libvirt/src/cpu/cpu_x86.c

1919 lines
47 KiB
C
Raw Normal View History

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
*
* Copyright (C) 2009-2011, 2013 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, see
* <http://www.gnu.org/licenses/>.
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
*
* Authors:
* Jiri Denemark <jdenemar@redhat.com>
*/
#include <config.h>
#include <stdint.h>
2012-12-12 17:59:27 +00:00
#include "virlog.h"
2012-12-12 18:06:53 +00:00
#include "viralloc.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"
#include "virbuffer.h"
#include "virendian.h"
#include "virstring.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
#define VIR_FROM_THIS VIR_FROM_CPU
#define VENDOR_STRING_LENGTH 12
static const struct cpuX86cpuid cpuidNull = { 0, 0, 0, 0, 0 };
static const virArch archs[] = { VIR_ARCH_I686, VIR_ARCH_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
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;
struct cpuX86Data *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;
const struct x86_vendor *vendor;
struct cpuX86Data *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 {
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
};
struct data_iterator {
struct cpuX86Data *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
#define DATA_ITERATOR_INIT(data) \
{ data, -1, false }
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);
}
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);
}
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;
}
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;
}
static void
x86cpuidAndBits(struct cpuX86cpuid *cpuid,
const struct cpuX86cpuid *mask)
{
cpuid->eax &= mask->eax;
cpuid->ebx &= mask->ebx;
cpuid->ecx &= mask->ecx;
cpuid->edx &= mask->edx;
}
/* skips all zero CPUID leafs */
static struct cpuX86cpuid *
x86DataCpuidNext(struct data_iterator *iterator)
{
struct cpuX86cpuid *ret;
struct cpuX86Data *data = iterator->data;
if (!data)
return NULL;
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 struct cpuX86Data *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
uint32_t function)
{
struct cpuX86cpuid *cpuids;
int len;
size_t 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 (function < CPUX86_EXTENDED) {
cpuids = data->basic;
len = data->basic_len;
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
i = function;
}
else {
cpuids = data->extended;
len = data->extended_len;
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
i = function - CPUX86_EXTENDED;
}
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(struct cpuX86Data *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
{
if (data == NULL)
return;
VIR_FREE(data->basic);
VIR_FREE(data->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
VIR_FREE(data);
}
2012-12-18 18:44:23 +00:00
static virCPUDataPtr
2013-07-16 12:39:40 +00:00
x86MakeCPUData(virArch arch, struct cpuX86Data **data)
{
2012-12-18 18:44:23 +00:00
virCPUDataPtr cpuData;
if (VIR_ALLOC(cpuData) < 0)
return NULL;
2013-07-16 12:39:40 +00:00
cpuData->arch = arch;
cpuData->data.x86 = *data;
*data = NULL;
return cpuData;
}
static void
2012-12-18 18:44:23 +00:00
x86FreeCPUData(virCPUDataPtr data)
{
if (!data)
return;
2013-07-16 12:39:40 +00:00
x86DataFree(data->data.x86);
VIR_FREE(data);
}
static struct cpuX86Data *
x86DataCopy(const struct cpuX86Data *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 cpuX86Data *copy = NULL;
size_t 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 (VIR_ALLOC(copy) < 0
|| VIR_ALLOC_N(copy->basic, data->basic_len) < 0
|| VIR_ALLOC_N(copy->extended, data->extended_len) < 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
x86DataFree(copy);
return NULL;
}
copy->basic_len = data->basic_len;
for (i = 0; i < data->basic_len; i++)
copy->basic[i] = data->basic[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
copy->extended_len = data->extended_len;
for (i = 0; i < data->extended_len; i++)
copy->extended[i] = data->extended[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
return copy;
}
static int
x86DataExpand(struct cpuX86Data *data,
int basic_by,
int extended_by)
{
size_t i;
if (basic_by > 0) {
size_t len = data->basic_len;
if (VIR_EXPAND_N(data->basic, data->basic_len, basic_by) < 0)
return -1;
for (i = 0; i < basic_by; i++)
data->basic[len + i].function = len + i;
}
if (extended_by > 0) {
size_t len = data->extended_len;
if (VIR_EXPAND_N(data->extended, data->extended_len, extended_by) < 0)
return -1;
for (i = 0; i < extended_by; i++)
data->extended[len + i].function = len + i + CPUX86_EXTENDED;
}
return 0;
}
static int
x86DataAddCpuid(struct cpuX86Data *data,
const struct cpuX86cpuid *cpuid)
{
unsigned int basic_by = 0;
unsigned int extended_by = 0;
struct cpuX86cpuid **cpuids;
unsigned int pos;
if (cpuid->function < CPUX86_EXTENDED) {
pos = cpuid->function;
basic_by = pos + 1 - data->basic_len;
cpuids = &data->basic;
} else {
pos = cpuid->function - CPUX86_EXTENDED;
extended_by = pos + 1 - data->extended_len;
cpuids = &data->extended;
}
if (x86DataExpand(data, basic_by, extended_by) < 0)
return -1;
x86cpuidSetBits((*cpuids) + pos, cpuid);
return 0;
}
static int
x86DataAdd(struct cpuX86Data *data1,
const struct cpuX86Data *data2)
{
size_t i;
if (x86DataExpand(data1,
data2->basic_len - data1->basic_len,
data2->extended_len - data1->extended_len) < 0)
return -1;
for (i = 0; i < data2->basic_len; i++) {
x86cpuidSetBits(data1->basic + i,
data2->basic + i);
}
for (i = 0; i < data2->extended_len; i++) {
x86cpuidSetBits(data1->extended + i,
data2->extended + i);
}
return 0;
}
static void
x86DataSubtract(struct cpuX86Data *data1,
const struct cpuX86Data *data2)
{
size_t i;
unsigned int len;
len = MIN(data1->basic_len, data2->basic_len);
for (i = 0; i < len; i++) {
x86cpuidClearBits(data1->basic + i,
data2->basic + i);
}
len = MIN(data1->extended_len, data2->extended_len);
for (i = 0; i < len; i++) {
x86cpuidClearBits(data1->extended + i,
data2->extended + i);
}
}
static void
x86DataIntersect(struct cpuX86Data *data1,
const struct cpuX86Data *data2)
{
struct data_iterator iter = DATA_ITERATOR_INIT(data1);
struct cpuX86cpuid *cpuid1;
struct cpuX86cpuid *cpuid2;
while ((cpuid1 = x86DataCpuidNext(&iter))) {
cpuid2 = x86DataCpuid(data2, cpuid1->function);
if (cpuid2)
x86cpuidAndBits(cpuid1, cpuid2);
else
x86cpuidClearBits(cpuid1, cpuid1);
}
}
static bool
x86DataIsEmpty(struct cpuX86Data *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 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
return x86DataCpuidNext(&iter) == 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
static bool
x86DataIsSubset(const struct cpuX86Data *data,
const struct cpuX86Data *subset)
{
struct data_iterator iter = DATA_ITERATOR_INIT((struct cpuX86Data *)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
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
}
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
}
/* also removes all detected features from data */
static int
x86DataToCPUFeatures(virCPUDefPtr cpu,
int policy,
struct cpuX86Data *data,
const struct x86_map *map)
{
const struct x86_feature *feature = map->features;
while (feature != NULL) {
if (x86DataIsSubset(data, feature->data)) {
x86DataSubtract(data, feature->data);
if (virCPUDefAddFeature(cpu, feature->name, policy) < 0)
return -1;
}
feature = feature->next;
}
return 0;
}
/* also removes bits corresponding to vendor string from data */
static const struct x86_vendor *
x86DataToVendor(struct cpuX86Data *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;
}
static virCPUDefPtr
x86DataToCPU(const struct cpuX86Data *data,
const struct x86_model *model,
const struct x86_map *map)
{
virCPUDefPtr cpu;
struct cpuX86Data *copy = NULL;
struct cpuX86Data *modelData = NULL;
const struct x86_vendor *vendor;
if (VIR_ALLOC(cpu) < 0 ||
VIR_STRDUP(cpu->model, model->name) < 0 ||
!(copy = x86DataCopy(data)) ||
!(modelData = x86DataCopy(model->data)))
goto error;
if ((vendor = x86DataToVendor(copy, map)) &&
VIR_STRDUP(cpu->vendor, vendor->name) < 0)
goto error;
x86DataSubtract(copy, modelData);
x86DataSubtract(modelData, data);
/* because feature policy is ignored for host CPU */
cpu->type = VIR_CPU_TYPE_GUEST;
if (x86DataToCPUFeatures(cpu, VIR_CPU_FEATURE_REQUIRE, copy, map) ||
x86DataToCPUFeatures(cpu, VIR_CPU_FEATURE_DISABLE, modelData, map))
goto error;
cleanup:
x86DataFree(modelData);
x86DataFree(copy);
return cpu;
error:
virCPUDefFree(cpu);
cpu = NULL;
goto cleanup;
}
static void
x86VendorFree(struct x86_vendor *vendor)
{
if (!vendor)
return;
VIR_FREE(vendor->name);
VIR_FREE(vendor);
}
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 error;
vendor->name = virXPathString("string(@name)", ctxt);
if (!vendor->name) {
virReportError(VIR_ERR_INTERNAL_ERROR,
"%s", _("Missing CPU vendor name"));
goto ignore;
}
if (x86VendorFind(map, vendor->name)) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("CPU vendor %s already defined"), vendor->name);
goto ignore;
}
string = virXPathString("string(@string)", ctxt);
if (!string) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Missing vendor string for CPU vendor %s"), vendor->name);
goto ignore;
}
if (strlen(string) != VENDOR_STRING_LENGTH) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Invalid CPU vendor string '%s'"), string);
goto ignore;
}
vendor->cpuid.function = 0;
vendor->cpuid.ebx = virReadBufInt32LE(string);
vendor->cpuid.edx = virReadBufInt32LE(string + 4);
vendor->cpuid.ecx = virReadBufInt32LE(string + 8);
if (!map->vendors) {
map->vendors = vendor;
} else {
vendor->next = map->vendors;
map->vendors = vendor;
}
out:
VIR_FREE(string);
return ret;
error:
ret = -1;
ignore:
x86VendorFree(vendor);
goto out;
}
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);
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 char *
x86FeatureNames(const struct x86_map *map,
const char *separator,
struct cpuX86Data *data)
{
virBuffer ret = VIR_BUFFER_INITIALIZER;
bool first = true;
struct x86_feature *next_feature = map->features;
virBufferAdd(&ret, "", 0);
while (next_feature) {
if (x86DataIsSubset(data, next_feature->data)) {
if (!first)
virBufferAdd(&ret, separator, -1);
else
first = false;
virBufferAdd(&ret, next_feature->name, -1);
}
next_feature = next_feature->next;
}
return virBufferContentAndReset(&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 int
x86FeatureLoad(xmlXPathContextPtr ctxt,
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;
xmlNodePtr ctxt_node = ctxt->node;
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;
size_t 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
int n;
if (!(feature = x86FeatureNew()))
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
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) {
virReportError(VIR_ERR_INTERNAL_ERROR,
"%s", _("Missing CPU feature name"));
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 ignore;
}
if (x86FeatureFind(map, feature->name)) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("CPU feature %s already defined"), feature->name);
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 ignore;
}
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++) {
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;
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) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Invalid cpuid[%zu] in %s feature"), i, feature->name);
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 ignore;
}
cpuid.function = fun;
cpuid.eax = eax;
cpuid.ebx = ebx;
cpuid.ecx = ecx;
cpuid.edx = edx;
if (x86DataAddCpuid(feature->data, &cpuid))
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 (map->features == NULL)
map->features = feature;
else {
feature->next = map->features;
map->features = feature;
}
out:
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;
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
ret = -1;
ignore:
x86FeatureFree(feature);
goto out;
}
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);
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 ||
VIR_STRDUP(copy->name, model->name) < 0 ||
!(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;
}
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;
size_t 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 (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) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Unknown CPU model %s"), cpu->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
goto error;
}
if ((model = x86ModelCopy(model)) == NULL)
goto error;
} else if (!(model = x86ModelNew())) {
goto error;
} else if (cpu->type == VIR_CPU_TYPE_HOST) {
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
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) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Unknown CPU feature %s"), cpu->features[i].name);
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;
}
if (x86DataAdd(model->data, feature->data))
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
}
return model;
error:
x86ModelFree(model);
return NULL;
}
static int
x86ModelSubtractCPU(struct x86_model *model,
const virCPUDefPtr cpu,
const struct x86_map *map)
{
const struct x86_model *cpu_model;
size_t i;
if (!(cpu_model = x86ModelFind(map, cpu->model))) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Unknown CPU model %s"),
cpu->model);
return -1;
}
x86DataSubtract(model->data, cpu_model->data);
for (i = 0; i < cpu->nfeatures; i++) {
const struct x86_feature *feature;
if (!(feature = x86FeatureFind(map, cpu->features[i].name))) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Unknown CPU feature %s"),
cpu->features[i].name);
return -1;
}
x86DataSubtract(model->data, feature->data);
}
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;
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;
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;
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;
}
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;
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,
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;
struct x86_model *model;
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;
size_t 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
int n;
if (!(model = x86ModelNew()))
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
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) {
virReportError(VIR_ERR_INTERNAL_ERROR,
"%s", _("Missing CPU model name"));
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 ignore;
}
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;
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) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Missing ancestor's name in CPU model %s"),
model->name);
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 ignore;
}
if ((ancestor = x86ModelFind(map, name)) == NULL) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Ancestor model %s not found for CPU model %s"),
name, model->name);
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(name);
goto ignore;
}
VIR_FREE(name);
model->vendor = ancestor->vendor;
x86DataFree(model->data);
if (!(model->data = x86DataCopy(ancestor->data)))
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 (virXPathBoolean("boolean(./vendor)", ctxt)) {
vendor = virXPathString("string(./vendor/@name)", ctxt);
if (!vendor) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Invalid vendor element in CPU model %s"),
model->name);
goto ignore;
}
if (!(model->vendor = x86VendorFind(map, vendor))) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Unknown vendor %s referenced by CPU model %s"),
vendor, model->name);
goto ignore;
}
}
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) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Missing feature name for CPU model %s"), model->name);
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 ignore;
}
if ((feature = x86FeatureFind(map, name)) == NULL) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Feature %s required by CPU model %s not found"),
name, model->name);
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(name);
goto ignore;
}
VIR_FREE(name);
if (x86DataAdd(model->data, feature->data))
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 (map->models == NULL)
map->models = model;
else {
model->next = map->models;
map->models = model;
}
out:
VIR_FREE(vendor);
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;
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
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);
}
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);
}
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)
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;
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;
}
/* A helper macro to exit the cpu computation function without writing
* redundant code:
* MSG: error message
* CPU_DEF: a struct cpuX86Data pointer with flags that are conflicting
* RET: return code to set
*
* This macro generates the error string outputs it into logs.
*/
#define virX86CpuIncompatible(MSG, CPU_DEF) \
do { \
char *flagsStr = NULL; \
if (!(flagsStr = x86FeatureNames(map, ", ", (CPU_DEF)))) { \
virReportOOMError(); \
goto error; \
} \
if (message && \
virAsprintf(message, "%s: %s", _(MSG), flagsStr) < 0) { \
VIR_FREE(flagsStr); \
goto error; \
} \
VIR_DEBUG("%s: %s", MSG, flagsStr); \
VIR_FREE(flagsStr); \
ret = VIR_CPU_COMPARE_INCOMPATIBLE; \
} while (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 virCPUCompareResult
x86Compute(virCPUDefPtr host,
virCPUDefPtr cpu,
2012-12-18 18:44:23 +00:00
virCPUDataPtr *guest,
char **message)
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;
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;
virCPUCompareResult ret;
enum compare_result result;
2013-07-16 12:39:40 +00:00
virArch arch;
size_t 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 != VIR_ARCH_NONE) {
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
bool found = false;
for (i = 0; i < ARRAY_CARDINALITY(archs); i++) {
if (archs[i] == 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
found = true;
break;
}
}
if (!found) {
VIR_DEBUG("CPU arch %s does not match host arch",
virArchToString(cpu->arch));
if (message &&
virAsprintf(message,
_("CPU arch %s does not match host arch"),
virArchToString(cpu->arch)) < 0)
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
return VIR_CPU_COMPARE_INCOMPATIBLE;
}
2013-07-16 12:39:40 +00:00
arch = cpu->arch;
} else {
arch = host->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
}
if (cpu->vendor &&
(!host->vendor || STRNEQ(cpu->vendor, host->vendor))) {
VIR_DEBUG("host CPU vendor does not match required CPU vendor %s",
cpu->vendor);
if (message &&
virAsprintf(message,
_("host CPU vendor does not match required "
"CPU vendor %s"),
cpu->vendor) < 0)
goto error;
return VIR_CPU_COMPARE_INCOMPATIBLE;
}
if (!(map = x86LoadMap()) ||
!(host_model = x86ModelFromCPU(host, map, VIR_CPU_FEATURE_REQUIRE)) ||
!(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;
x86DataIntersect(cpu_forbid->data, host_model->data);
if (!x86DataIsEmpty(cpu_forbid->data)) {
virX86CpuIncompatible(N_("Host CPU provides forbidden features"),
cpu_forbid->data);
goto out;
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
}
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) {
x86DataSubtract(cpu_require->data, host_model->data);
virX86CpuIncompatible(N_("Host CPU does not provide required "
"features"),
cpu_require->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 out;
}
ret = VIR_CPU_COMPARE_IDENTICAL;
if ((diff = x86ModelCopy(host_model)) == NULL)
goto error;
x86DataSubtract(diff->data, cpu_optional->data);
x86DataSubtract(diff->data, cpu_require->data);
x86DataSubtract(diff->data, cpu_disable->data);
x86DataSubtract(diff->data, cpu_force->data);
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) {
virX86CpuIncompatible(N_("Host CPU does not strictly match guest CPU: "
"Extra features"),
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
goto out;
}
if (guest != NULL) {
struct cpuX86Data *guestData;
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 ((guest_model = x86ModelCopy(host_model)) == NULL)
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 (cpu->type == VIR_CPU_TYPE_GUEST
&& cpu->match == VIR_CPU_MATCH_EXACT)
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
if (x86DataAdd(guest_model->data, cpu_force->data))
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
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
if (!(guestData = x86DataCopy(guest_model->data)) ||
2013-07-16 12:39:40 +00:00
!(*guest = x86MakeCPUData(arch, &guestData))) {
x86DataFree(guestData);
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
}
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;
error:
ret = VIR_CPU_COMPARE_ERROR;
goto out;
}
#undef virX86CpuIncompatible
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 virCPUCompareResult
x86Compare(virCPUDefPtr host,
virCPUDefPtr cpu)
{
return x86Compute(host, cpu, NULL, 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
}
static virCPUCompareResult
x86GuestData(virCPUDefPtr host,
virCPUDefPtr guest,
2012-12-18 18:44:23 +00:00
virCPUDataPtr *data,
char **message)
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 x86Compute(host, guest, data, message);
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
x86Decode(virCPUDefPtr cpu,
const struct cpuX86Data *data,
const char **models,
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;
virCPUDefPtr cpuCandidate;
virCPUDefPtr cpuModel = NULL;
size_t 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) {
if (!cpuModelIsAllowed(candidate->name, models, nmodels)) {
if (preferred && STREQ(candidate->name, preferred)) {
if (cpu->fallback != VIR_CPU_FALLBACK_ALLOW) {
virReportError(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);
}
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
}
if (!(cpuCandidate = x86DataToCPU(data, candidate, map)))
goto out;
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;
}
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;
}
}
}
if (preferred && STREQ(cpuCandidate->model, preferred)) {
virCPUDefFree(cpuModel);
cpuModel = cpuCandidate;
break;
}
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;
}
if (cpuModel == NULL) {
virReportError(VIR_ERR_INTERNAL_ERROR,
"%s", _("Cannot find suitable CPU model for given 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 out;
}
cpu->model = cpuModel->model;
cpu->vendor = cpuModel->vendor;
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);
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 int
x86DecodeCPUData(virCPUDefPtr cpu,
2012-12-18 18:44:23 +00:00
const virCPUDataPtr data,
const char **models,
unsigned int nmodels,
const char *preferred)
{
2013-07-16 12:39:40 +00:00
return x86Decode(cpu, data->data.x86, models, nmodels, 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
static struct cpuX86Data *
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
x86EncodePolicy(const virCPUDefPtr cpu,
const struct x86_map *map,
enum virCPUFeaturePolicy policy)
{
struct x86_model *model;
struct cpuX86Data *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
if (!(model = x86ModelFromCPU(cpu, map, policy)))
return NULL;
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
2013-07-16 12:39:40 +00:00
x86Encode(virArch arch,
const virCPUDefPtr cpu,
2012-12-18 18:44:23 +00:00
virCPUDataPtr *forced,
virCPUDataPtr *required,
virCPUDataPtr *optional,
virCPUDataPtr *disabled,
virCPUDataPtr *forbidden,
virCPUDataPtr *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;
struct cpuX86Data *data_forced = NULL;
struct cpuX86Data *data_required = NULL;
struct cpuX86Data *data_optional = NULL;
struct cpuX86Data *data_disabled = NULL;
struct cpuX86Data *data_forbidden = NULL;
struct cpuX86Data *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 (forced)
*forced = NULL;
if (required)
*required = NULL;
if (optional)
*optional = NULL;
if (disabled)
*disabled = NULL;
if (forbidden)
*forbidden = NULL;
if (vendor)
*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
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;
}
if (vendor) {
const struct x86_vendor *v = NULL;
if (cpu->vendor && !(v = x86VendorFind(map, cpu->vendor))) {
virReportError(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)) {
goto error;
}
}
if (forced &&
2013-07-16 12:39:40 +00:00
!(*forced = x86MakeCPUData(arch, &data_forced)))
goto error;
if (required &&
2013-07-16 12:39:40 +00:00
!(*required = x86MakeCPUData(arch, &data_required)))
goto error;
if (optional &&
2013-07-16 12:39:40 +00:00
!(*optional = x86MakeCPUData(arch, &data_optional)))
goto error;
if (disabled &&
2013-07-16 12:39:40 +00:00
!(*disabled = x86MakeCPUData(arch, &data_disabled)))
goto error;
if (forbidden &&
2013-07-16 12:39:40 +00:00
!(*forbidden = x86MakeCPUData(arch, &data_forbidden)))
goto error;
if (vendor &&
2013-07-16 12:39:40 +00:00
!(*vendor = x86MakeCPUData(arch, &data_vendor)))
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
ret = 0;
cleanup:
x86MapFree(map);
return ret;
error:
x86DataFree(data_forced);
x86DataFree(data_required);
x86DataFree(data_optional);
x86DataFree(data_disabled);
x86DataFree(data_forbidden);
x86DataFree(data_vendor);
if (forced)
x86FreeCPUData(*forced);
if (required)
x86FreeCPUData(*required);
if (optional)
x86FreeCPUData(*optional);
if (disabled)
x86FreeCPUData(*disabled);
if (forbidden)
x86FreeCPUData(*forbidden);
if (vendor)
x86FreeCPUData(*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)
{
# if __x86_64__
asm("xor %%ebx, %%ebx;" /* clear the other registers as some cpuid */
"xor %%ecx, %%ecx;" /* functions may use them as additional */
"xor %%edx, %%edx;" /* arguments */
"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
: "=a" (cpuid->eax),
"=b" (cpuid->ebx),
"=c" (cpuid->ecx),
"=d" (cpuid->edx)
: "a" (cpuid->function));
# 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;"
"xor %%ebx, %%ebx;" /* clear the other registers as some cpuid */
"xor %%ecx, %%ecx;" /* functions may use them as additional */
"xor %%edx, %%edx;" /* arguments */
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
"cpuid;"
"mov %%ebx, %1;"
"pop %%ebx;"
: "=a" (cpuid->eax),
"=r" (cpuid->ebx),
"=c" (cpuid->ecx),
"=d" (cpuid->edx)
: "a" (cpuid->function)
: "cc");
# 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)
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;
}
2012-12-18 18:44:23 +00:00
static virCPUDataPtr
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
x86NodeData(void)
{
2012-12-18 18:44:23 +00:00
virCPUDataPtr cpuData = NULL;
struct cpuX86Data *data;
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)
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;
if ((ret = cpuidSet(CPUX86_BASIC, &data->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;
data->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
if ((ret = cpuidSet(CPUX86_EXTENDED, &data->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;
data->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
2013-07-16 12:39:40 +00:00
if (!(cpuData = x86MakeCPUData(virArchFromHost(), &data)))
goto error;
return cpuData;
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:
x86DataFree(data);
return NULL;
}
#endif
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;
size_t i;
const struct x86_vendor *vendor = NULL;
struct x86_model *model = NULL;
bool outputVendor = true;
if (!(map = x86LoadMap()))
goto error;
if (!(base_model = x86ModelFromCPU(cpus[0], map, VIR_CPU_FEATURE_REQUIRE)))
goto error;
if (VIR_ALLOC(cpu) < 0)
goto error;
cpu->arch = cpus[0]->arch;
cpu->type = VIR_CPU_TYPE_GUEST;
cpu->match = VIR_CPU_MATCH_EXACT;
if (!cpus[0]->vendor)
outputVendor = false;
else if (!(vendor = x86VendorFind(map, cpus[0]->vendor))) {
virReportError(VIR_ERR_OPERATION_FAILED,
_("Unknown CPU vendor %s"), cpus[0]->vendor);
goto error;
}
for (i = 1; i < ncpus; i++) {
const char *vn = NULL;
if (!(model = x86ModelFromCPU(cpus[i], map, VIR_CPU_FEATURE_REQUIRE)))
goto error;
if (cpus[i]->vendor && model->vendor &&
STRNEQ(cpus[i]->vendor, model->vendor->name)) {
virReportError(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;
else {
outputVendor = false;
if (model->vendor)
vn = model->vendor->name;
}
if (vn) {
if (!vendor) {
if (!(vendor = x86VendorFind(map, vn))) {
virReportError(VIR_ERR_OPERATION_FAILED,
_("Unknown CPU vendor %s"), vn);
goto error;
}
} else if (STRNEQ(vendor->name, vn)) {
virReportError(VIR_ERR_OPERATION_FAILED,
"%s", _("CPU vendors do not match"));
goto error;
}
}
x86DataIntersect(base_model->data, model->data);
x86ModelFree(model);
model = NULL;
}
if (x86DataIsEmpty(base_model->data)) {
virReportError(VIR_ERR_OPERATION_FAILED,
"%s", _("CPUs are incompatible"));
goto error;
}
if (vendor && x86DataAddCpuid(base_model->data, &vendor->cpuid) < 0)
goto error;
if (x86Decode(cpu, base_model->data, models, nmodels, NULL) < 0)
goto error;
if (!outputVendor)
VIR_FREE(cpu->vendor);
cpu->arch = VIR_ARCH_NONE;
cleanup:
x86ModelFree(base_model);
x86MapFree(map);
return cpu;
error:
x86ModelFree(model);
virCPUDefFree(cpu);
cpu = NULL;
goto cleanup;
}
static int
x86UpdateCustom(virCPUDefPtr guest,
const virCPUDefPtr host)
{
int ret = -1;
size_t i;
struct x86_map *map;
struct x86_model *host_model = NULL;
if (!(map = x86LoadMap()) ||
!(host_model = x86ModelFromCPU(host, map, VIR_CPU_FEATURE_REQUIRE)))
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))) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Unknown CPU feature %s"),
guest->features[i].name);
goto cleanup;
}
if (x86DataIsSubset(host_model->data, feature->data))
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;
2013-07-15 13:40:40 +00:00
if (x86ModelSubtractCPU(host_model, guest, map) ||
x86DataToCPUFeatures(guest, VIR_CPU_FEATURE_REQUIRE,
host_model->data, map))
goto cleanup;
}
ret = 0;
cleanup:
x86MapFree(map);
x86ModelFree(host_model);
return ret;
}
static int
x86UpdateHostModel(virCPUDefPtr guest,
const virCPUDefPtr host)
{
virCPUDefPtr oldguest;
size_t i;
guest->match = VIR_CPU_MATCH_EXACT;
/* no updates are required */
if (guest->nfeatures == 0) {
virCPUDefFreeModel(guest);
return virCPUDefCopyModel(guest, host, true);
}
/* update the host model according to the desired configuration */
if (!(oldguest = virCPUDefCopy(guest)))
return -1;
virCPUDefFreeModel(guest);
if (virCPUDefCopyModel(guest, host, true) < 0)
return -1;
for (i = 0; i < oldguest->nfeatures; i++) {
if (virCPUDefUpdateFeature(guest,
oldguest->features[i].name,
oldguest->features[i].policy) < 0)
return -1;
}
return 0;
}
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:
return x86UpdateHostModel(guest, host);
case VIR_CPU_MODE_HOST_PASSTHROUGH:
guest->match = VIR_CPU_MATCH_MINIMUM;
virCPUDefFreeModel(guest);
return virCPUDefCopyModel(guest, host, true);
case VIR_CPU_MODE_LAST:
break;
}
virReportError(VIR_ERR_INTERNAL_ERROR,
_("Unexpected CPU mode: %d"), guest->mode);
return -1;
}
2012-12-18 18:44:23 +00:00
static int x86HasFeature(const virCPUDataPtr 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;
2013-07-16 12:39:40 +00:00
ret = x86DataIsSubset(data->data.x86, feature->data) ? 1 : 0;
cleanup:
x86MapFree(map);
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
struct cpuArchDriver cpuDriverX86 = {
.name = "x86",
.arch = archs,
.narch = ARRAY_CARDINALITY(archs),
.compare = x86Compare,
.decode = x86DecodeCPUData,
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
.encode = x86Encode,
.free = x86FreeCPUData,
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 HAVE_CPUID
.nodeData = x86NodeData,
#else
.nodeData = NULL,
#endif
.guestData = x86GuestData,
.baseline = x86Baseline,
.update = x86Update,
.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
};