From: Wei Liu Subject: Re: [PATCH] tools/libxc: Improve efficiency of xc_cpuid_apply_policy() Date: Tue, 6 Oct 2015 10:14:57 +0100 Message-ID: <20151006091457.GP29124@zion.uk.xensource.com> References: <1444050737-8437-1-git-send-email-andrew.cooper3@citrix.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Content-Disposition: inline In-Reply-To: <1444050737-8437-1-git-send-email-andrew.cooper3@citrix.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Andrew Cooper Cc: Wei Liu , Ian Jackson , Ian Campbell , Xen-devel List-Id: xen-devel@lists.xenproject.org On Mon, Oct 05, 2015 at 02:12:17PM +0100, Andrew Cooper wrote: > Having the internals of xc_cpuid_policy() make hypercalls to collect domain > information causes xc_cpuid_apply_policy() to be very inefficient. > > Re-order operations to collect all information at once at the outermost layer, > and pass a structure in to all cpuid policy generation functions. > > This removes several hypercalls (4 from HVM, 3 from PV) for each of the > up-to 108 leaves processed. > > No change in the eventual policy provided, although all the information > gathering how has (or has correct) error checking. > > Signed-off-by: Andrew Cooper > --- > CC: Ian Campbell > CC: Ian Jackson > CC: Wei Liu Acked-by: Wei Liu > --- > tools/libxc/xc_cpuid_x86.c | 245 +++++++++++++++++++++++++------------------- > 1 file changed, 141 insertions(+), 104 deletions(-) > > diff --git a/tools/libxc/xc_cpuid_x86.c b/tools/libxc/xc_cpuid_x86.c > index e146a3e..715f30c 100644 > --- a/tools/libxc/xc_cpuid_x86.c > +++ b/tools/libxc/xc_cpuid_x86.c > @@ -33,6 +33,27 @@ > #define DEF_MAX_INTELEXT 0x80000008u > #define DEF_MAX_AMDEXT 0x8000001cu > > +struct cpuid_domain_info > +{ > + enum > + { > + VENDOR_UNKNOWN, > + VENDOR_INTEL, > + VENDOR_AMD, > + } vendor; > + > + bool hvm; > + bool pvh; > + uint64_t xfeature_mask; > + > + /* PV-only information. */ > + bool pv64; > + > + /* HVM-only infomation. */ > + bool pae; > + bool nestedhvm; I would have used union for these fields. But I don't want to bikeshed too much on this. Definitely not a blocker for this patch. :-)