linux-kernel - Re: [PATCH v2] mm/gup: fix try_grab_compound_head() race with split_huge

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:   Thu, 17 Jun 2021 18:09:16 +0200
From:   Vlastimil Babka <vbabka@...e.cz>
To:     Yang Shi <shy828301@...il.com>
Cc:     Jann Horn <jannh@...gle.com>, John Hubbard <jhubbard@...dia.com>,
        Matthew Wilcox <willy@...radead.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Linux-MM <linux-mm@...ck.org>,
        kernel list <linux-kernel@...r.kernel.org>,
        "Kirill A . Shutemov" <kirill@...temov.name>,
        Jan Kara <jack@...e.cz>, stable <stable@...r.kernel.org>
Subject: Re: [PATCH v2] mm/gup: fix try_grab_compound_head() race with
 split_huge_page()

On 6/16/21 8:40 PM, Yang Shi wrote:
> On Wed, Jun 16, 2021 at 10:27 AM Vlastimil Babka <vbabka@...e.cz> wrote:
>>
>> On 6/16/21 1:10 AM, Yang Shi wrote:
>> > On Tue, Jun 15, 2021 at 5:10 AM Jann Horn <jannh@...gle.com> wrote:
>> >>
>> >
>> > It might be possible after Mel's "mm/page_alloc: Allow high-order
>> > pages to be stored on the per-cpu lists" patch
>> > (https://patchwork.kernel.org/project/linux-mm/patch/20210611135753.GC30378@techsingularity.net/).
>>
>> Those would be percpu indeed, but not "lockless, non-atomic", no? They are
>> protected by a local_lock.
> 
> The local_lock is *not* a lock on non-PREEMPT_RT kernel IIUC. It
> disables preempt and IRQ. But preempt disable is no-op on non-preempt
> kernel. IRQ disable can guarantee it is atomic context, but I'm not
> sure if it is equivalent to "atomic freelists" in Jann's context.

Hm right, I thought all locks had the acquire/release semantics, but this one is
just a barrier().