From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: linux-nfs-owner@vger.kernel.org Received: from fieldses.org ([174.143.236.118]:44390 "EHLO fieldses.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756674Ab2D3VpH (ORCPT ); Mon, 30 Apr 2012 17:45:07 -0400 Date: Mon, 30 Apr 2012 17:45:04 -0400 From: "J. Bruce Fields" To: Stanislav Kinsbursky Cc: "Trond.Myklebust@netapp.com" , "linux-nfs@vger.kernel.org" , "linux-kernel@vger.kernel.org" , "devel@openvz.org" Subject: Re: [PATCH 2/3] SUNRPC: check rpcbind clients usage counter before decrement Message-ID: <20120430214504.GD22323@fieldses.org> References: <20120425132314.13800.6813.stgit@localhost6.localdomain6> <20120425133749.13800.24003.stgit@localhost6.localdomain6> <20120427135507.GE14973@fieldses.org> <4F9AA84D.8090604@parallels.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <4F9AA84D.8090604@parallels.com> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Fri, Apr 27, 2012 at 06:08:13PM +0400, Stanislav Kinsbursky wrote: > On 27.04.2012 17:55, J. Bruce Fields wrote: > >On Wed, Apr 25, 2012 at 05:37:49PM +0400, Stanislav Kinsbursky wrote: > >>Registering service with svc_bind() can fail. In this case service will be > >>destroyed and during destruction it will try to unregister itself from rpcbind. > >>In this case unregister have to be skipped. > > > >Isn't this a preexisting bug, in which case perhaps Trond should at it > >to his list of bugs to submit now? > > > > Not it's not. > > Previously, in case of bind operations failure, rpcb_put_local() > wasn't called, but service data was destroyed instead: > > - if (svc_uses_rpcbind(serv)) { > - if (svc_rpcb_setup(serv, current->nsproxy->net_ns) < 0) { > - kfree(serv->sv_pools); > - kfree(serv); > - return NULL; > - } > - if (!serv->sv_shutdown) > - serv->sv_shutdown = svc_rpcb_cleanup; > - } > > Now (with svc_bind() introduction) service can be created, but > svc_bind() could fail. And thus svc_destroy() have to be called. > Which will call rpcb_put_local(). > > And the problem here that svc_bind() can fail before incrementing of > rpcb usage counter. OK, but then you're fixing a bug that you just introduced with the previous patch. So this should be combined with the previous patch. --b. > > >--b. > > > >> > >>Signed-off-by: Stanislav Kinsbursky > >> > >>--- > >> net/sunrpc/rpcb_clnt.c | 12 +++++++----- > >> 1 files changed, 7 insertions(+), 5 deletions(-) > >> > >>diff --git a/net/sunrpc/rpcb_clnt.c b/net/sunrpc/rpcb_clnt.c > >>index 78ac39f..4c38b33 100644 > >>--- a/net/sunrpc/rpcb_clnt.c > >>+++ b/net/sunrpc/rpcb_clnt.c > >>@@ -180,14 +180,16 @@ void rpcb_put_local(struct net *net) > >> struct sunrpc_net *sn = net_generic(net, sunrpc_net_id); > >> struct rpc_clnt *clnt = sn->rpcb_local_clnt; > >> struct rpc_clnt *clnt4 = sn->rpcb_local_clnt4; > >>- int shutdown; > >>+ int shutdown = 0; > >> > >> spin_lock(&sn->rpcb_clnt_lock); > >>- if (--sn->rpcb_users == 0) { > >>- sn->rpcb_local_clnt = NULL; > >>- sn->rpcb_local_clnt4 = NULL; > >>+ if (sn->rpcb_users) { > >>+ if (--sn->rpcb_users == 0) { > >>+ sn->rpcb_local_clnt = NULL; > >>+ sn->rpcb_local_clnt4 = NULL; > >>+ } > >>+ shutdown = !sn->rpcb_users; > >> } > >>- shutdown = !sn->rpcb_users; > >> spin_unlock(&sn->rpcb_clnt_lock); > >> > >> if (shutdown) { > >> > > > -- > Best regards, > Stanislav Kinsbursky