Response to Meltdown and Spectre

[Christian Weisgerber]

On 2018-02-01, "@lbutlr" <kremels at kreme.com> wrote:

> That seems highly unlikely. It will damage the role of Intel in
> the server market fora time, but the trouble is that AMD's behavior
> has been at least as bad as Intel's, if not worse, in regards to
> Meltdown, so there's not a clearly better choice even though the
> AMD chips have less issues.

AMD's initial response appeared to have been written by a PR person
who simply summarized the vulnerability information from the
Spectre/Meltdown papers and deployed the usual head-in-the-sand
position that there is no vulnerability until an exploit is
demonstrated.

AMD has always said that their x86 CPUs are not vulnerable to
Meltdown and nobody is contradicting them on this.  However, like
everybody else implementing speculative executaion, they are
vulnerable to Spectre variants 1 and 2.  The initial response
downplayed this dangerously, but they eventually admitted it.

The best reaction came from ARM.  They provided a COMPLETE list of
all their CPUs that are affected, and they documented another
vulnerability (Meltdown 3a, reading of supervisor registers from
user mode) that had not even been considered in the original research
papers.

-- 
Christian "naddy" Weisgerber                          naddy at mips.inka.de