linux-kernel - RE: [PATCH] tipc: ensure skb->lock is initialised

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]

Date:   Wed, 10 Jul 2019 13:10:20 +0000
From:   Jon Maloy <jon.maloy@...csson.com>
To:     Eric Dumazet <eric.dumazet@...il.com>,
        Chris Packham <Chris.Packham@...iedtelesis.co.nz>,
        "ying.xue@...driver.com" <ying.xue@...driver.com>,
        "davem@...emloft.net" <davem@...emloft.net>
CC:     "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        "tipc-discussion@...ts.sourceforge.net" 
        <tipc-discussion@...ts.sourceforge.net>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: RE: [PATCH] tipc: ensure skb->lock is initialised



> -----Original Message-----
> From: Eric Dumazet <eric.dumazet@...il.com>
> Sent: 10-Jul-19 04:00
> To: Jon Maloy <jon.maloy@...csson.com>; Eric Dumazet
> <eric.dumazet@...il.com>; Chris Packham
> <Chris.Packham@...iedtelesis.co.nz>; ying.xue@...driver.com;
> davem@...emloft.net
> Cc: netdev@...r.kernel.org; tipc-discussion@...ts.sourceforge.net; linux-
> kernel@...r.kernel.org
> Subject: Re: [PATCH] tipc: ensure skb->lock is initialised
> 
> 
> 
> On 7/9/19 10:15 PM, Jon Maloy wrote:
> >
> > It is not only for lockdep purposes, -it is essential.  But please provide details
> about where you see that more fixes are needed.
> >
> 
> Simple fact that you detect a problem only when skb_queue_purge() is called
> should talk by itself.
> 
> As I stated, there are many places where the list is manipulated _without_ its
> spinlock being held.

Yes, and that is the way it should be on the send path.

> 
> You want consistency, then
> 
> - grab the spinlock all the time.
> - Or do not ever use it.

That is exactly what we are doing. 
- The send path doesn't need the spinlock, and never grabs it.
- The receive path does need it, and always grabs it.

However, since we don't know from the beginning which path a created message will follow, we initialize the queue spinlock "just in case" when it is created, even though it may never be used later.
You can see this as a violation of the principle you are stating above, but it is a prize that is worth paying, given savings in code volume, complexity and performance.

> 
> Do not initialize the spinlock just in case a path will use skb_queue_purge()
> (instead of using __skb_queue_purge())

I am ok with that. I think we can agree that Chris goes for that solution, so we can get this bug fixed.

///jon