netdev - Re: [PATCH 4.14.y stable] xfrm: policy: remove pcpu policy cache

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]

Date:   Thu, 22 Aug 2019 15:37:11 +0200
From:   Florian Westphal <fw@...len.de>
To:     Greg KH <greg@...ah.com>
Cc:     Florian Westphal <fw@...len.de>, stable@...r.kernel.org,
        vakul.garg@....com, netdev@...r.kernel.org,
        Kristian Evensen <kristian.evensen@...il.com>,
        Steffen Klassert <steffen.klassert@...unet.com>
Subject: Re: [PATCH 4.14.y stable] xfrm: policy: remove pcpu policy cache

Greg KH <greg@...ah.com> wrote:
> On Thu, Aug 22, 2019 at 01:21:09PM +0200, Florian Westphal wrote:
> > commit e4db5b61c572475bbbcf63e3c8a2606bfccf2c9d upstream.
> > 
> > Kristian Evensen says:
> >   In a project I am involved in, we are running ipsec (Strongswan) on
> >   different mt7621-based routers. Each router is configured as an
> >   initiator and has around ~30 tunnels to different responders (running
> >   on misc. devices). Before the flow cache was removed (kernel 4.9), we
> >   got a combined throughput of around 70Mbit/s for all tunnels on one
> >   router. However, we recently switched to kernel 4.14 (4.14.48), and
> >   the total throughput is somewhere around 57Mbit/s (best-case). I.e., a
> >   drop of around 20%. Reverting the flow cache removal restores, as
> >   expected, performance levels to that of kernel 4.9.
> > 
> > When pcpu xdst exists, it has to be validated first before it can be
> > used.
> > 
> > A negative hit thus increases cost vs. no-cache.
> > 
> > As number of tunnels increases, hit rate decreases so this pcpu caching
> > isn't a viable strategy.
> > 
> > Furthermore, the xdst cache also needs to run with BH off, so when
> > removing this the bh disable/enable pairs can be removed too.
> > 
> > Kristian tested a 4.14.y backport of this change and reported
> > increased performance:
> > 
> >   In our tests, the throughput reduction has been reduced from around -20%
> >   to -5%. We also see that the overall throughput is independent of the
> >   number of tunnels, while before the throughput was reduced as the number
> >   of tunnels increased.
> > 
> > Reported-by: Kristian Evensen <kristian.evensen@...il.com>
> > Signed-off-by: Florian Westphal <fw@...len.de>
> > Signed-off-by: Steffen Klassert <steffen.klassert@...unet.com>
> > ---
> >  Vakul Garg reports traffic going via ipsec tunnels will cause the kernel
> >  to spin in an infinite loop due to xfrm policy reference count
> >  overflowing and becoming 0.
> >  The refcount leak is in the pcpu cache.  Instead of fixing this, just
> >  remove the pcpu cache -- its not present in any other stable release.
> >  Vakul reported that this patch fixes the problem.
> > 
> >  There are no major deviations from the upstream revert; conflicts
> >  were only due to context.
> 
> Now queued up, does 4.9.y also need this?

No, 4.14 was the first kernel with this thing and its already gone in
4.19.