[dpdk-dev] [PATCH] af_packet: make the device detachable

Iremonger, Bernard bernard.iremonger at intel.com
Tue Feb 9 10:34:11 CET 2016


 Hi Wojciech

<snip>

> > -----Original Message-----
> > From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Wojciech Zmuda
> > Sent: Tuesday, January 5, 2016 2:05 PM
> > To: dev at dpdk.org
> > Subject: [dpdk-dev] [PATCH] af_packet: make the device detachable
> >
> > Fix memory leak when detaching virtual device. Set dev_flags to
> > RTE_ETH_DEV_DETACHABLE and implement pmd_af_packet_drv.uninit
> method.
> > Copy device name to ethdev->data to make it compatibile with
> 
> Typo, compatibile should be compatible.
> 
> > rte_eth_dev_allocated().
> >
> > Signed-off-by: Wojciech Zmuda <woz at semihalf.com>
> > ---
> >  drivers/net/af_packet/rte_eth_af_packet.c | 29
> > ++++++++++++++++++++++++++++-
> >  1 file changed, 28 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/net/af_packet/rte_eth_af_packet.c
> > b/drivers/net/af_packet/rte_eth_af_packet.c
> > index 767f36b..7ef65ff 100644
> > --- a/drivers/net/af_packet/rte_eth_af_packet.c
> > +++ b/drivers/net/af_packet/rte_eth_af_packet.c
> > @@ -667,11 +667,13 @@ rte_pmd_init_internals(const char *name,
> >  	data->nb_tx_queues = (uint16_t)nb_queues;
> >  	data->dev_link = pmd_link;
> >  	data->mac_addrs = &(*internals)->eth_addr;
> > +	strncpy(data->name,
> > +		(*eth_dev)->data->name, strlen((*eth_dev)->data-
> > >name));
> >
> >  	(*eth_dev)->data = data;
> >  	(*eth_dev)->dev_ops = &ops;
> >  	(*eth_dev)->driver = NULL;
> > -	(*eth_dev)->data->dev_flags = 0;
> > +	(*eth_dev)->data->dev_flags = RTE_ETH_DEV_DETACHABLE;
> >  	(*eth_dev)->data->drv_name = drivername;
> >  	(*eth_dev)->data->kdrv = RTE_KDRV_NONE;
> >  	(*eth_dev)->data->numa_node = numa_node; @@ -836,10 +838,35
> @@ exit:
> >  	return ret;
> >  }
> >
> > +static int
> > +rte_pmd_af_packet_devuninit(const char *name) {
> > +	struct rte_eth_dev *eth_dev = NULL;
> > +
> > +	RTE_LOG(INFO, PMD, "Closing AF_PACKET ethdev on numa socket
> > %u\n",
> > +			rte_socket_id());
> > +
> > +	if (name == NULL)
> > +		return -1;
> > +
> > +	/* reserve an ethdev entry */
> 
> Above comment should be 	/* find the ethdev entry */
> 
> > +	eth_dev = rte_eth_dev_allocated(name);
> > +	if (eth_dev == NULL)
> > +		return -1;
> > +
> 
> In rte_pmd_init_internals() memory is allocated for rx and tx queues this
> should also be freed.
> 
> Line 606: rx_queue->rd = rte_zmalloc_socket(name, rdsize, 0, numa_node);
> Line 620:tx_queue->rd = rte_zmalloc_socket(name, rdsize, 0, numa_node);
> 
> 	rte_free((*internals)->rx_queue[q].rd);
> 	rte_free((*internals)->tx_queue[q].rd);
> 
> > +	rte_free(eth_dev->data->dev_private);
> > +	rte_free(eth_dev->data);
> > +
> > +	rte_eth_dev_release_port(eth_dev);
> > +
> > +	return 0;
> > +}
> > +
> >  static struct rte_driver pmd_af_packet_drv = {
> >  	.name = "eth_af_packet",
> >  	.type = PMD_VDEV,
> >  	.init = rte_pmd_af_packet_devinit,
> > +	.uninit = rte_pmd_af_packet_devuninit,
> >  };
> >
> >  PMD_REGISTER_DRIVER(pmd_af_packet_drv);
> > --
> > 1.9.1

Forgot to mention that there should probably be and entry in the 2.3 release notes, new features section, for this.

Regards,

Bernard.




More information about the dev mailing list