netdev - using rcu_read_lock() after calling dst_neigh

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: <ead29087-05cc-c586-3b7e-d37c2e6f753c@mellanox.com>
Date:   Tue, 10 Jan 2017 16:22:51 +0200
From:   Hadar Hen Zion <hadarh@...lanox.com>
To:     <davem@...emloft.net>
CC:     <netdev@...r.kernel.org>, <ogerlitz@...lanox.com>,
        <idosch@...lanox.com>
Subject: using rcu_read_lock() after calling dst_neigh_lookup

Hi Dave,

Drivers which are calling dst_neigh_lookup() are also using 
rcu_read_lock() before accessing the neigh pointer (and asking it's ll 
address data and its validity state).

You can find the same behavior in:

drivers/infiniband/core/addr.c, drivers/infiniband/hw/i40iw/i40iw_cm.c, 
drivers/infiniband/hw/nes/nes_cm.c, etc.

(the above locations are just an example).

While the documentation in neighbour.c says:

  "Neighbour entries are protected:
    - with reference count.
    - with rwlock neigh->lock
    Reference count prevents destruction.
    neigh->lock mainly serializes ll address data and its validity state."

So what is the right way to protect the neigh entry parameters? I 
couldn't find why rcu_read_lock() is helping here (dst_neigh_lookup 
already takes a reference on the neigh).

Thank you,

Hadar