linux-kernel - Re: [PATCH v4 09/12] x86/mtrr: construct a memory map with cache modes

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: <20230401142419.GCZCg+k9fgFPdAO4c/@fat_crate.local>
Date:   Sat, 1 Apr 2023 16:24:19 +0200
From:   Borislav Petkov <bp@...en8.de>
To:     Juergen Gross <jgross@...e.com>
Cc:     linux-kernel@...r.kernel.org, x86@...nel.org,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>,
        Dave Hansen <dave.hansen@...ux.intel.com>,
        "H. Peter Anvin" <hpa@...or.com>
Subject: Re: [PATCH v4 09/12] x86/mtrr: construct a memory map with cache
 modes

On Fri, Mar 31, 2023 at 03:23:13PM +0200, Juergen Gross wrote:
> In general the critical case is add_map_entry_at() returning 2 (in the
> case it is returning 1, the index can be set to -1, but there is always
> the "continue" statement right after that, which would execute the "i++"
> of the "for" statement).
> 
> add_map_entry_at() can return 2 only, if it detects "merge_prev" and
> "merge_next". "merge_prev" can be set only if the current index was > 0,
> which makes it impossible to return 2 if the index was 0.

Yeah, in the meantime I did add some debug printks in order to find my
way around that code...

> How should it be named? AMD TOP_MEM2 MSR?

It is already called that way - see "git grep TOP_MEM2" output.

> The problem isn't an added MTRR register, but a possibly replaced or removed
> one. Handling that is much more complicated, so I've chosen to do it the simple
> way.

Pls put that blurb over the function: needs to be called when MTRRs get
modified so that the map is kept valid, yadda yadda...

> In the end I'd expect setting of MTRRs to be a rare event, so there shouldn't be
> a performance issue with that approach.

Ack.

> Okay, I'll add some more comments.

Thx.

> OTOH, what was hard to write should be hard to read (just kidding).

I'm not kidding - it is not super easy.

:-)

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette