linux-kernel - Re: [PATCH net v1] netmem: fix skb_frag_address

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: <20250618142740.65203c69@kernel.org>
Date: Wed, 18 Jun 2025 14:27:40 -0700
From: Jakub Kicinski <kuba@...nel.org>
To: Mina Almasry <almasrymina@...gle.com>, Stanislav Fomichev
 <stfomichev@...il.com>
Cc: netdev@...r.kernel.org, linux-kernel@...r.kernel.org, "David S. Miller"
 <davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>, Paolo Abeni
 <pabeni@...hat.com>, Simon Horman <horms@...nel.org>, ap420073@...il.com
Subject: Re: [PATCH net v1] netmem: fix skb_frag_address_safe with
 unreadable skbs

On Tue, 17 Jun 2025 14:52:17 -0700 Mina Almasry wrote:
> > > Sorry, I realized right after hitting send, I'm missing:
> > >
> > > Fixes: 9f6b619edf2e ("net: support non paged skb frags")
> > >
> > > I can respin after the 24hr cooldown.  
> >
> > The function is used in five drivers, none of which support devmem tx,
> > does not look like there is a reason to route it via net.
> >
> > The change it self looks good, but not really sure it's needed.
> > skb_frag_address_safe is used in some pass-data-via-descriptor-ring mode,
> > I don't see 'modern' drivers (besides bnxt which added this support in 2015)
> > use it.  
> 
> Meh, a judgement call could be made here.  I've generally tried to
> make sure skb helpers are (unreadable) netmem compatible without a
> thorough analysis of all the callers to make sure they do or will one
> day use (unreadable) netmem. Seems better to me to fix this before
> some code path that plumbs unreadable memory to the helper is actually
> merged and that code starts crashing.

Fair points, tho I prefer the simple heuristic of "can it trigger on
net", otherwise it's really easy to waste time pondering each single
patch. I'll apply to net-next as is. Stanislav, do you want to ack?