linux-kernel - Re: [PATCH v2 2/5] x86/mce/inject: Warn the user on a not set valid bit in MCA

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]

Date:   Tue, 26 Oct 2021 13:53:32 -0500
From:   "Koralahalli Channabasappa, Smita" <skoralah@....com>
To:     Borislav Petkov <bp@...en8.de>
Cc:     Smita Koralahalli <Smita.KoralahalliChannabasappa@....com>,
        x86@...nel.org, linux-edac@...r.kernel.org,
        linux-kernel@...r.kernel.org, Tony Luck <tony.luck@...el.com>,
        "H . Peter Anvin" <hpa@...or.com>, yazen.ghannam@....com
Subject: Re: [PATCH v2 2/5] x86/mce/inject: Warn the user on a not set valid
 bit in MCA_STATUS

On 10/26/21 12:15 PM, Borislav Petkov wrote:

> On Tue, Oct 26, 2021 at 11:58:58AM -0500, Koralahalli Channabasappa, Smita wrote:
> Whoops, sorry about that.
>
> So let's analyze this properly - there are two cases:
>
> 1. warn if VAL=0: what does that bring us? The person doing the injection
> will simply have to set the valid bit and repeat the injection.
>
> I guess "maybe the user wants to inject with Val not set" doesn't make a
> whole lot of sense because nothing will happen - error will get ignored.
>
> So we can do all the warning we want - it will be useless and in some
> cases the user might not even see it.
>
> So it sounds to me like setting the valid bit directly makes a lot more
> sense.
>
> 2. Automatically set VAL=1 to correct any VAL=0 injections.
>
> Yes, we force the VAL bit to 1 and that is not what the user injected
> but the user injecting with VAL=0 will get ignored, i.e., it will be
> pointless.
>
> So we "help" here and set the valid bit.
>
> Anything else I'm missing?
>
> Sorry again for being back'n'forth on this.

Right. We are correcting VAL=0 injections made by the user by setting
valid bit unconditionally.

Not a problem. I could have broken this down in the comments before
coming up with this patch.

I will make the necessary changes and set the valid bit in the next
revision of the patch series.

Thanks,
Smita.