linux-kernel - Re: [PATCH v2 2/3] x86/vmemmap: Drop handling of 1GB vmemmap ranges

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]

Message-ID: <20210203141038.GA26693@linux>
Date:   Wed, 3 Feb 2021 15:10:42 +0100
From:   Oscar Salvador <osalvador@...e.de>
To:     David Hildenbrand <david@...hat.com>
Cc:     Andrew Morton <akpm@...ux-foundation.org>,
        Dave Hansen <dave.hansen@...ux.intel.com>,
        Andy Lutomirski <luto@...nel.org>,
        Peter Zijlstra <peterz@...radead.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        x86@...nel.org, "H . Peter Anvin" <hpa@...or.com>,
        Michal Hocko <mhocko@...nel.org>, linux-mm@...ck.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2 2/3] x86/vmemmap: Drop handling of 1GB vmemmap ranges

On Wed, Feb 03, 2021 at 02:33:56PM +0100, David Hildenbrand wrote:
> One problem I see with existing code / this change making more obvious is
> that when trying to remove in other granularity than we added (e.g., unplug
> a 128MB DIMM avaialble during boot), we remove the direct map of unrelated
> DIMMs.

So, let me see if I understand your concern.

We have a range that was mapped with 1GB page, and we try to remove
a 128MB chunk from it.
Yes, in that case we would clear the pud, and that is bad, so we should
keep the PAGE_ALIGNED checks.

Now, let us assume that scenario.
If you have a 1GB mapped range and you remove it in smaller chunks bit by bit
(e.g: 128M), the direct mapping of that range will never be cleared unless
I am missing something (and the pagetables won't be freed) ?

-- 
Oscar Salvador
SUSE L3