netdev - Re: [PATCH net v2 0/2] Revert the 'socket

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:   Tue, 5 May 2020 12:31:30 +0000
From:   "Nuernberger, Stefan" <snu@...zon.de>
To:     "Park, Seongjae" <sjpark@...zon.com>
CC:     "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "sj38.park@...il.com" <sj38.park@...il.com>,
        "stable@...r.kernel.org" <stable@...r.kernel.org>,
        "viro@...iv.linux.org.uk" <viro@...iv.linux.org.uk>,
        "kuba@...nel.org" <kuba@...nel.org>,
        "edumazet@...gle.com" <edumazet@...gle.com>,
        "Nuernberger, Stefan" <snu@...zon.de>,
        "sjpark@...zon.de" <sjpark@...zon.de>,
        "gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        "amit@...nel.org" <amit@...nel.org>,
        "davem@...emloft.net" <davem@...emloft.net>
Subject: Re: [PATCH net v2 0/2] Revert the 'socket_alloc' life cycle change

On Tue, 2020-05-05 at 13:54 +0200, SeongJae Park wrote:
> CC-ing stable@...r.kernel.org and adding some more explanations.
> 
> On Tue, 5 May 2020 10:10:33 +0200 SeongJae Park <sjpark@...zon.com>
> wrote:
> 
> > 
> > From: SeongJae Park <sjpark@...zon.de>
> > 
> > The commit 6d7855c54e1e ("sockfs: switch to ->free_inode()") made
> > the
> > deallocation of 'socket_alloc' to be done asynchronously using RCU,
> > as
> > same to 'sock.wq'.  And the following commit 333f7909a857
> > ("coallocate
> > socket_sq with socket itself") made those to have same life cycle.
> > 
> > The changes made the code much more simple, but also made
> > 'socket_alloc'
> > live longer than before.  For the reason, user programs intensively
> > repeating allocations and deallocations of sockets could cause
> > memory
> > pressure on recent kernels.
> I found this problem on a production virtual machine utilizing 4GB
> memory while
> running lebench[1].  The 'poll big' test of lebench opens 1000
> sockets, polls
> and closes those.  This test is repeated 10,000 times.  Therefore it
> should
> consume only 1000 'socket_alloc' objects at once.  As size of
> socket_alloc is
> about 800 Bytes, it's only 800 KiB.  However, on the recent kernels,
> it could
> consume up to 10,000,000 objects (about 8 GiB).  On the test machine,
> I
> confirmed it consuming about 4GB of the system memory and results in
> OOM.
> 
> [1] https://github.com/LinuxPerfStudy/LEBench
> 
> > 
> > 
> > To avoid the problem, this commit reverts the changes.
> I also tried to make fixup rather than reverts, but I couldn't easily
> find
> simple fixup.  As the commits 6d7855c54e1e and 333f7909a857 were for
> code
> refactoring rather than performance optimization, I thought
> introducing complex
> fixup for this problem would make no sense.  Meanwhile, the memory
> pressure
> regression could affect real machines.  To this end, I decided to
> quickly
> revert the commits first and consider better refactoring later.
> 

While lebench might be exercising a rather pathological case, the
increase in memory pressure is real. I am concerned that the OOM killer
is actually engaging and killing off processes when there are lots of
resources already marked for release. This might be true for other
lazy/delayed resource deallocation, too. This has obviously just become
too lazy currently.

So for both reverts:

Reviewed-by: Stefan Nuernberger <snu@...zon.com>

> 
> Thanks,
> SeongJae Park
> 
> > 
> > 
> > SeongJae Park (2):
> >   Revert "coallocate socket_wq with socket itself"
> >   Revert "sockfs: switch to ->free_inode()"
> > 
> >  drivers/net/tap.c      |  5 +++--
> >  drivers/net/tun.c      |  8 +++++---
> >  include/linux/if_tap.h |  1 +
> >  include/linux/net.h    |  4 ++--
> >  include/net/sock.h     |  4 ++--
> >  net/core/sock.c        |  2 +-
> >  net/socket.c           | 23 ++++++++++++++++-------
> >  7 files changed, 30 insertions(+), 17 deletions(-)
> > 



Amazon Development Center Germany GmbH
Krausenstr. 38
10117 Berlin
Geschaeftsfuehrung: Christian Schlaeger, Jonathan Weiss
Eingetragen am Amtsgericht Charlottenburg unter HRB 149173 B
Sitz: Berlin
Ust-ID: DE 289 237 879