On Tue, Mar 05, 2024 at 08:12:03AM +0800, Edward Adam Davis wrote:
> On Mon, 4 Mar 2024 17:07:07 +0000, Simon Horman wrote:
> > > If connection isn't established yet, get_mr() will fail, trigger connection after
> > > get_mr().
> > >
> > > Fixes: 584a8279a44a ("RDS: RDMA: return appropriate error on rdma map failures")
> > > Reported-and-tested-by: syzbot+d4faee732755bba9838e@xxxxxxxxxxxxxxxxxxxxxxxxx
> > > Signed-off-by: Edward Adam Davis <eadavis@xxxxxx>
> > > ---
> > > net/rds/rdma.c | 3 +++
> > > net/rds/send.c | 6 +-----
> > > 2 files changed, 4 insertions(+), 5 deletions(-)
> > >
> > > diff --git a/net/rds/rdma.c b/net/rds/rdma.c
> > > index fba82d36593a..a4e3c5de998b 100644
> > > --- a/net/rds/rdma.c
> > > +++ b/net/rds/rdma.c
> > > @@ -301,6 +301,9 @@ static int __rds_rdma_map(struct rds_sock *rs, struct rds_get_mr_args *args,
> > > kfree(sg);
> > > }
> > > ret = PTR_ERR(trans_private);
> > > + /* Trigger connection so that its ready for the next retry */
> > > + if (ret == -ENODEV)
> > > + rds_conn_connect_if_down(cp->cp_conn);
> >
> > Hi Edward,
> >
> > Elsewhere in this function it is assumed that cp may be NULL.
> > Does that need to be taken into account here too?
> Don't worry about this, if it is NULL, the get_mr() return value will not be -ENODEV.
Thanks, understood.
