netdev - Re: [RFC] net: skbuff: let struct skb

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 for Android: free password hash cracker in your pocket

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

Message-ID: <20230215095834.42f5e227@kernel.org>
Date:   Wed, 15 Feb 2023 09:58:34 -0800
From:   Jakub Kicinski <kuba@...nel.org>
To:     Paolo Abeni <pabeni@...hat.com>
Cc:     davem@...emloft.net, netdev@...r.kernel.org, edumazet@...gle.com,
        willemb@...gle.com, fw@...len.de
Subject: Re: [RFC] net: skbuff: let struct skb_ext live inside the head

On Wed, 15 Feb 2023 09:53:54 +0100 Paolo Abeni wrote:
> Still, if you are willing to experiment more this idea, I think you
> could save the extra cacheline miss encoding the 'alloc_mode' into the
> lower bits of skb->extensions (alike what _skb_refdst is doing with the
> SKB_DST_NOREF flag).

I thought I'd start with a simpler approach where allocation type 
is stored in the object itself, to limit the negative reactions :P
We could indeed save a cache miss (I think it'd be actually one fewer
miss than the current implementation, because most cases must end 
up looking at the skb_ext, f.e. to read the refcount. The fact that
SHARD_NOREF implies refcnt == 1 can save us reading from skb_ext).