[dpdk-dev] [PATCH] vhost: fix driver unregister for client mode

Ilya Maximets i.maximets at samsung.com
Thu Jul 21 10:33:13 CEST 2016


Thanks. Fixed.

Best regards, Ilya Maximets.

On 21.07.2016 11:24, Yuanhan Liu wrote:
> On Wed, Jul 20, 2016 at 11:32:43AM +0300, Ilya Maximets wrote:
>> Currently while calling of 'rte_vhost_driver_unregister()' connection
>> to QEMU will not be closed. This leads to inability to register driver
>> again and reconnect to same virtual machine.
>>
>> This scenario is reproducible with OVS. While executing of the following
>> command vhost port will be re-created (will be executed
>> 'rte_vhost_driver_register()' followed by 'rte_vhost_driver_unregister()')
>> network will be broken and QEMU possibly will crash:
>>
>> 	ovs-vsctl set Interface vhost1 ofport_request=15
>>
>> Fix this by closing all established connections on driver unregister and
>> removing of pending connections from reconnection list.
>>
>> Fixes: 64ab701c3d1e ("vhost: add vhost-user client mode")
>> Signed-off-by: Ilya Maximets <i.maximets at samsung.com>
> 
> Again, thanks for the fix.
> 
>> ---
>>
>> Patch prepared for master branch because dpdk-next-virtio doesn't contain
>> commit acbff5c67ea7 ("vhost: fix crash when exceeding file descriptors").
>> Porting to dpdk-next-virtio/master is trivial and may be performed on
>> demand.
> 
> Yeah, my bad, I haven't updated it after rc2, since Thomas no longer
> pull request from it.  Anyway, you just remind me that I should have
> done that.
> 
>>  /**
>>   * Unregister the specified vhost socket
>>   */
>> @@ -672,20 +700,34 @@ rte_vhost_driver_unregister(const char *path)
>>  {
>>  	int i;
>>  	int count;
>> +	struct vhost_user_connection *conn;
>>  
>>  	pthread_mutex_lock(&vhost_user.mutex);
>>  
>>  	for (i = 0; i < vhost_user.vsocket_cnt; i++) {
>> -		if (!strcmp(vhost_user.vsockets[i]->path, path)) {
>> -			if (vhost_user.vsockets[i]->is_server) {
>> -				fdset_del(&vhost_user.fdset,
>> -					vhost_user.vsockets[i]->listenfd);
>> -				close(vhost_user.vsockets[i]->listenfd);
>> +		struct vhost_user_socket *vsocket = vhost_user.vsockets[i];
>> +
>> +		if (!strcmp(vsocket->path, path)) {
>> +			if (vsocket->is_server) {
>> +				(void) fdset_del(&vhost_user.fdset,
>> +				                 vsocket->listenfd);
> 
> I would think the (void) cast is not neceessary here.
> 
>> +				close(vsocket->listenfd);
>>  				unlink(path);
>> +			} else if (vsocket->reconnect) {
>> +				vhost_user_remove_reconnect(vsocket);
>> +			}
>> +
>> +			conn = fdset_del(&vhost_user.fdset, vsocket->connfd);
>> +			if (conn) {
>> +				RTE_LOG(INFO, VHOST_CONFIG, "free connfd = %d"
>> +				    "for device '%s'\n", vsocket->connfd, path);
> 
> We should try not to break a log message into several lines, which
> hurts "git grep".  Here, it could be:
> 
> 				RTE_LOG(INFO, VHOST_CONFIG,
> 					"free connfd = %d for device '%s'\n",
> 					vsocket->connfd, path);
> 
> Besides the two minor nits,
> 
> Acked-by: Yuanhan Liu <yuanhan.liu at linux.intel.com>
> 
> 	--yliu
> 
> 


More information about the dev mailing list