linux-kernel - Re: [PATCH net] net/xen-netback: prevent UAF in xenvif_flush

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: <20240827071427.4c45fdb8@kernel.org>
Date: Tue, 27 Aug 2024 07:14:27 -0700
From: Jakub Kicinski <kuba@...nel.org>
To: Paolo Abeni <pabeni@...hat.com>
Cc: Jeongjun Park <aha310510@...il.com>, wei.liu@...nel.org, paul@....org,
 davem@...emloft.net, edumazet@...gle.com, madhuparnabhowmik04@...il.com,
 xen-devel@...ts.xenproject.org, netdev@...r.kernel.org,
 linux-kernel@...r.kernel.org
Subject: Re: [PATCH net] net/xen-netback: prevent UAF in xenvif_flush_hash()

On Tue, 27 Aug 2024 13:19:59 +0200 Paolo Abeni wrote:
> On 8/22/24 20:11, Jeongjun Park wrote:
> > During the list_for_each_entry_rcu iteration call of xenvif_flush_hash,
> > kfree_rcu does not exist inside the rcu read critical section, so if  
> 
> The above wording is confusing, do you mean "kfree_rcu does not exit 
> from "...?

I think they mean that kfree_rcu() is called without holding RCU read
lock..

> > kfree_rcu is called when the rcu grace period ends during the iteration,
> > UAF occurs when accessing head->next after the entry becomes free.  

.. so it can run immediately. Therefore the loop fetching head->next
may cause a UAF.

> The loop runs with irq disabled, the RCU critical section extends over 
> it, uninterrupted.

Is this an official RCU rule? I remember Paul told us it's the case for
softirq, but IDK if it is also for local IRQ disable.

> Do you have a splat for the reported UAF?
> 
> This does not look the correct solution.

The problem may not exist, but FWIW the change makes sense to me :)
We hold the write lock, and modify the list. for_each_entry_safe()
seems like a better fit than for_each_entry_rcu()