On Wed, 2025-03-26 at 06:13 -0400, Jeff Layton wrote:
> On Tue, 2025-03-25 at 18:35 -0400, trondmy@xxxxxxxxxx wrote:
> > From: Trond Myklebust <trond.myklebust@xxxxxxxxxxxxxxx>
> >
> > Replace the tests for the RPC client being shut down with tests for
> > whether the nfs_client is in an error state.
> >
> > Signed-off-by: Trond Myklebust <trond.myklebust@xxxxxxxxxxxxxxx>
> > ---
> > fs/nfs/nfs4proc.c | 2 +-
> > fs/nfs/nfs4state.c | 2 +-
> > 2 files changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
> > index 889511650ceb..50be54e0f578 100644
> > --- a/fs/nfs/nfs4proc.c
> > +++ b/fs/nfs/nfs4proc.c
> > @@ -9580,7 +9580,7 @@ static void nfs41_sequence_call_done(struct
> > rpc_task *task, void *data)
> > return;
> >
> > trace_nfs4_sequence(clp, task->tk_status);
> > - if (task->tk_status < 0 && !task->tk_client->cl_shutdown)
> > {
> > + if (task->tk_status < 0 && clp->cl_cons_state >= 0) {
> > dprintk("%s ERROR %d\n", __func__, task-
> > >tk_status);
> > if (refcount_read(&clp->cl_count) == 1)
> > return;
> > diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
> > index 542cdf71229f..f1f7eaa97973 100644
> > --- a/fs/nfs/nfs4state.c
> > +++ b/fs/nfs/nfs4state.c
> > @@ -1198,7 +1198,7 @@ void nfs4_schedule_state_manager(struct
> > nfs_client *clp)
> > struct rpc_clnt *clnt = clp->cl_rpcclient;
> > bool swapon = false;
> >
> > - if (clnt->cl_shutdown)
> > + if (clp->cl_cons_state < 0)
> > return;
> >
> > set_bit(NFS4CLNT_RUN_MANAGER, &clp->cl_state);
>
> One more thing:
>
> Do we need cl_shutdown at all? If we can replace these checks here
> with
> a check for cl_cons_state < 0, why not do the same in call_start()?
The struct nfs_client is a NFS level object. It can't be moved to the
RPC layer.
--
Trond Myklebust
Linux NFS client maintainer, Hammerspace
trond.myklebust@xxxxxxxxxxxxxxx
