netdev - Re: [PATCH net-next v2 0/2] Add support for DSFP transceiver type

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <a9835ab6-70a1-5a15-194e-977ff9c859ec@nvidia.com>
Date:   Thu, 26 Nov 2020 16:23:10 +0200
From:   Moshe Shemesh <moshe@...dia.com>
To:     Andrew Lunn <andrew@...n.ch>
CC:     Jakub Kicinski <kuba@...nel.org>,
        Moshe Shemesh <moshe@...lanox.com>,
        "David S. Miller" <davem@...emloft.net>,
        Adrian Pop <pop.adrian61@...il.com>,
        Michal Kubecek <mkubecek@...e.cz>, <netdev@...r.kernel.org>,
        "Vladyslav Tarasiuk" <vladyslavt@...dia.com>,
        Maxim Mikityanskiy <maximmi@...dia.com>
Subject: Re: [PATCH net-next v2 0/2] Add support for DSFP transceiver type


On 11/25/2020 4:18 PM, Andrew Lunn wrote:
> External email: Use caution opening links or attachments
>
>
>> OK, we will add API options to select bank and page to read any specific
>> page the user selects. So advanced user will use it get the optional pages
>> he needs, but what about non advanced user who wants to use the current API
>> with a current script for DSFP EEPROM. Isn't it better that he will get the
>> 5 mandatory pages then keep it not supported ?
> Users using ethtool will not see a difference. They get a dump of what
> ethtool knows how to decode. It should try the netlink API first, and
> then fall back to the old ioctl interface.
Yes, it makes sense that whenever command not supported by netlink API 
it falls back to old ioctl interface. As I see it we want here to add 
bank and page options to netlink API  to get data from specific page.
>
> If i was implementing the ethtool side of it, i would probably do some
> sort of caching system. We know page 0 should always exist, so
> pre-load that into the cache. Try the netlink API first. If that
> fails, use the ioctl interface. If the ioctl is used, put everything
> returned into the cache.
I am not sure what you mean by cache here. Don't you want to read page 0 
once you got the ethtool command to read from the module ? If not, then 
at what stage ?
>   The decoder can then start decoding, see what
> bits are set indicating other pages should be available. Ask for them
> from the cache. The netlink API can go fetch them and load them into
> the cache. If they cannot be loaded return ENODEV, and the decoder has
> to skip what it wanted to decode.

So the decoder should read page 0 and check according to page 0 and 
specification which pages should be present, right ?

What about the global offset that we currently got when user doesn't 
specify a page, do you mean that this global offset goes through the 
optional and non optional pages that exist and skip the ones that are 
missing according to the specific EEPROM ?

>   If you do it correctly, the decoder
> should not care about ioctl vs netlink.
>
> I can do a follow up patch for the generic SFP code in
> drivers/net/phy, once you have done the first implementation. But i
> only have a limited number of SFPs and most are 1G only. Russell King
> can hopefully test with his collection.
>
>       Andrew