diff options
| author | Paulo Alcantara <pc@manguebit.com> | 2024-11-26 15:55:53 -0300 |
|---|---|---|
| committer | Greg Kroah-Hartman <gregkh@linuxfoundation.org> | 2024-12-14 20:04:10 +0100 |
| commit | a21406d31414fd9eb199e2046df48873aaf58daf (patch) | |
| tree | e6ff7016d116f8da2aacffb58ffc38d464b88324 /scripts | |
| parent | 9ac8c1da9f82e622cbe1d4422848c324d3706787 (diff) | |
| download | linux-a21406d31414fd9eb199e2046df48873aaf58daf.tar.gz linux-a21406d31414fd9eb199e2046df48873aaf58daf.tar.bz2 linux-a21406d31414fd9eb199e2046df48873aaf58daf.zip | |
smb: client: don't try following DFS links in cifs_tree_connect()
[ Upstream commit 36008fe6e3dc588e5e9ceae6e82c7f69399eb5d8 ]
We can't properly support chasing DFS links in cifs_tree_connect()
because
(1) We don't support creating new sessions while we're reconnecting,
which would be required for DFS interlinks.
(2) ->is_path_accessible() can't be called from cifs_tree_connect()
as it would deadlock with smb2_reconnect(). This is required for
checking if new DFS target is a nested DFS link.
By unconditionally trying to get an DFS referral from new DFS target
isn't correct because if the new DFS target (interlink) is an DFS
standalone namespace, then we would end up getting -ELOOP and then
potentially leaving tcon disconnected.
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Diffstat (limited to 'scripts')
0 files changed, 0 insertions, 0 deletions
