linux-kernel - Re: [PATCH] mm/page_alloc: replace local

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]

Message-ID: <20220708155835.GK27531@techsingularity.net>
Date:   Fri, 8 Jul 2022 16:58:35 +0100
From:   Mel Gorman <mgorman@...hsingularity.net>
To:     Vlastimil Babka <vbabka@...e.cz>
Cc:     Andrew Morton <akpm@...ux-foundation.org>,
        Nicolas Saenz Julienne <nsaenzju@...hat.com>,
        Marcelo Tosatti <mtosatti@...hat.com>,
        Michal Hocko <mhocko@...nel.org>,
        Hugh Dickins <hughd@...gle.com>, Yu Zhao <yuzhao@...gle.com>,
        Marek Szyprowski <m.szyprowski@...sung.com>,
        LKML <linux-kernel@...r.kernel.org>,
        Linux-MM <linux-mm@...ck.org>
Subject: Re: [PATCH] mm/page_alloc: replace local_lock with normal spinlock
 -fix -fix

On Fri, Jul 08, 2022 at 04:54:47PM +0200, Vlastimil Babka wrote:
> On 7/8/22 16:44, Mel Gorman wrote:
> > pcpu_spin_unlock and pcpu_spin_unlock_irqrestore both unlock
> > pcp->lock and then enable preemption. This lacks symmetry against
> > both the pcpu_spin helpers and differs from how local_unlock_* is
> > implemented. While this is harmless, it's unnecessary and it's generally
> > better to unwind locks and preemption state in the reverse order as
> > they were acquired.
> 
> Hm I'm confused, it seems it's done in reverse order (which I agree with)
> before this -fix-fix, but not after it?
> 
> before, pcpu_spin_lock() (and variants) do pcpu_task_pin() and then
> spin_lock() (or variant), and pcpu_spin_unlock() does spin_unlock() and then
> pcpu_task_unpin(). That seems symmetrical, i.e. reverse order to me? And
> seems to match what local_lock family does too.
> 

You're not confused, I am. The patch and the changelog are outright brain
damage from excessive context switching and a sign that it's time for the
weekend to start.

Sorry for this absolute misfortune.

-- 
Mel Gorman
SUSE Labs