linux-kernel - Re: Re: [PATCH] [PATCH] mm: disable preemption before swapcache

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>] [day] [month] [year] [list]

Message-ID: <20180726074430.GV28386@dhcp22.suse.cz>
Date:   Thu, 26 Jul 2018 09:44:30 +0200
From:   Michal Hocko <mhocko@...nel.org>
To:     "zhaowuyun@...gtech.com" <zhaowuyun@...gtech.com>
Cc:     akpm <akpm@...ux-foundation.org>,
        mgorman <mgorman@...hsingularity.net>,
        minchan <minchan@...nel.org>, vinmenon <vinmenon@...eaurora.org>,
        hannes <hannes@...xchg.org>,
        "hillf.zj" <hillf.zj@...baba-inc.com>,
        linux-mm <linux-mm@...ck.org>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        Hugh Dickins <hughd@...gle.com>
Subject: Re: Re: [PATCH] [PATCH] mm: disable preemption before swapcache_free

On Thu 26-07-18 15:03:23, zhaowuyun@...gtech.com wrote:
> >On Thu 26-07-18 10:21:40, zhaowuyun@...gtech.com wrote:
> >[...]
> >> Our project really needs a fix to this issue
> >
> >Could you be more specific why? My understanding is that RT tasks
> >usually have all the memory mlocked otherwise all the real time
> >expectations are gone already.
> >--
> >Michal Hocko
> >SUSE Labs 
> 
> 
> The RT thread is created by a process with normal priority, and the process was sleep, 
> then some task needs the RT thread to do something, so the process create this thread, and set it to RT policy.
> I think that is the reason why RT task would read the swap.

OK I see. This design is quite fragile though. You are opening ticket to
priority inversions and what not.

Anyway, the underlying swap issue should be fixed. Unfortunatelly I do
not have a great idea how to do that properly.

-- 
Michal Hocko
SUSE Labs