netdev - Re: [PATCH net-next] Revert "net: init sk

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] [day] [month] [year] [list]

Message-ID: <5c1224cd-450a-f32b-c04c-bb83fc2137d0@gmail.com>
Date:   Tue, 24 Apr 2018 09:10:07 -0700
From:   Eric Dumazet <eric.dumazet@...il.com>
To:     Yafang Shao <laoar.shao@...il.com>,
        Eric Dumazet <eric.dumazet@...il.com>
Cc:     David Miller <davem@...emloft.net>, netdev@...r.kernel.org
Subject: Re: [PATCH net-next] Revert "net: init sk_cookie for inet socket"



On 04/24/2018 08:59 AM, Yafang Shao wrote:
> On Tue, Apr 24, 2018 at 11:49 PM, Eric Dumazet <eric.dumazet@...il.com> wrote:
>>
>>
>> On 04/24/2018 08:12 AM, Yafang Shao wrote:
>>> On Tue, Apr 24, 2018 at 8:38 PM, Eric Dumazet <eric.dumazet@...il.com> wrote:
>>>>
>>>>
>>>> On 04/24/2018 05:05 AM, Yafang Shao wrote:
>>>>> This revert commit <c6849a3ac17e> ("net: init sk_cookie for inet socket")
>>>>>
>>>>> Per discussion with Eric.
>>>>>
>>>>
>>>> I suggest you include a bit more details, about cache line false sharing.
>>>>
>>>
>>> Coud we adjust the struct common to avoid such kind of cache line
>>> false sharing ?
>>> I mean removing "atomic64_t  skc_cookie;" from struct sock_common and
>>> place it in struct inet_sock ?
>>
>> The false sharing is not there, it is on net->cookie_gen
>>
> 
> Yes.
> This is the current issue.
> May be we should adjust struct net as well.

This field will still need to be modified by many cpus.

Its exact placement in memory wont avoid false sharing and stalls.

> 
> Regarding sk_cookie, as it is only used by inet_sock now, may be it is
> better placed in srtuct inet_sock ?
>

You are mistaken.

It is used on all sockets really (including request_sock and timewait)

ss -temoia   will give you socket ids for all sockets types.