On Tue, Aug 12, 2014 at 03:31:43PM -0400, Trond Myklebust wrote: > On Tue, Aug 12, 2014 at 3:16 PM, Bruce Fields <bfields@xxxxxxxxxxxx> wrote: > > On Sun, Aug 03, 2014 at 01:03:06PM -0400, Trond Myklebust wrote: > >> We already determined that there was enough wspace when we > >> called svc_xprt_enqueue. > > > > So xpo_has_wspace may have returned true then, but I don't see what > > guarantees it still would now. Couldn't another server thread have also > > run svc_recv() and the atomic_add(rqstp->rq_reserved, > > &xprt->xpt_reserved) between the svc_xprt_enqueue call and now? > > The point is that all this is just a heuristic: the TCP send window > can collapse at any time. So rather than waste cycles doing multiple > redundant tests that ultimately mean nothing when you call sendmsg(), > do it once and be damned... OK, makes sense. --b. -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
