[dpdk-dev] [PATCH 2/3] app/testpmd: fix slave device isn't released

[lihuisong (C)]

在 2022/2/4 20:14, Ferruh Yigit 写道:
> On 10/25/2021 7:39 AM, Min Hu (Connor) wrote:
>> From: Huisong Li <lihuisong at huawei.com>
>>
>> Currently, some eth devices are added to bond device, these devices 
>> are not
>> released when the quit command is executed in testpmd. This patch 
>> adds the
>> release operation for all active slaves under a bond device.
>>
>
> 'close_port()' function traverses all ports, when bonding is close, if it
> removes the slaves, won't 'close_port()' close slave devices?
Yes
>
> If so this prevents, 'cl_quit' global variable.
The variable is used to ensure that the slave ports are not closed when the
bonding port is closed, and are closed when testpmd quit or is killed.
>
>> Fixes: 0e545d3047fe ("app/testpmd: check stopping port is not in 
>> bonding")
>> Cc: stable at dpdk.org
>>
>> Signed-off-by: Huisong Li <lihuisong at huawei.com>
>> Signed-off-by: Min Hu (Connor) <humin29 at huawei.com>
>> ---
>>   app/test-pmd/cmdline.c |  1 +
>>   app/test-pmd/testpmd.c | 67 ++++++++++++++++++++++++++++++++++++------
>>   app/test-pmd/testpmd.h |  1 +
>>   3 files changed, 60 insertions(+), 9 deletions(-)
>>
>> diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
>> index 5bfb4b509b..98236a8be3 100644
>> --- a/app/test-pmd/cmdline.c
>> +++ b/app/test-pmd/cmdline.c
>> @@ -8743,6 +8743,7 @@ static void cmd_quit_parsed(__rte_unused void 
>> *parsed_result,
>>                   __rte_unused void *data)
>>   {
>>       cmdline_quit(cl);
>> +    cl_quit = 1;
>>   }
>>     cmdline_parse_token_string_t cmd_quit_quit =
>> diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c
>> index d6b9ebc4dd..fea9744bd6 100644
>> --- a/app/test-pmd/testpmd.c
>> +++ b/app/test-pmd/testpmd.c
>> @@ -221,6 +221,7 @@ unsigned int xstats_display_num; /**< Size of 
>> extended statistics to show */
>>    * option. Set flag to exit stats period loop after received 
>> SIGINT/SIGTERM.
>>    */
>>   uint8_t f_quit;
>> +uint8_t cl_quit; /* Quit testpmd from cmdline. */
>>     /*
>>    * Max Rx frame size, set by '--max-pkt-len' parameter.
>> @@ -613,15 +614,6 @@ eth_dev_start_mp(uint16_t port_id)
>>       return 0;
>>   }
>>   -static int
>> -eth_dev_stop_mp(uint16_t port_id)
>> -{
>> -    if (is_proc_primary())
>> -        return rte_eth_dev_stop(port_id);
>> -
>> -    return 0;
>> -}
>> -
>>   static void
>>   mempool_free_mp(struct rte_mempool *mp)
>>   {
>> @@ -3123,6 +3115,55 @@ remove_invalid_ports(void)
>>       nb_cfg_ports = nb_fwd_ports;
>>   }
>>   +#ifdef RTE_NET_BOND
>> +static void
>> +handle_bonding_slave_device(portid_t bond_pid)
>
> 'handle' in the function is not very specific, it is not clear
> what this function does.
ok
>
>> +{
>> +    portid_t slave_pids[RTE_MAX_ETHPORTS];
>> +    struct rte_port *port;
>> +    portid_t slave_pid;
>> +    int num_slaves;
>> +    int i;
>> +
>> +    num_slaves = rte_eth_bond_slaves_get(bond_pid, slave_pids,
>> +                         RTE_MAX_ETHPORTS);
>> +    if (num_slaves < 0) {
>> +        fprintf(stderr, "Failed to get slave list for port = %u\n",
>> +            bond_pid);
>> +        return;
>> +    }
>> +
>> +    for (i = 0; i < num_slaves; i++) {
>> +        slave_pid = slave_pids[i];
>> +        /* Before removing a slave device, stop the slave device. */
>> +        if (port_is_started(slave_pid) == 1) {
>> +            if (rte_eth_dev_stop(slave_pid) != 0)
>> +                fprintf(stderr, "rte_eth_dev_stop failed for port 
>> %u\n",
>> +                    slave_pid);
>> +
>> +            port = &ports[slave_pid];
>> +            if (rte_atomic16_cmpset(&(port->port_status),
>> +                RTE_PORT_STARTED, RTE_PORT_STOPPED) == 0)
>> +                fprintf(stderr, "Port %u can not be set into 
>> stopped\n",
>> +                    slave_pid);
>> +        }
>> +
>> +        /*
>> +         * Remove the slave from a bonded device in case of failing to
>> +         * close bond device.
>> +         */
>> +        if (rte_eth_bond_slave_remove(bond_pid, slave_pid) != 0)
>
> Similar to Aman's comment in previous patch, if a bonding device is 
> removed
> shouldn't the bonding PMD stop the slaves and removed them, instead of 
> application?

I agree. I plan to remove them, and move the operations of clearing the 
slave flag

and closing slave port to after the bonding port is closed.

>
>> +            fprintf(stderr, "Failed to remove slave %u from master 
>> port = %u\n",
>> +                slave_pid, bond_pid);
>> +        clear_port_slave_flag(slave_pid);
>> +
>> +        /* Close slave device when testpmd quit or is killed. */
>> +        if (cl_quit == 1 || f_quit == 1)
>> +            rte_eth_dev_close(slave_pid);
>> +    }
>> +}
>> +#endif
>> +
>>   void
>>   close_port(portid_t pid)
>>   {
>> @@ -3161,6 +3202,14 @@ close_port(portid_t pid)
>>             if (is_proc_primary()) {
>>               port_flow_flush(pi);
>> +#ifdef RTE_NET_BOND
>> +            /*
>> +             * If this port is bond device, all slaves under the
>> +             * device need to be removed or closed.
>> +             */
>> +            if (port->bond_flag == 1)
>> +                handle_bonding_slave_device(pi);
>> +#endif
>>               port_flex_item_flush(pi);
>>               rte_eth_dev_close(pi);
>>           }
>> diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h
>> index ad3b4f875c..216fc41432 100644
>> --- a/app/test-pmd/testpmd.h
>> +++ b/app/test-pmd/testpmd.h
>> @@ -27,6 +27,7 @@
>>   #define RTE_PORT_STARTED        (uint16_t)1
>>   #define RTE_PORT_CLOSED         (uint16_t)2
>>   #define RTE_PORT_HANDLING       (uint16_t)3
>> +extern uint8_t cl_quit;
>>     /*
>>    * It is used to allocate the memory for hash key.
>
> .