[PATCH v4 1/2] ethdev: fix one address occupies two indexes in MAC addrs

[lihuisong (C)]

在 2022/6/2 21:54, Andrew Rybchenko 写道:
> Cc more driver maintainers
>
> On 6/2/22 06:16, lihuisong (C) wrote:
>>
>> 在 2022/6/2 1:49, Andrew Rybchenko 写道:
>>> On 6/1/22 09:39, Min Hu (Connor) wrote:
>>>> From: Huisong Li <lihuisong at huawei.com>
>>>>
>>>> The dev->data->mac_addrs[0] will be changed to a new MAC address when
>>>> applications modify the default MAC address by
>>>> rte_eth_dev_default_mac_addr_set(). However, if the new default one 
>>>> has
>>>> been added as a non-default MAC address by 
>>>> rte_eth_dev_mac_addr_add(), the
>>>> the rte_eth_dev_default_mac_addr_set() doesn't remove it from the 
>>>> mac_addrs
>>>> list. As a result, one MAC address occupies two indexes in the 
>>>> list. Like:
>>>> add(MAC1)
>>>> add(MAC2)
>>>> add(MAC3)
>>>> add(MAC4)
>>>> set_default(MAC3)
>>>> default=MAC3, filters=MAC1, MAC2, MAC3, MAC4
>>>>
>>>> In addition, some PMDs, such as i40e, ice, hns3 and so on, do 
>>>> remove the
>>>> old default MAC when set default MAC. If user continues to do
>>>> set_default(MAC5), and the mac_addrs list is default=MAC5, 
>>>> filters=(MAC1,
>>>> MAC2, MAC3, MAC4). At this moment, user can still view MAC3 from 
>>>> the list,
>>>> but packets with MAC3 aren't actually received by the PMD.
>>>
>>> IMHO, the main problem is inconsistency which exists right now.
>>> rte_eth_dev_mac_addr_add() checks for duplicate MAC addition
>>> including the default one (index zero) and extends the entry
>>> pool mask (including zero entry case).
>>>
>>> However, the patch above does not extend zero entry pool mask.
>>> So, the result will depend on order which is bad:
>>> A. Set default to A, add MAC A with pool 2 => pool mask has 2
>>> B. Add MAC A with pool 2, set default to A => pool mask is empty
>>>
>> I don't know how this MAC pool works in which driver.
>
> Me too
>
>> However, the 'eth_dev_mac_restore' API show that 1) the default MAC 
>> has only pool zero
>> if set it by the 'mac_addr_add', 2) the default one hasn't pool 
>> information if set it
>> by 'default_mac_addr_set'.
>>
>> Do you mean we should inherit its pool mask in this case?
>
> I simply want to make it consistent
I know what you mean. However, from the implementations mentioned above, 
it seems that
drivers used pool mask don't care the pool mask of the default MAC.
Do we need to set pool mask for the default MAC?
>
>>> Am I missing something in the code?
>>> What is the right/intended behaviour?
>>>
>>>>
>>>> Fixes: 854d8ad4ef68 ("ethdev: add default mac address modifier")
>>>> Cc: stable at dpdk.org
>>>>
>>>> Signed-off-by: Huisong Li <lihuisong at huawei.com>
>>>> Signed-off-by: Min Hu <humin29 at huawei.com>
>>>> ---
>>>>   lib/ethdev/rte_ethdev.c | 39 +++++++++++++++++++++++++++++++++++++--
>>>>   1 file changed, 37 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/lib/ethdev/rte_ethdev.c b/lib/ethdev/rte_ethdev.c
>>>> index 46c088dc88..fc9ca8d6fd 100644
>>>> --- a/lib/ethdev/rte_ethdev.c
>>>> +++ b/lib/ethdev/rte_ethdev.c
>>>> @@ -4260,7 +4260,10 @@ rte_eth_dev_mac_addr_remove(uint16_t 
>>>> port_id, struct rte_ether_addr *addr)
>>>>   int
>>>>   rte_eth_dev_default_mac_addr_set(uint16_t port_id, struct 
>>>> rte_ether_addr *addr)
>>>>   {
>>>> +    uint64_t mac_pool_sel_bk = 0;
>>>>       struct rte_eth_dev *dev;
>>>> +    uint32_t pool;
>>>> +    int index;
>>>>       int ret;
>>>>         RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
>>>> @@ -4278,16 +4281,48 @@ rte_eth_dev_default_mac_addr_set(uint16_t 
>>>> port_id, struct rte_ether_addr *addr)
>>>> RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->mac_addr_set, -ENOTSUP);
>>>>   +    /*
>>>> +     * If the address has been added as a non-default MAC address by
>>>> +     * rte_eth_dev_mac_addr_add API, it should be removed from
>>>> +     * dev->data->mac_addrs[].
>>>> +     */
>>>> +    index = eth_dev_get_mac_addr_index(port_id, addr);
>>>> +    if (index > 0) {
>>>> +        /* remove address in NIC data structure */
>>>> +        mac_pool_sel_bk = dev->data->mac_pool_sel[index];
>>>> +        ret = rte_eth_dev_mac_addr_remove(port_id, addr);
>>>> +        if (ret < 0) {
>>>> +            RTE_ETHDEV_LOG(ERR,
>>>> +            "Delete MAC address from the MAC list of ethdev port 
>>>> %u.\n",
>>>> +            port_id);
>>>> +            return ret;
>>>> +        }
>>>> +        /* reset pool bitmap */
>>>> +        dev->data->mac_pool_sel[index] = 0;
>>>> +    }
>>>> +
>>>>       ret = (*dev->dev_ops->mac_addr_set)(dev, addr);
>>>>       if (ret < 0)
>>>> -        return ret;
>>>> +        goto back;
>>>>         /* Update default address in NIC data structure */
>>>>       rte_ether_addr_copy(addr, &dev->data->mac_addrs[0]);
>>>>         return 0;
>>>> -}
>>>>   +back:
>>>> +    if (index > 0) {
>>>> +        pool = 0;
>>>> +        do {
>>>> +            if (mac_pool_sel_bk & UINT64_C(1))
>>>> +                rte_eth_dev_mac_addr_add(port_id, addr, pool);
>>>
>>> Don't we want to have at least error logs in the case of rollback
>>> failure here?
>> It doesn't feel necessary. It may trigger the printing of a large 
>> number of error logs
>> in abnormal scenarios.
>
> Otherwise how will user know that rollback failed and configuration
> is inconsistent?
Ack.
>
>>>
>>>> +            mac_pool_sel_bk >>= 1;
>>>> +            pool++;
>>>> +        } while (mac_pool_sel_bk);
>>>
>>> Please, compare vs 0 explicitly.
>> Ack
>>>
>>>> +    }
>>>> +
>>>> +    return ret;
>>>> +}
>>>>     /*
>>>>    * Returns index into MAC address array of addr. Use 
>>>> 00:00:00:00:00:00 to find
>>>
>>> .
>
> .