On Tue, Sep 14, 2010 at 6:39 AM, Jeff Layton <jlayton@xxxxxxxxxx> wrote:
> On Tue, 14 Sep 2010 15:18:32 +0530
> Suresh Jayaraman <sjayaraman@xxxxxxx> wrote:
>
>> On 09/13/2010 11:32 PM, Jeff Layton wrote:
>> > cifs_get_smb_ses must be called on a server pointer on which it holds an
>> > active reference. It first does a search for an existing SMB session. If
>> > it finds one, it'll put the server reference and then try to ensure that
>> > the negprot is done, etc.
>> >
>> > If it encounters an error at that point then it'll return an error.
>> > There's a potential problem here though. When cifs_get_smb_ses returns
>> > an error, the caller will also put the TCP server reference leading to a
>> > double-put.
>> >
>> > Fix this by having cifs_get_smb_ses only put the server reference if
>> > it found an existing session that it could use and isn't returning an
>> > error.
>> >
>> > Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx>
>> > ---
>> > fs/cifs/connect.c | 6 +++---
>> > 1 files changed, 3 insertions(+), 3 deletions(-)
>> >
>> > diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
>> > index 67dad54..88c84a3 100644
>> > --- a/fs/cifs/connect.c
>> > +++ b/fs/cifs/connect.c
>> > @@ -1706,9 +1706,6 @@ cifs_get_smb_ses(struct TCP_Server_Info *server, struct smb_vol *volume_info)
>> > if (ses) {
>> > cFYI(1, "Existing smb sess found (status=%d)", ses->status);
>> >
>> > - /* existing SMB ses has a server reference already */
>> > - cifs_put_tcp_session(server);
>> > -
>> > mutex_lock(&ses->session_mutex);
>> > rc = cifs_negotiate_protocol(xid, ses);
>> > if (rc) {
>> > @@ -1731,6 +1728,9 @@ cifs_get_smb_ses(struct TCP_Server_Info *server, struct smb_vol *volume_info)
>> > }
>> > }
>> > mutex_unlock(&ses->session_mutex);
>> > +
>> > + /* existing SMB ses has a server reference already */
>> > + cifs_put_tcp_session(server);
>> > FreeXid(xid);
>> > return ses;
>> > }
>>
>> Looks correct to me.
>>
>> Reviewed-by: Suresh Jayaraman <sjayaraman@xxxxxxx>
>
> Thanks. In hindsight, I should have probably sent this to stable too.
>
> Steve, would you like me to resend and cc stable? We probably want this
> in 2.6.36 too, if possible.
Yes - I agree makes sense. Any other candidate patches for 2.6.36?
--
Thanks,
Steve
--
To unsubscribe from this list: send the line "unsubscribe linux-cifs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
