[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <9098b8dd-65d4-b629-d6ab-6e2fdd041148@oss.nxp.com>
Date: Wed, 27 Sep 2023 15:33:02 +0300
From: "Radu Pirea (OSS)" <radu-nicolae.pirea@....nxp.com>
To: Igor Russkikh <irusskikh@...vell.com>,
Sabrina Dubroca <sd@...asysnail.net>, Dmitry Bezrukov <dbezrukov@...vell.com>
Cc: andrew@...n.ch, hkallweit1@...il.com, linux@...linux.org.uk,
davem@...emloft.net, edumazet@...gle.com, kuba@...nel.org,
pabeni@...hat.com, richardcochran@...il.com,
sebastian.tobuschat@....nxp.com, netdev@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [EXT] Re: [PATCH net-next v5 3/7] net: macsec: indicate next pn
update when offloading
On 26.09.2023 15:16, Igor Russkikh wrote:
> Hi guys,
>
>> On 21.09.2023 18:11, Sabrina Dubroca wrote:
>>> 2023-09-20, 12:22:33 +0300, Radu Pirea (NXP OSS) wrote:
>>>> Indicate next PN update using update_pn flag in macsec_context.
>>>> Offloaded MACsec implementations does not know whether or not the
>>>> MACSEC_SA_ATTR_PN attribute was passed for an SA update and assume
>>>> that next PN should always updated, but this is not always true.
>>>
>>> This should probably go through net so that we can fix some drivers
>>> that are currently doing the wrong thing. octeontx2 should be
>>> fixable. atlantic looks like it would reset the PN to whatever was
>>> read during the last dump, and it's unclear if that can be fixed
>>> (AFAIU set_egress_sa_record writes the whole config at once). mscc
>
> Thats correct, atlantic hardware requires full table to be in data buffer registers.
> I really doubt its possible to skip PN writing.
>
>>> doesn't seem to modify the PN (even if requested -- should it should
>>> reject the update), and mlx5 doesn't allow PN update (by storing the
>>> initial value of next_pn on SA creation).
>>
>> I updated octeontx2, mssc and mlx5. Atlantic is unclear.
>>
>> Mark, Igor, in the atlantic MACsec driver, can the SAs be updated
>> without a PN update?
>
> Reviewed the code and the docs I have - my view is it can not.
> All the packed record in macsec_api.c:set_egress_sa_record is expected by hardware in full.
Ok. I will not touch atlantic.
--
Radu P.
Powered by blists - more mailing lists