[dpdk-dev] [PATCH v2 4/4] app/testpmd: show example to handle hot unplug

[Matan Azrad]

Hi Jeff

Please see comments...

From: Jeff Guo
> Sent: Friday, June 22, 2018 2:51 PM
> To: stephen at networkplumber.org; bruce.richardson at intel.com;
> ferruh.yigit at intel.com; konstantin.ananyev at intel.com;
> gaetan.rivet at 6wind.com; jingjing.wu at intel.com; Thomas Monjalon
> <thomas at monjalon.net>; Mordechay Haimovsky <motih at mellanox.com>;
> Matan Azrad <matan at mellanox.com>; harry.van.haaren at intel.com;
> qi.z.zhang at intel.com; shaopeng.he at intel.com
> Cc: jblunck at infradead.org; shreyansh.jain at nxp.com; dev at dpdk.org;
> jia.guo at intel.com; helin.zhang at intel.com
> Subject: [PATCH v2 4/4] app/testpmd: show example to handle hot unplug
> 
> Use testpmd for example, to show how an application smoothly handle failure
> when device being hot unplug. If app have enabled the device event monitor
> and register the hot plug event’s callback before running, once app detect the
> removal event, the callback would be called. It will first stop the packet
> forwarding, then stop the port, close the port, and finally detach the port to
> remove the device out from the device lists.
> 
> Signed-off-by: Jeff Guo <jia.guo at intel.com>
> ---
> v2->v1(v21):
> rebase testpmd code
> ---
>  app/test-pmd/testpmd.c | 25 ++++++++++++++++++++-----
>  1 file changed, 20 insertions(+), 5 deletions(-)
> 
> diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c index
> 24c1998..286f242 100644
> --- a/app/test-pmd/testpmd.c
> +++ b/app/test-pmd/testpmd.c
> @@ -1951,9 +1951,10 @@ eth_dev_event_callback_unregister(void)
>  void
>  attach_port(char *identifier)
>  {
> -	portid_t pi = 0;
>  	unsigned int socket_id;
> 
> +	portid_t pi = rte_eth_dev_count_avail();
> +
>  	printf("Attaching a new port...\n");
> 
>  	if (identifier == NULL) {
> @@ -2125,16 +2126,25 @@ check_all_ports_link_status(uint32_t port_mask)
> static void  rmv_event_callback(void *arg)  {

There is a race between ethdev RMV event to the EAL remove event, I think the application must synchronize it if both are configured.

> +	struct rte_eth_dev *dev;
> +
>  	int need_to_start = 0;
>  	int org_no_link_check = no_link_check;
>  	portid_t port_id = (intptr_t)arg;
> 
>  	RTE_ETH_VALID_PORTID_OR_RET(port_id);
> +	dev = &rte_eth_devices[port_id];
> +
> +	if (dev->state == RTE_ETH_DEV_UNUSED)
> +		return;

Can you explain why do you check the state?
Doesn't RTE_ETH_VALID_PORTID_OR_RET do it?

> +	printf("removing device %s\n", dev->device->name);
> 
>  	if (!test_done && port_is_forwarding(port_id)) {
>  		need_to_start = 1;
>  		stop_packet_forwarding();
>  	}
> +
>  	no_link_check = 1;
>  	stop_port(port_id);
>  	no_link_check = org_no_link_check;
> @@ -2196,6 +2206,9 @@ static void
>  eth_dev_event_callback(char *device_name, enum rte_dev_event_type type,
>  			     __rte_unused void *arg)
>  {
> +	uint16_t port_id;
> +	int ret;
> +
>  	if (type >= RTE_DEV_EVENT_MAX) {
>  		fprintf(stderr, "%s called upon invalid event %d\n",
>  			__func__, type);
> @@ -2206,9 +2219,12 @@ eth_dev_event_callback(char *device_name, enum
> rte_dev_event_type type,
>  	case RTE_DEV_EVENT_REMOVE:
>  		RTE_LOG(ERR, EAL, "The device: %s has been removed!\n",
>  			device_name);
> -		/* TODO: After finish failure handle, begin to stop
> -		 * packet forward, stop port, close port, detach port.
> -		 */
> +		ret = rte_eth_dev_get_port_by_name(device_name, &port_id);
> +		if (ret) {
> +			printf("can not get port by device %s!\n",
> device_name);
> +			return;
> +		}
> +		rmv_event_callback((void *)(intptr_t)port_id);
>  		break;
>  	case RTE_DEV_EVENT_ADD:
>  		RTE_LOG(ERR, EAL, "The device: %s has been added!\n", @@ -
> 2736,7 +2752,6 @@ main(int argc, char** argv)
>  			return -1;
>  		}
>  		eth_dev_event_callback_register();
> -
>  	}
> 
>  	if (start_port(RTE_PORT_ALL) != 0)
> --
> 2.7.4