linux-kernel - Re: [PATCH hotfix 6.12] mm, mmap: limit THP aligment of anonymous mappings to PMD-aligned sizes

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <e423577c-c1b9-4420-9237-271321a9738d@lucifer.local>
Date: Thu, 24 Oct 2024 17:17:45 +0100
From: Lorenzo Stoakes <lorenzo.stoakes@...cle.com>
To: Vlastimil Babka <vbabka@...e.cz>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
        "Liam R. Howlett" <Liam.Howlett@...cle.com>,
        Jann Horn <jannh@...gle.com>,
        Thorsten Leemhuis <regressions@...mhuis.info>, linux-mm@...ck.org,
        linux-kernel@...r.kernel.org, Petr Tesarik <ptesarik@...e.com>,
        Michael Matz <matz@...e.de>,
        Gabriel Krisman Bertazi <gabriel@...sman.be>,
        Matthias Bodenbinder <matthias@...enbinder.de>, stable@...r.kernel.org,
        Rik van Riel <riel@...riel.com>,
        Yang Shi <yang@...amperecomputing.com>,
        Matthew Wilcox <willy@...radead.org>
Subject: Re: [PATCH hotfix 6.12] mm, mmap: limit THP aligment of anonymous
 mappings to PMD-aligned sizes

On Thu, Oct 24, 2024 at 06:04:41PM +0200, Vlastimil Babka wrote:

> Petr suggested the same, but changing  __thp_get_unmapped_area() affects FS
> THP's and the proposed check seemed wrong to me:
>
> https://lore.kernel.org/all/9d7c73f6-1e1a-458b-93c6-3b44959022e0@suse.cz/
>
> While it could be fixed, I'm still not sure if we want to restrict FS THPs
> the same as anonymous THPs. AFAIU even small mappings of a range from a file
> should be aligned properly to make it possible for a large range from the
> same file (that includes the smaller range) mapped elsewhere to be THP
> backed? I mean we can investigate it further, but for the regression fix to
> backported to stable kernels it seemed more safe to address only the case
> that was changed by commit efa7df3e3bb5 specifically, i.e. anonymous mappings.
>

Ack, yeah totally agreed - sorry I missed the fs usage before, see my 2nd
reply. I had wrongly assumed this was only used in 1 place, where it would
be sensible to move the check, however with fs using it of course it's not.

Gave an R-b tag on other reply so this patch LGTM! :)

Cheers for finding this utterly critical fix!