netdev - Re: netmem series needs some love and Acks from MM folks

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: <72e9f23c-58ca-49c3-baee-f76b91d093ff@gmail.com>
Date: Wed, 18 Jun 2025 08:51:32 +0100
From: Pavel Begunkov <asml.silence@...il.com>
To: Byungchul Park <byungchul@...com>, David Hildenbrand <david@...hat.com>
Cc: Harry Yoo <harry.yoo@...cle.com>, Mina Almasry <almasrymina@...gle.com>,
 willy@...radead.org, Jakub Kicinski <kuba@...nel.org>,
 netdev@...r.kernel.org, linux-kernel@...r.kernel.org, linux-mm@...ck.org,
 kernel_team@...ynix.com, ilias.apalodimas@...aro.org, hawk@...nel.org,
 akpm@...ux-foundation.org, davem@...emloft.net, john.fastabend@...il.com,
 andrew+netdev@...n.ch, toke@...hat.com, tariqt@...dia.com,
 edumazet@...gle.com, pabeni@...hat.com, saeedm@...dia.com, leon@...nel.org,
 ast@...nel.org, daniel@...earbox.net, lorenzo.stoakes@...cle.com,
 Liam.Howlett@...cle.com, vbabka@...e.cz, rppt@...nel.org, surenb@...gle.com,
 mhocko@...e.com, horms@...nel.org, linux-rdma@...r.kernel.org,
 bpf@...r.kernel.org, vishal.moola@...il.com
Subject: Re: netmem series needs some love and Acks from MM folks

On 6/18/25 01:08, Byungchul Park wrote:
> On Tue, Jun 17, 2025 at 06:09:36PM +0200, David Hildenbrand wrote:
...>> 1) have a dedicated data-structure we will allocate alter dynamically.
>>
>> 2) Make it overlay "struct page" for now in a way that doesn't break things
>>
>> 3) Convert all users of "struct page" to the new data-structure
>>
>> Later, the memdesc data-structure will then actually come be allocated
>> dynamically, so "struct page" content will not apply anymore, and we can
>> shrink "struct page".
>>
>>
>> What I see in this patch is exactly 1) and 2).
>>
>> I am not 100% sure about existing "struct net_iov" and how that
>> interacts with "struct page" overlay. I suspects it's just a dynamically
>> allocated structure?
>>
>> Because this patch changes the layout of "struct net_iov", which is a
>> bit confusing at first sight?
> 
> The changes of the layout was asked by network folks, that was to split
> the struct net_iov fields to two, netmem_desc and net_iov specific ones.

That's right, and the moved fields should never be touched without
checking that it's a net_iov first. All of it will eventually converge
to netmem_desc without net_iov sticking out where it shouldn't, but my
personal preference would be to get this merged, as it's a good enough
base, and do the rest on top.

> How to organize struct net_iov further is up to the network folks, but
> I believe the current layout should be the first step.
-- 
Pavel Begunkov