linux-kernel - Re: [PATCH RFC] ashmem: Fix lockdep RECLAIM

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:   Tue, 6 Feb 2018 14:32:13 -0800
From:   Joel Fernandes <joelaf@...gle.com>
To:     Minchan Kim <minchan@...nel.org>
Cc:     LKML <linux-kernel@...r.kernel.org>,
        Peter Zilstra <peterz@...radead.org>,
        Michal Hocko <mhocko@...nel.org>,
        "open list:MEMORY MANAGEMENT" <linux-mm@...ck.org>
Subject: Re: [PATCH RFC] ashmem: Fix lockdep RECLAIM_FS false positive

Hi Minchan,

On Tue, Feb 6, 2018 at 2:01 PM, Minchan Kim <minchan@...nel.org> wrote:
[...]
> On Mon, Feb 05, 2018 at 04:49:03PM -0800, Joel Fernandes wrote:
>> During invocation of ashmem shrinker under memory pressure, ashmem
>> calls into VFS code via vfs_fallocate. We however make sure we
>> don't enter it if the allocation was GFP_FS to prevent looping
>> into filesystem code. However lockdep doesn't know this and prints
>> a lockdep splat as below.
>>
>> This patch fixes the issue by releasing the reclaim_fs lock after
>> checking for GFP_FS but before calling into the VFS path, and
>> reacquiring it after so that lockdep can continue reporting any
>> reclaim issues later.
>
> At first glance, it looks reasonable. However, Couldn't we return
> just 0 in ashmem_shrink_count when the context is under FS?
>

We're already checking if GFP_FS in ashmem_shrink_scan and bailing out
though, did I miss something?

The problem is not that there is a deadlock that occurs, the problem
that even when we're not under FS, lockdep reports an issue that can't
happen. The fix is for the lockdep false positive that occurs.

thanks,

- Joel