From: Michael Chan Subject: [PATCH net-next 1/3] cnic: Use proper ulp_ops for per device operations. Date: Mon, 17 Mar 2014 19:19:06 -0800 Message-ID: <1395112748-21965-2-git-send-email-mchan@broadcom.com> References: <1395112748-21965-1-git-send-email-mchan@broadcom.com> Mime-Version: 1.0 Content-Type: text/plain Cc: , To: Return-path: Received: from mail-gw2-out.broadcom.com ([216.31.210.63]:54062 "EHLO mail-gw2-out.broadcom.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753493AbaCRCAQ (ORCPT ); Mon, 17 Mar 2014 22:00:16 -0400 In-Reply-To: <1395112748-21965-1-git-send-email-mchan@broadcom.com> Sender: netdev-owner@vger.kernel.org List-ID: For per device operations, cnic needs to dereference the RCU protected cp->ulp_ops instead of the global cnic_ulp_tbl. In 2 locations, cnic_send_nlmsg() and cnic_copy_ulp_stats(), it was referencing the global table. If the device has been unregistered and these functions are still being called (very unlikely scenarios), it could lead to NULL pointer dereference. Signed-off-by: Michael Chan --- drivers/net/ethernet/broadcom/cnic.c | 5 +++-- 1 files changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/broadcom/cnic.c b/drivers/net/ethernet/broadcom/cnic.c index fcf9105..2cb2482 100644 --- a/drivers/net/ethernet/broadcom/cnic.c +++ b/drivers/net/ethernet/broadcom/cnic.c @@ -342,7 +342,7 @@ static int cnic_send_nlmsg(struct cnic_local *cp, u32 type, while (retry < 3) { rc = 0; rcu_read_lock(); - ulp_ops = rcu_dereference(cnic_ulp_tbl[CNIC_ULP_ISCSI]); + ulp_ops = rcu_dereference(cp->ulp_ops[CNIC_ULP_ISCSI]); if (ulp_ops) rc = ulp_ops->iscsi_nl_send_msg( cp->ulp_handle[CNIC_ULP_ISCSI], @@ -3244,7 +3244,8 @@ static int cnic_copy_ulp_stats(struct cnic_dev *dev, int ulp_type) int rc; mutex_lock(&cnic_lock); - ulp_ops = cnic_ulp_tbl_prot(ulp_type); + ulp_ops = rcu_dereference_protected(cp->ulp_ops[ulp_type], + lockdep_is_held(&cnic_lock)); if (ulp_ops && ulp_ops->cnic_get_stats) rc = ulp_ops->cnic_get_stats(cp->ulp_handle[ulp_type]); else -- 1.7.1