FYI this patch did not fix mount after resume from suspend.
I still see "CIFS: VFS: cifs_tree_connect: could not find superblock:
-22" errors with the cifs mount unusable after resume.
Regards,
Satadru
On Wed, Mar 16, 2022 at 11:44 PM ronnie sahlberg
<ronniesahlberg@xxxxxxxxx> wrote:
>
> On Thu, Mar 17, 2022 at 1:20 PM Steve French <smfrench@xxxxxxxxx> wrote:
> >
> > cifssmb3: fix incorrect session setup check for multiuser mounts
>
> If it fixes multiuser then Acked-by me.
> We are so close to rc8 that we can not do intrusive changes, so if
> it fixes it short term.
> For longer term, post rc8 we need to rewrite the statemaching completely
> and separate out "what happens in server->tcpStatus" as one statemachine and
> "what happens in server->status" as a separate one. Right now it is a mess.
>
>
> >
> > A recent change to how the SMB3 server (socket) and session status
> > is managed regressed multiuser mounts by changing the check
> > for whether session setup is needed to the socket (TCP_Server_info)
> > structure instead of the session struct (cifs_ses). Add additional
> > check in cifs_setup_sesion to fix this.
> >
> > Fixes: 73f9bfbe3d81 ("cifs: maintain a state machine for tcp/smb/tcon sessions")
> > Reported-by: Ronnie Sahlberg <lsahlber@xxxxxxxxxx>
> > Signed-off-by: Steve French <stfrench@xxxxxxxxxxxxx>
> > ---
> > fs/cifs/connect.c | 3 ++-
> > 1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
> > index 053cb449eb16..d3020abfe404 100644
> > --- a/fs/cifs/connect.c
> > +++ b/fs/cifs/connect.c
> > @@ -3924,7 +3924,8 @@ cifs_setup_session(const unsigned int xid,
> > struct cifs_ses *ses,
> >
> > /* only send once per connect */
> > spin_lock(&cifs_tcp_ses_lock);
> > - if (server->tcpStatus != CifsNeedSessSetup) {
> > + if ((server->tcpStatus != CifsNeedSessSetup) &&
> > + (ses->status == CifsGood)) {
> > spin_unlock(&cifs_tcp_ses_lock);
> > return 0;
> > }
> >
> > --
> > Thanks,
> >
> > Steve
