netdev - Re: [PATCH v7 11/15] io_uring/zcrx: implement zerocopy receive pp memory provider

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: <20241115151428.6f1e1aba@kernel.org>
Date: Fri, 15 Nov 2024 15:14:28 -0800
From: Jakub Kicinski <kuba@...nel.org>
To: Mina Almasry <almasrymina@...gle.com>
Cc: David Wei <dw@...idwei.uk>, Pavel Begunkov <asml.silence@...il.com>,
 io-uring@...r.kernel.org, netdev@...r.kernel.org, Jens Axboe
 <axboe@...nel.dk>, Paolo Abeni <pabeni@...hat.com>, "David S. Miller"
 <davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>, Jesper Dangaard
 Brouer <hawk@...nel.org>, David Ahern <dsahern@...nel.org>, Stanislav
 Fomichev <stfomichev@...il.com>, Joe Damato <jdamato@...tly.com>, Pedro
 Tammela <pctammela@...atatu.com>
Subject: Re: [PATCH v7 11/15] io_uring/zcrx: implement zerocopy receive pp
 memory provider

On Thu, 14 Nov 2024 12:56:14 -0800 Mina Almasry wrote:
> But I've been bringing this up a lot in the past (and offering
> alternatives that don't introduce this overloading) and I think this
> conversation has run its course. I'm unsure about this approach and
> this could use another pair of eyes. Jakub, sorry to bother you but
> you probably are the one that reviewed the whole net_iov stuff most
> closely. Any chance you would take a look and provide direction here?
> Maybe my concern is overblown...

Sorry I haven't read this code very closely (still!?) really hoping 
for next version to come during the merge window when time is more
abundant :S

From scanning the quoted context I gather you're asking about using 
the elevated ->pp_ref_count for user-owned pages? If yes - I also
find that part.. borderline incorrect. The page pool stats will show
these pages as allocated which normally means held by the driver or 
the stack. Pages given to the user should effectively leave the pool.
So I'm guessing this is some optimization. Same for patch 8.

But let me not get sucked into this before we wrap up the net-next PR..