Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 30 Apr 2006 11:50:30 +0200
From:      Marius Nuennerich <marius.nuennerich@gmx.net>
To:        David Xu <davidxu@freebsd.org>
Cc:        Igor Sysoev <is@rambler-co.ru>, freebsd-current@freebsd.org
Subject:   Re: FreeBSD-SA-06:14.fpu
Message-ID:  <20060430115030.69402983@sol>
In-Reply-To: <44547C85.7040107@freebsd.org>
References:  <20060430115932.Y39336@is.park.rambler.ru> <44547C85.7040107@freebsd.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 30 Apr 2006 16:59:49 +0800
David Xu <davidxu@freebsd.org> wrote:

> Igor Sysoev wrote:
> > 
> > The last security advisory FreeBSD-SA-06:14.fpu
> > ftp://ftp.freebsd.org/pub/FreeBSD/CERT/advisories/FreeBSD-SA-06:14.fpu.asc
> > that fixes very doubtful security bug in AMD CPUs, also
> > adds unnecessary penalty in FPU context switch for all other
> > SSE-enabled CPUs.
> > 
> > 
> > Igor Sysoev
> > http://sysoev.ru/en/
> 
> Probably it should only be applied to AMD CPU but not Intel and others,
> it is easy to check cpu vendor and put a
> if (bug_fxsave)
> 	fpu_clean_state();
> 
> in file npx.c.
> 
> David Xu

Shouldn't we use an existing variable which is already in L1 instead of
the new dummy_variable? Or isn't this noticable in performance?

btw Linux does it this way:
> http://www.kernel.org/git/?p=linux/kernel/git/stable/linux-2.6.16.y.git;a=commitdiff;h=7466f9e72dac13452d871a3fb72fc7bd9c93c864

regards
Marius

P.S. I have no idea what variable we could use.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20060430115030.69402983>