuse(struct vm_fault *vmf, struct page *page,
  *   held to the old page, as well as updating the rmap.
  * - In any case, unlock the PTL and drop the reference we took to the old page.
  */
-static int wp_page_copy(struct vm_fault *vmf, struct page *old_page)
+static int wp_page_copy(struct vm_fault *vmf)
 {
        struct vm_area_struct *vma = vmf->vma;
        struct mm_struct *mm = vma->vm_mm;
+       struct page *old_page = vmf->page;
        struct page *new_page = NULL;
        pte_t entry;
        int page_copied = 0;
@@ -2305,26 +2307,25 @@ static int wp_pfn_shared(struct vm_fault *vmf)
                        return 0;
                }
        }
-       return wp_page_reuse(vmf, NULL, 0, 0);
+       return wp_page_reuse(vmf, 0, 0);
 }
 
-static int wp_page_shared(struct vm_fault *vmf, struct page *old_page)
+static int wp_page_shared(struct vm_fault *vmf)
        __releases(vmf->ptl)
 {
        struct vm_area_struct *vma = vmf->vma;
        int page_mkwrite = 0;
 
-       get_page(old_page);
+       get_page(vmf->page);
 
        if (vma->vm_ops && vma->vm_ops->page_mkwrite) {
                int tmp;
 
                pte_unmap_unlock(vmf->pte, vmf->ptl);
-               vmf->page = old_page;
                tmp = do_page_mkwrite(vmf);
                if (unlikely(!tmp || (tmp &
                                      (VM_FAULT_ERROR | VM_FAULT_NOPAGE)))) {
-                       put_page(old_page);
+                       put_page(vmf->page);
                        return tmp;
                }
                /*
@@ -2336,15 +2337,15 @@ static int wp_page_shared(struct vm_fault *vmf, struct page *old_page)
                vmf->pte = pte_offset_map_lock(vma->vm_mm, vmf->pmd,
                                                vmf->address, &vmf->ptl);
                if (!pte_same(*vmf->pte, vmf->orig_pte)) {
-                       unlock_page(old_page);
+                       unlock_page(vmf->page);
                        pte_unmap_unlock(vmf->pte, vmf->ptl);
-                       put_page(old_page);
+                       put_page(vmf->page);
                        return 0;
                }
                page_mkwrite = 1;
        }
 
-       return wp_page_reuse(vmf, old_page, page_mkwrite, 1);
+       return wp_page_reuse(vmf, page_mkwrite, 1);
 }
 
 /*
@@ -2369,10 +2370,9 @@ static int do_wp_page(struct vm_fault *vmf)
        __releases(vmf->ptl)
 {
        struct vm_area_struct *vma = vmf->vma;
-       struct page *old_page;
 
-       old_page = vm_normal_page(vma, vmf->address, vmf->orig_pte);
-       if (!old_page) {
+       vmf->page = vm_normal_page(vma, vmf->address, vmf->orig_pte);
+       if (!vmf->page) {
                /*
                 * VM_MIXEDMAP !pfn_valid() case, or VM_SOFTDIRTY clear on a
                 * VM_PFNMAP VMA.
@@ -2385,30 +2385,30 @@ static int do_wp_page(struct vm_fault *vmf)
                        return wp_pfn_shared(vmf);
 
                pte_unmap_unlock(vmf->pte, vmf->ptl);
-               return wp_page_copy(vmf, old_page);
+               return wp_page_copy(vmf);
        }
 
        /*
         * Take out anonymous pages first, anonymous shared vmas are
         * not dirty accountable.
         */
-       if (PageAnon(old_page) && !PageKsm(old_page)) {
+       if (PageAnon(vmf->page) && !PageKsm(vmf->page)) {
                int total_mapcount;
-               if (!trylock_page(old_page)) {
-                       get_page(old_page);
+               if (!trylock_page(vmf->page)) {
+                       get_page(vmf->page);
                        pte_unmap_unlock(vmf->pte, vmf->ptl);
-                       lock_page(old_page);
+                       lock_page(vmf->page);
                        vmf->pte = pte_offset_map_lock(vma->vm_mm, vmf->pmd,
                                        vmf->address, &vmf->ptl);
                        if (!pte_same(*vmf->pte, vmf->orig_pte)) {
-                               unlock_page(old_page);
+                               unlock_page(vmf->page);
                                pte_unmap_unlock(vmf->pte, vmf->ptl);
-                               put_page(old_page);
+                               put_page(vmf->page);
                                return 0;
                        }
-                       put_page(old_page);
+                       put_page(vmf->page);
                }
-               if (reuse_swap_page(old_page, &total_mapcount)) {
+               if (reuse_swap_page(vmf->page, &total_mapcount)) {
                        if (total_mapcount == 1) {
                                /*
                                 * The page is all ours. Move it to
@@ -2417,24 +2417,24 @@ static int do_wp_page(struct vm_fault *vmf)
                                 * Protected against the rmap code by
                                 * the page lock.
                                 */
-                               page_move_anon_rmap(old_page, vma);
+                               page_move_anon_rmap(vmf->page, vma);
                        }
-                       unlock_page(old_page);
-                       return wp_page_reuse(vmf, old_page, 0, 0);
+                       unlock_page(vmf->page);
+                       return wp_page_reuse(vmf, 0, 0);
                }
-               unlock_page(old_page);
+               unlock_page(vmf->page);
        } else if (unlikely((vma->vm_flags & (VM_WRITE|VM_SHARED)) ==
                                        (VM_WRITE|VM_SHARED))) {
-               return wp_page_shared(vmf, old_page);
+               return wp_page_shared(vmf);
        }
 
        /*
         * Ok, we need to copy. Oh, well..
         */
-       get_page(old_page);
+       get_page(vmf->page);
 
        pte_unmap_unlock(vmf->pte, vmf->ptl);
-       return wp_page_copy(vmf, old_page);
+       return wp_page_copy(vmf);
 }
 
 static void unmap_mapping_range_vma(struct vm_area_struct *vma,