From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753613Ab0IMPxZ (ORCPT ); Mon, 13 Sep 2010 11:53:25 -0400 Received: from caiajhbdcbhh.dreamhost.com ([208.97.132.177]:50124 "EHLO homiemail-a61.g.dreamhost.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1752219Ab0IMPxY (ORCPT ); Mon, 13 Sep 2010 11:53:24 -0400 Subject: [PATCH] net/9p: fix memory handling/allocation in rdma_request() From: Davidlohr Bueso Reply-To: dave@gnu.org To: ericvh@gmail.com, rminnich@sandia.gov, lucho@ionkov.net Cc: v9fs-developer@lists.sourceforge.net, LKML Content-Type: text/plain; charset="UTF-8" Date: Mon, 13 Sep 2010 11:53:18 -0400 Message-ID: <1284393198.2055.1.camel@cowboy> Mime-Version: 1.0 X-Mailer: Evolution 2.28.3 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org [PATCH] net/9p: fix memory handling/allocation in rdma_request() Return -ENOMEM when erroring on kmalloc and fix memory leaks when returning on error. Signed-off-by: Davidlohr Bueso --- net/9p/trans_rdma.c | 29 ++++++++++++++++++----------- 1 files changed, 18 insertions(+), 11 deletions(-) diff --git a/net/9p/trans_rdma.c b/net/9p/trans_rdma.c index 0ea20c3..17c5ba7 100644 --- a/net/9p/trans_rdma.c +++ b/net/9p/trans_rdma.c @@ -426,8 +426,10 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req) /* Allocate an fcall for the reply */ rpl_context = kmalloc(sizeof *rpl_context, GFP_KERNEL); - if (!rpl_context) + if (!rpl_context) { + err = -ENOMEM; goto err_close; + } /* * If the request has a buffer, steal it, otherwise @@ -445,8 +447,8 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req) } rpl_context->rc = req->rc; if (!rpl_context->rc) { - kfree(rpl_context); - goto err_close; + err = -ENOMEM; + goto err_free2; } /* @@ -458,11 +460,8 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req) */ if (atomic_inc_return(&rdma->rq_count) <= rdma->rq_depth) { err = post_recv(client, rpl_context); - if (err) { - kfree(rpl_context->rc); - kfree(rpl_context); - goto err_close; - } + if (err) + goto err_free1; } else atomic_dec(&rdma->rq_count); @@ -471,8 +470,10 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req) /* Post the request */ c = kmalloc(sizeof *c, GFP_KERNEL); - if (!c) - goto err_close; + if (!c) { + err = -ENOMEM; + goto err_free1; + } c->req = req; c->busa = ib_dma_map_single(rdma->cm_id->device, @@ -499,9 +500,15 @@ static int rdma_request(struct p9_client *client, struct p9_req_t *req) return ib_post_send(rdma->qp, &wr, &bad_wr); error: + kfree(c); + kfree(rpl_context->rc); + kfree(rpl_context); P9_DPRINTK(P9_DEBUG_ERROR, "EIO\n"); return -EIO; - + err_free1: + kfree(rpl_context->rc); + err_free2: + kfree(rpl_context); err_close: spin_lock_irqsave(&rdma->req_lock, flags); if (rdma->state < P9_RDMA_CLOSING) { -- 1.7.0.4