[MODERATED] Re: Slow Randomizing Boosts Denial of Service - Bulletin #1

[mark gross <mgross@linux.intel.com>]

On Mon, Apr 20, 2020 at 11:45:06PM +0200, speck for Thomas Gleixner wrote:
> Folks,
> 
> I've merged the V8 series with the polishing done by Borislav into the
> master branch (based on 5.7-rc2)  of the speck git repository at:
> 
>    cvs.ou.linutronix.de:linux/speck/linux
> 
> The commits are considered stable from now on. Any further tweaks need
> to go on top.
> 
> The stable branches linux-5.6.y, linux-5.4.y, linux-4.19.y and
> linux-4.14.y have been updated with backports. Kernels prior to 4.14 are
> left as an exercise to the members of the kernel necrophilia cult as
> usual.

We have ran all these through the test cases I used in my development. They all
pass.

FWIW I did backports from 4.14.y to 4.9 (easy) and 4.4 (needed to cherry pic an
extra patch to make it work)  These also pass testing.

Should I post them to this list or sit on them until disclosure?

Thank you for all your help on this effort!  I do appreciate it.

--mark

> A tarball with git bundles is attached.
> 
> The subject line of this mail is a reference to the following part of
> the documentation:
> 
>   Executing RDRAND, RDSEED or EGETKEY will delay memory accesses from
>   other logical processors that miss their core caches, with an impact
>   similar to legacy locked cache-line-split accesses.
> 
> IOW, yet another DoS tool. Oh well...
> 
> Thanks,
> 
>         tglx
>