Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
/*
|
|
|
|
* cpu_x86.c: CPU driver for CPUs with x86 compatible CPUID instruction
|
|
|
|
*
|
2011-01-10 23:56:36 +00:00
|
|
|
* Copyright (C) 2009-2011 Red Hat, Inc.
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
*
|
|
|
|
* This library is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
|
|
|
* version 2.1 of the License, or (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This library is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* Lesser General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU Lesser General Public
|
|
|
|
* License along with this library; if not, write to the Free Software
|
|
|
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
|
|
|
*
|
|
|
|
* Authors:
|
|
|
|
* Jiri Denemark <jdenemar@redhat.com>
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <config.h>
|
|
|
|
|
|
|
|
#include <stdint.h>
|
|
|
|
|
|
|
|
#include "logging.h"
|
|
|
|
#include "memory.h"
|
2010-04-14 15:41:32 +00:00
|
|
|
#include "util.h"
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
#include "cpu.h"
|
|
|
|
#include "cpu_map.h"
|
|
|
|
#include "cpu_x86.h"
|
|
|
|
|
|
|
|
|
|
|
|
#define VIR_FROM_THIS VIR_FROM_CPU
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
#define VENDOR_STRING_LENGTH 12
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
static const struct cpuX86cpuid cpuidNull = { 0, 0, 0, 0, 0 };
|
2010-07-02 15:51:59 +00:00
|
|
|
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
static const char *archs[] = { "i686", "x86_64" };
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
struct x86_vendor {
|
|
|
|
char *name;
|
|
|
|
struct cpuX86cpuid cpuid;
|
|
|
|
|
|
|
|
struct x86_vendor *next;
|
|
|
|
};
|
|
|
|
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
struct x86_feature {
|
|
|
|
char *name;
|
2010-06-30 11:08:57 +00:00
|
|
|
union cpuData *data;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
struct x86_feature *next;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct x86_model {
|
|
|
|
char *name;
|
2010-07-02 15:51:59 +00:00
|
|
|
const struct x86_vendor *vendor;
|
2010-06-30 11:08:57 +00:00
|
|
|
union cpuData *data;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
struct x86_model *next;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct x86_map {
|
2010-07-02 15:51:59 +00:00
|
|
|
struct x86_vendor *vendors;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
struct x86_feature *features;
|
|
|
|
struct x86_model *models;
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
enum compare_result {
|
|
|
|
SUBSET,
|
|
|
|
EQUAL,
|
|
|
|
SUPERSET,
|
|
|
|
UNRELATED
|
|
|
|
};
|
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
struct data_iterator {
|
|
|
|
union cpuData *data;
|
|
|
|
int pos;
|
|
|
|
bool extended;
|
|
|
|
};
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
#define DATA_ITERATOR_INIT(data) \
|
|
|
|
{ data, -1, false }
|
|
|
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
x86DataIteratorInit(struct data_iterator *iter,
|
|
|
|
union cpuData *data)
|
|
|
|
{
|
|
|
|
struct data_iterator init = DATA_ITERATOR_INIT(data);
|
|
|
|
|
|
|
|
*iter = init;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
static int
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
x86cpuidMatch(const struct cpuX86cpuid *cpuid1,
|
|
|
|
const struct cpuX86cpuid *cpuid2)
|
|
|
|
{
|
|
|
|
return (cpuid1->eax == cpuid2->eax &&
|
|
|
|
cpuid1->ebx == cpuid2->ebx &&
|
|
|
|
cpuid1->ecx == cpuid2->ecx &&
|
|
|
|
cpuid1->edx == cpuid2->edx);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
static int
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
x86cpuidMatchMasked(const struct cpuX86cpuid *cpuid,
|
|
|
|
const struct cpuX86cpuid *mask)
|
|
|
|
{
|
|
|
|
return ((cpuid->eax & mask->eax) == mask->eax &&
|
|
|
|
(cpuid->ebx & mask->ebx) == mask->ebx &&
|
|
|
|
(cpuid->ecx & mask->ecx) == mask->ecx &&
|
|
|
|
(cpuid->edx & mask->edx) == mask->edx);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
static int
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
x86cpuidMatchAny(const struct cpuX86cpuid *cpuid,
|
|
|
|
const struct cpuX86cpuid *mask)
|
|
|
|
{
|
|
|
|
return ((cpuid->eax & mask->eax) ||
|
|
|
|
(cpuid->ebx & mask->ebx) ||
|
|
|
|
(cpuid->ecx & mask->ecx) ||
|
|
|
|
(cpuid->edx & mask->edx));
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
static void
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
x86cpuidSetBits(struct cpuX86cpuid *cpuid,
|
|
|
|
const struct cpuX86cpuid *mask)
|
|
|
|
{
|
|
|
|
cpuid->eax |= mask->eax;
|
|
|
|
cpuid->ebx |= mask->ebx;
|
|
|
|
cpuid->ecx |= mask->ecx;
|
|
|
|
cpuid->edx |= mask->edx;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
static void
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
x86cpuidClearBits(struct cpuX86cpuid *cpuid,
|
|
|
|
const struct cpuX86cpuid *mask)
|
|
|
|
{
|
|
|
|
cpuid->eax &= ~mask->eax;
|
|
|
|
cpuid->ebx &= ~mask->ebx;
|
|
|
|
cpuid->ecx &= ~mask->ecx;
|
|
|
|
cpuid->edx &= ~mask->edx;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
static void
|
2010-01-27 13:33:20 +00:00
|
|
|
x86cpuidAndBits(struct cpuX86cpuid *cpuid,
|
|
|
|
const struct cpuX86cpuid *mask)
|
|
|
|
{
|
|
|
|
cpuid->eax &= mask->eax;
|
|
|
|
cpuid->ebx &= mask->ebx;
|
|
|
|
cpuid->ecx &= mask->ecx;
|
|
|
|
cpuid->edx &= mask->edx;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
/* skips all zero CPUID leafs */
|
|
|
|
static struct cpuX86cpuid *
|
|
|
|
x86DataCpuidNext(struct data_iterator *iterator)
|
|
|
|
{
|
|
|
|
struct cpuX86cpuid *ret;
|
|
|
|
struct cpuX86Data *data;
|
|
|
|
|
|
|
|
if (!iterator->data)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
data = &iterator->data->x86;
|
|
|
|
|
|
|
|
do {
|
|
|
|
ret = NULL;
|
|
|
|
iterator->pos++;
|
|
|
|
|
|
|
|
if (!iterator->extended) {
|
|
|
|
if (iterator->pos < data->basic_len)
|
|
|
|
ret = data->basic + iterator->pos;
|
|
|
|
else {
|
|
|
|
iterator->extended = true;
|
|
|
|
iterator->pos = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (iterator->extended && iterator->pos < data->extended_len) {
|
|
|
|
ret = data->extended + iterator->pos;
|
|
|
|
}
|
|
|
|
} while (ret && x86cpuidMatch(ret, &cpuidNull));
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
static struct cpuX86cpuid *
|
|
|
|
x86DataCpuid(const union cpuData *data,
|
|
|
|
uint32_t function)
|
|
|
|
{
|
|
|
|
struct cpuX86cpuid *cpuids;
|
|
|
|
int len;
|
|
|
|
unsigned int i;
|
|
|
|
|
|
|
|
if (function < CPUX86_EXTENDED) {
|
|
|
|
cpuids = data->x86.basic;
|
|
|
|
len = data->x86.basic_len;
|
|
|
|
i = function;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
cpuids = data->x86.extended;
|
|
|
|
len = data->x86.extended_len;
|
|
|
|
i = function - CPUX86_EXTENDED;
|
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if (i < len && !x86cpuidMatch(cpuids + i, &cpuidNull))
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
return cpuids + i;
|
|
|
|
else
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
x86DataFree(union cpuData *data)
|
|
|
|
{
|
|
|
|
if (data == NULL)
|
|
|
|
return;
|
|
|
|
|
|
|
|
VIR_FREE(data->x86.basic);
|
|
|
|
VIR_FREE(data->x86.extended);
|
|
|
|
VIR_FREE(data);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static union cpuData *
|
|
|
|
x86DataCopy(const union cpuData *data)
|
|
|
|
{
|
|
|
|
union cpuData *copy = NULL;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
if (VIR_ALLOC(copy) < 0
|
|
|
|
|| VIR_ALLOC_N(copy->x86.basic, data->x86.basic_len) < 0
|
|
|
|
|| VIR_ALLOC_N(copy->x86.extended, data->x86.extended_len) < 0) {
|
|
|
|
x86DataFree(copy);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
copy->x86.basic_len = data->x86.basic_len;
|
|
|
|
for (i = 0; i < data->x86.basic_len; i++)
|
|
|
|
copy->x86.basic[i] = data->x86.basic[i];
|
|
|
|
|
|
|
|
copy->x86.extended_len = data->x86.extended_len;
|
|
|
|
for (i = 0; i < data->x86.extended_len; i++)
|
|
|
|
copy->x86.extended[i] = data->x86.extended[i];
|
|
|
|
|
|
|
|
return copy;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
static int
|
|
|
|
x86DataExpand(union cpuData *data,
|
|
|
|
int basic_by,
|
|
|
|
int extended_by)
|
|
|
|
{
|
|
|
|
size_t i;
|
|
|
|
|
|
|
|
if (basic_by > 0) {
|
|
|
|
size_t len = data->x86.basic_len;
|
|
|
|
if (VIR_EXPAND_N(data->x86.basic, data->x86.basic_len, basic_by) < 0)
|
|
|
|
goto no_memory;
|
|
|
|
|
|
|
|
for (i = 0; i < basic_by; i++)
|
|
|
|
data->x86.basic[len + i].function = len + i;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (extended_by > 0) {
|
|
|
|
size_t len = data->x86.extended_len;
|
|
|
|
if (VIR_EXPAND_N(data->x86.extended, data->x86.extended_len, extended_by) < 0)
|
|
|
|
goto no_memory;
|
|
|
|
|
|
|
|
for (i = 0; i < extended_by; i++)
|
|
|
|
data->x86.extended[len + i].function = len + i + CPUX86_EXTENDED;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
no_memory:
|
|
|
|
virReportOOMError();
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
static int
|
|
|
|
x86DataAddCpuid(union cpuData *data,
|
|
|
|
const struct cpuX86cpuid *cpuid)
|
|
|
|
{
|
2010-06-30 11:08:57 +00:00
|
|
|
unsigned int basic_by = 0;
|
|
|
|
unsigned int extended_by = 0;
|
2010-07-02 15:51:59 +00:00
|
|
|
struct cpuX86cpuid **cpuids;
|
|
|
|
unsigned int pos;
|
|
|
|
|
|
|
|
if (cpuid->function < CPUX86_EXTENDED) {
|
|
|
|
pos = cpuid->function;
|
2010-06-30 11:08:57 +00:00
|
|
|
basic_by = pos + 1 - data->x86.basic_len;
|
2010-07-02 15:51:59 +00:00
|
|
|
cpuids = &data->x86.basic;
|
|
|
|
} else {
|
|
|
|
pos = cpuid->function - CPUX86_EXTENDED;
|
2010-06-30 11:08:57 +00:00
|
|
|
extended_by = pos + 1 - data->x86.extended_len;
|
2010-07-02 15:51:59 +00:00
|
|
|
cpuids = &data->x86.extended;
|
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if (x86DataExpand(data, basic_by, extended_by) < 0)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
x86cpuidSetBits((*cpuids) + pos, cpuid);
|
2010-07-02 15:51:59 +00:00
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static int
|
|
|
|
x86DataAdd(union cpuData *data1,
|
|
|
|
const union cpuData *data2)
|
|
|
|
{
|
|
|
|
unsigned int i;
|
|
|
|
|
|
|
|
if (x86DataExpand(data1,
|
|
|
|
data2->x86.basic_len - data1->x86.basic_len,
|
|
|
|
data2->x86.extended_len - data1->x86.extended_len) < 0)
|
|
|
|
return -1;
|
2010-07-02 15:51:59 +00:00
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
for (i = 0; i < data2->x86.basic_len; i++) {
|
|
|
|
x86cpuidSetBits(data1->x86.basic + i,
|
|
|
|
data2->x86.basic + i);
|
2010-07-02 15:51:59 +00:00
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
for (i = 0; i < data2->x86.extended_len; i++) {
|
|
|
|
x86cpuidSetBits(data1->x86.extended + i,
|
|
|
|
data2->x86.extended + i);
|
|
|
|
}
|
2010-07-02 15:51:59 +00:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-04-14 15:41:32 +00:00
|
|
|
static void
|
|
|
|
x86DataSubtract(union cpuData *data1,
|
|
|
|
const union cpuData *data2)
|
|
|
|
{
|
|
|
|
unsigned int i;
|
|
|
|
unsigned int len;
|
|
|
|
|
|
|
|
len = MIN(data1->x86.basic_len, data2->x86.basic_len);
|
|
|
|
for (i = 0; i < len; i++) {
|
|
|
|
x86cpuidClearBits(data1->x86.basic + i,
|
|
|
|
data2->x86.basic + i);
|
|
|
|
}
|
|
|
|
|
|
|
|
len = MIN(data1->x86.extended_len, data2->x86.extended_len);
|
|
|
|
for (i = 0; i < len; i++) {
|
|
|
|
x86cpuidClearBits(data1->x86.extended + i,
|
|
|
|
data2->x86.extended + i);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
static void
|
|
|
|
x86DataIntersect(union cpuData *data1,
|
|
|
|
const union cpuData *data2)
|
2010-07-02 15:51:40 +00:00
|
|
|
{
|
2010-06-30 11:08:57 +00:00
|
|
|
struct data_iterator iter = DATA_ITERATOR_INIT(data1);
|
|
|
|
struct cpuX86cpuid *cpuid1;
|
|
|
|
struct cpuX86cpuid *cpuid2;
|
2010-07-02 15:51:40 +00:00
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
while ((cpuid1 = x86DataCpuidNext(&iter))) {
|
|
|
|
cpuid2 = x86DataCpuid(data2, cpuid1->function);
|
|
|
|
if (cpuid2)
|
|
|
|
x86cpuidAndBits(cpuid1, cpuid2);
|
|
|
|
else
|
|
|
|
x86cpuidClearBits(cpuid1, cpuid1);
|
2010-07-02 15:51:40 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
static bool
|
|
|
|
x86DataIsEmpty(union cpuData *data)
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
{
|
2010-06-30 11:08:57 +00:00
|
|
|
struct data_iterator iter = DATA_ITERATOR_INIT(data);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
2012-03-22 11:33:35 +00:00
|
|
|
return x86DataCpuidNext(&iter) == NULL;
|
2010-06-30 11:08:57 +00:00
|
|
|
}
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
static bool
|
|
|
|
x86DataIsSubset(const union cpuData *data,
|
|
|
|
const union cpuData *subset)
|
|
|
|
{
|
|
|
|
|
|
|
|
struct data_iterator iter = DATA_ITERATOR_INIT((union cpuData *) subset);
|
|
|
|
const struct cpuX86cpuid *cpuid;
|
|
|
|
const struct cpuX86cpuid *cpuidSubset;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
while ((cpuidSubset = x86DataCpuidNext(&iter))) {
|
|
|
|
if (!(cpuid = x86DataCpuid(data, cpuidSubset->function)) ||
|
|
|
|
!x86cpuidMatchMasked(cpuid, cpuidSubset))
|
|
|
|
return false;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
return true;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-03-23 08:32:50 +00:00
|
|
|
/* also removes all detected features from data */
|
|
|
|
static int
|
|
|
|
x86DataToCPUFeatures(virCPUDefPtr cpu,
|
|
|
|
int policy,
|
|
|
|
union cpuData *data,
|
|
|
|
const struct x86_map *map)
|
|
|
|
{
|
|
|
|
const struct x86_feature *feature = map->features;
|
|
|
|
|
|
|
|
while (feature != NULL) {
|
2010-06-30 11:08:57 +00:00
|
|
|
if (x86DataIsSubset(data, feature->data)) {
|
|
|
|
x86DataSubtract(data, feature->data);
|
|
|
|
if (virCPUDefAddFeature(cpu, feature->name, policy) < 0)
|
|
|
|
return -1;
|
2010-03-23 08:32:50 +00:00
|
|
|
}
|
|
|
|
feature = feature->next;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
/* also removes bits corresponding to vendor string from data */
|
|
|
|
static const struct x86_vendor *
|
|
|
|
x86DataToVendor(union cpuData *data,
|
|
|
|
const struct x86_map *map)
|
|
|
|
{
|
|
|
|
const struct x86_vendor *vendor = map->vendors;
|
|
|
|
struct cpuX86cpuid *cpuid;
|
|
|
|
|
|
|
|
while (vendor) {
|
|
|
|
if ((cpuid = x86DataCpuid(data, vendor->cpuid.function)) &&
|
|
|
|
x86cpuidMatchMasked(cpuid, &vendor->cpuid)) {
|
|
|
|
x86cpuidClearBits(cpuid, &vendor->cpuid);
|
|
|
|
return vendor;
|
|
|
|
}
|
|
|
|
vendor = vendor->next;
|
|
|
|
}
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-01-15 15:58:59 +00:00
|
|
|
static virCPUDefPtr
|
|
|
|
x86DataToCPU(const union cpuData *data,
|
|
|
|
const struct x86_model *model,
|
|
|
|
const struct x86_map *map)
|
|
|
|
{
|
|
|
|
virCPUDefPtr cpu;
|
2010-04-14 15:41:32 +00:00
|
|
|
union cpuData *copy = NULL;
|
|
|
|
union cpuData *modelData = NULL;
|
2010-07-02 15:51:59 +00:00
|
|
|
const struct x86_vendor *vendor;
|
2010-01-15 15:58:59 +00:00
|
|
|
|
|
|
|
if (VIR_ALLOC(cpu) < 0 ||
|
2010-04-14 15:41:32 +00:00
|
|
|
!(cpu->model = strdup(model->name)) ||
|
|
|
|
!(copy = x86DataCopy(data)) ||
|
2010-06-30 11:08:57 +00:00
|
|
|
!(modelData = x86DataCopy(model->data)))
|
2010-01-15 15:58:59 +00:00
|
|
|
goto no_memory;
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
if ((vendor = x86DataToVendor(copy, map)) &&
|
|
|
|
!(cpu->vendor = strdup(vendor->name)))
|
|
|
|
goto no_memory;
|
|
|
|
|
2010-04-14 15:41:32 +00:00
|
|
|
x86DataSubtract(copy, modelData);
|
|
|
|
x86DataSubtract(modelData, data);
|
|
|
|
|
|
|
|
/* because feature policy is ignored for host CPU */
|
|
|
|
cpu->type = VIR_CPU_TYPE_GUEST;
|
2010-01-15 15:58:59 +00:00
|
|
|
|
2010-04-14 15:41:32 +00:00
|
|
|
if (x86DataToCPUFeatures(cpu, VIR_CPU_FEATURE_REQUIRE, copy, map) ||
|
|
|
|
x86DataToCPUFeatures(cpu, VIR_CPU_FEATURE_DISABLE, modelData, map))
|
2010-03-23 08:32:50 +00:00
|
|
|
goto error;
|
2010-01-15 15:58:59 +00:00
|
|
|
|
|
|
|
cleanup:
|
2010-04-14 15:41:32 +00:00
|
|
|
x86DataFree(modelData);
|
|
|
|
x86DataFree(copy);
|
2010-01-15 15:58:59 +00:00
|
|
|
return cpu;
|
|
|
|
|
|
|
|
no_memory:
|
2010-02-04 18:19:08 +00:00
|
|
|
virReportOOMError();
|
2010-01-15 15:58:59 +00:00
|
|
|
error:
|
|
|
|
virCPUDefFree(cpu);
|
|
|
|
cpu = NULL;
|
|
|
|
goto cleanup;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
static void
|
|
|
|
x86VendorFree(struct x86_vendor *vendor)
|
|
|
|
{
|
|
|
|
if (!vendor)
|
|
|
|
return;
|
|
|
|
|
|
|
|
VIR_FREE(vendor->name);
|
|
|
|
VIR_FREE(vendor);
|
2010-06-30 11:08:57 +00:00
|
|
|
}
|
2010-07-02 15:51:59 +00:00
|
|
|
|
|
|
|
|
|
|
|
static struct x86_vendor *
|
|
|
|
x86VendorFind(const struct x86_map *map,
|
|
|
|
const char *name)
|
|
|
|
{
|
|
|
|
struct x86_vendor *vendor;
|
|
|
|
|
|
|
|
vendor = map->vendors;
|
|
|
|
while (vendor) {
|
|
|
|
if (STREQ(vendor->name, name))
|
|
|
|
return vendor;
|
|
|
|
|
|
|
|
vendor = vendor->next;
|
|
|
|
}
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static int
|
|
|
|
x86VendorLoad(xmlXPathContextPtr ctxt,
|
|
|
|
struct x86_map *map)
|
|
|
|
{
|
|
|
|
struct x86_vendor *vendor = NULL;
|
|
|
|
char *string = NULL;
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
if (VIR_ALLOC(vendor) < 0)
|
|
|
|
goto no_memory;
|
|
|
|
|
|
|
|
vendor->name = virXPathString("string(@name)", ctxt);
|
|
|
|
if (!vendor->name) {
|
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
|
|
|
"%s", _("Missing CPU vendor name"));
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (x86VendorFind(map, vendor->name)) {
|
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
|
|
|
_("CPU vendor %s already defined"), vendor->name);
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
|
|
|
|
string = virXPathString("string(@string)", ctxt);
|
|
|
|
if (!string) {
|
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
|
|
|
_("Missing vendor string for CPU vendor %s"), vendor->name);
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
if (strlen(string) != VENDOR_STRING_LENGTH) {
|
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
|
|
|
_("Invalid CPU vendor string '%s'"), string);
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
|
|
|
|
vendor->cpuid.function = 0;
|
|
|
|
vendor->cpuid.ebx = (string[0] ) |
|
|
|
|
(string[1] << 8) |
|
|
|
|
(string[2] << 16) |
|
|
|
|
(string[3] << 24);
|
|
|
|
vendor->cpuid.edx = (string[4] ) |
|
|
|
|
(string[5] << 8) |
|
|
|
|
(string[6] << 16) |
|
|
|
|
(string[7] << 24);
|
|
|
|
vendor->cpuid.ecx = (string[8] ) |
|
|
|
|
(string[9] << 8) |
|
|
|
|
(string[10] << 16) |
|
|
|
|
(string[11] << 24);
|
|
|
|
|
|
|
|
if (!map->vendors)
|
|
|
|
map->vendors = vendor;
|
|
|
|
else {
|
|
|
|
vendor->next = map->vendors;
|
|
|
|
map->vendors = vendor;
|
|
|
|
}
|
|
|
|
|
|
|
|
out:
|
|
|
|
VIR_FREE(string);
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
no_memory:
|
|
|
|
virReportOOMError();
|
|
|
|
ret = -1;
|
|
|
|
ignore:
|
|
|
|
x86VendorFree(vendor);
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
static struct x86_feature *
|
|
|
|
x86FeatureNew(void)
|
|
|
|
{
|
|
|
|
struct x86_feature *feature;
|
|
|
|
|
|
|
|
if (VIR_ALLOC(feature) < 0)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
if (VIR_ALLOC(feature->data) < 0) {
|
|
|
|
VIR_FREE(feature);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
return feature;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
static void
|
|
|
|
x86FeatureFree(struct x86_feature *feature)
|
|
|
|
{
|
|
|
|
if (feature == NULL)
|
|
|
|
return;
|
|
|
|
|
|
|
|
VIR_FREE(feature->name);
|
2010-06-30 11:08:57 +00:00
|
|
|
x86DataFree(feature->data);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
VIR_FREE(feature);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static struct x86_feature *
|
|
|
|
x86FeatureFind(const struct x86_map *map,
|
|
|
|
const char *name)
|
|
|
|
{
|
|
|
|
struct x86_feature *feature;
|
|
|
|
|
|
|
|
feature = map->features;
|
|
|
|
while (feature != NULL) {
|
|
|
|
if (STREQ(feature->name, name))
|
|
|
|
return feature;
|
|
|
|
|
|
|
|
feature = feature->next;
|
|
|
|
}
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static int
|
|
|
|
x86FeatureLoad(xmlXPathContextPtr ctxt,
|
2010-07-02 15:51:59 +00:00
|
|
|
struct x86_map *map)
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
{
|
|
|
|
xmlNodePtr *nodes = NULL;
|
2009-12-21 18:12:45 +00:00
|
|
|
xmlNodePtr ctxt_node = ctxt->node;
|
2010-06-30 11:08:57 +00:00
|
|
|
struct x86_feature *feature;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
int ret = 0;
|
|
|
|
int i;
|
|
|
|
int n;
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if (!(feature = x86FeatureNew()))
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
goto no_memory;
|
|
|
|
|
2010-02-04 21:52:34 +00:00
|
|
|
feature->name = virXPathString("string(@name)", ctxt);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
if (feature->name == NULL) {
|
2010-02-10 12:22:36 +00:00
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
"%s", _("Missing CPU feature name"));
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (x86FeatureFind(map, feature->name)) {
|
2010-02-10 12:22:36 +00:00
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
_("CPU feature %s already defined"), feature->name);
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
|
2010-02-04 21:52:34 +00:00
|
|
|
n = virXPathNodeSet("./cpuid", ctxt, &nodes);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
if (n < 0)
|
|
|
|
goto ignore;
|
|
|
|
|
|
|
|
for (i = 0; i < n; i++) {
|
2010-06-30 11:08:57 +00:00
|
|
|
struct cpuX86cpuid cpuid;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
unsigned long fun, eax, ebx, ecx, edx;
|
|
|
|
int ret_fun, ret_eax, ret_ebx, ret_ecx, ret_edx;
|
|
|
|
|
|
|
|
ctxt->node = nodes[i];
|
|
|
|
fun = eax = ebx = ecx = edx = 0;
|
2010-02-04 21:52:34 +00:00
|
|
|
ret_fun = virXPathULongHex("string(@function)", ctxt, &fun);
|
|
|
|
ret_eax = virXPathULongHex("string(@eax)", ctxt, &eax);
|
|
|
|
ret_ebx = virXPathULongHex("string(@ebx)", ctxt, &ebx);
|
|
|
|
ret_ecx = virXPathULongHex("string(@ecx)", ctxt, &ecx);
|
|
|
|
ret_edx = virXPathULongHex("string(@edx)", ctxt, &edx);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
if (ret_fun < 0 || ret_eax == -2 || ret_ebx == -2
|
|
|
|
|| ret_ecx == -2 || ret_edx == -2) {
|
2010-02-10 12:22:36 +00:00
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
_("Invalid cpuid[%d] in %s feature"), i, feature->name);
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
cpuid.function = fun;
|
|
|
|
cpuid.eax = eax;
|
|
|
|
cpuid.ebx = ebx;
|
|
|
|
cpuid.ecx = ecx;
|
|
|
|
cpuid.edx = edx;
|
|
|
|
|
|
|
|
if (x86DataAddCpuid(feature->data, &cpuid))
|
|
|
|
goto no_memory;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if (map->features == NULL)
|
|
|
|
map->features = feature;
|
|
|
|
else {
|
|
|
|
feature->next = map->features;
|
|
|
|
map->features = feature;
|
|
|
|
}
|
|
|
|
|
|
|
|
out:
|
2009-12-21 18:12:45 +00:00
|
|
|
ctxt->node = ctxt_node;
|
|
|
|
VIR_FREE(nodes);
|
|
|
|
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
return ret;
|
|
|
|
|
|
|
|
no_memory:
|
2010-02-04 18:19:08 +00:00
|
|
|
virReportOOMError();
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
ret = -1;
|
|
|
|
|
|
|
|
ignore:
|
|
|
|
x86FeatureFree(feature);
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
static struct x86_model *
|
|
|
|
x86ModelNew(void)
|
|
|
|
{
|
|
|
|
struct x86_model *model;
|
|
|
|
|
|
|
|
if (VIR_ALLOC(model) < 0)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
if (VIR_ALLOC(model->data) < 0) {
|
|
|
|
VIR_FREE(model);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
return model;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
static void
|
|
|
|
x86ModelFree(struct x86_model *model)
|
|
|
|
{
|
|
|
|
if (model == NULL)
|
|
|
|
return;
|
|
|
|
|
|
|
|
VIR_FREE(model->name);
|
2010-06-30 11:08:57 +00:00
|
|
|
x86DataFree(model->data);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
VIR_FREE(model);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static struct x86_model *
|
|
|
|
x86ModelCopy(const struct x86_model *model)
|
|
|
|
{
|
|
|
|
struct x86_model *copy;
|
|
|
|
|
|
|
|
if (VIR_ALLOC(copy) < 0
|
2010-06-30 11:08:57 +00:00
|
|
|
|| !(copy->name = strdup(model->name))
|
|
|
|
|| !(copy->data = x86DataCopy(model->data))) {
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
x86ModelFree(copy);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
copy->vendor = model->vendor;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
return copy;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static struct x86_model *
|
|
|
|
x86ModelFind(const struct x86_map *map,
|
|
|
|
const char *name)
|
|
|
|
{
|
|
|
|
struct x86_model *model;
|
|
|
|
|
|
|
|
model = map->models;
|
|
|
|
while (model != NULL) {
|
|
|
|
if (STREQ(model->name, name))
|
|
|
|
return model;
|
|
|
|
|
|
|
|
model = model->next;
|
|
|
|
}
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static struct x86_model *
|
|
|
|
x86ModelFromCPU(const virCPUDefPtr cpu,
|
|
|
|
const struct x86_map *map,
|
|
|
|
int policy)
|
|
|
|
{
|
|
|
|
struct x86_model *model = NULL;
|
|
|
|
int i;
|
|
|
|
|
2010-07-12 12:41:36 +00:00
|
|
|
if (policy == VIR_CPU_FEATURE_REQUIRE) {
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
if ((model = x86ModelFind(map, cpu->model)) == NULL) {
|
2010-02-10 12:22:36 +00:00
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
_("Unknown CPU model %s"), cpu->model);
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
if ((model = x86ModelCopy(model)) == NULL)
|
|
|
|
goto no_memory;
|
2010-06-30 11:08:57 +00:00
|
|
|
} else if (!(model = x86ModelNew())) {
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
goto no_memory;
|
2010-06-30 11:08:57 +00:00
|
|
|
} else if (cpu->type == VIR_CPU_TYPE_HOST) {
|
2010-07-12 12:41:36 +00:00
|
|
|
return model;
|
2010-06-30 11:08:57 +00:00
|
|
|
}
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
for (i = 0; i < cpu->nfeatures; i++) {
|
|
|
|
const struct x86_feature *feature;
|
|
|
|
|
|
|
|
if (cpu->type == VIR_CPU_TYPE_GUEST
|
|
|
|
&& cpu->features[i].policy != policy)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
if ((feature = x86FeatureFind(map, cpu->features[i].name)) == NULL) {
|
2010-02-10 12:22:36 +00:00
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
_("Unknown CPU feature %s"), cpu->features[i].name);
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if (x86DataAdd(model->data, feature->data))
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
goto no_memory;
|
|
|
|
}
|
|
|
|
|
|
|
|
return model;
|
|
|
|
|
|
|
|
no_memory:
|
2010-02-04 18:19:08 +00:00
|
|
|
virReportOOMError();
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
error:
|
|
|
|
x86ModelFree(model);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-03-23 08:32:50 +00:00
|
|
|
static int
|
|
|
|
x86ModelSubtractCPU(struct x86_model *model,
|
|
|
|
const virCPUDefPtr cpu,
|
|
|
|
const struct x86_map *map)
|
|
|
|
{
|
|
|
|
const struct x86_model *cpu_model;
|
|
|
|
unsigned int i;
|
|
|
|
|
|
|
|
if (!(cpu_model = x86ModelFind(map, cpu->model))) {
|
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
|
|
|
_("Unknown CPU model %s"),
|
|
|
|
cpu->model);
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
x86DataSubtract(model->data, cpu_model->data);
|
2010-03-23 08:32:50 +00:00
|
|
|
|
|
|
|
for (i = 0; i < cpu->nfeatures; i++) {
|
|
|
|
const struct x86_feature *feature;
|
|
|
|
|
|
|
|
if (!(feature = x86FeatureFind(map, cpu->features[i].name))) {
|
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
|
|
|
_("Unknown CPU feature %s"),
|
|
|
|
cpu->features[i].name);
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
x86DataSubtract(model->data, feature->data);
|
2010-03-23 08:32:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
static enum compare_result
|
|
|
|
x86ModelCompare(const struct x86_model *model1,
|
|
|
|
const struct x86_model *model2)
|
|
|
|
{
|
|
|
|
enum compare_result result = EQUAL;
|
2010-06-30 11:08:57 +00:00
|
|
|
struct data_iterator iter1 = DATA_ITERATOR_INIT(model1->data);
|
|
|
|
struct data_iterator iter2 = DATA_ITERATOR_INIT(model2->data);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
struct cpuX86cpuid *cpuid1;
|
|
|
|
struct cpuX86cpuid *cpuid2;
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
while ((cpuid1 = x86DataCpuidNext(&iter1))) {
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
enum compare_result match = SUPERSET;
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if ((cpuid2 = x86DataCpuid(model2->data, cpuid1->function))) {
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
if (x86cpuidMatch(cpuid1, cpuid2))
|
|
|
|
continue;
|
|
|
|
else if (!x86cpuidMatchMasked(cpuid1, cpuid2))
|
|
|
|
match = SUBSET;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (result == EQUAL)
|
|
|
|
result = match;
|
|
|
|
else if (result != match)
|
|
|
|
return UNRELATED;
|
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
while ((cpuid2 = x86DataCpuidNext(&iter2))) {
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
enum compare_result match = SUBSET;
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if ((cpuid1 = x86DataCpuid(model1->data, cpuid2->function))) {
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
if (x86cpuidMatch(cpuid2, cpuid1))
|
|
|
|
continue;
|
|
|
|
else if (!x86cpuidMatchMasked(cpuid2, cpuid1))
|
|
|
|
match = SUPERSET;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (result == EQUAL)
|
|
|
|
result = match;
|
|
|
|
else if (result != match)
|
|
|
|
return UNRELATED;
|
|
|
|
}
|
|
|
|
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static int
|
|
|
|
x86ModelLoad(xmlXPathContextPtr ctxt,
|
2010-07-02 15:51:59 +00:00
|
|
|
struct x86_map *map)
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
{
|
|
|
|
xmlNodePtr *nodes = NULL;
|
2010-06-30 11:08:57 +00:00
|
|
|
struct x86_model *model;
|
2010-07-02 15:51:59 +00:00
|
|
|
char *vendor = NULL;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
int ret = 0;
|
|
|
|
int i;
|
|
|
|
int n;
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if (!(model = x86ModelNew()))
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
goto no_memory;
|
|
|
|
|
2010-02-04 21:52:34 +00:00
|
|
|
model->name = virXPathString("string(@name)", ctxt);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
if (model->name == NULL) {
|
2010-02-10 12:22:36 +00:00
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
"%s", _("Missing CPU model name"));
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
|
2010-02-04 21:52:34 +00:00
|
|
|
if (virXPathNode("./model", ctxt) != NULL) {
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
const struct x86_model *ancestor;
|
|
|
|
char *name;
|
|
|
|
|
2010-02-04 21:52:34 +00:00
|
|
|
name = virXPathString("string(./model/@name)", ctxt);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
if (name == NULL) {
|
2010-02-10 12:22:36 +00:00
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
_("Missing ancestor's name in CPU model %s"),
|
|
|
|
model->name);
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
|
|
|
|
if ((ancestor = x86ModelFind(map, name)) == NULL) {
|
2010-02-10 12:22:36 +00:00
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
_("Ancestor model %s not found for CPU model %s"),
|
|
|
|
name, model->name);
|
|
|
|
VIR_FREE(name);
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
|
|
|
|
VIR_FREE(name);
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
model->vendor = ancestor->vendor;
|
2011-01-10 23:56:36 +00:00
|
|
|
x86DataFree(model->data);
|
2010-06-30 11:08:57 +00:00
|
|
|
if (!(model->data = x86DataCopy(ancestor->data)))
|
|
|
|
goto no_memory;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
}
|
|
|
|
|
2010-10-13 09:42:19 +00:00
|
|
|
if (virXPathBoolean("boolean(./vendor)", ctxt)) {
|
|
|
|
vendor = virXPathString("string(./vendor/@name)", ctxt);
|
|
|
|
if (!vendor) {
|
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
|
|
|
_("Invalid vendor element in CPU model %s"),
|
|
|
|
model->name);
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
if (!(model->vendor = x86VendorFind(map, vendor))) {
|
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
|
|
|
_("Unknown vendor %s referenced by CPU model %s"),
|
|
|
|
vendor, model->name);
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2010-02-04 21:52:34 +00:00
|
|
|
n = virXPathNodeSet("./feature", ctxt, &nodes);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
if (n < 0)
|
|
|
|
goto ignore;
|
|
|
|
|
|
|
|
for (i = 0; i < n; i++) {
|
|
|
|
const struct x86_feature *feature;
|
|
|
|
char *name;
|
|
|
|
|
|
|
|
if ((name = virXMLPropString(nodes[i], "name")) == NULL) {
|
2010-02-10 12:22:36 +00:00
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
_("Missing feature name for CPU model %s"), model->name);
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
|
|
|
|
if ((feature = x86FeatureFind(map, name)) == NULL) {
|
2010-02-10 12:22:36 +00:00
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
_("Feature %s required by CPU model %s not found"),
|
|
|
|
name, model->name);
|
|
|
|
VIR_FREE(name);
|
|
|
|
goto ignore;
|
|
|
|
}
|
|
|
|
VIR_FREE(name);
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if (x86DataAdd(model->data, feature->data))
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
goto no_memory;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (map->models == NULL)
|
|
|
|
map->models = model;
|
|
|
|
else {
|
|
|
|
model->next = map->models;
|
|
|
|
map->models = model;
|
|
|
|
}
|
|
|
|
|
|
|
|
out:
|
2010-07-02 15:51:59 +00:00
|
|
|
VIR_FREE(vendor);
|
2009-12-21 18:12:45 +00:00
|
|
|
VIR_FREE(nodes);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
return ret;
|
|
|
|
|
|
|
|
no_memory:
|
2010-02-04 18:19:08 +00:00
|
|
|
virReportOOMError();
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
ret = -1;
|
|
|
|
|
|
|
|
ignore:
|
|
|
|
x86ModelFree(model);
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
x86MapFree(struct x86_map *map)
|
|
|
|
{
|
|
|
|
if (map == NULL)
|
|
|
|
return;
|
|
|
|
|
|
|
|
while (map->features != NULL) {
|
|
|
|
struct x86_feature *feature = map->features;
|
|
|
|
map->features = feature->next;
|
|
|
|
x86FeatureFree(feature);
|
|
|
|
}
|
|
|
|
|
|
|
|
while (map->models != NULL) {
|
|
|
|
struct x86_model *model = map->models;
|
|
|
|
map->models = model->next;
|
|
|
|
x86ModelFree(model);
|
|
|
|
}
|
|
|
|
|
2010-08-12 20:30:11 +00:00
|
|
|
while (map->vendors != NULL) {
|
|
|
|
struct x86_vendor *vendor = map->vendors;
|
|
|
|
map->vendors = vendor->next;
|
|
|
|
x86VendorFree(vendor);
|
|
|
|
}
|
|
|
|
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
VIR_FREE(map);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
static int
|
|
|
|
x86MapLoadCallback(enum cpuMapElement element,
|
|
|
|
xmlXPathContextPtr ctxt,
|
|
|
|
void *data)
|
|
|
|
{
|
|
|
|
struct x86_map *map = data;
|
|
|
|
|
|
|
|
switch (element) {
|
|
|
|
case CPU_MAP_ELEMENT_VENDOR:
|
|
|
|
return x86VendorLoad(ctxt, map);
|
|
|
|
case CPU_MAP_ELEMENT_FEATURE:
|
|
|
|
return x86FeatureLoad(ctxt, map);
|
|
|
|
case CPU_MAP_ELEMENT_MODEL:
|
|
|
|
return x86ModelLoad(ctxt, map);
|
|
|
|
case CPU_MAP_ELEMENT_LAST:
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
static struct x86_map *
|
|
|
|
x86LoadMap(void)
|
|
|
|
{
|
|
|
|
struct x86_map *map;
|
|
|
|
|
|
|
|
if (VIR_ALLOC(map) < 0) {
|
2010-02-04 18:19:08 +00:00
|
|
|
virReportOOMError();
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
if (cpuMapLoad("x86", x86MapLoadCallback, map) < 0)
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
goto error;
|
|
|
|
|
|
|
|
return map;
|
|
|
|
|
|
|
|
error:
|
|
|
|
x86MapFree(map);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static virCPUCompareResult
|
|
|
|
x86Compute(virCPUDefPtr host,
|
|
|
|
virCPUDefPtr cpu,
|
|
|
|
union cpuData **guest)
|
|
|
|
{
|
|
|
|
struct x86_map *map = NULL;
|
|
|
|
struct x86_model *host_model = NULL;
|
|
|
|
struct x86_model *cpu_force = NULL;
|
|
|
|
struct x86_model *cpu_require = NULL;
|
|
|
|
struct x86_model *cpu_optional = NULL;
|
|
|
|
struct x86_model *cpu_disable = NULL;
|
|
|
|
struct x86_model *cpu_forbid = NULL;
|
|
|
|
struct x86_model *diff = NULL;
|
|
|
|
struct x86_model *guest_model = NULL;
|
2010-06-30 11:08:57 +00:00
|
|
|
struct data_iterator iter;
|
|
|
|
const struct cpuX86cpuid *cpuid;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
virCPUCompareResult ret;
|
|
|
|
enum compare_result result;
|
2010-04-07 12:57:16 +00:00
|
|
|
unsigned int i;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
if (cpu->arch != NULL) {
|
|
|
|
bool found = false;
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_CARDINALITY(archs); i++) {
|
|
|
|
if (STREQ(archs[i], cpu->arch)) {
|
|
|
|
found = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2010-01-12 14:25:44 +00:00
|
|
|
if (!found) {
|
|
|
|
VIR_DEBUG("CPU arch %s does not match host arch", cpu->arch);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
return VIR_CPU_COMPARE_INCOMPATIBLE;
|
2010-01-12 14:25:44 +00:00
|
|
|
}
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
}
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
if (cpu->vendor &&
|
|
|
|
(!host->vendor || STRNEQ(cpu->vendor, host->vendor))) {
|
|
|
|
VIR_DEBUG("host CPU vendor does not match required CPU vendor %s",
|
|
|
|
cpu->vendor);
|
|
|
|
return VIR_CPU_COMPARE_INCOMPATIBLE;
|
|
|
|
}
|
|
|
|
|
2010-04-07 12:57:16 +00:00
|
|
|
if (!(map = x86LoadMap()) ||
|
2010-07-12 12:41:36 +00:00
|
|
|
!(host_model = x86ModelFromCPU(host, map, VIR_CPU_FEATURE_REQUIRE)) ||
|
2010-04-07 12:57:16 +00:00
|
|
|
!(cpu_force = x86ModelFromCPU(cpu, map, VIR_CPU_FEATURE_FORCE)) ||
|
|
|
|
!(cpu_require = x86ModelFromCPU(cpu, map, VIR_CPU_FEATURE_REQUIRE)) ||
|
|
|
|
!(cpu_optional = x86ModelFromCPU(cpu, map, VIR_CPU_FEATURE_OPTIONAL)) ||
|
|
|
|
!(cpu_disable = x86ModelFromCPU(cpu, map, VIR_CPU_FEATURE_DISABLE)) ||
|
|
|
|
!(cpu_forbid = x86ModelFromCPU(cpu, map, VIR_CPU_FEATURE_FORBID)))
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
goto error;
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
x86DataIteratorInit(&iter, cpu_forbid->data);
|
|
|
|
while ((cpuid = x86DataCpuidNext(&iter))) {
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
const struct cpuX86cpuid *cpuid2;
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
cpuid2 = x86DataCpuid(host_model->data, cpuid->function);
|
|
|
|
if (cpuid2 != NULL && x86cpuidMatchAny(cpuid2, cpuid)) {
|
2010-01-12 14:25:44 +00:00
|
|
|
VIR_DEBUG("Host CPU provides forbidden features in CPUID function 0x%x",
|
2010-06-30 11:08:57 +00:00
|
|
|
cpuid->function);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
ret = VIR_CPU_COMPARE_INCOMPATIBLE;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
x86DataSubtract(cpu_require->data, cpu_disable->data);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
result = x86ModelCompare(host_model, cpu_require);
|
|
|
|
if (result == SUBSET || result == UNRELATED) {
|
2011-05-09 09:24:09 +00:00
|
|
|
VIR_DEBUG("Host CPU does not provide all required features");
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
ret = VIR_CPU_COMPARE_INCOMPATIBLE;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = VIR_CPU_COMPARE_IDENTICAL;
|
|
|
|
|
2010-04-07 12:57:16 +00:00
|
|
|
if ((diff = x86ModelCopy(host_model)) == NULL)
|
|
|
|
goto no_memory;
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
x86DataSubtract(diff->data, cpu_optional->data);
|
|
|
|
x86DataSubtract(diff->data, cpu_require->data);
|
|
|
|
x86DataSubtract(diff->data, cpu_disable->data);
|
|
|
|
x86DataSubtract(diff->data, cpu_force->data);
|
2010-04-07 12:57:16 +00:00
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if (!x86DataIsEmpty(diff->data))
|
|
|
|
ret = VIR_CPU_COMPARE_SUPERSET;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
if (ret == VIR_CPU_COMPARE_SUPERSET
|
|
|
|
&& cpu->type == VIR_CPU_TYPE_GUEST
|
|
|
|
&& cpu->match == VIR_CPU_MATCH_STRICT) {
|
2011-05-09 09:24:09 +00:00
|
|
|
VIR_DEBUG("Host CPU does not strictly match guest CPU");
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
ret = VIR_CPU_COMPARE_INCOMPATIBLE;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (guest != NULL) {
|
|
|
|
if ((guest_model = x86ModelCopy(host_model)) == NULL)
|
|
|
|
goto no_memory;
|
|
|
|
|
|
|
|
if (cpu->type == VIR_CPU_TYPE_GUEST
|
|
|
|
&& cpu->match == VIR_CPU_MATCH_EXACT)
|
2010-06-30 11:08:57 +00:00
|
|
|
x86DataSubtract(guest_model->data, diff->data);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if (x86DataAdd(guest_model->data, cpu_force->data))
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
goto no_memory;
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
x86DataSubtract(guest_model->data, cpu_disable->data);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if ((*guest = x86DataCopy(guest_model->data)) == NULL)
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
goto no_memory;
|
|
|
|
}
|
|
|
|
|
|
|
|
out:
|
|
|
|
x86MapFree(map);
|
|
|
|
x86ModelFree(host_model);
|
|
|
|
x86ModelFree(diff);
|
|
|
|
x86ModelFree(cpu_force);
|
|
|
|
x86ModelFree(cpu_require);
|
|
|
|
x86ModelFree(cpu_optional);
|
|
|
|
x86ModelFree(cpu_disable);
|
|
|
|
x86ModelFree(cpu_forbid);
|
|
|
|
x86ModelFree(guest_model);
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
no_memory:
|
2010-02-04 18:19:08 +00:00
|
|
|
virReportOOMError();
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
error:
|
|
|
|
ret = VIR_CPU_COMPARE_ERROR;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static virCPUCompareResult
|
|
|
|
x86Compare(virCPUDefPtr host,
|
|
|
|
virCPUDefPtr cpu)
|
|
|
|
{
|
|
|
|
return x86Compute(host, cpu, NULL);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static virCPUCompareResult
|
|
|
|
x86GuestData(virCPUDefPtr host,
|
|
|
|
virCPUDefPtr guest,
|
|
|
|
union cpuData **data)
|
|
|
|
{
|
|
|
|
return x86Compute(host, guest, data);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static int
|
|
|
|
x86Decode(virCPUDefPtr cpu,
|
|
|
|
const union cpuData *data,
|
2010-02-11 09:47:43 +00:00
|
|
|
const char **models,
|
2010-04-15 10:06:13 +00:00
|
|
|
unsigned int nmodels,
|
|
|
|
const char *preferred)
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
{
|
|
|
|
int ret = -1;
|
|
|
|
struct x86_map *map;
|
|
|
|
const struct x86_model *candidate;
|
2010-01-15 15:58:59 +00:00
|
|
|
virCPUDefPtr cpuCandidate;
|
|
|
|
virCPUDefPtr cpuModel = NULL;
|
2010-04-14 15:41:32 +00:00
|
|
|
unsigned int i;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
if (data == NULL || (map = x86LoadMap()) == NULL)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
candidate = map->models;
|
|
|
|
while (candidate != NULL) {
|
2010-01-15 15:58:59 +00:00
|
|
|
bool allowed = (models == NULL);
|
|
|
|
|
|
|
|
for (i = 0; i < nmodels; i++) {
|
2010-02-28 12:34:06 +00:00
|
|
|
if (models && models[i] && STREQ(models[i], candidate->name)) {
|
2010-01-15 15:58:59 +00:00
|
|
|
allowed = true;
|
|
|
|
break;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
}
|
2010-01-15 15:58:59 +00:00
|
|
|
}
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
2010-01-15 15:58:59 +00:00
|
|
|
if (!allowed) {
|
2011-12-21 13:27:16 +00:00
|
|
|
if (preferred && STREQ(candidate->name, preferred)) {
|
|
|
|
if (cpu->fallback != VIR_CPU_FALLBACK_ALLOW) {
|
|
|
|
virCPUReportError(VIR_ERR_CONFIG_UNSUPPORTED,
|
|
|
|
_("CPU model %s is not supported by hypervisor"),
|
|
|
|
preferred);
|
|
|
|
goto out;
|
|
|
|
} else {
|
|
|
|
VIR_WARN("Preferred CPU model %s not allowed by"
|
|
|
|
" hypervisor; closest supported model will be"
|
|
|
|
" used", preferred);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
VIR_DEBUG("CPU model %s not allowed by hypervisor; ignoring",
|
|
|
|
candidate->name);
|
|
|
|
}
|
2010-01-15 15:58:59 +00:00
|
|
|
goto next;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
}
|
|
|
|
|
2010-01-15 15:58:59 +00:00
|
|
|
if (!(cpuCandidate = x86DataToCPU(data, candidate, map)))
|
|
|
|
goto out;
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
if (candidate->vendor && cpuCandidate->vendor &&
|
|
|
|
STRNEQ(candidate->vendor->name, cpuCandidate->vendor)) {
|
|
|
|
VIR_DEBUG("CPU vendor %s of model %s differs from %s; ignoring",
|
|
|
|
candidate->vendor->name, candidate->name,
|
|
|
|
cpuCandidate->vendor);
|
|
|
|
virCPUDefFree(cpuCandidate);
|
|
|
|
goto next;
|
|
|
|
}
|
|
|
|
|
2010-04-14 15:41:32 +00:00
|
|
|
if (cpu->type == VIR_CPU_TYPE_HOST) {
|
|
|
|
cpuCandidate->type = VIR_CPU_TYPE_HOST;
|
|
|
|
for (i = 0; i < cpuCandidate->nfeatures; i++) {
|
|
|
|
switch (cpuCandidate->features[i].policy) {
|
|
|
|
case VIR_CPU_FEATURE_DISABLE:
|
|
|
|
virCPUDefFree(cpuCandidate);
|
|
|
|
goto next;
|
|
|
|
default:
|
|
|
|
cpuCandidate->features[i].policy = -1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2010-04-15 10:06:13 +00:00
|
|
|
if (preferred && STREQ(cpuCandidate->model, preferred)) {
|
|
|
|
virCPUDefFree(cpuModel);
|
|
|
|
cpuModel = cpuCandidate;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2010-01-15 15:58:59 +00:00
|
|
|
if (cpuModel == NULL
|
|
|
|
|| cpuModel->nfeatures > cpuCandidate->nfeatures) {
|
|
|
|
virCPUDefFree(cpuModel);
|
|
|
|
cpuModel = cpuCandidate;
|
|
|
|
} else
|
|
|
|
virCPUDefFree(cpuCandidate);
|
|
|
|
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
next:
|
|
|
|
candidate = candidate->next;
|
|
|
|
}
|
|
|
|
|
2010-01-15 15:58:59 +00:00
|
|
|
if (cpuModel == NULL) {
|
2010-02-10 12:22:36 +00:00
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
"%s", _("Cannot find suitable CPU model for given data"));
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2010-01-15 15:58:59 +00:00
|
|
|
cpu->model = cpuModel->model;
|
2010-07-02 15:51:59 +00:00
|
|
|
cpu->vendor = cpuModel->vendor;
|
2010-01-15 15:58:59 +00:00
|
|
|
cpu->nfeatures = cpuModel->nfeatures;
|
|
|
|
cpu->features = cpuModel->features;
|
|
|
|
VIR_FREE(cpuModel);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
ret = 0;
|
|
|
|
|
|
|
|
out:
|
|
|
|
x86MapFree(map);
|
2010-01-15 15:58:59 +00:00
|
|
|
virCPUDefFree(cpuModel);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static union cpuData *
|
|
|
|
x86EncodePolicy(const virCPUDefPtr cpu,
|
|
|
|
const struct x86_map *map,
|
|
|
|
enum virCPUFeaturePolicy policy)
|
|
|
|
{
|
|
|
|
struct x86_model *model;
|
|
|
|
union cpuData *data = NULL;
|
|
|
|
|
|
|
|
if (!(model = x86ModelFromCPU(cpu, map, policy)))
|
|
|
|
return NULL;
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
data = model->data;
|
|
|
|
model->data = NULL;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
x86ModelFree(model);
|
|
|
|
|
|
|
|
return data;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static int
|
|
|
|
x86Encode(const virCPUDefPtr cpu,
|
|
|
|
union cpuData **forced,
|
|
|
|
union cpuData **required,
|
|
|
|
union cpuData **optional,
|
|
|
|
union cpuData **disabled,
|
2010-07-02 15:51:59 +00:00
|
|
|
union cpuData **forbidden,
|
|
|
|
union cpuData **vendor)
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
{
|
|
|
|
struct x86_map *map = NULL;
|
|
|
|
union cpuData *data_forced = NULL;
|
|
|
|
union cpuData *data_required = NULL;
|
|
|
|
union cpuData *data_optional = NULL;
|
|
|
|
union cpuData *data_disabled = NULL;
|
|
|
|
union cpuData *data_forbidden = NULL;
|
2010-07-02 15:51:59 +00:00
|
|
|
union cpuData *data_vendor = NULL;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
int ret = -1;
|
|
|
|
|
|
|
|
if ((map = x86LoadMap()) == NULL)
|
|
|
|
goto error;
|
|
|
|
|
|
|
|
if (forced) {
|
|
|
|
data_forced = x86EncodePolicy(cpu, map, VIR_CPU_FEATURE_FORCE);
|
|
|
|
if (!data_forced)
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (required) {
|
|
|
|
data_required = x86EncodePolicy(cpu, map, VIR_CPU_FEATURE_REQUIRE);
|
|
|
|
if (!data_required)
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (optional) {
|
|
|
|
data_optional = x86EncodePolicy(cpu, map, VIR_CPU_FEATURE_OPTIONAL);
|
|
|
|
if (!data_optional)
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (disabled) {
|
|
|
|
data_disabled = x86EncodePolicy(cpu, map, VIR_CPU_FEATURE_DISABLE);
|
|
|
|
if (!data_disabled)
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (forbidden) {
|
|
|
|
data_forbidden = x86EncodePolicy(cpu, map, VIR_CPU_FEATURE_FORBID);
|
|
|
|
if (!data_forbidden)
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
if (vendor) {
|
|
|
|
const struct x86_vendor *v = NULL;
|
|
|
|
|
|
|
|
if (cpu->vendor && !(v = x86VendorFind(map, cpu->vendor))) {
|
|
|
|
virCPUReportError(VIR_ERR_OPERATION_FAILED,
|
|
|
|
_("CPU vendor %s not found"), cpu->vendor);
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (v &&
|
|
|
|
(VIR_ALLOC(data_vendor) < 0 ||
|
|
|
|
x86DataAddCpuid(data_vendor, &v->cpuid) < 0)) {
|
|
|
|
virReportOOMError();
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
if (forced)
|
|
|
|
*forced = data_forced;
|
|
|
|
if (required)
|
|
|
|
*required = data_required;
|
|
|
|
if (optional)
|
|
|
|
*optional = data_optional;
|
|
|
|
if (disabled)
|
|
|
|
*disabled = data_disabled;
|
|
|
|
if (forbidden)
|
|
|
|
*forbidden = data_forbidden;
|
2010-07-02 15:51:59 +00:00
|
|
|
if (vendor)
|
|
|
|
*vendor = data_vendor;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
ret = 0;
|
|
|
|
|
|
|
|
cleanup:
|
|
|
|
x86MapFree(map);
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
error:
|
|
|
|
x86DataFree(data_forced);
|
|
|
|
x86DataFree(data_required);
|
|
|
|
x86DataFree(data_optional);
|
|
|
|
x86DataFree(data_disabled);
|
|
|
|
x86DataFree(data_forbidden);
|
2010-07-02 15:51:59 +00:00
|
|
|
x86DataFree(data_vendor);
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
goto cleanup;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
#if HAVE_CPUID
|
|
|
|
static inline void
|
|
|
|
cpuidCall(struct cpuX86cpuid *cpuid)
|
|
|
|
{
|
2010-03-09 18:22:22 +00:00
|
|
|
# if __x86_64__
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
asm("cpuid"
|
|
|
|
: "=a" (cpuid->eax),
|
|
|
|
"=b" (cpuid->ebx),
|
|
|
|
"=c" (cpuid->ecx),
|
|
|
|
"=d" (cpuid->edx)
|
|
|
|
: "a" (cpuid->function));
|
2010-03-09 18:22:22 +00:00
|
|
|
# else
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
/* we need to avoid direct use of ebx for CPUID output as it is used
|
|
|
|
* for global offset table on i386 with -fPIC
|
|
|
|
*/
|
|
|
|
asm("push %%ebx;"
|
|
|
|
"cpuid;"
|
|
|
|
"mov %%ebx, %1;"
|
|
|
|
"pop %%ebx;"
|
|
|
|
: "=a" (cpuid->eax),
|
|
|
|
"=r" (cpuid->ebx),
|
|
|
|
"=c" (cpuid->ecx),
|
|
|
|
"=d" (cpuid->edx)
|
|
|
|
: "a" (cpuid->function)
|
|
|
|
: "cc");
|
2010-03-09 18:22:22 +00:00
|
|
|
# endif
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static int
|
|
|
|
cpuidSet(uint32_t base, struct cpuX86cpuid **set)
|
|
|
|
{
|
|
|
|
uint32_t max;
|
|
|
|
uint32_t i;
|
|
|
|
struct cpuX86cpuid cpuid = { base, 0, 0, 0, 0 };
|
|
|
|
|
|
|
|
cpuidCall(&cpuid);
|
|
|
|
max = cpuid.eax - base;
|
|
|
|
|
|
|
|
if (VIR_ALLOC_N(*set, max + 1) < 0) {
|
2010-02-04 18:19:08 +00:00
|
|
|
virReportOOMError();
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i <= max; i++) {
|
|
|
|
cpuid.function = base | i;
|
|
|
|
cpuidCall(&cpuid);
|
|
|
|
(*set)[i] = cpuid;
|
|
|
|
}
|
|
|
|
|
|
|
|
return max + 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static union cpuData *
|
|
|
|
x86NodeData(void)
|
|
|
|
{
|
|
|
|
union cpuData *data;
|
2010-06-30 11:08:57 +00:00
|
|
|
int ret;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
if (VIR_ALLOC(data) < 0) {
|
2010-02-04 18:19:08 +00:00
|
|
|
virReportOOMError();
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if ((ret = cpuidSet(CPUX86_BASIC, &data->x86.basic)) < 0)
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
goto error;
|
2010-06-30 11:08:57 +00:00
|
|
|
data->x86.basic_len = ret;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if ((ret = cpuidSet(CPUX86_EXTENDED, &data->x86.extended)) < 0)
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
goto error;
|
2010-06-30 11:08:57 +00:00
|
|
|
data->x86.extended_len = ret;
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
|
|
|
|
return data;
|
|
|
|
|
|
|
|
error:
|
|
|
|
x86DataFree(data);
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
2010-01-27 13:33:20 +00:00
|
|
|
static virCPUDefPtr
|
|
|
|
x86Baseline(virCPUDefPtr *cpus,
|
|
|
|
unsigned int ncpus,
|
|
|
|
const char **models,
|
|
|
|
unsigned int nmodels)
|
|
|
|
{
|
|
|
|
struct x86_map *map = NULL;
|
|
|
|
struct x86_model *base_model = NULL;
|
|
|
|
virCPUDefPtr cpu = NULL;
|
|
|
|
unsigned int i;
|
2010-07-02 15:51:59 +00:00
|
|
|
const struct x86_vendor *vendor = NULL;
|
|
|
|
struct x86_model *model = NULL;
|
2010-10-13 10:26:22 +00:00
|
|
|
bool outputVendor = true;
|
2010-01-27 13:33:20 +00:00
|
|
|
|
|
|
|
if (!(map = x86LoadMap()))
|
|
|
|
goto error;
|
|
|
|
|
2010-07-12 12:41:36 +00:00
|
|
|
if (!(base_model = x86ModelFromCPU(cpus[0], map, VIR_CPU_FEATURE_REQUIRE)))
|
2010-01-27 13:33:20 +00:00
|
|
|
goto error;
|
|
|
|
|
|
|
|
if (VIR_ALLOC(cpu) < 0 ||
|
|
|
|
!(cpu->arch = strdup(cpus[0]->arch)))
|
|
|
|
goto no_memory;
|
2010-04-14 15:41:32 +00:00
|
|
|
cpu->type = VIR_CPU_TYPE_GUEST;
|
|
|
|
cpu->match = VIR_CPU_MATCH_EXACT;
|
2010-01-27 13:33:20 +00:00
|
|
|
|
2010-10-13 10:26:22 +00:00
|
|
|
if (!cpus[0]->vendor)
|
|
|
|
outputVendor = false;
|
|
|
|
else if (!(vendor = x86VendorFind(map, cpus[0]->vendor))) {
|
2010-07-02 15:51:59 +00:00
|
|
|
virCPUReportError(VIR_ERR_OPERATION_FAILED,
|
|
|
|
_("Unknown CPU vendor %s"), cpus[0]->vendor);
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
2010-01-27 13:33:20 +00:00
|
|
|
for (i = 1; i < ncpus; i++) {
|
2010-07-02 15:51:59 +00:00
|
|
|
const char *vn = NULL;
|
|
|
|
|
2010-07-12 12:41:36 +00:00
|
|
|
if (!(model = x86ModelFromCPU(cpus[i], map, VIR_CPU_FEATURE_REQUIRE)))
|
2010-01-27 13:33:20 +00:00
|
|
|
goto error;
|
|
|
|
|
2010-07-02 15:51:59 +00:00
|
|
|
if (cpus[i]->vendor && model->vendor &&
|
|
|
|
STRNEQ(cpus[i]->vendor, model->vendor->name)) {
|
|
|
|
virCPUReportError(VIR_ERR_OPERATION_FAILED,
|
|
|
|
_("CPU vendor %s of model %s differs from vendor %s"),
|
|
|
|
model->vendor->name, model->name, cpus[i]->vendor);
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (cpus[i]->vendor)
|
|
|
|
vn = cpus[i]->vendor;
|
2010-10-13 10:26:22 +00:00
|
|
|
else {
|
|
|
|
outputVendor = false;
|
|
|
|
if (model->vendor)
|
|
|
|
vn = model->vendor->name;
|
|
|
|
}
|
2010-07-02 15:51:59 +00:00
|
|
|
|
|
|
|
if (vn) {
|
|
|
|
if (!vendor) {
|
|
|
|
if (!(vendor = x86VendorFind(map, vn))) {
|
|
|
|
virCPUReportError(VIR_ERR_OPERATION_FAILED,
|
|
|
|
_("Unknown CPU vendor %s"), vn);
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
} else if (STRNEQ(vendor->name, vn)) {
|
|
|
|
virCPUReportError(VIR_ERR_OPERATION_FAILED,
|
|
|
|
"%s", _("CPU vendors do not match"));
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
x86DataIntersect(base_model->data, model->data);
|
2010-01-27 13:33:20 +00:00
|
|
|
x86ModelFree(model);
|
2010-07-02 15:51:59 +00:00
|
|
|
model = NULL;
|
2010-01-27 13:33:20 +00:00
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if (x86DataIsEmpty(base_model->data)) {
|
2010-07-02 15:51:40 +00:00
|
|
|
virCPUReportError(VIR_ERR_OPERATION_FAILED,
|
|
|
|
"%s", _("CPUs are incompatible"));
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if (vendor && x86DataAddCpuid(base_model->data, &vendor->cpuid) < 0)
|
2010-07-02 15:51:59 +00:00
|
|
|
goto no_memory;
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if (x86Decode(cpu, base_model->data, models, nmodels, NULL) < 0)
|
2010-01-27 13:33:20 +00:00
|
|
|
goto error;
|
|
|
|
|
2010-10-13 10:26:22 +00:00
|
|
|
if (!outputVendor)
|
|
|
|
VIR_FREE(cpu->vendor);
|
|
|
|
|
2010-07-02 15:51:20 +00:00
|
|
|
VIR_FREE(cpu->arch);
|
|
|
|
|
2010-01-27 13:33:20 +00:00
|
|
|
cleanup:
|
|
|
|
x86ModelFree(base_model);
|
|
|
|
x86MapFree(map);
|
|
|
|
|
|
|
|
return cpu;
|
|
|
|
|
|
|
|
no_memory:
|
|
|
|
virReportOOMError();
|
|
|
|
error:
|
2010-07-02 15:51:59 +00:00
|
|
|
x86ModelFree(model);
|
2010-01-27 13:33:20 +00:00
|
|
|
virCPUDefFree(cpu);
|
|
|
|
cpu = NULL;
|
|
|
|
goto cleanup;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2010-03-23 08:32:50 +00:00
|
|
|
static int
|
2011-12-19 14:41:16 +00:00
|
|
|
x86UpdateCustom(virCPUDefPtr guest,
|
|
|
|
const virCPUDefPtr host)
|
2010-03-23 08:32:50 +00:00
|
|
|
{
|
|
|
|
int ret = -1;
|
|
|
|
unsigned int i;
|
|
|
|
struct x86_map *map;
|
|
|
|
struct x86_model *host_model = NULL;
|
|
|
|
|
|
|
|
if (!(map = x86LoadMap()) ||
|
2010-07-12 12:41:36 +00:00
|
|
|
!(host_model = x86ModelFromCPU(host, map, VIR_CPU_FEATURE_REQUIRE)))
|
2010-03-23 08:32:50 +00:00
|
|
|
goto cleanup;
|
|
|
|
|
|
|
|
for (i = 0; i < guest->nfeatures; i++) {
|
|
|
|
if (guest->features[i].policy == VIR_CPU_FEATURE_OPTIONAL) {
|
|
|
|
const struct x86_feature *feature;
|
|
|
|
if (!(feature = x86FeatureFind(map, guest->features[i].name))) {
|
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
|
|
|
_("Unknown CPU feature %s"),
|
|
|
|
guest->features[i].name);
|
|
|
|
goto cleanup;
|
|
|
|
}
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
if (x86DataIsSubset(host_model->data, feature->data))
|
2010-03-23 08:32:50 +00:00
|
|
|
guest->features[i].policy = VIR_CPU_FEATURE_REQUIRE;
|
|
|
|
else
|
|
|
|
guest->features[i].policy = VIR_CPU_FEATURE_DISABLE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (guest->match == VIR_CPU_MATCH_MINIMUM) {
|
|
|
|
guest->match = VIR_CPU_MATCH_EXACT;
|
|
|
|
if (x86ModelSubtractCPU(host_model, guest, map)
|
2010-06-30 11:08:57 +00:00
|
|
|
|| x86DataToCPUFeatures(guest, VIR_CPU_FEATURE_REQUIRE,
|
|
|
|
host_model->data, map))
|
2010-03-23 08:32:50 +00:00
|
|
|
goto cleanup;
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = 0;
|
|
|
|
|
|
|
|
cleanup:
|
|
|
|
x86MapFree(map);
|
|
|
|
x86ModelFree(host_model);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2011-12-19 14:41:16 +00:00
|
|
|
static int
|
|
|
|
x86Update(virCPUDefPtr guest,
|
|
|
|
const virCPUDefPtr host)
|
|
|
|
{
|
|
|
|
switch ((enum virCPUMode) guest->mode) {
|
|
|
|
case VIR_CPU_MODE_CUSTOM:
|
|
|
|
return x86UpdateCustom(guest, host);
|
|
|
|
|
|
|
|
case VIR_CPU_MODE_HOST_MODEL:
|
|
|
|
case VIR_CPU_MODE_HOST_PASSTHROUGH:
|
|
|
|
if (guest->mode == VIR_CPU_MODE_HOST_MODEL)
|
|
|
|
guest->match = VIR_CPU_MATCH_EXACT;
|
|
|
|
else
|
|
|
|
guest->match = VIR_CPU_MATCH_MINIMUM;
|
|
|
|
virCPUDefFreeModel(guest);
|
|
|
|
return virCPUDefCopyModel(guest, host, true);
|
|
|
|
|
|
|
|
case VIR_CPU_MODE_LAST:
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
virCPUReportError(VIR_ERR_INTERNAL_ERROR,
|
|
|
|
_("Unexpected CPU mode: %d"), guest->mode);
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
2010-09-22 11:47:09 +00:00
|
|
|
static int x86HasFeature(const union cpuData *data,
|
|
|
|
const char *name)
|
|
|
|
{
|
|
|
|
struct x86_map *map;
|
|
|
|
struct x86_feature *feature;
|
|
|
|
int ret = -1;
|
|
|
|
|
|
|
|
if (!(map = x86LoadMap()))
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
if (!(feature = x86FeatureFind(map, name)))
|
|
|
|
goto cleanup;
|
|
|
|
|
2010-06-30 11:08:57 +00:00
|
|
|
ret = x86DataIsSubset(data, feature->data) ? 1 : 0;
|
2010-09-22 11:47:09 +00:00
|
|
|
|
|
|
|
cleanup:
|
|
|
|
x86MapFree(map);
|
|
|
|
return ret;
|
|
|
|
}
|
2010-03-23 08:32:50 +00:00
|
|
|
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
struct cpuArchDriver cpuDriverX86 = {
|
|
|
|
.name = "x86",
|
|
|
|
.arch = archs,
|
|
|
|
.narch = ARRAY_CARDINALITY(archs),
|
|
|
|
.compare = x86Compare,
|
|
|
|
.decode = x86Decode,
|
|
|
|
.encode = x86Encode,
|
|
|
|
.free = x86DataFree,
|
|
|
|
#if HAVE_CPUID
|
|
|
|
.nodeData = x86NodeData,
|
|
|
|
#else
|
|
|
|
.nodeData = NULL,
|
|
|
|
#endif
|
2010-01-25 17:27:56 +00:00
|
|
|
.guestData = x86GuestData,
|
2010-01-27 13:33:20 +00:00
|
|
|
.baseline = x86Baseline,
|
2010-03-23 08:32:50 +00:00
|
|
|
.update = x86Update,
|
2010-09-22 11:47:09 +00:00
|
|
|
.hasFeature = x86HasFeature,
|
Adds CPU selection infrastructure
Each driver supporting CPU selection must fill in host CPU capabilities.
When filling them, drivers for hypervisors running on the same node as
libvirtd can use cpuNodeData() to obtain raw CPU data. Other drivers,
such as VMware, need to implement their own way of getting such data.
Raw data can be decoded into virCPUDefPtr using cpuDecode() function.
When implementing virConnectCompareCPU(), a hypervisor driver can just
call cpuCompareXML() function with host CPU capabilities.
For each guest for which a driver supports selecting CPU models, it must
set the appropriate feature in guest's capabilities:
virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0)
Actions needed when a domain is being created depend on whether the
hypervisor understands raw CPU data (currently CPUID for i686, x86_64
architectures) or symbolic names has to be used.
Typical use by hypervisors which prefer CPUID (such as VMware and Xen):
- convert guest CPU configuration from domain's XML into a set of raw
data structures each representing one of the feature policies:
cpuEncode(conn, architecture, guest_cpu_config,
&forced_data, &required_data, &optional_data,
&disabled_data, &forbidden_data)
- create a mask or whatever the hypervisor expects to see and pass it
to the hypervisor
Typical use by hypervisors with symbolic model names (such as QEMU):
- get raw CPU data for a computed guest CPU:
cpuGuestData(conn, host_cpu, guest_cpu_config, &data)
- decode raw data into virCPUDefPtr with a possible restriction on
allowed model names:
cpuDecode(conn, guest, data, n_allowed_models, allowed_models)
- pass guest->model and guest->features to the hypervisor
* src/cpu/cpu.c src/cpu/cpu.h src/cpu/cpu_generic.c
src/cpu/cpu_generic.h src/cpu/cpu_map.c src/cpu/cpu_map.h
src/cpu/cpu_x86.c src/cpu/cpu_x86.h src/cpu/cpu_x86_data.h
* configure.in: check for CPUID instruction
* src/Makefile.am: glue the new files in
* src/libvirt_private.syms: add new private symbols
* po/POTFILES.in: add new cpu files containing translatable strings
2009-12-18 15:02:11 +00:00
|
|
|
};
|