netdev - Re: [RFC PATCH net-next 9/9] net-sysfs: remove the use of rtnl_trylock/restart

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <YV1kjDfTE2N2XK9J@dhcp22.suse.cz>
Date:   Wed, 6 Oct 2021 10:55:40 +0200
From:   Michal Hocko <mhocko@...e.com>
To:     Antoine Tenart <atenart@...nel.org>
Cc:     davem@...emloft.net, kuba@...nel.org, pabeni@...hat.com,
        gregkh@...uxfoundation.org, ebiederm@...ssion.com,
        stephen@...workplumber.org, herbert@...dor.apana.org.au,
        juri.lelli@...hat.com, netdev@...r.kernel.org
Subject: Re: [RFC PATCH net-next 9/9] net-sysfs: remove the use of
 rtnl_trylock/restart_syscall

On Wed 06-10-21 10:03:41, Antoine Tenart wrote:
> Quoting Michal Hocko (2021-10-06 08:45:58)
> > On Tue 28-09-21 14:55:00, Antoine Tenart wrote:
> > > The ABBA deadlock avoided by using rtnl_trylock and restart_syscall was
> > > fixed in previous commits, we can now remove the use of this
> > > trylock/restart logic and have net-sysfs operations not spinning when
> > > rtnl is already taken.
> > 
> > Shouldn't those lock be interruptible or killable at least? As mentioned
> > in other reply we are seeing multiple a contention on some sysfs file
> > because mlx driver tends to do some heavy lifting in its speed callback
> > so it would be great to be able to interact with waiters during that
> > time.
> 
> Haven't thought much about this, but it should be possible to use
> rtnl_lock_killable. I think this should be a patch on its own with its
> own justification though (to help bisecting). But that is easy to do
> once the trylock logic is gone.

Agreed

-- 
Michal Hocko
SUSE Labs