Re: [dpdk-dev] [PATCH v4 2/2] ethdev: update qfi definition

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

On 4/13/2021 10:24 AM, Ori Kam wrote:
> Hi Raslan,
> 
>> -----Original Message-----
>> From: Raslan Darawsheh <rasland@nvidia.com>
>> Subject: RE: [PATCH v4 2/2] ethdev: update qfi definition
>>
>> Hi,
>>
>>> -----Original Message-----
>>> From: Ferruh Yigit <ferruh.yigit@intel.com>
>>> Sent: Tuesday, April 6, 2021 7:10 PM
>>> To: Raslan Darawsheh <rasland@nvidia.com>; dev@dpdk.org; Ori Kam
>>> <orika@nvidia.com>; andrew.rybchenko@oktetlabs.ru
>>> Cc: ivan.malov@oktetlabs.ru; ying.a.wang@intel.com;
>>> olivier.matz@6wind.com; Slava Ovsiienko <viacheslavo@nvidia.com>; Shiri
>>> Kuzin <shirik@nvidia.com>; stable@dpdk.org; David Marchand
>>> <david.marchand@redhat.com>; NBU-Contact-Thomas Monjalon
>>> <thomas@monjalon.net>
>>> Subject: Re: [PATCH v4 2/2] ethdev: update qfi definition
>>>
>>> On 4/4/2021 8:45 AM, Raslan Darawsheh wrote:
>>>> qfi field is 8 bits which represent single bit for
>>>> PPP (paging Policy Presence) single bit for RQI
>>>> (Reflective QoS Indicator) and 6 bits for qfi
>>>> (QoS Flow Identifier) based on RFC 38415-g30
>>>>
>>>> This update the doxygen format and the mask for qfi
>>>> to properly identify the full 8 bits of the field.
>>>>
>>>> note: changing the default mask would cause different
>>>> patterns generated by testpmd.
>>>>
>>>> Fixes: 346553db5bd1 ("ethdev: add GTP extension header to flow API")
>>>> Cc: ying.a.wang@intel.com
>>>> Cc: stable@dpdk.org
>>>>
>>>> Signed-off-by: Raslan Darawsheh <rasland@nvidia.com>
>>>> ---
>>>>    doc/guides/testpmd_app_ug/testpmd_funcs.rst |  3 ++-
>>>>    lib/librte_ethdev/rte_flow.h                | 20 +++++++++++++++++---
>>>>    2 files changed, 19 insertions(+), 4 deletions(-)
>>>>
>>>> diff --git a/doc/guides/testpmd_app_ug/testpmd_funcs.rst
>>> b/doc/guides/testpmd_app_ug/testpmd_funcs.rst
>>>> index f59eb8a27d..dd39c4c3c2 100644
>>>> --- a/doc/guides/testpmd_app_ug/testpmd_funcs.rst
>>>> +++ b/doc/guides/testpmd_app_ug/testpmd_funcs.rst
>>>> @@ -3742,7 +3742,8 @@ This section lists supported pattern items and
>>> their attributes, if any.
>>>>    - ``gtp_psc``: match GTP PDU extension header with type 0x85.
>>>>
>>>>      - ``pdu_type {unsigned}``: PDU type.
>>>> -  - ``qfi {unsigned}``: QoS flow identifier.
>>>> +
>>>> +  - ``qfi {unsigned}``: PPP, RQI and QoS flow identifier.
>>>>
>>>>    - ``pppoes``, ``pppoed``: match PPPoE header.
>>>>
>>>> diff --git a/lib/librte_ethdev/rte_flow.h b/lib/librte_ethdev/rte_flow.h
>>>> index 6cc57136ac..20b4389429 100644
>>>> --- a/lib/librte_ethdev/rte_flow.h
>>>> +++ b/lib/librte_ethdev/rte_flow.h
>>>> @@ -20,6 +20,7 @@
>>>>    #include <rte_arp.h>
>>>>    #include <rte_common.h>
>>>>    #include <rte_ether.h>
>>>> +#include <rte_gtp.h>
>>>>    #include <rte_icmp.h>
>>>>    #include <rte_ip.h>
>>>>    #include <rte_sctp.h>
>>>> @@ -1421,16 +1422,29 @@ static const struct rte_flow_item_meta
>>> rte_flow_item_meta_mask = {
>>>>     *
>>>>     * Matches a GTP PDU extension header with type 0x85.
>>>>     */
>>>> +RTE_STD_C11
>>>>    struct rte_flow_item_gtp_psc {
>>>> -	uint8_t pdu_type; /**< PDU type. */
>>>> -	uint8_t qfi; /**< QoS flow identifier. */
>>>> +	union {
>>>> +		struct {
>>>> +			/*
>>>> +			 * These fields are retained for compatibility.
>>>> +			 * Please switch to the new header field below.
>>>> +			 */
>>>> +			uint8_t pdu_type; /**< PDU type. */
>>>> +			uint8_t qfi; /**< PPP, RQI, QoS flow identifier. */
>>>> +
>>>> +		};
>>>> +		struct rte_gtp_psc_hdr hdr;
>>>> +	};
>>>>    };
>>>
>>> This will change the struct size even with union, since old version is missing
>>> all fields from protocol header. Struct size will go from 2 --> 5 bytes [1].
>>>
>>> Since this is public struct, we can't change its size.
>>>
>>> @Ori, Andrew,
>>>
>>> Do you have a suggestion for next step?
>>>
> 
> I think Ferruh is right, and I think that we should at this point just update the documentation.
> Sorry for the detour
> Just small comment about the original  patch.
> I don’t think you should change the default mask since it may break existing apps.
> 

I will continue with first patch [1], and will update the protocol reference in 
the commit log.

Meanwhile adding 'struct rte_gtp_psc_hdr' (patch v4 1/2) work can continue 
separately.
And when it is added we can send a deprecation notice to update 'struct 
rte_flow_item_gtp_psc' and finally update it on v21.11 to have 'struct 
rte_gtp_psc_hdr' in it. Makes sense?

[1]
https://patches.dpdk.org/project/dpdk/patch/20210323121134.19113-1-rasland@nvidia.com/

>>> - We can still add the "struct rte_gtp_psc_hdr", and add a deprecation notice
>>> for "struct rte_flow_item_gtp_psc" to say it will use  "struct
>>> rte_gtp_psc_hdr"
>>> on 21.11.
>>>
>>> - And for this release use the Raslan's first version:
>>>     -	uint8_t qfi; /**< QoS flow identifier. */
>>>     +	uint8_t qfi; /**< PPP, RQI, QoS flow identifier. */
>>>
>>>
>>> Does it make sense? Any comments?
>>>
>> @Ori Kam, @andrew.rybchenko@oktetlabs.ru,
>> This is a kind remainder of this patch,
>>
>>>
>>>
>>> [1]
>>> struct rte_gtp_psc_hdr {
>>>           uint8_t                    ext_hdr_len;          /*     0     1 */
>>>           uint8_t                    type:4;               /*     1: 0  1 */
>>>           uint8_t                    qmp:1;                /*     1: 4  1 */
>>>
>>>           /* XXX 3 bits hole, try to pack */
>>>
>>>           union {
>>>                   struct {
>>>                           uint8_t    snp:1;                /*     2: 0  1 */
>>>                           uint8_t    spare_dl1:2;          /*     2: 1  1 */
>>>                   };                                       /*     2     1 */
>>>                   struct {
>>>                           uint8_t    dl_delay_ind:1;       /*     2: 0  1 */
>>>                           uint8_t    ul_delay_ind:1;       /*     2: 1  1 */
>>>                           uint8_t    snp_ul1:1;            /*     2: 2  1 */
>>>                   };                                       /*     2     1 */
>>>           };                                               /*     2     1 */
>>>           union {
>>>                   struct {
>>>                           uint8_t    ppp:1;                /*     3: 0  1 */
>>>                           uint8_t    rqi:1;                /*     3: 1  1 */
>>>                   };                                       /*     3     1 */
>>>                   struct {
>>>                           uint8_t    n_delay_ind:1;        /*     3: 0  1 */
>>>                           uint8_t    spare_ul2:1;          /*     3: 1  1 */
>>>                   };                                       /*     3     1 */
>>>           };                                               /*     3     1 */
>>>           uint8_t                    qfi:6;                /*     4: 0  1 */
>>>
>>>           /* XXX 2 bits hole, try to pack */
>>>
>>>           uint8_t                    data[];               /*     5     0 */
>>>
>>>           /* size: 5, cachelines: 1, members: 7 */
>>>           /* sum members: 3 */
>>>           /* sum bitfield members: 11 bits, bit holes: 2, sum bit holes: 5 bits */
>>>           /* last cacheline: 5 bytes */
>>> };
>>>
>>>>
>>>>    /** Default mask for RTE_FLOW_ITEM_TYPE_GTP_PSC. */
>>>>    #ifndef __cplusplus
>>>>    static const struct rte_flow_item_gtp_psc
>>>>    rte_flow_item_gtp_psc_mask = {
>>>> -	.qfi = 0x3f,
>>>> +	.hdr.ppp = 1,
>>>> +	.hdr.rqi = 1,
>>>> +	.hdr.qfi = 0x3f,
>>>>    };
>>>>    #endif
>>>>
>>>>
>