linux-kernel - Re: Re: [PATCH v5 1/1] mm/memory-failure: disable unpoison once hw error happens

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:   Wed, 15 Jun 2022 07:39:20 +0000
From:   HORIGUCHI NAOYA(堀口　直也) 
        <naoya.horiguchi@....com>
To:     Oscar Salvador <osalvador@...e.de>,
        zhenwei pi <pizhenwei@...edance.com>
CC:     "akpm@...ux-foundation.org" <akpm@...ux-foundation.org>,
        "linux-mm@...ck.org" <linux-mm@...ck.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "david@...hat.com" <david@...hat.com>,
        "linmiaohe@...wei.com" <linmiaohe@...wei.com>,
        "gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>
Subject: Re: Re: [PATCH v5 1/1] mm/memory-failure: disable unpoison once hw
 error happens

On Wed, Jun 15, 2022 at 07:52:06AM +0200, Oscar Salvador wrote:
> On Wed, Jun 15, 2022 at 01:18:23PM +0800, zhenwei pi wrote:
> > Hi,
> > 
> > Because memory_failure() may be called by hardware error randomly,
> > hw_memory_failure should be protected by mf_mutex to avoid this case:
> > int unpoison_memory(unsigned long pfn)
> > {
> >     ...
> >     if (hw_memory_failure) {
> >     }
> >     ... --> memory_failure() happens, and mark hw_memory_failure as true
> >     mutex_lock(&mf_mutex);

I think that this race can cause the reported problem (hw_memory_failure is
unreliable outside mf_mutex), so we need put the check in mf_mutex for the proper fix.

Thanks,
Naoya Horiguchi

> 
> Yeah, I am aware of that.
> But once memory_failure() sets hw_memory_failure to true, it does not really matter
> whether unpoison_memory() checks that while holding or not the lock, does it?
> 
> Note that it does not really matter in the end, but I am just curious whether
> there is any strong impediment to that. 
> 
> 
> -- 
> Oscar Salvador
> SUSE Labs