Hash: SHA256

On Fri, Oct 14, 2016 at 09:37:55AM -0700, Eric Shelton wrote:
> On Tuesday, December 1, 2015 at 1:09:06 PM UTC-5, Eric Shelton wrote:
> >
> > Marek,
> >
> > Just a heads up that I think I sorted out what needs to happen for Skylake 
> > to work with Qubes:
> >
> > https://groups.google.com/d/msg/qubes-users/UXP8RJffpsY/k15G1txyAgAJ
> >
> > Summary: (1) a linux command line option is required, and (2) i915 
> > requires firmware blobs for Skylake, which need to be included in initramfs.
> >
> > Also, although not necessary to get things running, attached is a recent 
> > patch from xen-devel to get Xen's mwait-idle working for Skylake, in case 
> > you want to make use of it.
> >
> > Best,
> > Eric
> >
> >
> After doing a fresh install of Qubes 3.2 on a Skylake system, I noticed 
> that 'xl dmesg' was reporting "mwait-idle: does not run on family 6 model 
> 94".  As I mentioned above, a patch is available to address this.  Since 
> then, the above patch was committed to xen.git 
> (4c6cd64519f9bc270a7278128c94e4b66e3d2077).
> There are also commits for mwait-idle updates to Kaby Lake 
> (1008063e1cd15257875b9db18eede710077adea5) and Skylake Xeon 
> (9062fd6bfa5211398582780d89d655cff9f9ed8c).


> While you are at it, you might want to pull in a couple of subsequent 
> Skylake-related commits: 5a211704e8813c4890c8ce8dc4189d1dfb35ecd0 
> ("mwait-idle: prevent SKL-H boot failure when C8+C9+C10 enabled") 

Backporting this one to Xen 4.6 is non-trivial, as
xen/include/public/arch-x86/cpufeatureset.h isn't there. And to make
things harder, "CPUID level 0x00000007:0.ebx" values are moved to word 5
from word 7 in the commit introducing cpufeatureset.h, without any
explanation. In fact I'm not sure if that isn't a bug.

So, I'll leave it out, as there is a workaround: either specify
max_cstate=7, or enable SGX in BIOS.

Hmm, but actually this may make things worse: instead of unsupported
mwait-idle, backporting only first patches may result in unbootable
system (unless mentioned workaround will be applied). So, maybe better
do not apply any of those?

> and 372ad59dd0e7a3df0bd46ec3c8b934d739eb07b5 ("mwait-idle: add a missing 
> __init annotation").

And this one is a fix for 5a211704e8813c4890c8ce8dc4189d1dfb35ecd0.

> There is another Skylake-related commit unrelated to mwait-idle 
> (c52319642bb9069436d4aec75361049f5fac63ed), but I don't know if this is 
> actually likely to cause any issues.

This also is tricky to backport to 4.6.

- -- 
Best Regards,
Marek Marczykowski-Górecki
Invisible Things Lab
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
Version: GnuPG v2


You received this message because you are subscribed to the Google Groups 
"qubes-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to qubes-devel+unsubscr...@googlegroups.com.
To post to this group, send email to qubes-devel@googlegroups.com.
To view this discussion on the web visit 
For more options, visit https://groups.google.com/d/optout.

Reply via email to