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-next>] [day] [month] [year] [list]
Message-ID: <20241204221254.3537932-1-sbrivio@redhat.com>
Date: Wed,  4 Dec 2024 23:12:52 +0100
From: Stefano Brivio <sbrivio@...hat.com>
To: Willem de Bruijn <willemdebruijn.kernel@...il.com>
Cc: Eric Dumazet <edumazet@...gle.com>,
	netdev@...r.kernel.org,
	Kuniyuki Iwashima <kuniyu@...zon.com>,
	Mike Manning <mvrmanning@...il.com>,
	David Gibson <david@...son.dropbear.id.au>,
	Paul Holzinger <pholzing@...hat.com>,
	Philo Lu <lulie@...ux.alibaba.com>,
	Cambda Zhu <cambda@...ux.alibaba.com>,
	Fred Chen <fred.cc@...baba-inc.com>,
	Yubing Qiu <yubing.qiuyubing@...baba-inc.com>
Subject: [PATCH net-next 0/2] Fix race between datagram socket address change and rehash

Patch 2/2 fixes a race condition in the lookup of datagram sockets
between address change (triggered by connect()) and rehashing. The
issue occurs regardless of the type of hash, that is, it happens with
updates to the secondary hash as well as to the newly introduced
four-tuple hash.

Patch 1/2 is a small optimisation to simplify 2/2.

This is essentially a rebase onto current net-next of the RFC I
originally posted against 'net', before the 6.13 merge window,
at:

  https://lore.kernel.org/netdev/20241114215414.3357873-1-sbrivio@redhat.com/

with a couple of minor changes described in the single patch messages.

The rebase is not trivial as four-tuple hashes were introduced
meanwhile, with commits 1b29a730ef8b ("ipv6/udp: Add 4-tuple hash for
connected socket"), 78c91ae2c6de ("ipv4/udp: Add 4-tuple hash for
connected socket"), and their dependencies, but the race condition,
described in detail in the commit message for 1/2, is exactly the same
as before.

Stefano Brivio (2):
  datagram: Rehash sockets only if local address changed for their
    family
  datagram, udp: Set local address and rehash socket atomically against
    lookup

 include/net/inet_hashtables.h | 13 ++++++
 include/net/sock.h            |  2 +-
 include/net/udp.h             |  3 +-
 net/core/sock.c               | 12 ++++-
 net/ipv4/datagram.c           |  7 +--
 net/ipv4/inet_hashtables.c    | 13 ------
 net/ipv4/udp.c                | 84 +++++++++++++++++++++++------------
 net/ipv4/udp_impl.h           |  2 +-
 net/ipv4/udplite.c            |  2 +-
 net/ipv6/datagram.c           | 30 +++++++++----
 net/ipv6/udp.c                | 31 +++++++------
 net/ipv6/udp_impl.h           |  2 +-
 net/ipv6/udplite.c            |  2 +-
 13 files changed, 130 insertions(+), 73 deletions(-)

-- 
2.40.1


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ