linux-kernel - Re: [PATCH] arch/x86: reset MXCSR to default in kernel_fpu

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] [day] [month] [year] [list]

Date:   Thu, 11 Jun 2020 08:50:18 -0600
From:   Shuah Khan <skhan@...uxfoundation.org>
To:     Andy Lutomirski <luto@...capital.net>,
        Borislav Petkov <bp@...en8.de>
Cc:     Petteri Aimonen <jpa@....mail.kapsi.fi>,
        Andy Lutomirski <luto@...nel.org>,
        Shuah Khan <shuah@...nel.org>,
        Dave Hansen <dave.hansen@...el.com>,
        "H. Peter Anvin" <hpa@...or.com>, x86-ml <x86@...nel.org>,
        lkml <linux-kernel@...r.kernel.org>, skhan@...uxfoundation.org
Subject: Re: [PATCH] arch/x86: reset MXCSR to default in kernel_fpu_begin()

On 6/11/20 8:19 AM, Andy Lutomirski wrote:
> 
> 
>> On Jun 11, 2020, at 2:18 AM, Borislav Petkov <bp@...en8.de> wrote:
>>
>> On Thu, Jun 11, 2020 at 10:36:08AM +0300, Petteri Aimonen wrote:
>>> Hi,
>>>
>>>> How about putting the file you frob in
>>>> /sys/kernel/debug/selftest_helpers/something_or_other.  The idea would
>>>> be that /sys/kernel/debug/selftest_helpers would be a general place
>>>> for kernel helpers needed to make selftests work.
>>>
>>> Seems like this is the consensus for now.
>>>
>>> Any opinions on whether the module should remove "selftest_helpers"
>>> directory on unloading, or not?
>>>
>>> 1) Removing would break if other test modules will use the same dir.
>>> 2) Not removing will leave the directory dangling.
>>> 3) Remove only if empty is one option, though I'm unsure how to
>>>    cleanly check if debugfs directory is empty.
>>> 4) E.g. /sys/kernel/debug/x86/ is created centrally and a symbol is
>>>    exported for its dentry. But I'm not sure if it really makes sense
>>>    to add another exported symbol just for selftest_helpers.
>>
>> I'd say you do the simple thing and cleanup after you're done, i.e.,
>> remove the dir. When something else starts using it, then it would need
>> to be taught to deal with multiple users.
> 
> Seems good to me. Let’s have at least two users before we go nuts with the architecture :)
> 

I am good with this. Let's start with simpler approach and build.
Please cc linux-kselftest and add me to the patch thread.

Thanks for this work. I think this will help expand coverage.

thanks,
-- Shuah