Re: [dpdk-dev] [dpdk-stable] [PATCH] net: fix Tx VLAN flag for offload emulation

[Ferruh Yigit <ferruh.yigit@intel.com>]

On 3/26/2019 6:01 PM, Chas Williams wrote:
> 
> 
> On 3/26/19 12:50 PM, Ferruh Yigit wrote:
>> On 3/25/2019 3:05 PM, Chas Williams wrote:
>>> From: Bill Hong <bhong@brocade.com>
>>>
>>> A PMD might use rte_vlan_insert to implement Tx VLAN offload.  Typically
>>> the PMD will insert the VLAN header in the transmit path and then attempt
>>> to send the packets. If this fails, the packets are returned to
>>> the application which may attempt to send these packets again. If the
>>> PKT_TX_VLAN flag is not cleared, the transmit path may attempt to insert
>>> the VLAN header again.
>>>
>>> Fixes: 47aa48b969f8 ("net: fix stripped VLAN flag for offload emulation");
>>> Cc: stable@dpdk.org
>>>
>>> Signed-off-by: Bill Hong <bhong@brocade.com>
>>> Signed-off-by: Chas Williams <chas3@att.com>
>>> ---
>>>   lib/librte_net/rte_ether.h | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/lib/librte_net/rte_ether.h b/lib/librte_net/rte_ether.h
>>> index c2c5e249f..e0d831113 100644
>>> --- a/lib/librte_net/rte_ether.h
>>> +++ b/lib/librte_net/rte_ether.h
>>> @@ -408,7 +408,7 @@ static inline int rte_vlan_insert(struct rte_mbuf **m)
>>>   	vh = (struct vlan_hdr *) (nh + 1);
>>>   	vh->vlan_tci = rte_cpu_to_be_16((*m)->vlan_tci);
>>>   
>>> -	(*m)->ol_flags &= ~PKT_RX_VLAN_STRIPPED;
>>> +	(*m)->ol_flags &= ~(PKT_RX_VLAN_STRIPPED | PKT_TX_VLAN);
>>
>> Hi Chas,
>>
>> Looks reasonable, AFAIK, PKT_TX_VLAN flag means 'mbuf->vlan_tci' has valid value
>> and a request from application to driver to insert VLAN information.
>> After successful insertion flag can be removed.
>>
>> But in mbuf header, flags documented as:
>> PKT_TX_VLAN: TX packet is a 802.1q VLAN packet.
>> PKT_TX_QINQ: TX packet with double VLAN inserted.

Hi Olivier,

Any comment on them, should they be fixed?

> 
> Note the usage slightly below:
> 
> 	/**
> 	 * Bitmask of all supported packet Tx offload features flags,
> 	 * which can be set for packet.
> 	 */
> 	#define PKT_TX_OFFLOAD_MASK (    \
> 			PKT_TX_OUTER_IPV6 |      \
> 			PKT_TX_OUTER_IPV4 |      \
> 			PKT_TX_OUTER_IP_CKSUM |  \
> 			PKT_TX_VLAN_PKT |        \
> 
> So this implies that PKT_TX_VLAN_PKT (which is PKT_TX_VLAN's old name)
> is an offload feature, not that the actual packet has an 802.1q header
> (offloaded or otherwise).
> 
> The same seems to apply for PKT_TX_QINQ.

Agree, I just would like to confirm, so lgtm:
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

> 
>> Hi Oliver,
>>
>> Is above comments correct, or I am missing something J
>>
>>
>> Also here cleaning 'PKT_RX_VLAN_STRIPPED' looks like to say 'vlan' information
>> is not stripped but in the packet data, but 'RX' in this context can be
>> confusing, should we have a more generic 'PKT_VLAN_STRIPPED' ?
>>