kclient: some writepage cleanup

diff --git a/src/TODO b/src/TODO
index 328fe83a7d8a84e9c32fcddee6c4e466c772b274..b818e989a96e59a3fe1c3c55fc7878bc6e75925b 100644 (file)
--- a/src/TODO
+++ b/src/TODO
@@ -18,8 +18,10 @@ big items
 
 
 snaps on kclient
-/- we need a rw snap mutex, so that pending io blocks while we update snap contexts
-  - we can push down to a per-realm rw mutex later
+
+- short reads not allowed..
+- short writes ...?
+
 - need to do ORDERSNAP flag
 - to get sync write retry and delayed flushsnap for sync writes
 

diff --git a/src/kernel/addr.c b/src/kernel/addr.c
index 6418031383527f033e84b1fe9a50ed5180a2aa91..41135f59d2051304e9daaf838bc8e15755d5f60b 100644 (file)
--- a/src/kernel/addr.c
+++ b/src/kernel/addr.c
@@ -287,7 +287,7 @@ static int ceph_readpages(struct file *file, struct address_space *mapping,
  * ceph_writepage:
  *  clear dirty page, and set the writeback flag in the radix tree.
  *  to actually write data to the remote OSDs.
- * leave pages locked.
+ * leave page locked.
  */
 static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
 {
@@ -298,11 +298,14 @@ static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
        int len = PAGE_CACHE_SIZE;
        loff_t i_size;
        int err = 0;
-       int was_dirty;
        struct ceph_snap_context *snapc;
 
-       if (!page->mapping || !page->mapping->host)
+       dout(10, "writepage %p idx %lu\n", page, page->index);
+
+       if (!page->mapping || !page->mapping->host) {
+               dout(10, "writepage %p - no mapping\n", page);
                return -EFAULT;
+       }
        inode = page->mapping->host;
        ci = ceph_inode(inode);
        osdc = &ceph_inode_to_client(inode)->osdc;
@@ -311,31 +314,32 @@ static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
        i_size = i_size_read(inode);
        if (i_size < page_off + len)
                len = i_size - page_off;
-
-       dout(10, "ceph_writepage inode %p page %p index %lu on %llu~%u\n",
+       snapc = (void *)page->private;
+       dout(10, "writepage inode %p page %p index %lu on %llu~%u\n",
             inode, page, page->index, page_off, len);
 
-       page_cache_get(page);
-       was_dirty = PageDirty(page);
-       snapc = (void *)page->private;
        set_page_writeback(page);
-       err = ceph_osdc_writepages(osdc, ceph_vino(inode), &ci->i_layout,
-                                  (void *)page->private,
-                                  page_off, len, &page, 1);
-       if (err >= 0) {
-               if (was_dirty) {
-                       dout(20, "cleaned page %p\n", page);
+       if (snapc) {
+               err = ceph_osdc_writepages(osdc, ceph_vino(inode),
+                                          &ci->i_layout, snapc,
+                                          page_off, len, &page, 1);
+               if (err >= 0) {
+                       dout(20, "writepage cleaned page %p\n", page);
+                       page->private = 0;
+                       ClearPagePrivate(page);
                        ceph_put_wrbuffer_cap_refs(ci, 1, snapc);
+                       ceph_put_snap_context(snapc);
+                       err = 0;  /* vfs expects us to return 0 */
+               } else {
+                       dout(20, "writepage redirtying page %p\n", page);
+                       ceph_redirty_page(page->mapping, page);
+                       if (wbc)
+                               wbc->pages_skipped++;
                }
-               SetPageUptodate(page);
-               err = 0;  /* vfs expects us to return 0 */
        } else {
-               if (wbc)
-                       wbc->pages_skipped++;
-               ceph_set_page_dirty(page, snapc);
+               dout(20, "writepage %p noop (no snapc)\n", page);
        }
        end_page_writeback(page);
-       page_cache_release(page);
        return err;
 }