linux-kernel - Re: [PATCH] mm: implement "memory.oops_if_bad

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: <aHAGeikybkJrPp6Y@tiehlicka>
Date: Thu, 10 Jul 2025 20:29:14 +0200
From: Michal Hocko <mhocko@...e.com>
To: Lorenzo Stoakes <lorenzo.stoakes@...cle.com>
Cc: Alexey Dobriyan <adobriyan@...il.com>, akpm@...ux-foundation.org,
	linux-kernel@...r.kernel.org, linux-mm@...ck.org,
	David Hildenbrand <david@...hat.com>,
	"Liam R. Howlett" <Liam.Howlett@...cle.com>,
	Vlastimil Babka <vbabka@...e.cz>, Mike Rapoport <rppt@...nel.org>,
	Suren Baghdasaryan <surenb@...gle.com>
Subject: Re: [PATCH] mm: implement "memory.oops_if_bad_pte=1" boot option

On Thu 10-07-25 18:02:07, Lorenzo Stoakes wrote:
> OK I wasn't clear enough I guess - NAK.

Completely Agreed!
Because....

> This is not upstreamable, nor anything like it.
> 
> On Thu, Jul 10, 2025 at 07:57:00PM +0300, Alexey Dobriyan wrote:
> > On Thu, Jul 10, 2025 at 05:16:52PM +0100, Lorenzo Stoakes wrote:
[...]
> > > You seem to be using BUG_ON() to _maybe_ cause a panic, maybe not, but by
> > > doing this you're inferring that there's unrecoverable system instability,
> > > which isf clearly not the case.

.... of exactly this. I believe we have already/finally established that
BUG_ON (not even VM_BUG_ON) is a sensible debugging tool. In this
particular case it is not even clear when the page table got corrupted
and it could have happened loong before we notice that so crash dumping
right away doesn't really guarantee anything.

So if this really helped in some specific situations and there is hope
it might help in the future then I believe
[...]
> > > Overall I suspect there's one single case you're worried about, that really
> > > you want to put a WARN_ON_ONCE() against - then you can panic_on_warn and
> > > get what you want.

is exactly what you should do. But even then dump_stack should be
dropped to not duplicate the information printed etc.

-- 
Michal Hocko
SUSE Labs