linux-kernel - Re: [PATCH -next] mm/hotplug: silence a lockdep splat with printk()

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-next>] [day] [month] [year] [list]

Date:   Wed, 15 Jan 2020 18:02:35 +0100
From:   Petr Mladek <pmladek@...e.com>
To:     Qian Cai <cai@....pw>
Cc:     Michal Hocko <mhocko@...nel.org>,
        David Hildenbrand <david@...hat.com>,
        akpm@...ux-foundation.org, sergey.senozhatsky.work@...il.com,
        rostedt@...dmis.org, peterz@...radead.org, linux-mm@...ck.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH -next] mm/hotplug: silence a lockdep splat with printk()

On Wed 2020-01-15 06:49:03, Qian Cai wrote:
> 
> 
> > On Jan 15, 2020, at 4:52 AM, Petr Mladek <pmladek@...e.com> wrote:
> > 
> > I could understand that Michal is against hack in -mm code that
> > would just hide a false positive warning.
> 
> Well, I don’t have any confidence to say everything this patch is
> trying to fix is false positives.

You look at this from a wrong angle. AFAIK, all lockdep reports pasted
in the below mentioned thread were false positives. Now, this patch
complicates an already complicated -mm code to hide the warning
and fix theoretical problems.

I suggest to disable lockdep around the safe allocation in the console
initialization code. Then we will see if there are other locations
that trigger this lockdep warning. It is trivial and will not
complicate the code because of false positives.

> I have been spent the last a few months to research this, so
> I don’t feel like to do this again.
> 
> https://lore.kernel.org/linux-mm/1570633715.5937.10.camel@lca.pw/

Have you tried to disable lockdep around the problematic allocation?

Have you seen other lockdep reports caused by exactly this printk()
in the allocator code?

My big problem with this patch is that the commit message does not
contain any lockdep report. It will complicate removing the hack
when it is not longer needed. Nobody will know what was the exact
problem and if it is safe to get removed. I believe that printk()
will offload console handling rather sooner than later and this
extra logic will not be necessary.

Best Regards,
Petr