Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 20 May 2019 19:55:32 -0700
From:      Mel Pilgrim <list_freebsd@bluerosetech.com>
To:        Tim Daneliuk <tundra@tundraware.com>, freebsd-questions@freebsd.org
Subject:   Re: Unclear On The New MDS Patch
Message-ID:  <17c04ae2-9b32-3cbe-93c9-819af21b4ff8@bluerosetech.com>
In-Reply-To: <ab6a8ca9-9c5a-251d-85df-74e8a5bb7410@tundraware.com>
References:  <db4ca28f-feed-5c71-b178-d3a750f1ed28@tundraware.com> <20190519230343.4340af09@gumby.homeunix.com> <ab6a8ca9-9c5a-251d-85df-74e8a5bb7410@tundraware.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On 2019-05-20 11:11, Tim Daneliuk wrote:
> What about cloud based servers like Digital Ocean FreeBSD droplets?  Does
> microcode updating even make sense in that context since the underlying
> system is actually what touches the hardware?

Short answer: no.

Longer answer:

Microcode updates use the WRMSR (WRite Model Specific Register) 
instruction, which requires ring 0.  Intel and AMD virtualization both 
emulate ring 0 for guests by pushing the physical ring 0 into a layer 
only accessible by the VMM and faking it for guests so they can operate 
without paravirtualization.  This is visible with software like 
msr-tools, where the rdmsr tool works fine but wrmsr appears to have no 
effect when run on a guest.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?17c04ae2-9b32-3cbe-93c9-819af21b4ff8>