qemu/hw/i386/pc_q35.c

351 lines
11 KiB
C
Raw Normal View History

/*
* Q35 chipset based pc system emulator
*
* Copyright (c) 2003-2004 Fabrice Bellard
* Copyright (c) 2009, 2010
* Isaku Yamahata <yamahata at valinux co jp>
* VA Linux Systems Japan K.K.
* Copyright (C) 2012 Jason Baron <jbaron@redhat.com>
*
* This is based on pc.c, but heavily modified.
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
* THE SOFTWARE.
*/
#include "hw/hw.h"
#include "hw/loader.h"
#include "sysemu/arch_init.h"
#include "hw/i2c/smbus.h"
#include "hw/boards.h"
#include "hw/timer/mc146818rtc.h"
#include "hw/xen/xen.h"
#include "sysemu/kvm.h"
#include "hw/kvm/clock.h"
#include "hw/pci-host/q35.h"
#include "exec/address-spaces.h"
#include "hw/i386/ich9.h"
#include "hw/i386/smbios.h"
#include "hw/ide/pci.h"
#include "hw/ide/ahci.h"
#include "hw/usb.h"
#include "hw/cpu/icc_bus.h"
/* ICH9 AHCI has 6 ports */
#define MAX_SATA_PORTS 6
static bool has_pci_info;
i386: ACPI table generation code from seabios This adds C code for generating ACPI tables at runtime, imported from seabios git tree commit 51684b7ced75fb76776e8ee84833fcfb6ecf12dd Although ACPI tables come from a system BIOS on real hw, it makes sense that the ACPI tables are coupled with the virtual machine, since they have to abstract the x86 machine to the OS's. This is widely desired as a way to avoid the churn and proliferation of QEMU-specific interfaces associated with ACPI tables in bios code. Notes: As BIOS can reprogram devices prior to loading ACPI tables, we pre-format ACPI tables but defer loading hardware configuration there until tables are loaded. The code structure was intentionally kept as close to the seabios original as possible, to simplify comparison and making sure we didn't lose anything in translation. Minor code duplication results, to help ensure there are no functional regressions, I think it's better to merge it like this and do more code changes in follow-up patches. Cross-version compatibility concerns have been addressed: ACPI tables are exposed to guest as FW_CFG entries. When running with -M 1.5 and older, this patch disables ACPI table generation, and doesn't expose ACPI tables to guest. As table content is likely to change over time, the following measures are taken to simplify cross-version migration: - All tables besides the RSDP are packed in a single FW CFG entry. This entry size is currently 23K. We round it up to 64K to avoid too much churn there. - Tables are placed in special ROM blob (not mapped into guest memory) which is automatically migrated together with the guest, same as BIOS code. - Offsets where hardware configuration is loaded in ACPI tables are also migrated, this is in case future ACPI changes make us rearrange the tables in memory. This patch reuses some code from SeaBIOS, which was originally under LGPLv2 and then relicensed to GPLv3 or LGPLv3, in QEMU under GPLv2+. This relicensing has been acked by all contributors that had contributed to the code since the v2->v3 relicense. ACKs approving the v2+ relicensing are listed below. The list might include ACKs from people not holding copyright on any parts of the reused code, but it's better to err on the side of caution and include them. Affected SeaBIOS files (GPLv2+ license headers added) <http://thread.gmane.org/gmane.comp.bios.coreboot.seabios/5949>: src/acpi-dsdt-cpu-hotplug.dsl src/acpi-dsdt-dbug.dsl src/acpi-dsdt-hpet.dsl src/acpi-dsdt-isa.dsl src/acpi-dsdt-pci-crs.dsl src/acpi.c src/acpi.h src/ssdt-misc.dsl src/ssdt-pcihp.dsl src/ssdt-proc.dsl tools/acpi_extract.py tools/acpi_extract_preprocess.py Each one of the listed people agreed to the following: > If you allow the use of your contribution in QEMU under the > terms of GPLv2 or later as proposed by this patch, > please respond to this mail including the line: > > Acked-by: Name <email address> Acked-by: Gerd Hoffmann <kraxel@redhat.com> Acked-by: Jan Kiszka <jan.kiszka@siemens.com> Acked-by: Jason Baron <jbaron@akamai.com> Acked-by: David Woodhouse <David.Woodhouse@intel.com> Acked-by: Gleb Natapov <gleb@redhat.com> Acked-by: Marcelo Tosatti <mtosatti@redhat.com> Acked-by: Dave Frodin <dave.frodin@se-eng.com> Acked-by: Paolo Bonzini <pbonzini@redhat.com> Acked-by: Kevin O'Connor <kevin@koconnor.net> Acked-by: Laszlo Ersek <lersek@redhat.com> Acked-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Acked-by: Isaku Yamahata <yamahata@valinux.co.jp> Acked-by: Magnus Christensson <magnus.christensson@intel.com> Acked-by: Hu Tao <hutao@cn.fujitsu.com> Acked-by: Eduardo Habkost <ehabkost@redhat.com> Reviewed-by: Gerd Hoffmann <kraxel@redhat.com> Tested-by: Gerd Hoffmann <kraxel@redhat.com> Reviewed-by: Igor Mammedov <imammedo@redhat.com> Tested-by: Igor Mammedov <imammedo@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2013-07-24 19:56:14 +04:00
static bool has_acpi_build = true;
static bool smbios_type1_defaults = true;
/* PC hardware initialisation */
static void pc_q35_init(QEMUMachineInitArgs *args)
{
ram_addr_t below_4g_mem_size, above_4g_mem_size;
Q35PCIHost *q35_host;
PCIHostState *phb;
PCIBus *host_bus;
PCIDevice *lpc;
BusState *idebus[MAX_SATA_PORTS];
ISADevice *rtc_state;
ISADevice *floppy;
MemoryRegion *pci_memory;
MemoryRegion *rom_memory;
MemoryRegion *ram_memory;
GSIState *gsi_state;
ISABus *isa_bus;
int pci_enabled = 1;
qemu_irq *cpu_irq;
qemu_irq *gsi;
qemu_irq *i8259;
int i;
ICH9LPCState *ich9_lpc;
PCIDevice *ahci;
DeviceState *icc_bridge;
PcGuestInfo *guest_info;
if (xen_enabled() && xen_hvm_init(&ram_memory) != 0) {
fprintf(stderr, "xen hardware virtual machine initialisation failed\n");
exit(1);
}
icc_bridge = qdev_create(NULL, TYPE_ICC_BRIDGE);
object_property_add_child(qdev_get_machine(), "icc-bridge",
OBJECT(icc_bridge), NULL);
pc_cpus_init(args->cpu_model, icc_bridge);
pc_acpi_init("q35-acpi-dsdt.aml");
kvmclock_create();
if (args->ram_size >= 0xb0000000) {
above_4g_mem_size = args->ram_size - 0xb0000000;
below_4g_mem_size = 0xb0000000;
} else {
above_4g_mem_size = 0;
below_4g_mem_size = args->ram_size;
}
/* pci enabled */
if (pci_enabled) {
pci_memory = g_new(MemoryRegion, 1);
memory_region_init(pci_memory, NULL, "pci", UINT64_MAX);
rom_memory = pci_memory;
} else {
pci_memory = NULL;
rom_memory = get_system_memory();
}
guest_info = pc_guest_info_init(below_4g_mem_size, above_4g_mem_size);
guest_info->has_pci_info = has_pci_info;
guest_info->isapc_ram_fw = false;
i386: ACPI table generation code from seabios This adds C code for generating ACPI tables at runtime, imported from seabios git tree commit 51684b7ced75fb76776e8ee84833fcfb6ecf12dd Although ACPI tables come from a system BIOS on real hw, it makes sense that the ACPI tables are coupled with the virtual machine, since they have to abstract the x86 machine to the OS's. This is widely desired as a way to avoid the churn and proliferation of QEMU-specific interfaces associated with ACPI tables in bios code. Notes: As BIOS can reprogram devices prior to loading ACPI tables, we pre-format ACPI tables but defer loading hardware configuration there until tables are loaded. The code structure was intentionally kept as close to the seabios original as possible, to simplify comparison and making sure we didn't lose anything in translation. Minor code duplication results, to help ensure there are no functional regressions, I think it's better to merge it like this and do more code changes in follow-up patches. Cross-version compatibility concerns have been addressed: ACPI tables are exposed to guest as FW_CFG entries. When running with -M 1.5 and older, this patch disables ACPI table generation, and doesn't expose ACPI tables to guest. As table content is likely to change over time, the following measures are taken to simplify cross-version migration: - All tables besides the RSDP are packed in a single FW CFG entry. This entry size is currently 23K. We round it up to 64K to avoid too much churn there. - Tables are placed in special ROM blob (not mapped into guest memory) which is automatically migrated together with the guest, same as BIOS code. - Offsets where hardware configuration is loaded in ACPI tables are also migrated, this is in case future ACPI changes make us rearrange the tables in memory. This patch reuses some code from SeaBIOS, which was originally under LGPLv2 and then relicensed to GPLv3 or LGPLv3, in QEMU under GPLv2+. This relicensing has been acked by all contributors that had contributed to the code since the v2->v3 relicense. ACKs approving the v2+ relicensing are listed below. The list might include ACKs from people not holding copyright on any parts of the reused code, but it's better to err on the side of caution and include them. Affected SeaBIOS files (GPLv2+ license headers added) <http://thread.gmane.org/gmane.comp.bios.coreboot.seabios/5949>: src/acpi-dsdt-cpu-hotplug.dsl src/acpi-dsdt-dbug.dsl src/acpi-dsdt-hpet.dsl src/acpi-dsdt-isa.dsl src/acpi-dsdt-pci-crs.dsl src/acpi.c src/acpi.h src/ssdt-misc.dsl src/ssdt-pcihp.dsl src/ssdt-proc.dsl tools/acpi_extract.py tools/acpi_extract_preprocess.py Each one of the listed people agreed to the following: > If you allow the use of your contribution in QEMU under the > terms of GPLv2 or later as proposed by this patch, > please respond to this mail including the line: > > Acked-by: Name <email address> Acked-by: Gerd Hoffmann <kraxel@redhat.com> Acked-by: Jan Kiszka <jan.kiszka@siemens.com> Acked-by: Jason Baron <jbaron@akamai.com> Acked-by: David Woodhouse <David.Woodhouse@intel.com> Acked-by: Gleb Natapov <gleb@redhat.com> Acked-by: Marcelo Tosatti <mtosatti@redhat.com> Acked-by: Dave Frodin <dave.frodin@se-eng.com> Acked-by: Paolo Bonzini <pbonzini@redhat.com> Acked-by: Kevin O'Connor <kevin@koconnor.net> Acked-by: Laszlo Ersek <lersek@redhat.com> Acked-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Acked-by: Isaku Yamahata <yamahata@valinux.co.jp> Acked-by: Magnus Christensson <magnus.christensson@intel.com> Acked-by: Hu Tao <hutao@cn.fujitsu.com> Acked-by: Eduardo Habkost <ehabkost@redhat.com> Reviewed-by: Gerd Hoffmann <kraxel@redhat.com> Tested-by: Gerd Hoffmann <kraxel@redhat.com> Reviewed-by: Igor Mammedov <imammedo@redhat.com> Tested-by: Igor Mammedov <imammedo@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2013-07-24 19:56:14 +04:00
guest_info->has_acpi_build = has_acpi_build;
if (smbios_type1_defaults) {
/* These values are guest ABI, do not change */
smbios_set_type1_defaults("QEMU", "Standard PC (Q35 + ICH9, 2009)",
args->machine->name);
}
/* allocate ram and load rom/bios */
if (!xen_enabled()) {
pc_memory_init(get_system_memory(),
args->kernel_filename, args->kernel_cmdline,
args->initrd_filename,
below_4g_mem_size, above_4g_mem_size,
rom_memory, &ram_memory, guest_info);
}
/* irq lines */
gsi_state = g_malloc0(sizeof(*gsi_state));
if (kvm_irqchip_in_kernel()) {
kvm_pc_setup_irq_routing(pci_enabled);
gsi = qemu_allocate_irqs(kvm_pc_gsi_handler, gsi_state,
GSI_NUM_PINS);
} else {
gsi = qemu_allocate_irqs(gsi_handler, gsi_state, GSI_NUM_PINS);
}
/* create pci host bus */
q35_host = Q35_HOST_DEVICE(qdev_create(NULL, TYPE_Q35_HOST_DEVICE));
object_property_add_child(qdev_get_machine(), "q35", OBJECT(q35_host), NULL);
q35_host->mch.ram_memory = ram_memory;
q35_host->mch.pci_address_space = pci_memory;
q35_host->mch.system_memory = get_system_memory();
q35_host->mch.address_space_io = get_system_io();
q35_host->mch.below_4g_mem_size = below_4g_mem_size;
q35_host->mch.above_4g_mem_size = above_4g_mem_size;
q35_host->mch.guest_info = guest_info;
/* pci */
qdev_init_nofail(DEVICE(q35_host));
phb = PCI_HOST_BRIDGE(q35_host);
host_bus = phb->bus;
/* create ISA bus */
lpc = pci_create_simple_multifunction(host_bus, PCI_DEVFN(ICH9_LPC_DEV,
ICH9_LPC_FUNC), true,
TYPE_ICH9_LPC_DEVICE);
ich9_lpc = ICH9_LPC_DEVICE(lpc);
ich9_lpc->pic = gsi;
ich9_lpc->ioapic = gsi_state->ioapic_irq;
pci_bus_irqs(host_bus, ich9_lpc_set_irq, ich9_lpc_map_irq, ich9_lpc,
ICH9_LPC_NB_PIRQS);
pci_bus_set_route_irq_fn(host_bus, ich9_route_intx_pin_to_irq);
isa_bus = ich9_lpc->isa_bus;
/*end early*/
isa_bus_irqs(isa_bus, gsi);
if (kvm_irqchip_in_kernel()) {
i8259 = kvm_i8259_init(isa_bus);
} else if (xen_enabled()) {
i8259 = xen_interrupt_controller_init();
} else {
cpu_irq = pc_allocate_cpu_irq();
i8259 = i8259_init(isa_bus, cpu_irq[0]);
}
for (i = 0; i < ISA_NUM_IRQS; i++) {
gsi_state->i8259_irq[i] = i8259[i];
}
if (pci_enabled) {
ioapic_init_gsi(gsi_state, NULL);
}
qdev_init_nofail(icc_bridge);
pc_register_ferr_irq(gsi[13]);
/* init basic PC hardware */
pc_basic_device_init(isa_bus, gsi, &rtc_state, &floppy, false, 0xff0104);
/* connect pm stuff to lpc */
ich9_lpc_pm_init(lpc);
/* ahci and SATA device, for q35 1 ahci controller is built-in */
ahci = pci_create_simple_multifunction(host_bus,
PCI_DEVFN(ICH9_SATA1_DEV,
ICH9_SATA1_FUNC),
true, "ich9-ahci");
idebus[0] = qdev_get_child_bus(&ahci->qdev, "ide.0");
idebus[1] = qdev_get_child_bus(&ahci->qdev, "ide.1");
if (usb_enabled(false)) {
/* Should we create 6 UHCI according to ich9 spec? */
ehci_create_ich9_with_companions(host_bus, 0x1d);
}
/* TODO: Populate SPD eeprom data. */
smbus_eeprom_init(ich9_smb_init(host_bus,
PCI_DEVFN(ICH9_SMB_DEV, ICH9_SMB_FUNC),
0xb100),
8, NULL, 0);
pc_cmos_init(below_4g_mem_size, above_4g_mem_size, args->boot_order,
floppy, idebus[0], idebus[1], rtc_state);
/* the rest devices to which pci devfn is automatically assigned */
pc_vga_init(isa_bus, host_bus);
pc_nic_init(isa_bus, host_bus);
if (pci_enabled) {
pc_pci_device_init(host_bus);
}
}
static void pc_compat_1_7(QEMUMachineInitArgs *args)
{
smbios_type1_defaults = false;
}
static void pc_compat_1_6(QEMUMachineInitArgs *args)
{
pc_compat_1_7(args);
has_pci_info = false;
rom_file_in_ram = false;
i386: ACPI table generation code from seabios This adds C code for generating ACPI tables at runtime, imported from seabios git tree commit 51684b7ced75fb76776e8ee84833fcfb6ecf12dd Although ACPI tables come from a system BIOS on real hw, it makes sense that the ACPI tables are coupled with the virtual machine, since they have to abstract the x86 machine to the OS's. This is widely desired as a way to avoid the churn and proliferation of QEMU-specific interfaces associated with ACPI tables in bios code. Notes: As BIOS can reprogram devices prior to loading ACPI tables, we pre-format ACPI tables but defer loading hardware configuration there until tables are loaded. The code structure was intentionally kept as close to the seabios original as possible, to simplify comparison and making sure we didn't lose anything in translation. Minor code duplication results, to help ensure there are no functional regressions, I think it's better to merge it like this and do more code changes in follow-up patches. Cross-version compatibility concerns have been addressed: ACPI tables are exposed to guest as FW_CFG entries. When running with -M 1.5 and older, this patch disables ACPI table generation, and doesn't expose ACPI tables to guest. As table content is likely to change over time, the following measures are taken to simplify cross-version migration: - All tables besides the RSDP are packed in a single FW CFG entry. This entry size is currently 23K. We round it up to 64K to avoid too much churn there. - Tables are placed in special ROM blob (not mapped into guest memory) which is automatically migrated together with the guest, same as BIOS code. - Offsets where hardware configuration is loaded in ACPI tables are also migrated, this is in case future ACPI changes make us rearrange the tables in memory. This patch reuses some code from SeaBIOS, which was originally under LGPLv2 and then relicensed to GPLv3 or LGPLv3, in QEMU under GPLv2+. This relicensing has been acked by all contributors that had contributed to the code since the v2->v3 relicense. ACKs approving the v2+ relicensing are listed below. The list might include ACKs from people not holding copyright on any parts of the reused code, but it's better to err on the side of caution and include them. Affected SeaBIOS files (GPLv2+ license headers added) <http://thread.gmane.org/gmane.comp.bios.coreboot.seabios/5949>: src/acpi-dsdt-cpu-hotplug.dsl src/acpi-dsdt-dbug.dsl src/acpi-dsdt-hpet.dsl src/acpi-dsdt-isa.dsl src/acpi-dsdt-pci-crs.dsl src/acpi.c src/acpi.h src/ssdt-misc.dsl src/ssdt-pcihp.dsl src/ssdt-proc.dsl tools/acpi_extract.py tools/acpi_extract_preprocess.py Each one of the listed people agreed to the following: > If you allow the use of your contribution in QEMU under the > terms of GPLv2 or later as proposed by this patch, > please respond to this mail including the line: > > Acked-by: Name <email address> Acked-by: Gerd Hoffmann <kraxel@redhat.com> Acked-by: Jan Kiszka <jan.kiszka@siemens.com> Acked-by: Jason Baron <jbaron@akamai.com> Acked-by: David Woodhouse <David.Woodhouse@intel.com> Acked-by: Gleb Natapov <gleb@redhat.com> Acked-by: Marcelo Tosatti <mtosatti@redhat.com> Acked-by: Dave Frodin <dave.frodin@se-eng.com> Acked-by: Paolo Bonzini <pbonzini@redhat.com> Acked-by: Kevin O'Connor <kevin@koconnor.net> Acked-by: Laszlo Ersek <lersek@redhat.com> Acked-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Acked-by: Isaku Yamahata <yamahata@valinux.co.jp> Acked-by: Magnus Christensson <magnus.christensson@intel.com> Acked-by: Hu Tao <hutao@cn.fujitsu.com> Acked-by: Eduardo Habkost <ehabkost@redhat.com> Reviewed-by: Gerd Hoffmann <kraxel@redhat.com> Tested-by: Gerd Hoffmann <kraxel@redhat.com> Reviewed-by: Igor Mammedov <imammedo@redhat.com> Tested-by: Igor Mammedov <imammedo@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2013-07-24 19:56:14 +04:00
has_acpi_build = false;
}
static void pc_compat_1_5(QEMUMachineInitArgs *args)
{
pc_compat_1_6(args);
}
static void pc_compat_1_4(QEMUMachineInitArgs *args)
pc: Kill the "use flash device for BIOS unless KVM" misfeature Use of a flash memory device for the BIOS was added in series "[PATCH v10 0/8] PC system flash support", commit 4732dca..1b89faf, v1.1. Flash vs. ROM is a guest-visible difference. Thus, flash use had to be suppressed for machine types pc-1.0 and older. This was accomplished by adding a dummy device "pc-sysfw" with property "rom_only": * Non-zero rom_only means "use ROM". Default for pc-1.0 and older. * Zero rom_only means "maybe use flash". Default for newer machines. Not only is the dummy device ugly, it was also retroactively added to the older machine types! Fortunately, it's not guest-visible (thus no immediate guest ABI breakage), and has no vmstate (thus no immediate migration breakage). Breakage occurs only if the user unwisely enables flash by setting rom_only to zero. Patch review FAIL #1. Why "maybe use flash"? Flash didn't (and still doesn't) work with KVM. Therefore, rom_only=0 really means "use flash, except when KVM is enabled, use ROM". This is a Bad Idea, because it makes enabling/ disabling KVM guest-visible. Patch review FAIL #2. Aside: it also precludes migrating between KVM on and off, but that's not possible for other reasons anyway. Fix as follows: 1. Change the meaning of rom_only=0 to mean "use flash, no ifs, buts, or maybes" for pc-i440fx-1.5 and pc-q35-1.5. Don't change anything for older machines (to remain bug-compatible). 2. Change the default value from 0 to 1 for these machines. Necessary, because 0 doesn't work with KVM. Once it does, we can flip the default back to 0. 3. Don't revert the retroactive addition of device "pc-sysfw" to older machine types. Seems not worth the trouble. 4. Add a TODO comment asking for device "pc-sysfw" to be dropped once flash works with KVM. Net effect is that you get a BIOS ROM again even when KVM is disabled, just like for machines predating the introduction of flash. To get flash instead, use "--global pc-sysfw.rom_only=0". Signed-off-by: Markus Armbruster <armbru@redhat.com> Message-id: 1365780303-26398-4-git-send-email-armbru@redhat.com Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2013-04-12 19:25:03 +04:00
{
pc_compat_1_5(args);
x86_cpu_compat_set_features("n270", FEAT_1_ECX, 0, CPUID_EXT_MOVBE);
x86_cpu_compat_set_features("Westmere", FEAT_1_ECX, 0, CPUID_EXT_PCLMULQDQ);
}
static void pc_q35_init_1_7(QEMUMachineInitArgs *args)
{
pc_compat_1_7(args);
pc_q35_init(args);
}
static void pc_q35_init_1_6(QEMUMachineInitArgs *args)
{
pc_compat_1_6(args);
pc_q35_init(args);
}
static void pc_q35_init_1_5(QEMUMachineInitArgs *args)
{
pc_compat_1_5(args);
pc_q35_init(args);
}
static void pc_q35_init_1_4(QEMUMachineInitArgs *args)
{
pc_compat_1_4(args);
pc_q35_init(args);
pc: Kill the "use flash device for BIOS unless KVM" misfeature Use of a flash memory device for the BIOS was added in series "[PATCH v10 0/8] PC system flash support", commit 4732dca..1b89faf, v1.1. Flash vs. ROM is a guest-visible difference. Thus, flash use had to be suppressed for machine types pc-1.0 and older. This was accomplished by adding a dummy device "pc-sysfw" with property "rom_only": * Non-zero rom_only means "use ROM". Default for pc-1.0 and older. * Zero rom_only means "maybe use flash". Default for newer machines. Not only is the dummy device ugly, it was also retroactively added to the older machine types! Fortunately, it's not guest-visible (thus no immediate guest ABI breakage), and has no vmstate (thus no immediate migration breakage). Breakage occurs only if the user unwisely enables flash by setting rom_only to zero. Patch review FAIL #1. Why "maybe use flash"? Flash didn't (and still doesn't) work with KVM. Therefore, rom_only=0 really means "use flash, except when KVM is enabled, use ROM". This is a Bad Idea, because it makes enabling/ disabling KVM guest-visible. Patch review FAIL #2. Aside: it also precludes migrating between KVM on and off, but that's not possible for other reasons anyway. Fix as follows: 1. Change the meaning of rom_only=0 to mean "use flash, no ifs, buts, or maybes" for pc-i440fx-1.5 and pc-q35-1.5. Don't change anything for older machines (to remain bug-compatible). 2. Change the default value from 0 to 1 for these machines. Necessary, because 0 doesn't work with KVM. Once it does, we can flip the default back to 0. 3. Don't revert the retroactive addition of device "pc-sysfw" to older machine types. Seems not worth the trouble. 4. Add a TODO comment asking for device "pc-sysfw" to be dropped once flash works with KVM. Net effect is that you get a BIOS ROM again even when KVM is disabled, just like for machines predating the introduction of flash. To get flash instead, use "--global pc-sysfw.rom_only=0". Signed-off-by: Markus Armbruster <armbru@redhat.com> Message-id: 1365780303-26398-4-git-send-email-armbru@redhat.com Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2013-04-12 19:25:03 +04:00
}
#define PC_Q35_MACHINE_OPTIONS \
PC_DEFAULT_MACHINE_OPTIONS, \
.desc = "Standard PC (Q35 + ICH9, 2009)", \
.hot_add_cpu = pc_hot_add_cpu
#define PC_Q35_2_0_MACHINE_OPTIONS \
PC_Q35_MACHINE_OPTIONS, \
.default_machine_opts = "firmware=bios-256k.bin"
static QEMUMachine pc_q35_machine_v2_0 = {
PC_Q35_2_0_MACHINE_OPTIONS,
.name = "pc-q35-2.0",
.alias = "q35",
.init = pc_q35_init,
};
#define PC_Q35_1_7_MACHINE_OPTIONS PC_Q35_MACHINE_OPTIONS
static QEMUMachine pc_q35_machine_v1_7 = {
PC_Q35_1_7_MACHINE_OPTIONS,
.name = "pc-q35-1.7",
.init = pc_q35_init_1_7,
.compat_props = (GlobalProperty[]) {
PC_Q35_COMPAT_1_7,
{ /* end of list */ }
},
};
#define PC_Q35_1_6_MACHINE_OPTIONS PC_Q35_MACHINE_OPTIONS
static QEMUMachine pc_q35_machine_v1_6 = {
PC_Q35_1_6_MACHINE_OPTIONS,
.name = "pc-q35-1.6",
.init = pc_q35_init_1_6,
.compat_props = (GlobalProperty[]) {
PC_Q35_COMPAT_1_6,
{ /* end of list */ }
},
};
static QEMUMachine pc_q35_machine_v1_5 = {
PC_Q35_1_6_MACHINE_OPTIONS,
.name = "pc-q35-1.5",
.init = pc_q35_init_1_5,
.compat_props = (GlobalProperty[]) {
PC_Q35_COMPAT_1_5,
{ /* end of list */ }
},
};
#define PC_Q35_1_4_MACHINE_OPTIONS \
PC_Q35_1_6_MACHINE_OPTIONS, \
.hot_add_cpu = NULL
static QEMUMachine pc_q35_machine_v1_4 = {
PC_Q35_1_4_MACHINE_OPTIONS,
.name = "pc-q35-1.4",
pc: Kill the "use flash device for BIOS unless KVM" misfeature Use of a flash memory device for the BIOS was added in series "[PATCH v10 0/8] PC system flash support", commit 4732dca..1b89faf, v1.1. Flash vs. ROM is a guest-visible difference. Thus, flash use had to be suppressed for machine types pc-1.0 and older. This was accomplished by adding a dummy device "pc-sysfw" with property "rom_only": * Non-zero rom_only means "use ROM". Default for pc-1.0 and older. * Zero rom_only means "maybe use flash". Default for newer machines. Not only is the dummy device ugly, it was also retroactively added to the older machine types! Fortunately, it's not guest-visible (thus no immediate guest ABI breakage), and has no vmstate (thus no immediate migration breakage). Breakage occurs only if the user unwisely enables flash by setting rom_only to zero. Patch review FAIL #1. Why "maybe use flash"? Flash didn't (and still doesn't) work with KVM. Therefore, rom_only=0 really means "use flash, except when KVM is enabled, use ROM". This is a Bad Idea, because it makes enabling/ disabling KVM guest-visible. Patch review FAIL #2. Aside: it also precludes migrating between KVM on and off, but that's not possible for other reasons anyway. Fix as follows: 1. Change the meaning of rom_only=0 to mean "use flash, no ifs, buts, or maybes" for pc-i440fx-1.5 and pc-q35-1.5. Don't change anything for older machines (to remain bug-compatible). 2. Change the default value from 0 to 1 for these machines. Necessary, because 0 doesn't work with KVM. Once it does, we can flip the default back to 0. 3. Don't revert the retroactive addition of device "pc-sysfw" to older machine types. Seems not worth the trouble. 4. Add a TODO comment asking for device "pc-sysfw" to be dropped once flash works with KVM. Net effect is that you get a BIOS ROM again even when KVM is disabled, just like for machines predating the introduction of flash. To get flash instead, use "--global pc-sysfw.rom_only=0". Signed-off-by: Markus Armbruster <armbru@redhat.com> Message-id: 1365780303-26398-4-git-send-email-armbru@redhat.com Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
2013-04-12 19:25:03 +04:00
.init = pc_q35_init_1_4,
.compat_props = (GlobalProperty[]) {
PC_COMPAT_1_4,
{ /* end of list */ }
},
};
static void pc_q35_machine_init(void)
{
qemu_register_machine(&pc_q35_machine_v2_0);
qemu_register_machine(&pc_q35_machine_v1_7);
qemu_register_machine(&pc_q35_machine_v1_6);
qemu_register_machine(&pc_q35_machine_v1_5);
qemu_register_machine(&pc_q35_machine_v1_4);
}
machine_init(pc_q35_machine_init);