Re: [PATCH 2/2] x86: Add "virt flag" in /proc/cpuinfo

["Yang, Sheng" <sheng.yang@intel.com>]

On Friday 27 June 2008 22:50:06 Aurelien Jarno wrote:
> On Fri, Jun 27, 2008 at 11:31:38AM +0800, Yang, Sheng wrote:
> > On Friday 27 June 2008 11:26:15 Jeremy Fitzhardinge wrote:
> > > Yang, Sheng wrote:
> > > > From 54b1bb9fe5d2fe40fc047b43dd4e1a480d41a977 Mon Sep 17
> > > > 00:00:00 2001 From: Sheng Yang <sheng.yang@intel.com>
> > > > Date: Tue, 24 Jun 2008 17:03:17 +0800
> > > > Subject: [PATCH] x86: Add "virt flag" in /proc/cpuinfo
> > > >
> > > > The hardware virtualization technology evolves very fast. But
> > > > currently it's hard to tell if your CPU support a certain
> > > > kind of HW technology without dig into the source code.
> > > >
> > > > The patch add a new item under /proc/cpuinfo, named "virt
> > > > flag". The "virt flag" got the similar function as "flag". It
> > > > is used to indicate what features does this CPU supported. It
> > > > don't cover all features but only the important ones.
> > >
> > > A cpu feature is a cpu feature.  I'd prefer to see all this in
> > > "flags:".
> > >
> > >     J
> >
> > But I think (as I said before)
> >
> > 1. The standard flag covered upper level of cpu capability, they
> > are covered by CPUID. And virt flag was enabled by vmx/svm (we
> > can leave it blank also), and covered by MSR. It's very
> > different.
> >
> > 2. If we add virtual feature to standard flag, I am afraid it
> > would grow too fast, though we just add some key feature to it.
>
> What about dumping the virtualisation flags to the kernel log when
> the vmx/svm module is loaded?
>
> Also do we have an idea of the number of flags that would appear on
> a current CPU?

I think dmesg is not so reliable...

At the time, I exported 5 flags for vmx, and 3 of them can be find in 
current CPU.

My originally patch add a attribute on sysfs for kvm. But later Avi 
suggest to use /proc/cpuinfo, for virtualization feature is becoming 
more important part of CPU. And I agreed with him.

--
Thanks
Yang, Sheng