smb3: fix broken cached reads when posix locks

commit e4be320eeca842a3d7648258ee3673f1755a5a59 upstream.

Mandatory locking is enforced for cached reads, which violates
default posix semantics, and also it is enforced inconsistently.
This affected recent versions of libreoffice, and can be
demonstrated by opening a file twice from the same client,
locking it from handle one and trying to read from it from
handle two (which fails, returning EACCES).

There is already a mount option "forcemandatorylock"
(which defaults to off), so with this change only when the user
intentionally specifies "forcemandatorylock" on mount will we
break posix semantics on read to a locked range (ie we will
only fail in this case, if the user mounts with
"forcemandatorylock").

An earlier patch fixed the write path.

Fixes: 85160e03a79e ("CIFS: Implement caching mechanism for mandatory brlocks")
Cc: stable@vger.kernel.org
Cc: Pavel Shilovsky <piastryyy@gmail.com>
Reviewed-by: David Howells <dhowells@redhat.com>
Reported-by: abartlet@samba.org
Reported-by: Kevin Ottens <kevin.ottens@enioka.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

diff --git a/fs/smb/client/file.c b/fs/smb/client/file.c
index 06a0667f8ff201b108669b632915d108195a17e9..b202eac6584e13be1c39d49cc7a97cb355a746b6 100644 (file)
--- a/fs/smb/client/file.c
+++ b/fs/smb/client/file.c
@@ -2911,9 +2911,7 @@ cifs_strict_readv(struct kiocb *iocb, struct iov_iter *to)
        if (!CIFS_CACHE_READ(cinode))
                return netfs_unbuffered_read_iter(iocb, to);
 
-       if (cap_unix(tcon->ses) &&
-           (CIFS_UNIX_FCNTL_CAP & le64_to_cpu(tcon->fsUnixInfo.Capability)) &&
-           ((cifs_sb->mnt_cifs_flags & CIFS_MOUNT_NOPOSIXBRL) == 0)) {
+       if ((cifs_sb->mnt_cifs_flags & CIFS_MOUNT_NOPOSIXBRL) == 0) {
                if (iocb->ki_flags & IOCB_DIRECT)
                        return netfs_unbuffered_read_iter(iocb, to);
                return netfs_buffered_read_iter(iocb, to);