OSDN Git Service

ceph: fix double page_unlock() in page_mkwrite()
authorYan, Zheng <zyan@redhat.com>
Thu, 17 Dec 2015 07:36:39 +0000 (15:36 +0800)
committerIlya Dryomov <idryomov@gmail.com>
Thu, 21 Jan 2016 18:36:07 +0000 (19:36 +0100)
ceph_update_writeable_page() unlocks the page on errors, so
page_mkwrite() should not unlock the page again.

Signed-off-by: Yan, Zheng <zyan@redhat.com>
fs/ceph/addr.c

index b7d218a..ce3c773 100644 (file)
@@ -1378,11 +1378,13 @@ static int ceph_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 
        ret = VM_FAULT_NOPAGE;
        if ((off > size) ||
-           (page->mapping != inode->i_mapping))
+           (page->mapping != inode->i_mapping)) {
+               unlock_page(page);
                goto out;
+       }
 
        ret = ceph_update_writeable_page(vma->vm_file, off, len, page);
-       if (ret == 0) {
+       if (ret >= 0) {
                /* success.  we'll keep the page locked. */
                set_page_dirty(page);
                ret = VM_FAULT_LOCKED;
@@ -1393,8 +1395,6 @@ static int ceph_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
                        ret = VM_FAULT_SIGBUS;
        }
 out:
-       if (ret != VM_FAULT_LOCKED)
-               unlock_page(page);
        if (ret == VM_FAULT_LOCKED ||
            ci->i_inline_version != CEPH_INLINE_NONE) {
                int dirty;