From mboxrd@z Thu Jan 1 00:00:00 1970 From: Maxime Coquelin Subject: [PATCH v2 3/3] vhost: don't take access_lock on VHOST_USER_RESET_OWNER Date: Mon, 12 Feb 2018 16:46:12 +0100 Message-ID: <20180212154612.5297-4-maxime.coquelin@redhat.com> References: <20180212154612.5297-1-maxime.coquelin@redhat.com> Cc: dev@dpdk.org, stable@dpdk.org, zhihong.wang@intel.com, qian.q.xu@intel.com, lei.a.yao@intel.com, Maxime Coquelin To: tiwei.bie@intel.com, yliu@fridaylinux.org, ferruh.yigit@intel.com, victork@redhat.com, thomas@monjalon.net, olivier.matz@6wind.com, jianfeng.tan@intel.com Return-path: In-Reply-To: <20180212154612.5297-1-maxime.coquelin@redhat.com> List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" A deadlock happens when handling VHOST_USER_RESET_OWNER request for the same reason the lock is not taken for VHOST_USER_GET_VRING_BASE. It is safe not to take the lock, as the queues are no more used by the application when the virtqueues and the device are reset. Fixes: a3688046995f ("vhost: protect active rings from async ring changes") Cc: stable@dpdk.org Cc: Victor Kaplansky Reviewed-by: Tiwei Bie Reviewed-by: Jianfeng Tan Signed-off-by: Maxime Coquelin --- lib/librte_vhost/vhost_user.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/lib/librte_vhost/vhost_user.c b/lib/librte_vhost/vhost_user.c index 65ee33919..90ed2112e 100644 --- a/lib/librte_vhost/vhost_user.c +++ b/lib/librte_vhost/vhost_user.c @@ -1348,16 +1348,16 @@ vhost_user_msg_handler(int vid, int fd) } /* - * Note: we don't lock all queues on VHOST_USER_GET_VRING_BASE, - * since it is sent when virtio stops and device is destroyed. - * destroy_device waits for queues to be inactive, so it is safe. - * Otherwise taking the access_lock would cause a dead lock. + * Note: we don't lock all queues on VHOST_USER_GET_VRING_BASE + * and VHOST_USER_RESET_OWNER, since it is sent when virtio stops + * and device is destroyed. destroy_device waits for queues to be + * inactive, so it is safe. Otherwise taking the access_lock + * would cause a dead lock. */ switch (msg.request.master) { case VHOST_USER_SET_FEATURES: case VHOST_USER_SET_PROTOCOL_FEATURES: case VHOST_USER_SET_OWNER: - case VHOST_USER_RESET_OWNER: case VHOST_USER_SET_MEM_TABLE: case VHOST_USER_SET_LOG_BASE: case VHOST_USER_SET_LOG_FD: -- 2.14.3