ceph: invalidate pages that beyond EOF in ceph_writepages_start()

Dirty pages can be associated with different capsnap. Different capsnap
may have different EOF value. So invalidating dirty pages according to
the largest EOF value is wrong. Dirty pages beyond EOF, but associated
with other capsnap, do not get invalidated.

Signed-off-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

fs/ceph/addr.c

@@ -801,7 +801,7 @@ static int ceph_writepages_start(struct address_space *mapping,
 	struct ceph_osd_request *req = NULL;
 	struct ceph_writeback_ctl ceph_wbc;
 	bool should_loop, range_whole = false;
-	bool stop, done = false;
+	bool done = false;
 
 	dout("writepages_start %p (mode=%s)\n", inode,
 	     wbc->sync_mode == WB_SYNC_NONE ? "NONE" :
@@ -865,8 +865,7 @@ static int ceph_writepages_start(struct address_space *mapping,
 	ceph_put_snap_context(last_snapc);
 	last_snapc = snapc;
 
-	stop = false;
-	while (!stop && index <= end) {
+	while (!done && index <= end) {
 		int num_ops = 0, op_idx;
 		unsigned i, pvec_pages, max_pages, locked_pages = 0;
 		struct page **pages = NULL, **data_pages;
@@ -899,16 +898,26 @@ static int ceph_writepages_start(struct address_space *mapping,
 				unlock_page(page);
 				continue;
 			}
-			if (strip_unit_end && (page->index > strip_unit_end)) {
-				dout("end of strip unit %p\n", page);
+			/* only if matching snap context */
+			pgsnapc = page_snap_context(page);
+			if (pgsnapc != snapc) {
+				dout("page snapc %p %lld != oldest %p %lld\n",
+				     pgsnapc, pgsnapc->seq, snapc, snapc->seq);
 				unlock_page(page);
-				break;
+				continue;
 			}
 			if (page_offset(page) >= ceph_wbc.i_size) {
 				dout("%p page eof %llu\n",
 				     page, ceph_wbc.i_size);
-				/* not done if range_cyclic */
-				stop = true;
+				if (ceph_wbc.size_stable ||
+				    page_offset(page) >= i_size_read(inode))
+					mapping->a_ops->invalidatepage(page,
+								0, PAGE_SIZE);
+				unlock_page(page);
+				continue;
+			}
+			if (strip_unit_end && (page->index > strip_unit_end)) {
+				dout("end of strip unit %p\n", page);
 				unlock_page(page);
 				break;
 			}
@@ -922,15 +931,6 @@ static int ceph_writepages_start(struct address_space *mapping,
 				wait_on_page_writeback(page);
 			}
 
-			/* only if matching snap context */
-			pgsnapc = page_snap_context(page);
-			if (pgsnapc != snapc) {
-				dout("page snapc %p %lld != oldest %p %lld\n",
-				     pgsnapc, pgsnapc->seq, snapc, snapc->seq);
-				unlock_page(page);
-				continue;
-			}
-
 			if (!clear_page_dirty_for_io(page)) {
 				dout("%p !clear_page_dirty_for_io\n", page);
 				unlock_page(page);
@@ -1143,7 +1143,7 @@ static int ceph_writepages_start(struct address_space *mapping,
 		 * we tagged for writeback prior to entering this loop.
 		 */
 		if (wbc->nr_to_write <= 0 && wbc->sync_mode == WB_SYNC_NONE)
-			done = stop = true;
+			done = true;
 
 release_pvec_pages:
 		dout("pagevec_release on %d pages (%p)\n", (int)pvec.nr,

fs/ceph/inode.c

@@ -1867,20 +1867,9 @@ void __ceph_do_pending_vmtruncate(struct inode *inode)
 	 * possibly truncate them.. so write AND block!
 	 */
 	if (ci->i_wrbuffer_ref_head < ci->i_wrbuffer_ref) {
-		struct ceph_cap_snap *capsnap;
-		to = ci->i_truncate_size;
-		list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) {
-			// MDS should have revoked Frw caps
-			WARN_ON_ONCE(capsnap->writing);
-			if (capsnap->dirty_pages && capsnap->size > to)
-				to = capsnap->size;
-		}
 		spin_unlock(&ci->i_ceph_lock);
 		dout("__do_pending_vmtruncate %p flushing snaps first\n",
 		     inode);
-
-		truncate_pagecache(inode, to);
-
 		filemap_write_and_wait_range(&inode->i_data, 0,
 					     inode->i_sb->s_maxbytes);
 		goto retry;