[linux-cifs-client] [PATCH 4/4] cifs: flush data on any setattr

[Jeff Layton]

We already flush all the dirty pages for an inode before doing
ATTR_SIZE and ATTR_MTIME changes. There's another problem though -- if
we change the mode so that the file becomes read-only then we may not
be able to write data to it after a reconnect.

Fix this by just going back to flushing all the dirty data on any
setattr call. There are probably some cases that can be optimized out,
but I'm not sure they're worthwhile and we need to consider them more
carefully to make sure that we don't cause regressions if we have
to reconnect before writeback occurs.

Signed-off-by: Jeff Layton <jlayton at redhat.com>
---
 fs/cifs/inode.c |   58 ++++++++++++++++++++++++++++--------------------------
 1 files changed, 30 insertions(+), 28 deletions(-)

diff --git a/fs/cifs/inode.c b/fs/cifs/inode.c
index 3525ea5..ed7f048 100644
--- a/fs/cifs/inode.c
+++ b/fs/cifs/inode.c
@@ -1793,20 +1793,21 @@ cifs_setattr_unix(struct dentry *direntry, struct iattr *attrs)
 		goto out;
 	}
 
-	if ((attrs->ia_valid & ATTR_MTIME) || (attrs->ia_valid & ATTR_SIZE)) {
-		/*
-		   Flush data before changing file size or changing the last
-		   write time of the file on the server. If the
-		   flush returns error, store it to report later and continue.
-		   BB: This should be smarter. Why bother flushing pages that
-		   will be truncated anyway? Also, should we error out here if
-		   the flush returns error?
-		 */
-		rc = filemap_write_and_wait(inode->i_mapping);
-		if (rc != 0) {
-			cifsInode->write_behind_rc = rc;
-			rc = 0;
-		}
+	/*
+	 * Attempt to flush data before changing attributes. We need to do
+	 * this for ATTR_SIZE and ATTR_MTIME for sure, and if we change the
+	 * ownership or mode then we may also need to do this. Here, we take
+	 * the safe way out and just do the flush on all setattr requests. If
+	 * the flush returns error, store it to report later and continue.
+	 *
+	 * BB: This should be smarter. Why bother flushing pages that
+	 * will be truncated anyway? Also, should we error out here if
+	 * the flush returns error?
+	 */
+	rc = filemap_write_and_wait(inode->i_mapping);
+	if (rc != 0) {
+		cifsInode->write_behind_rc = rc;
+		rc = 0;
 	}
 
 	if (attrs->ia_valid & ATTR_SIZE) {
@@ -1904,20 +1905,21 @@ cifs_setattr_nounix(struct dentry *direntry, struct iattr *attrs)
 		return -ENOMEM;
 	}
 
-	if ((attrs->ia_valid & ATTR_MTIME) || (attrs->ia_valid & ATTR_SIZE)) {
-		/*
-		   Flush data before changing file size or changing the last
-		   write time of the file on the server. If the
-		   flush returns error, store it to report later and continue.
-		   BB: This should be smarter. Why bother flushing pages that
-		   will be truncated anyway? Also, should we error out here if
-		   the flush returns error?
-		 */
-		rc = filemap_write_and_wait(inode->i_mapping);
-		if (rc != 0) {
-			cifsInode->write_behind_rc = rc;
-			rc = 0;
-		}
+	/*
+	 * Attempt to flush data before changing attributes. We need to do
+	 * this for ATTR_SIZE and ATTR_MTIME for sure, and if we change the
+	 * ownership or mode then we may also need to do this. Here, we take
+	 * the safe way out and just do the flush on all setattr requests. If
+	 * the flush returns error, store it to report later and continue.
+	 *
+	 * BB: This should be smarter. Why bother flushing pages that
+	 * will be truncated anyway? Also, should we error out here if
+	 * the flush returns error?
+	 */
+	rc = filemap_write_and_wait(inode->i_mapping);
+	if (rc != 0) {
+		cifsInode->write_behind_rc = rc;
+		rc = 0;
 	}
 
 	if (attrs->ia_valid & ATTR_SIZE) {
-- 
1.5.5.6