linux-kernel - Re: [PATCH] mm/kasan: make quarantine_lock a raw_spinlock

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]

Message-ID: <20181013135058.GC4931@worktop.programming.kicks-ass.net>
Date:   Sat, 13 Oct 2018 15:50:58 +0200
From:   Peter Zijlstra <peterz@...radead.org>
To:     Andrew Morton <akpm@...ux-foundation.org>
Cc:     Sebastian Andrzej Siewior <bigeasy@...utronix.de>,
        Dmitry Vyukov <dvyukov@...gle.com>,
        Clark Williams <williams@...hat.com>,
        Alexander Potapenko <glider@...gle.com>,
        kasan-dev <kasan-dev@...glegroups.com>,
        Linux-MM <linux-mm@...ck.org>,
        LKML <linux-kernel@...r.kernel.org>,
        linux-rt-users@...r.kernel.org,
        Thomas Gleixner <tglx@...utronix.de>
Subject: Re: [PATCH] mm/kasan: make quarantine_lock a raw_spinlock_t

On Fri, Oct 12, 2018 at 04:56:55PM -0700, Andrew Morton wrote:
> There are several reasons for using raw_*, so an explanatory comment at
> each site is called for.
> 
> However it would be smarter to stop "using raw_* for several reasons". 
> Instead, create a differently named variant for each such reason.  ie, do
> 
> /*
>  * Nice comment goes here.  It explains all the possible reasons why -rt
>  * might use a raw_spin_lock when a spin_lock could otherwise be used.
>  */
> #define raw_spin_lock_for_rt	raw_spinlock
> 
> Then use raw_spin_lock_for_rt() at all such sites.

The whole raw_spinlock_t is for RT, no other reason. It is the one true
spinlock.

>From this, it naturally follows that:

 - nesting order: raw_spinlock_t < spinlock_t < mutex_t
 - raw_spinlock_t sections must be bounded

The patch under discussion is the result of the nesting order rule; and
is allowed to violate the second rule, by virtue of it being debug code.

There are no other reasons; and I'm somewhat confused by what you
propose.