[linux-cifs-client] non-direct reading
diff mbox

Message ID 49D218F1.3080006@etersoft.ru
State New, archived
Headers show

Commit Message

Pavel Shilovsky March 31, 2009, 1:21 p.m. UTC
Hello!

I found an interesting bug in cifs driver. When we don't use "direct" 
option during mounting, we read only from cache. My tests prove that it 
isn't right - if client don't keep oplock and inode->clientCanCacheRead 
== false, we should read from server if we want to get valid data.

Here is patch, that provide reading from cache only if it's valid.


 static loff_t cifs_llseek(struct file *file, loff_t offset, int origin)
 {
        /* origin == SEEK_END => we must revalidate the cached file 
length */
@@ -733,7 +747,7 @@ const struct inode_operations cifs_symlink_inode_ops = {
 const struct file_operations cifs_file_ops = {
        .read = do_sync_read,
        .write = do_sync_write,
-       .aio_read = generic_file_aio_read,
+       .aio_read = cifs_file_aio_read,
        .aio_write = cifs_file_aio_write,
        .open = cifs_open,
        .release = cifs_close,


Signed-off-by: Pavel Shilovsky <piastryyy@gmail.com>



--
Best regards,
Pavel Shilovsky.

Patch
diff mbox

diff --git a/fs/cifs/cifsfs.c b/fs/cifs/cifsfs.c
index 3ae62fb..7fe72d4 100644
--- a/fs/cifs/cifsfs.c
+++ b/fs/cifs/cifsfs.c
@@ -622,6 +622,20 @@  static ssize_t cifs_file_aio_write(struct kiocb 
*iocb, const struct iovec *iov,
        return written;
 }

+static ssize_t cifs_file_aio_read(struct kiocb *iocb, const struct 
iovec *iov,
+                                 unsigned long nr_segs, loff_t pos)
+{
+       struct inode *inode = iocb->ki_filp->f_path.dentry->d_inode;
+       ssize_t read;
+
+       if (CIFS_I(inode)->clientCanCacheRead)
+               read = generic_file_aio_read(iocb, iov, nr_segs, pos);
+       else
+               read = cifs_user_read(iocb->ki_filp, iov->iov_base, 
iov->iov_len, &pos);
+       return read;
+}
+
+