On Mon, 2018-09-03 at 13:11 -0400, Chuck Lever wrote:
> > On Sep 3, 2018, at 11:29 AM, Trond Myklebust <trondmy@xxxxxxxxx>
> > wrote:
> >
> > Currently, we grab the socket bit lock before we allow the message
> > to be XDR encoded. That significantly slows down the transmission
> > rate, since we serialise on a potentially blocking operation.
>
> Which operation blocks, and how often?
RPCSEC_GSS allocates memory when doing privacy encoding, for instance.
> > Signed-off-by: Trond Myklebust <trond.myklebust@xxxxxxxxxxxxxxx>
> > ---
> > net/sunrpc/clnt.c | 8 ++++----
> > 1 file changed, 4 insertions(+), 4 deletions(-)
> >
> > diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
> > index e5ac35e803ad..66ec61347716 100644
> > --- a/net/sunrpc/clnt.c
> > +++ b/net/sunrpc/clnt.c
> > @@ -1949,9 +1949,6 @@ call_transmit(struct rpc_task *task)
> > task->tk_action = call_status;
> > if (task->tk_status < 0)
> > return;
> > - if (!xprt_prepare_transmit(task))
> > - return;
> > - task->tk_action = call_transmit_status;
> > /* Encode here so that rpcsec_gss can use correct sequence
> > number. */
> > if (rpc_task_need_encode(task)) {
> > rpc_xdr_encode(task);
> > @@ -1965,8 +1962,11 @@ call_transmit(struct rpc_task *task)
> > return;
> > }
> > }
> > + if (!xprt_prepare_transmit(task))
> > + return;
> > + task->tk_action = call_transmit_status;
> > xprt_transmit(task);
> > - if (task->tk_status < 0)
> > + if (task->tk_status < 0) {
>
> The added curly bracket seems incorrect.
Oops. Yes, that's a rebase brain-fart... Thanks for spotting it!
>
>
> > return;
> > if (is_retrans)
> > task->tk_client->cl_stats->rpcretrans++;
> > --
> > 2.17.1
> >
>
> --
> Chuck Lever
> chucklever@xxxxxxxxx
>
>
>
