linux-kernel - Re: [PATCH] x86/cacheinfo: Don't use cpu_llc_shared_map for !CONFIG

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: <20220818122925.GA8507@linuxonhyperv3.guj3yctzbm1etfxqx2vob5hsef.xx.internal.cloudapp.net>
Date:   Thu, 18 Aug 2022 05:29:25 -0700
From:   Saurabh Singh Sengar <ssengar@...ux.microsoft.com>
To:     Borislav Petkov <bp@...en8.de>
Cc:     ssengar@...rosoft.com, mikelley@...rosoft.com, tglx@...utronix.de,
        mingo@...hat.com, dave.hansen@...ux.intel.com, x86@...nel.org,
        hpa@...or.com, peterz@...radead.org, tim.c.chen@...ux.intel.com,
        will@...nel.org, song.bao.hua@...ilicon.com,
        suravee.suthikulpanit@....com, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] x86/cacheinfo: Don't use cpu_llc_shared_map for
 !CONFIG_SMP

On Thu, Aug 18, 2022 at 01:57:47PM +0200, Borislav Petkov wrote:
> On Wed, Aug 17, 2022 at 09:52:25PM -0700, Saurabh Singh Sengar wrote:
> > Shall I mention here "Non-SMP AMD processor" instead ?
> 
> You should explain what that is.
> 
> Is that a CONFIG_SMP=n kernel which you boot on a AMD CPU?

Yes, thats correct.

> 
> IOW, how do I reproduce the issue you're describing below, here locally?

Boot latest linux kernel on AMD CPU, having CONGIG_SMP=n.
Once booted execute lscpu (just lscpu without any argument ), this will cause segfault.
Please make a note issue is observed with 2.34 version (default lscpu version with
Ubuntu 20.04.4) of lscpu.

Also, if for the above test we take linux kernel code before below commit,
shared_cpu_map file gets created and lscpu just works fine.
b81dce77ced ("cpumask: Fix invalid uniprocessor mask assumption")

> 
> Send dmesg pls.
> 

dmesg :
We get only 2 lines of dmesg for this segfault.
Step 1: dmesg -c
Step 2: lscpu
Step 3: dmesg
below is the output of step 3.

Dmesg:
[ 6951.530138] lscpu[99034]: segfault at 0 ip 00005557bc9c35ba sp 00007ffced2681c8 error 4 in lscpu[5557bc9c0000+e000]
[ 6951.530145] Code: 4d 39 c2 76 33 48 c1 e9 06 49 89 c8 4c 8d 2c cd 00 00 00 00 44 89 c9 4f 8b 04 c3 83 e1 3f 4d 0f a3 c8 73 14 4c 8b 02 4d 8b 00 <4f> 8b 04 28 49 d3 e8 41 83 e0 01 44 01 c3 48 89 c1 49 39 c4 74 38

> Also, what is the use case?
> 

Explained above, putting here more precise steps:
Step 1: Build linux kernel for AMD processor supporting L3 cache with CONFIG_SMP=n.
Step 2: Boot this kernel.
Step 3: Once in command prompt, enter lscpu

If lspu is version 2.34 like the one I have (default with Ubuntu 20.04.4),
segfault will be observed.

> Why would you even build with SMP off?

This is caught in our testing.

> 
> Feel free to explain more verbosely what you're trying to accomplish.
> 
> Thx.
> 
> -- 
> Regards/Gruss,
>     Boris.
> 
> https://people.kernel.org/tglx/notes-about-netiquette

Please let me know if any thing is unclear or need more info reproducing the scenario.

- Saurabh