linux-kernel - Re: [PATCH 1/2] memblock: show a warning if allocation in KHO scratch fails

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: <aC3wx5erybg00SaQ@kernel.org>
Date: Wed, 21 May 2025 18:27:03 +0300
From: Mike Rapoport <rppt@...nel.org>
To: Oscar Salvador <osalvador@...e.de>
Cc: Changyuan Lyu <changyuanl@...gle.com>, akpm@...ux-foundation.org,
	bhe@...hat.com, chrisl@...nel.org, graf@...zon.com,
	jasonmiu@...gle.com, kexec@...ts.infradead.org,
	linux-kernel@...r.kernel.org, linux-mm@...ck.org,
	pasha.tatashin@...een.com
Subject: Re: [PATCH 1/2] memblock: show a warning if allocation in KHO
 scratch fails

On Wed, May 21, 2025 at 10:48:55AM +0200, Oscar Salvador wrote:
> On Wed, May 21, 2025 at 10:43:15AM +0300, Mike Rapoport wrote:
> > I think we should just make sparse_init_nid() panic or at least change 
> > "sparse_init_nid: node[0] memory map backing failed. Some memory will not be available."
> > to something more visible and clear. 
> 
> Panicking the system seems a bit too harsh.
> Those sections will not be initialized, and sure you will lose some memory,
> but still.
> 
> I think that making sure that subsection_map_init() does not access
> non-initialized values is enough.

It's not only subsection_map_init(), next failing one is
memmap_init_range() and maybe there's more, but we can audit and fix them.
I believe all those accesses are at init time because after system is
booted we are careful to avoid accessing absent sections.

> Because wrt. error message, I am not sure it can get more clear that we
> failed we allocate memory to back the section and so that section will
> not be activated :-)

Add a dump_stack()? ;-)

> -- 
> Oscar Salvador
> SUSE Labs
> 

-- 
Sincerely yours,
Mike.