[PATCH v5 2/2] drivers/net: return number of types in get supported types

[Ferruh Yigit]

On 1/23/2024 2:59 PM, Ferruh Yigit wrote:
> On 1/23/2024 12:07 PM, Power, Ciara wrote:
>> Hi Ferruh,
>>
>>> -----Original Message-----
>>> From: Ferruh Yigit <ferruh.yigit at amd.com>
>>> Sent: Friday, January 19, 2024 2:51 PM
>>> To: Sivaramakrishnan, VenkatX <venkatx.sivaramakrishnan at intel.com>; Igor
>>> Russkikh <irusskikh at marvell.com>; Selwin Sebastian
>>> <selwin.sebastian at amd.com>; Ajit Khaparde
>>> <ajit.khaparde at broadcom.com>; Somnath Kotur
>>> <somnath.kotur at broadcom.com>; Nithin Dabilpuram
>>> <ndabilpuram at marvell.com>; Kiran Kumar K <kirankumark at marvell.com>;
>>> Sunil Kumar Kori <skori at marvell.com>; Satha Rao
>>> <skoteshwar at marvell.com>; Zhang, Yuying <yuying.zhang at intel.com>; Xing,
>>> Beilei <beilei.xing at intel.com>; Rahul Lakkireddy
>>> <rahul.lakkireddy at chelsio.com>; Hemant Agrawal
>>> <hemant.agrawal at nxp.com>; Sachin Saxena <sachin.saxena at nxp.com>; Su,
>>> Simei <simei.su at intel.com>; Wu, Wenjun1 <wenjun1.wu at intel.com>;
>>> Gagandeep Singh <g.singh at nxp.com>; John Daley <johndale at cisco.com>;
>>> Hyong Youb Kim <hyonkim at cisco.com>; Gaetan Rivet <grive at u256.net>;
>>> Zhang, Qi Z <qi.z.zhang at intel.com>; Wang, Xiao W <xiao.w.wang at intel.com>;
>>> Jie Hai <haijie1 at huawei.com>; Yisen Zhuang <yisen.zhuang at huawei.com>;
>>> Wu, Jingjing <jingjing.wu at intel.com>; Yang, Qiming
>>> <qiming.yang at intel.com>; Guo, Junfeng <junfeng.guo at intel.com>; Andrew
>>> Boyer <andrew.boyer at amd.com>; Long Li <longli at microsoft.com>; Matan
>>> Azrad <matan at nvidia.com>; Viacheslav Ovsiienko <viacheslavo at nvidia.com>;
>>> Dariusz Sosnowski <dsosnowski at nvidia.com>; Ori Kam <orika at nvidia.com>;
>>> Suanming Mou <suanmingm at nvidia.com>; Chaoyong He
>>> <chaoyong.he at corigine.com>; Jiawen Wu <jiawenwu at trustnetic.com>;
>>> Harman Kalra <hkalra at marvell.com>; Devendra Singh Rawat
>>> <dsinghrawat at marvell.com>; Alok Prasad <palok at marvell.com>; Andrew
>>> Rybchenko <andrew.rybchenko at oktetlabs.ru>; Jerin Jacob
>>> <jerinj at marvell.com>; Maciej Czekaj <mczekaj at marvell.com>; Jian Wang
>>> <jianwang at trustnetic.com>; Behrens, Jochen <jbehrens at vmware.com>;
>>> Thomas Monjalon <thomas at monjalon.net>
>>> Cc: dev at dpdk.org; Power, Ciara <ciara.power at intel.com>
>>> Subject: Re: [PATCH v5 2/2] drivers/net: return number of types in get
>>> supported types
>>>
>>> On 1/18/2024 12:07 PM, Sivaramakrishnan Venkat wrote:
>>>> Missing "RTE_PTYPE_UNKNOWN" ptype causes buffer overflow.
>>>> Enhance code such that the dev_supported_ptypes_get() function pointer
>>>> now returns  the number of elements to eliminate the need for
>>>> "RTE_PTYPE_UNKNOWN" as the last item.
>>>>
>>>> Signed-off-by: Sivaramakrishnan Venkat
>>>> <venkatx.sivaramakrishnan at intel.com>
>>>>
>>>> --
>>>>   v5:
>>>>      - modified commit message.
>>>>      - tidied formatting of code.
>>>>      - added doxygen comment.
>>>>   v4:
>>>>      - split into two patches, one for backporting and another one for
>>>>        upstream rework.
>>>>   v3:
>>>>      - reworked the function to return number of elements and remove the
>>>>        need for RTE_PTYPE_UNKNOWN in list.
>>>>   v2:
>>>>      - extended fix for multiple drivers.
>>>> ---
>>>
>>> <...>
>>>
>>>>  59 files changed, 188 insertions(+), 141 deletions(-)
>>>>
>>>
>>> Some driver still have the flag:
>>> - drivers/net/mvneta/mvneta_ethdev.c
>>> - drivers/net/mvpp2/mrvl_ethdev.c
>>> - pfe
>>> - dpaa
>>> - drivers/net/thunderx/nicvf_ethdev.c
>>> - drivers/net/nfp/nfp_net_common.c
>>>
>>> Above seems the ones updated in previous patch, flags added in previous
>>> patch should be removed in this one.
>>>
>>>
>>> And following seems missed and still has the flag:
>>>
>>> - drivers/net/ngbe/ngbe_ptypes.c
>>>
>>> <...>
>>>
>>>> @@ -3971,9 +3975,6 @@ rte_eth_dev_set_ptypes(uint16_t port_id,
>>> uint32_t ptype_mask,
>>>>  		}
>>>>  	}
>>>>
>>>> -	if (set_ptypes != NULL && j < num)
>>>> -		set_ptypes[j] = RTE_PTYPE_UNKNOWN;
>>>> -
>>>>
>>>
>>> This change is new in this version and not mentioned in the changelog.
>>>
>>> 'rte_eth_dev_set_ptypes()' returns 'set_ptypes' that terminated with
>>> 'RTE_PTYPE_UNKNOWN', this is how that API works.
>>> Why changing it in this patch?
>>
>> Apologies, yes, we missed this in the changelog.
>>
>> For the change itself, if we are removing the need for RTE_PTYPE_UNKNOWN in the supported ptypes lists to mark the last element, do we still need to add it here when setting ptypes list?
>> Maybe a misunderstanding on my part - I thought it would be the same for both cases.
>>
>>
> 
> They are two different APIs, and 'rte_eth_dev_set_ptypes()' returns
> ptypes to user that will be terminated by RTE_PTYPE_UNKNOWN, so to not
> break the user code we can't change it.
> 

A little more details if helps:
'rte_eth_dev_get_supported_ptypes()', fills the '*ptypes' and function
return value is number of elements in the '*ptypes' array. There is no
requirement that last element of '*ptypes' will be 'RTE_PTYPE_UNKNOWN',
and if you check the current implementation, it is not.

'rte_eth_dev_set_ptypes()', fills the '*set_ptypes' array and function
return value is function success status. User can deduce the size of
'*set_ptypes' by 'RTE_PTYPE_UNKNOWN' marker at the end of the array.
Requirement is last element should be 'RTE_PTYPE_UNKNOWN', and this
documented in API.

We are preserving above API behavior and only changing ethdev - driver
interface.