Date:   Wed, 22 Feb 2023 23:26:46 +0100
From:   Borislav Petkov <bp@alien8.de>
To:     Andrew Cooper <andrew.cooper3@citrix.com>
Cc:     Tavis Ormandy <taviso@gmail.com>, linux-kernel@vger.kernel.org,
        x86@kernel.org
Subject: Re: x86: AMD Zen2 ymm registers rolling back
Message-ID: <Y/aWputqWVAN0cJt@zn.tnic>
References: <Y/W4x7/KFqmDmmR7@thinkstation.cmpxchg8b.net>
 <Y/XTT59OrLw2as4R@zn.tnic>
 <Y/Xc+yMzI83WZ4V1@zn.tnic>
 <0371ec3d-0899-f94a-7f21-21d805df2927@citrix.com>
 <Y/Xp73KJe3c/1jrn@zn.tnic>
 <Y/aIjUr78yd9U+wl@thinkstation.cmpxchg8b.net>
 <f074e03d-a991-23cd-80d7-162067143034@citrix.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <f074e03d-a991-23cd-80d7-162067143034@citrix.com>
Precedence: bulk

On Wed, Feb 22, 2023 at 10:17:35PM +0000, Andrew Cooper wrote:
> I don't see anything in your code which would cause a change in MXCSR. 
> Which most likely means there's a path in the kernel modifying MXCSR and
> that doesn't sound like a thing that ought to be happening by default.

We were just talking about this internally too. We could trace it to see
what fiddles with MXCSR but meh, bigger fish to fry.

-- 
Regards/Gruss,
    Boris.

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