[dpdk-dev] [PATCH v3] vhost: fix connect hang in client mode

Yuanhan Liu yuanhan.liu at linux.intel.com
Thu Jul 21 15:56:15 CEST 2016


On Thu, Jul 21, 2016 at 04:43:25PM +0300, Ilya Maximets wrote:
> 
> 
> On 21.07.2016 16:35, Yuanhan Liu wrote:
> > On Thu, Jul 21, 2016 at 04:19:35PM +0300, Ilya Maximets wrote:
> >> If something abnormal happened to QEMU, 'connect()' can block calling
> >> thread (e.g. main thread of OVS) forever or for a really long time.
> >> This can break whole application or block the reconnection thread.
> >>
> >> Example with OVS:
> >>
> >> 	ovs_rcu(urcu2)|WARN|blocked 512000 ms waiting for main to quiesce
> >> 	(gdb) bt
> >> 	#0  connect () from /lib64/libpthread.so.0
> >> 	#1  vhost_user_create_client (vsocket=0xa816e0)
> >> 	#2  rte_vhost_driver_register
> >> 	#3  netdev_dpdk_vhost_user_construct
> >> 	#4  netdev_open (name=0xa664b0 "vhost1")
> >> 	[...]
> >> 	#11 main
> >>
> >> Fix that by setting non-blocking mode for client sockets for connection.
> >>
> >> Fixes: 64ab701c3d1e ("vhost: add vhost-user client mode")
> >>
> >> Signed-off-by: Ilya Maximets <i.maximets at samsung.com>
> > 
> > Acked-by: Yuanhan Liu <yuanhan.liu at linux.intel.com>
> > 
> > One help I'd like to ask is that I'd appriciate if you could do the test
> > to make sure that your 2 (latest) patches fix the two issues you reported.
> > 
> > You might have already done that; I just want to make sure.
> 
> I've performed the test with 'ofport_request' script before sending patches.
> And currently test still works. No leaks of descriptors, no hangs,
> no QEMU crashes observed.
> Sometimes network device breaks on QEMU side, but it's QEMU issue. In this
> case I'm receiving following message from DPDK's vhost:
> 
> VHOST_CONFIG: vhost-user client: socket created, fd: 28
> VHOST_CONFIG: failed to connect to /vhost1: Resource temporarily unavailable
> VHOST_CONFIG: /vhost1: reconnecting...
> 
> Before the 'hang' patch there was hang of main thread.
> 
> After QEMU restart all works normally. OVS restart not required.

Good to know and appreciate that!

	--yliu



More information about the dev mailing list