2 * QEMU PC System Emulator
4 * Copyright (c) 2003-2004 Fabrice Bellard
6 * Permission is hereby granted, free of charge, to any person obtaining a copy
7 * of this software and associated documentation files (the "Software"), to deal
8 * in the Software without restriction, including without limitation the rights
9 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
10 * copies of the Software, and to permit persons to whom the Software is
11 * furnished to do so, subject to the following conditions:
13 * The above copyright notice and this permission notice shall be included in
14 * all copies or substantial portions of the Software.
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
19 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
25 #include "hw/i386/pc.h"
26 #include "hw/char/serial.h"
27 #include "hw/i386/apic.h"
28 #include "hw/i386/topology.h"
29 #include "sysemu/cpus.h"
30 #include "hw/block/fdc.h"
32 #include "hw/pci/pci.h"
33 #include "hw/pci/pci_bus.h"
34 #include "hw/nvram/fw_cfg.h"
35 #include "hw/timer/hpet.h"
36 #include "hw/smbios/smbios.h"
37 #include "hw/loader.h"
39 #include "multiboot.h"
40 #include "hw/timer/mc146818rtc.h"
41 #include "hw/timer/i8254.h"
42 #include "hw/audio/pcspk.h"
43 #include "hw/pci/msi.h"
44 #include "hw/sysbus.h"
45 #include "sysemu/sysemu.h"
46 #include "sysemu/numa.h"
47 #include "sysemu/kvm.h"
48 #include "sysemu/qtest.h"
50 #include "hw/xen/xen.h"
51 #include "sysemu/block-backend.h"
52 #include "hw/block/block.h"
53 #include "ui/qemu-spice.h"
54 #include "exec/memory.h"
55 #include "exec/address-spaces.h"
56 #include "sysemu/arch_init.h"
57 #include "qemu/bitmap.h"
58 #include "qemu/config-file.h"
59 #include "qemu/error-report.h"
60 #include "hw/acpi/acpi.h"
61 #include "hw/acpi/cpu_hotplug.h"
62 #include "hw/boards.h"
63 #include "hw/pci/pci_host.h"
64 #include "acpi-build.h"
65 #include "hw/mem/pc-dimm.h"
66 #include "qapi/visitor.h"
67 #include "qapi-visit.h"
70 /* debug PC/ISA interrupts */
74 #define DPRINTF(fmt, ...) \
75 do { printf("CPUIRQ: " fmt , ## __VA_ARGS__); } while (0)
77 #define DPRINTF(fmt, ...)
80 #define BIOS_CFG_IOPORT 0x510
81 #define FW_CFG_ACPI_TABLES (FW_CFG_ARCH_LOCAL + 0)
82 #define FW_CFG_SMBIOS_ENTRIES (FW_CFG_ARCH_LOCAL + 1)
83 #define FW_CFG_IRQ0_OVERRIDE (FW_CFG_ARCH_LOCAL + 2)
84 #define FW_CFG_E820_TABLE (FW_CFG_ARCH_LOCAL + 3)
85 #define FW_CFG_HPET (FW_CFG_ARCH_LOCAL + 4)
87 #define E820_NR_ENTRIES 16
93 } QEMU_PACKED
__attribute((__aligned__(4)));
97 struct e820_entry entry
[E820_NR_ENTRIES
];
98 } QEMU_PACKED
__attribute((__aligned__(4)));
100 static struct e820_table e820_reserve
;
101 static struct e820_entry
*e820_table
;
102 static unsigned e820_entries
;
103 struct hpet_fw_config hpet_cfg
= {.count
= UINT8_MAX
};
105 void gsi_handler(void *opaque
, int n
, int level
)
107 GSIState
*s
= opaque
;
109 DPRINTF("pc: %s GSI %d\n", level
? "raising" : "lowering", n
);
110 if (n
< ISA_NUM_IRQS
) {
111 qemu_set_irq(s
->i8259_irq
[n
], level
);
113 qemu_set_irq(s
->ioapic_irq
[n
], level
);
116 static void ioport80_write(void *opaque
, hwaddr addr
, uint64_t data
,
121 static uint64_t ioport80_read(void *opaque
, hwaddr addr
, unsigned size
)
123 return 0xffffffffffffffffULL
;
126 /* MSDOS compatibility mode FPU exception support */
127 static qemu_irq ferr_irq
;
129 void pc_register_ferr_irq(qemu_irq irq
)
134 /* XXX: add IGNNE support */
135 void cpu_set_ferr(CPUX86State
*s
)
137 qemu_irq_raise(ferr_irq
);
140 static void ioportF0_write(void *opaque
, hwaddr addr
, uint64_t data
,
143 qemu_irq_lower(ferr_irq
);
146 static uint64_t ioportF0_read(void *opaque
, hwaddr addr
, unsigned size
)
148 return 0xffffffffffffffffULL
;
152 uint64_t cpu_get_tsc(CPUX86State
*env
)
154 return cpu_get_ticks();
158 int cpu_get_pic_interrupt(CPUX86State
*env
)
160 X86CPU
*cpu
= x86_env_get_cpu(env
);
163 intno
= apic_get_interrupt(cpu
->apic_state
);
167 /* read the irq from the PIC */
168 if (!apic_accept_pic_intr(cpu
->apic_state
)) {
172 intno
= pic_read_irq(isa_pic
);
176 static void pic_irq_request(void *opaque
, int irq
, int level
)
178 CPUState
*cs
= first_cpu
;
179 X86CPU
*cpu
= X86_CPU(cs
);
181 DPRINTF("pic_irqs: %s irq %d\n", level
? "raise" : "lower", irq
);
182 if (cpu
->apic_state
) {
185 if (apic_accept_pic_intr(cpu
->apic_state
)) {
186 apic_deliver_pic_intr(cpu
->apic_state
, level
);
191 cpu_interrupt(cs
, CPU_INTERRUPT_HARD
);
193 cpu_reset_interrupt(cs
, CPU_INTERRUPT_HARD
);
198 /* PC cmos mappings */
200 #define REG_EQUIPMENT_BYTE 0x14
202 static int cmos_get_fd_drive_type(FDriveType fd0
)
208 /* 1.44 Mb 3"5 drive */
212 /* 2.88 Mb 3"5 drive */
216 /* 1.2 Mb 5"5 drive */
219 case FDRIVE_DRV_NONE
:
227 static void cmos_init_hd(ISADevice
*s
, int type_ofs
, int info_ofs
,
228 int16_t cylinders
, int8_t heads
, int8_t sectors
)
230 rtc_set_memory(s
, type_ofs
, 47);
231 rtc_set_memory(s
, info_ofs
, cylinders
);
232 rtc_set_memory(s
, info_ofs
+ 1, cylinders
>> 8);
233 rtc_set_memory(s
, info_ofs
+ 2, heads
);
234 rtc_set_memory(s
, info_ofs
+ 3, 0xff);
235 rtc_set_memory(s
, info_ofs
+ 4, 0xff);
236 rtc_set_memory(s
, info_ofs
+ 5, 0xc0 | ((heads
> 8) << 3));
237 rtc_set_memory(s
, info_ofs
+ 6, cylinders
);
238 rtc_set_memory(s
, info_ofs
+ 7, cylinders
>> 8);
239 rtc_set_memory(s
, info_ofs
+ 8, sectors
);
242 /* convert boot_device letter to something recognizable by the bios */
243 static int boot_device2nibble(char boot_device
)
245 switch(boot_device
) {
248 return 0x01; /* floppy boot */
250 return 0x02; /* hard drive boot */
252 return 0x03; /* CD-ROM boot */
254 return 0x04; /* Network boot */
259 static void set_boot_dev(ISADevice
*s
, const char *boot_device
, Error
**errp
)
261 #define PC_MAX_BOOT_DEVICES 3
262 int nbds
, bds
[3] = { 0, };
265 nbds
= strlen(boot_device
);
266 if (nbds
> PC_MAX_BOOT_DEVICES
) {
267 error_setg(errp
, "Too many boot devices for PC");
270 for (i
= 0; i
< nbds
; i
++) {
271 bds
[i
] = boot_device2nibble(boot_device
[i
]);
273 error_setg(errp
, "Invalid boot device for PC: '%c'",
278 rtc_set_memory(s
, 0x3d, (bds
[1] << 4) | bds
[0]);
279 rtc_set_memory(s
, 0x38, (bds
[2] << 4) | (fd_bootchk
? 0x0 : 0x1));
282 static void pc_boot_set(void *opaque
, const char *boot_device
, Error
**errp
)
284 set_boot_dev(opaque
, boot_device
, errp
);
287 static void pc_cmos_init_floppy(ISADevice
*rtc_state
, ISADevice
*floppy
)
290 FDriveType fd_type
[2] = { FDRIVE_DRV_NONE
, FDRIVE_DRV_NONE
};
294 for (i
= 0; i
< 2; i
++) {
295 fd_type
[i
] = isa_fdc_get_drive_type(floppy
, i
);
298 val
= (cmos_get_fd_drive_type(fd_type
[0]) << 4) |
299 cmos_get_fd_drive_type(fd_type
[1]);
300 rtc_set_memory(rtc_state
, 0x10, val
);
302 val
= rtc_get_memory(rtc_state
, REG_EQUIPMENT_BYTE
);
304 if (fd_type
[0] < FDRIVE_DRV_NONE
) {
307 if (fd_type
[1] < FDRIVE_DRV_NONE
) {
314 val
|= 0x01; /* 1 drive, ready for boot */
317 val
|= 0x41; /* 2 drives, ready for boot */
320 rtc_set_memory(rtc_state
, REG_EQUIPMENT_BYTE
, val
);
323 typedef struct pc_cmos_init_late_arg
{
324 ISADevice
*rtc_state
;
326 } pc_cmos_init_late_arg
;
328 typedef struct check_fdc_state
{
333 static int check_fdc(Object
*obj
, void *opaque
)
335 CheckFdcState
*state
= opaque
;
338 Error
*local_err
= NULL
;
340 fdc
= object_dynamic_cast(obj
, TYPE_ISA_FDC
);
345 iobase
= object_property_get_int(obj
, "iobase", &local_err
);
346 if (local_err
|| iobase
!= 0x3f0) {
347 error_free(local_err
);
352 state
->multiple
= true;
354 state
->floppy
= ISA_DEVICE(obj
);
359 static const char * const fdc_container_path
[] = {
360 "/unattached", "/peripheral", "/peripheral-anon"
364 * Locate the FDC at IO address 0x3f0, in order to configure the CMOS registers
367 ISADevice
*pc_find_fdc0(void)
371 CheckFdcState state
= { 0 };
373 for (i
= 0; i
< ARRAY_SIZE(fdc_container_path
); i
++) {
374 container
= container_get(qdev_get_machine(), fdc_container_path
[i
]);
375 object_child_foreach(container
, check_fdc
, &state
);
378 if (state
.multiple
) {
379 error_report("warning: multiple floppy disk controllers with "
380 "iobase=0x3f0 have been found");
381 error_printf("the one being picked for CMOS setup might not reflect "
388 static void pc_cmos_init_late(void *opaque
)
390 pc_cmos_init_late_arg
*arg
= opaque
;
391 ISADevice
*s
= arg
->rtc_state
;
393 int8_t heads
, sectors
;
398 if (ide_get_geometry(arg
->idebus
[0], 0,
399 &cylinders
, &heads
, §ors
) >= 0) {
400 cmos_init_hd(s
, 0x19, 0x1b, cylinders
, heads
, sectors
);
403 if (ide_get_geometry(arg
->idebus
[0], 1,
404 &cylinders
, &heads
, §ors
) >= 0) {
405 cmos_init_hd(s
, 0x1a, 0x24, cylinders
, heads
, sectors
);
408 rtc_set_memory(s
, 0x12, val
);
411 for (i
= 0; i
< 4; i
++) {
412 /* NOTE: ide_get_geometry() returns the physical
413 geometry. It is always such that: 1 <= sects <= 63, 1
414 <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
415 geometry can be different if a translation is done. */
416 if (ide_get_geometry(arg
->idebus
[i
/ 2], i
% 2,
417 &cylinders
, &heads
, §ors
) >= 0) {
418 trans
= ide_get_bios_chs_trans(arg
->idebus
[i
/ 2], i
% 2) - 1;
419 assert((trans
& ~3) == 0);
420 val
|= trans
<< (i
* 2);
423 rtc_set_memory(s
, 0x39, val
);
425 pc_cmos_init_floppy(s
, pc_find_fdc0());
427 qemu_unregister_reset(pc_cmos_init_late
, opaque
);
430 void pc_cmos_init(PCMachineState
*pcms
,
431 BusState
*idebus0
, BusState
*idebus1
,
435 static pc_cmos_init_late_arg arg
;
437 /* various important CMOS locations needed by PC/Bochs bios */
440 /* base memory (first MiB) */
441 val
= MIN(pcms
->below_4g_mem_size
/ 1024, 640);
442 rtc_set_memory(s
, 0x15, val
);
443 rtc_set_memory(s
, 0x16, val
>> 8);
444 /* extended memory (next 64MiB) */
445 if (pcms
->below_4g_mem_size
> 1024 * 1024) {
446 val
= (pcms
->below_4g_mem_size
- 1024 * 1024) / 1024;
452 rtc_set_memory(s
, 0x17, val
);
453 rtc_set_memory(s
, 0x18, val
>> 8);
454 rtc_set_memory(s
, 0x30, val
);
455 rtc_set_memory(s
, 0x31, val
>> 8);
456 /* memory between 16MiB and 4GiB */
457 if (pcms
->below_4g_mem_size
> 16 * 1024 * 1024) {
458 val
= (pcms
->below_4g_mem_size
- 16 * 1024 * 1024) / 65536;
464 rtc_set_memory(s
, 0x34, val
);
465 rtc_set_memory(s
, 0x35, val
>> 8);
466 /* memory above 4GiB */
467 val
= pcms
->above_4g_mem_size
/ 65536;
468 rtc_set_memory(s
, 0x5b, val
);
469 rtc_set_memory(s
, 0x5c, val
>> 8);
470 rtc_set_memory(s
, 0x5d, val
>> 16);
472 /* set the number of CPU */
473 rtc_set_memory(s
, 0x5f, smp_cpus
- 1);
475 object_property_add_link(OBJECT(pcms
), "rtc_state",
477 (Object
**)&pcms
->rtc
,
478 object_property_allow_set_link
,
479 OBJ_PROP_LINK_UNREF_ON_RELEASE
, &error_abort
);
480 object_property_set_link(OBJECT(pcms
), OBJECT(s
),
481 "rtc_state", &error_abort
);
483 set_boot_dev(s
, MACHINE(pcms
)->boot_order
, &error_fatal
);
486 val
|= 0x02; /* FPU is there */
487 val
|= 0x04; /* PS/2 mouse installed */
488 rtc_set_memory(s
, REG_EQUIPMENT_BYTE
, val
);
490 /* hard drives and FDC */
492 arg
.idebus
[0] = idebus0
;
493 arg
.idebus
[1] = idebus1
;
494 qemu_register_reset(pc_cmos_init_late
, &arg
);
497 #define TYPE_PORT92 "port92"
498 #define PORT92(obj) OBJECT_CHECK(Port92State, (obj), TYPE_PORT92)
500 /* port 92 stuff: could be split off */
501 typedef struct Port92State
{
502 ISADevice parent_obj
;
509 static void port92_write(void *opaque
, hwaddr addr
, uint64_t val
,
512 Port92State
*s
= opaque
;
513 int oldval
= s
->outport
;
515 DPRINTF("port92: write 0x%02" PRIx64
"\n", val
);
517 qemu_set_irq(*s
->a20_out
, (val
>> 1) & 1);
518 if ((val
& 1) && !(oldval
& 1)) {
519 qemu_system_reset_request();
523 static uint64_t port92_read(void *opaque
, hwaddr addr
,
526 Port92State
*s
= opaque
;
530 DPRINTF("port92: read 0x%02x\n", ret
);
534 static void port92_init(ISADevice
*dev
, qemu_irq
*a20_out
)
536 Port92State
*s
= PORT92(dev
);
538 s
->a20_out
= a20_out
;
541 static const VMStateDescription vmstate_port92_isa
= {
544 .minimum_version_id
= 1,
545 .fields
= (VMStateField
[]) {
546 VMSTATE_UINT8(outport
, Port92State
),
547 VMSTATE_END_OF_LIST()
551 static void port92_reset(DeviceState
*d
)
553 Port92State
*s
= PORT92(d
);
558 static const MemoryRegionOps port92_ops
= {
560 .write
= port92_write
,
562 .min_access_size
= 1,
563 .max_access_size
= 1,
565 .endianness
= DEVICE_LITTLE_ENDIAN
,
568 static void port92_initfn(Object
*obj
)
570 Port92State
*s
= PORT92(obj
);
572 memory_region_init_io(&s
->io
, OBJECT(s
), &port92_ops
, s
, "port92", 1);
577 static void port92_realizefn(DeviceState
*dev
, Error
**errp
)
579 ISADevice
*isadev
= ISA_DEVICE(dev
);
580 Port92State
*s
= PORT92(dev
);
582 isa_register_ioport(isadev
, &s
->io
, 0x92);
585 static void port92_class_initfn(ObjectClass
*klass
, void *data
)
587 DeviceClass
*dc
= DEVICE_CLASS(klass
);
589 dc
->realize
= port92_realizefn
;
590 dc
->reset
= port92_reset
;
591 dc
->vmsd
= &vmstate_port92_isa
;
593 * Reason: unlike ordinary ISA devices, this one needs additional
594 * wiring: its A20 output line needs to be wired up by
597 dc
->cannot_instantiate_with_device_add_yet
= true;
600 static const TypeInfo port92_info
= {
602 .parent
= TYPE_ISA_DEVICE
,
603 .instance_size
= sizeof(Port92State
),
604 .instance_init
= port92_initfn
,
605 .class_init
= port92_class_initfn
,
608 static void port92_register_types(void)
610 type_register_static(&port92_info
);
613 type_init(port92_register_types
)
615 static void handle_a20_line_change(void *opaque
, int irq
, int level
)
617 X86CPU
*cpu
= opaque
;
619 /* XXX: send to all CPUs ? */
620 /* XXX: add logic to handle multiple A20 line sources */
621 x86_cpu_set_a20(cpu
, level
);
624 int e820_add_entry(uint64_t address
, uint64_t length
, uint32_t type
)
626 int index
= le32_to_cpu(e820_reserve
.count
);
627 struct e820_entry
*entry
;
629 if (type
!= E820_RAM
) {
630 /* old FW_CFG_E820_TABLE entry -- reservations only */
631 if (index
>= E820_NR_ENTRIES
) {
634 entry
= &e820_reserve
.entry
[index
++];
636 entry
->address
= cpu_to_le64(address
);
637 entry
->length
= cpu_to_le64(length
);
638 entry
->type
= cpu_to_le32(type
);
640 e820_reserve
.count
= cpu_to_le32(index
);
643 /* new "etc/e820" file -- include ram too */
644 e820_table
= g_renew(struct e820_entry
, e820_table
, e820_entries
+ 1);
645 e820_table
[e820_entries
].address
= cpu_to_le64(address
);
646 e820_table
[e820_entries
].length
= cpu_to_le64(length
);
647 e820_table
[e820_entries
].type
= cpu_to_le32(type
);
653 int e820_get_num_entries(void)
658 bool e820_get_entry(int idx
, uint32_t type
, uint64_t *address
, uint64_t *length
)
660 if (idx
< e820_entries
&& e820_table
[idx
].type
== cpu_to_le32(type
)) {
661 *address
= le64_to_cpu(e820_table
[idx
].address
);
662 *length
= le64_to_cpu(e820_table
[idx
].length
);
668 /* Enables contiguous-apic-ID mode, for compatibility */
669 static bool compat_apic_id_mode
;
671 void enable_compat_apic_id_mode(void)
673 compat_apic_id_mode
= true;
676 /* Calculates initial APIC ID for a specific CPU index
678 * Currently we need to be able to calculate the APIC ID from the CPU index
679 * alone (without requiring a CPU object), as the QEMU<->Seabios interfaces have
680 * no concept of "CPU index", and the NUMA tables on fw_cfg need the APIC ID of
681 * all CPUs up to max_cpus.
683 static uint32_t x86_cpu_apic_id_from_index(unsigned int cpu_index
)
688 correct_id
= x86_apicid_from_cpu_idx(smp_cores
, smp_threads
, cpu_index
);
689 if (compat_apic_id_mode
) {
690 if (cpu_index
!= correct_id
&& !warned
&& !qtest_enabled()) {
691 error_report("APIC IDs set in compatibility mode, "
692 "CPU topology won't match the configuration");
701 /* Calculates the limit to CPU APIC ID values
703 * This function returns the limit for the APIC ID value, so that all
704 * CPU APIC IDs are < pc_apic_id_limit().
706 * This is used for FW_CFG_MAX_CPUS. See comments on bochs_bios_init().
708 static unsigned int pc_apic_id_limit(unsigned int max_cpus
)
710 return x86_cpu_apic_id_from_index(max_cpus
- 1) + 1;
713 static void pc_build_smbios(FWCfgState
*fw_cfg
)
715 uint8_t *smbios_tables
, *smbios_anchor
;
716 size_t smbios_tables_len
, smbios_anchor_len
;
717 struct smbios_phys_mem_area
*mem_array
;
718 unsigned i
, array_count
;
720 smbios_tables
= smbios_get_table_legacy(&smbios_tables_len
);
722 fw_cfg_add_bytes(fw_cfg
, FW_CFG_SMBIOS_ENTRIES
,
723 smbios_tables
, smbios_tables_len
);
726 /* build the array of physical mem area from e820 table */
727 mem_array
= g_malloc0(sizeof(*mem_array
) * e820_get_num_entries());
728 for (i
= 0, array_count
= 0; i
< e820_get_num_entries(); i
++) {
731 if (e820_get_entry(i
, E820_RAM
, &addr
, &len
)) {
732 mem_array
[array_count
].address
= addr
;
733 mem_array
[array_count
].length
= len
;
737 smbios_get_tables(mem_array
, array_count
,
738 &smbios_tables
, &smbios_tables_len
,
739 &smbios_anchor
, &smbios_anchor_len
);
743 fw_cfg_add_file(fw_cfg
, "etc/smbios/smbios-tables",
744 smbios_tables
, smbios_tables_len
);
745 fw_cfg_add_file(fw_cfg
, "etc/smbios/smbios-anchor",
746 smbios_anchor
, smbios_anchor_len
);
750 static FWCfgState
*bochs_bios_init(AddressSpace
*as
)
753 uint64_t *numa_fw_cfg
;
755 unsigned int apic_id_limit
= pc_apic_id_limit(max_cpus
);
757 fw_cfg
= fw_cfg_init_io_dma(BIOS_CFG_IOPORT
, BIOS_CFG_IOPORT
+ 4, as
);
759 /* FW_CFG_MAX_CPUS is a bit confusing/problematic on x86:
761 * SeaBIOS needs FW_CFG_MAX_CPUS for CPU hotplug, but the CPU hotplug
762 * QEMU<->SeaBIOS interface is not based on the "CPU index", but on the APIC
763 * ID of hotplugged CPUs[1]. This means that FW_CFG_MAX_CPUS is not the
764 * "maximum number of CPUs", but the "limit to the APIC ID values SeaBIOS
767 * So, this means we must not use max_cpus, here, but the maximum possible
768 * APIC ID value, plus one.
770 * [1] The only kind of "CPU identifier" used between SeaBIOS and QEMU is
771 * the APIC ID, not the "CPU index"
773 fw_cfg_add_i16(fw_cfg
, FW_CFG_MAX_CPUS
, (uint16_t)apic_id_limit
);
774 fw_cfg_add_i64(fw_cfg
, FW_CFG_RAM_SIZE
, (uint64_t)ram_size
);
775 fw_cfg_add_bytes(fw_cfg
, FW_CFG_ACPI_TABLES
,
776 acpi_tables
, acpi_tables_len
);
777 fw_cfg_add_i32(fw_cfg
, FW_CFG_IRQ0_OVERRIDE
, kvm_allows_irq0_override());
779 pc_build_smbios(fw_cfg
);
781 fw_cfg_add_bytes(fw_cfg
, FW_CFG_E820_TABLE
,
782 &e820_reserve
, sizeof(e820_reserve
));
783 fw_cfg_add_file(fw_cfg
, "etc/e820", e820_table
,
784 sizeof(struct e820_entry
) * e820_entries
);
786 fw_cfg_add_bytes(fw_cfg
, FW_CFG_HPET
, &hpet_cfg
, sizeof(hpet_cfg
));
787 /* allocate memory for the NUMA channel: one (64bit) word for the number
788 * of nodes, one word for each VCPU->node and one word for each node to
789 * hold the amount of memory.
791 numa_fw_cfg
= g_new0(uint64_t, 1 + apic_id_limit
+ nb_numa_nodes
);
792 numa_fw_cfg
[0] = cpu_to_le64(nb_numa_nodes
);
793 for (i
= 0; i
< max_cpus
; i
++) {
794 unsigned int apic_id
= x86_cpu_apic_id_from_index(i
);
795 assert(apic_id
< apic_id_limit
);
796 for (j
= 0; j
< nb_numa_nodes
; j
++) {
797 if (test_bit(i
, numa_info
[j
].node_cpu
)) {
798 numa_fw_cfg
[apic_id
+ 1] = cpu_to_le64(j
);
803 for (i
= 0; i
< nb_numa_nodes
; i
++) {
804 numa_fw_cfg
[apic_id_limit
+ 1 + i
] = cpu_to_le64(numa_info
[i
].node_mem
);
806 fw_cfg_add_bytes(fw_cfg
, FW_CFG_NUMA
, numa_fw_cfg
,
807 (1 + apic_id_limit
+ nb_numa_nodes
) *
808 sizeof(*numa_fw_cfg
));
813 static long get_file_size(FILE *f
)
817 /* XXX: on Unix systems, using fstat() probably makes more sense */
820 fseek(f
, 0, SEEK_END
);
822 fseek(f
, where
, SEEK_SET
);
827 static void load_linux(PCMachineState
*pcms
,
831 int setup_size
, kernel_size
, initrd_size
= 0, cmdline_size
;
833 uint8_t header
[8192], *setup
, *kernel
, *initrd_data
;
834 hwaddr real_addr
, prot_addr
, cmdline_addr
, initrd_addr
= 0;
837 MachineState
*machine
= MACHINE(pcms
);
838 PCMachineClass
*pcmc
= PC_MACHINE_GET_CLASS(pcms
);
839 const char *kernel_filename
= machine
->kernel_filename
;
840 const char *initrd_filename
= machine
->initrd_filename
;
841 const char *kernel_cmdline
= machine
->kernel_cmdline
;
843 /* Align to 16 bytes as a paranoia measure */
844 cmdline_size
= (strlen(kernel_cmdline
)+16) & ~15;
846 /* load the kernel header */
847 f
= fopen(kernel_filename
, "rb");
848 if (!f
|| !(kernel_size
= get_file_size(f
)) ||
849 fread(header
, 1, MIN(ARRAY_SIZE(header
), kernel_size
), f
) !=
850 MIN(ARRAY_SIZE(header
), kernel_size
)) {
851 fprintf(stderr
, "qemu: could not load kernel '%s': %s\n",
852 kernel_filename
, strerror(errno
));
856 /* kernel protocol version */
858 fprintf(stderr
, "header magic: %#x\n", ldl_p(header
+0x202));
860 if (ldl_p(header
+0x202) == 0x53726448) {
861 protocol
= lduw_p(header
+0x206);
863 /* This looks like a multiboot kernel. If it is, let's stop
864 treating it like a Linux kernel. */
865 if (load_multiboot(fw_cfg
, f
, kernel_filename
, initrd_filename
,
866 kernel_cmdline
, kernel_size
, header
)) {
872 if (protocol
< 0x200 || !(header
[0x211] & 0x01)) {
875 cmdline_addr
= 0x9a000 - cmdline_size
;
877 } else if (protocol
< 0x202) {
878 /* High but ancient kernel */
880 cmdline_addr
= 0x9a000 - cmdline_size
;
881 prot_addr
= 0x100000;
883 /* High and recent kernel */
885 cmdline_addr
= 0x20000;
886 prot_addr
= 0x100000;
891 "qemu: real_addr = 0x" TARGET_FMT_plx
"\n"
892 "qemu: cmdline_addr = 0x" TARGET_FMT_plx
"\n"
893 "qemu: prot_addr = 0x" TARGET_FMT_plx
"\n",
899 /* highest address for loading the initrd */
900 if (protocol
>= 0x203) {
901 initrd_max
= ldl_p(header
+0x22c);
903 initrd_max
= 0x37ffffff;
906 if (initrd_max
>= pcms
->below_4g_mem_size
- pcmc
->acpi_data_size
) {
907 initrd_max
= pcms
->below_4g_mem_size
- pcmc
->acpi_data_size
- 1;
910 fw_cfg_add_i32(fw_cfg
, FW_CFG_CMDLINE_ADDR
, cmdline_addr
);
911 fw_cfg_add_i32(fw_cfg
, FW_CFG_CMDLINE_SIZE
, strlen(kernel_cmdline
)+1);
912 fw_cfg_add_string(fw_cfg
, FW_CFG_CMDLINE_DATA
, kernel_cmdline
);
914 if (protocol
>= 0x202) {
915 stl_p(header
+0x228, cmdline_addr
);
917 stw_p(header
+0x20, 0xA33F);
918 stw_p(header
+0x22, cmdline_addr
-real_addr
);
921 /* handle vga= parameter */
922 vmode
= strstr(kernel_cmdline
, "vga=");
924 unsigned int video_mode
;
927 if (!strncmp(vmode
, "normal", 6)) {
929 } else if (!strncmp(vmode
, "ext", 3)) {
931 } else if (!strncmp(vmode
, "ask", 3)) {
934 video_mode
= strtol(vmode
, NULL
, 0);
936 stw_p(header
+0x1fa, video_mode
);
940 /* High nybble = B reserved for QEMU; low nybble is revision number.
941 If this code is substantially changed, you may want to consider
942 incrementing the revision. */
943 if (protocol
>= 0x200) {
944 header
[0x210] = 0xB0;
947 if (protocol
>= 0x201) {
948 header
[0x211] |= 0x80; /* CAN_USE_HEAP */
949 stw_p(header
+0x224, cmdline_addr
-real_addr
-0x200);
953 if (initrd_filename
) {
954 if (protocol
< 0x200) {
955 fprintf(stderr
, "qemu: linux kernel too old to load a ram disk\n");
959 initrd_size
= get_image_size(initrd_filename
);
960 if (initrd_size
< 0) {
961 fprintf(stderr
, "qemu: error reading initrd %s: %s\n",
962 initrd_filename
, strerror(errno
));
966 initrd_addr
= (initrd_max
-initrd_size
) & ~4095;
968 initrd_data
= g_malloc(initrd_size
);
969 load_image(initrd_filename
, initrd_data
);
971 fw_cfg_add_i32(fw_cfg
, FW_CFG_INITRD_ADDR
, initrd_addr
);
972 fw_cfg_add_i32(fw_cfg
, FW_CFG_INITRD_SIZE
, initrd_size
);
973 fw_cfg_add_bytes(fw_cfg
, FW_CFG_INITRD_DATA
, initrd_data
, initrd_size
);
975 stl_p(header
+0x218, initrd_addr
);
976 stl_p(header
+0x21c, initrd_size
);
979 /* load kernel and setup */
980 setup_size
= header
[0x1f1];
981 if (setup_size
== 0) {
984 setup_size
= (setup_size
+1)*512;
985 if (setup_size
> kernel_size
) {
986 fprintf(stderr
, "qemu: invalid kernel header\n");
989 kernel_size
-= setup_size
;
991 setup
= g_malloc(setup_size
);
992 kernel
= g_malloc(kernel_size
);
993 fseek(f
, 0, SEEK_SET
);
994 if (fread(setup
, 1, setup_size
, f
) != setup_size
) {
995 fprintf(stderr
, "fread() failed\n");
998 if (fread(kernel
, 1, kernel_size
, f
) != kernel_size
) {
999 fprintf(stderr
, "fread() failed\n");
1003 memcpy(setup
, header
, MIN(sizeof(header
), setup_size
));
1005 fw_cfg_add_i32(fw_cfg
, FW_CFG_KERNEL_ADDR
, prot_addr
);
1006 fw_cfg_add_i32(fw_cfg
, FW_CFG_KERNEL_SIZE
, kernel_size
);
1007 fw_cfg_add_bytes(fw_cfg
, FW_CFG_KERNEL_DATA
, kernel
, kernel_size
);
1009 fw_cfg_add_i32(fw_cfg
, FW_CFG_SETUP_ADDR
, real_addr
);
1010 fw_cfg_add_i32(fw_cfg
, FW_CFG_SETUP_SIZE
, setup_size
);
1011 fw_cfg_add_bytes(fw_cfg
, FW_CFG_SETUP_DATA
, setup
, setup_size
);
1013 option_rom
[nb_option_roms
].name
= "linuxboot.bin";
1014 option_rom
[nb_option_roms
].bootindex
= 0;
1018 #define NE2000_NB_MAX 6
1020 static const int ne2000_io
[NE2000_NB_MAX
] = { 0x300, 0x320, 0x340, 0x360,
1022 static const int ne2000_irq
[NE2000_NB_MAX
] = { 9, 10, 11, 3, 4, 5 };
1024 void pc_init_ne2k_isa(ISABus
*bus
, NICInfo
*nd
)
1026 static int nb_ne2k
= 0;
1028 if (nb_ne2k
== NE2000_NB_MAX
)
1030 isa_ne2000_init(bus
, ne2000_io
[nb_ne2k
],
1031 ne2000_irq
[nb_ne2k
], nd
);
1035 DeviceState
*cpu_get_current_apic(void)
1038 X86CPU
*cpu
= X86_CPU(current_cpu
);
1039 return cpu
->apic_state
;
1045 void pc_acpi_smi_interrupt(void *opaque
, int irq
, int level
)
1047 X86CPU
*cpu
= opaque
;
1050 cpu_interrupt(CPU(cpu
), CPU_INTERRUPT_SMI
);
1054 static X86CPU
*pc_new_cpu(const char *cpu_model
, int64_t apic_id
,
1058 Error
*local_err
= NULL
;
1060 cpu
= cpu_x86_create(cpu_model
, &local_err
);
1061 if (local_err
!= NULL
) {
1065 object_property_set_int(OBJECT(cpu
), apic_id
, "apic-id", &local_err
);
1066 object_property_set_bool(OBJECT(cpu
), true, "realized", &local_err
);
1070 error_propagate(errp
, local_err
);
1071 object_unref(OBJECT(cpu
));
1077 void pc_hot_add_cpu(const int64_t id
, Error
**errp
)
1080 MachineState
*machine
= MACHINE(qdev_get_machine());
1081 int64_t apic_id
= x86_cpu_apic_id_from_index(id
);
1082 Error
*local_err
= NULL
;
1085 error_setg(errp
, "Invalid CPU id: %" PRIi64
, id
);
1089 if (cpu_exists(apic_id
)) {
1090 error_setg(errp
, "Unable to add CPU: %" PRIi64
1091 ", it already exists", id
);
1095 if (id
>= max_cpus
) {
1096 error_setg(errp
, "Unable to add CPU: %" PRIi64
1097 ", max allowed: %d", id
, max_cpus
- 1);
1101 if (apic_id
>= ACPI_CPU_HOTPLUG_ID_LIMIT
) {
1102 error_setg(errp
, "Unable to add CPU: %" PRIi64
1103 ", resulting APIC ID (%" PRIi64
") is too large",
1108 cpu
= pc_new_cpu(machine
->cpu_model
, apic_id
, &local_err
);
1110 error_propagate(errp
, local_err
);
1113 object_unref(OBJECT(cpu
));
1116 void pc_cpus_init(PCMachineState
*pcms
)
1120 MachineState
*machine
= MACHINE(pcms
);
1121 unsigned long apic_id_limit
;
1124 if (machine
->cpu_model
== NULL
) {
1125 #ifdef TARGET_X86_64
1126 machine
->cpu_model
= "qemu64";
1128 machine
->cpu_model
= "qemu32";
1132 apic_id_limit
= pc_apic_id_limit(max_cpus
);
1133 if (apic_id_limit
> ACPI_CPU_HOTPLUG_ID_LIMIT
) {
1134 error_report("max_cpus is too large. APIC ID of last CPU is %lu",
1139 for (i
= 0; i
< smp_cpus
; i
++) {
1140 cpu
= pc_new_cpu(machine
->cpu_model
, x86_cpu_apic_id_from_index(i
),
1142 object_unref(OBJECT(cpu
));
1145 /* tell smbios about cpuid version and features */
1146 smbios_set_cpuid(cpu
->env
.cpuid_version
, cpu
->env
.features
[FEAT_1_EDX
]);
1149 /* pci-info ROM file. Little endian format */
1150 typedef struct PcRomPciInfo
{
1157 typedef struct PcGuestInfoState
{
1159 Notifier machine_done
;
1163 void pc_guest_info_machine_done(Notifier
*notifier
, void *data
)
1165 PcGuestInfoState
*guest_info_state
= container_of(notifier
,
1168 PCIBus
*bus
= PC_MACHINE(qdev_get_machine())->bus
;
1171 int extra_hosts
= 0;
1173 QLIST_FOREACH(bus
, &bus
->child
, sibling
) {
1174 /* look for expander root buses */
1175 if (pci_bus_is_root(bus
)) {
1179 if (extra_hosts
&& guest_info_state
->info
.fw_cfg
) {
1180 uint64_t *val
= g_malloc(sizeof(*val
));
1181 *val
= cpu_to_le64(extra_hosts
);
1182 fw_cfg_add_file(guest_info_state
->info
.fw_cfg
,
1183 "etc/extra-pci-roots", val
, sizeof(*val
));
1187 acpi_setup(&guest_info_state
->info
);
1190 PcGuestInfo
*pc_guest_info_init(PCMachineState
*pcms
)
1192 PcGuestInfoState
*guest_info_state
= g_malloc0(sizeof *guest_info_state
);
1193 PcGuestInfo
*guest_info
= &guest_info_state
->info
;
1196 guest_info
->ram_size_below_4g
= pcms
->below_4g_mem_size
;
1197 guest_info
->ram_size
= pcms
->below_4g_mem_size
+ pcms
->above_4g_mem_size
;
1198 guest_info
->apic_id_limit
= pc_apic_id_limit(max_cpus
);
1199 guest_info
->apic_xrupt_override
= kvm_allows_irq0_override();
1200 guest_info
->numa_nodes
= nb_numa_nodes
;
1201 guest_info
->node_mem
= g_malloc0(guest_info
->numa_nodes
*
1202 sizeof *guest_info
->node_mem
);
1203 for (i
= 0; i
< nb_numa_nodes
; i
++) {
1204 guest_info
->node_mem
[i
] = numa_info
[i
].node_mem
;
1207 guest_info
->node_cpu
= g_malloc0(guest_info
->apic_id_limit
*
1208 sizeof *guest_info
->node_cpu
);
1210 for (i
= 0; i
< max_cpus
; i
++) {
1211 unsigned int apic_id
= x86_cpu_apic_id_from_index(i
);
1212 assert(apic_id
< guest_info
->apic_id_limit
);
1213 for (j
= 0; j
< nb_numa_nodes
; j
++) {
1214 if (test_bit(i
, numa_info
[j
].node_cpu
)) {
1215 guest_info
->node_cpu
[apic_id
] = j
;
1221 guest_info_state
->machine_done
.notify
= pc_guest_info_machine_done
;
1222 qemu_add_machine_init_done_notifier(&guest_info_state
->machine_done
);
1226 /* setup pci memory address space mapping into system address space */
1227 void pc_pci_as_mapping_init(Object
*owner
, MemoryRegion
*system_memory
,
1228 MemoryRegion
*pci_address_space
)
1230 /* Set to lower priority than RAM */
1231 memory_region_add_subregion_overlap(system_memory
, 0x0,
1232 pci_address_space
, -1);
1235 void pc_acpi_init(const char *default_dsdt
)
1239 if (acpi_tables
!= NULL
) {
1240 /* manually set via -acpitable, leave it alone */
1244 filename
= qemu_find_file(QEMU_FILE_TYPE_BIOS
, default_dsdt
);
1245 if (filename
== NULL
) {
1246 fprintf(stderr
, "WARNING: failed to find %s\n", default_dsdt
);
1248 QemuOpts
*opts
= qemu_opts_create(qemu_find_opts("acpi"), NULL
, 0,
1252 qemu_opt_set(opts
, "file", filename
, &error_abort
);
1254 acpi_table_add_builtin(opts
, &err
);
1256 error_reportf_err(err
, "WARNING: failed to load %s: ",
1263 FWCfgState
*xen_load_linux(PCMachineState
*pcms
,
1264 PcGuestInfo
*guest_info
)
1269 assert(MACHINE(pcms
)->kernel_filename
!= NULL
);
1271 fw_cfg
= fw_cfg_init_io(BIOS_CFG_IOPORT
);
1274 load_linux(pcms
, fw_cfg
);
1275 for (i
= 0; i
< nb_option_roms
; i
++) {
1276 assert(!strcmp(option_rom
[i
].name
, "linuxboot.bin") ||
1277 !strcmp(option_rom
[i
].name
, "multiboot.bin"));
1278 rom_add_option(option_rom
[i
].name
, option_rom
[i
].bootindex
);
1280 guest_info
->fw_cfg
= fw_cfg
;
1284 FWCfgState
*pc_memory_init(PCMachineState
*pcms
,
1285 MemoryRegion
*system_memory
,
1286 MemoryRegion
*rom_memory
,
1287 MemoryRegion
**ram_memory
,
1288 PcGuestInfo
*guest_info
)
1291 MemoryRegion
*ram
, *option_rom_mr
;
1292 MemoryRegion
*ram_below_4g
, *ram_above_4g
;
1294 MachineState
*machine
= MACHINE(pcms
);
1295 PCMachineClass
*pcmc
= PC_MACHINE_GET_CLASS(pcms
);
1297 assert(machine
->ram_size
== pcms
->below_4g_mem_size
+
1298 pcms
->above_4g_mem_size
);
1300 linux_boot
= (machine
->kernel_filename
!= NULL
);
1302 /* Allocate RAM. We allocate it as a single memory region and use
1303 * aliases to address portions of it, mostly for backwards compatibility
1304 * with older qemus that used qemu_ram_alloc().
1306 ram
= g_malloc(sizeof(*ram
));
1307 memory_region_allocate_system_memory(ram
, NULL
, "pc.ram",
1310 ram_below_4g
= g_malloc(sizeof(*ram_below_4g
));
1311 memory_region_init_alias(ram_below_4g
, NULL
, "ram-below-4g", ram
,
1312 0, pcms
->below_4g_mem_size
);
1313 memory_region_add_subregion(system_memory
, 0, ram_below_4g
);
1314 e820_add_entry(0, pcms
->below_4g_mem_size
, E820_RAM
);
1315 if (pcms
->above_4g_mem_size
> 0) {
1316 ram_above_4g
= g_malloc(sizeof(*ram_above_4g
));
1317 memory_region_init_alias(ram_above_4g
, NULL
, "ram-above-4g", ram
,
1318 pcms
->below_4g_mem_size
,
1319 pcms
->above_4g_mem_size
);
1320 memory_region_add_subregion(system_memory
, 0x100000000ULL
,
1322 e820_add_entry(0x100000000ULL
, pcms
->above_4g_mem_size
, E820_RAM
);
1325 if (!guest_info
->has_reserved_memory
&&
1326 (machine
->ram_slots
||
1327 (machine
->maxram_size
> machine
->ram_size
))) {
1328 MachineClass
*mc
= MACHINE_GET_CLASS(machine
);
1330 error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
1335 /* initialize hotplug memory address space */
1336 if (guest_info
->has_reserved_memory
&&
1337 (machine
->ram_size
< machine
->maxram_size
)) {
1338 ram_addr_t hotplug_mem_size
=
1339 machine
->maxram_size
- machine
->ram_size
;
1341 if (machine
->ram_slots
> ACPI_MAX_RAM_SLOTS
) {
1342 error_report("unsupported amount of memory slots: %"PRIu64
,
1343 machine
->ram_slots
);
1347 if (QEMU_ALIGN_UP(machine
->maxram_size
,
1348 TARGET_PAGE_SIZE
) != machine
->maxram_size
) {
1349 error_report("maximum memory size must by aligned to multiple of "
1350 "%d bytes", TARGET_PAGE_SIZE
);
1354 pcms
->hotplug_memory
.base
=
1355 ROUND_UP(0x100000000ULL
+ pcms
->above_4g_mem_size
, 1ULL << 30);
1357 if (pcmc
->enforce_aligned_dimm
) {
1358 /* size hotplug region assuming 1G page max alignment per slot */
1359 hotplug_mem_size
+= (1ULL << 30) * machine
->ram_slots
;
1362 if ((pcms
->hotplug_memory
.base
+ hotplug_mem_size
) <
1364 error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT
,
1365 machine
->maxram_size
);
1369 memory_region_init(&pcms
->hotplug_memory
.mr
, OBJECT(pcms
),
1370 "hotplug-memory", hotplug_mem_size
);
1371 memory_region_add_subregion(system_memory
, pcms
->hotplug_memory
.base
,
1372 &pcms
->hotplug_memory
.mr
);
1375 /* Initialize PC system firmware */
1376 pc_system_firmware_init(rom_memory
, guest_info
->isapc_ram_fw
);
1378 option_rom_mr
= g_malloc(sizeof(*option_rom_mr
));
1379 memory_region_init_ram(option_rom_mr
, NULL
, "pc.rom", PC_ROM_SIZE
,
1381 vmstate_register_ram_global(option_rom_mr
);
1382 memory_region_add_subregion_overlap(rom_memory
,
1387 fw_cfg
= bochs_bios_init(&address_space_memory
);
1391 if (guest_info
->has_reserved_memory
&& pcms
->hotplug_memory
.base
) {
1392 uint64_t *val
= g_malloc(sizeof(*val
));
1393 PCMachineClass
*pcmc
= PC_MACHINE_GET_CLASS(pcms
);
1394 uint64_t res_mem_end
= pcms
->hotplug_memory
.base
;
1396 if (!pcmc
->broken_reserved_end
) {
1397 res_mem_end
+= memory_region_size(&pcms
->hotplug_memory
.mr
);
1399 *val
= cpu_to_le64(ROUND_UP(res_mem_end
, 0x1ULL
<< 30));
1400 fw_cfg_add_file(fw_cfg
, "etc/reserved-memory-end", val
, sizeof(*val
));
1404 load_linux(pcms
, fw_cfg
);
1407 for (i
= 0; i
< nb_option_roms
; i
++) {
1408 rom_add_option(option_rom
[i
].name
, option_rom
[i
].bootindex
);
1410 guest_info
->fw_cfg
= fw_cfg
;
1414 qemu_irq
pc_allocate_cpu_irq(void)
1416 return qemu_allocate_irq(pic_irq_request
, NULL
, 0);
1419 DeviceState
*pc_vga_init(ISABus
*isa_bus
, PCIBus
*pci_bus
)
1421 DeviceState
*dev
= NULL
;
1424 PCIDevice
*pcidev
= pci_vga_init(pci_bus
);
1425 dev
= pcidev
? &pcidev
->qdev
: NULL
;
1426 } else if (isa_bus
) {
1427 ISADevice
*isadev
= isa_vga_init(isa_bus
);
1428 dev
= isadev
? DEVICE(isadev
) : NULL
;
1433 static const MemoryRegionOps ioport80_io_ops
= {
1434 .write
= ioport80_write
,
1435 .read
= ioport80_read
,
1436 .endianness
= DEVICE_NATIVE_ENDIAN
,
1438 .min_access_size
= 1,
1439 .max_access_size
= 1,
1443 static const MemoryRegionOps ioportF0_io_ops
= {
1444 .write
= ioportF0_write
,
1445 .read
= ioportF0_read
,
1446 .endianness
= DEVICE_NATIVE_ENDIAN
,
1448 .min_access_size
= 1,
1449 .max_access_size
= 1,
1453 void pc_basic_device_init(ISABus
*isa_bus
, qemu_irq
*gsi
,
1454 ISADevice
**rtc_state
,
1460 DriveInfo
*fd
[MAX_FD
];
1461 DeviceState
*hpet
= NULL
;
1462 int pit_isa_irq
= 0;
1463 qemu_irq pit_alt_irq
= NULL
;
1464 qemu_irq rtc_irq
= NULL
;
1466 ISADevice
*i8042
, *port92
, *vmmouse
, *pit
= NULL
;
1467 MemoryRegion
*ioport80_io
= g_new(MemoryRegion
, 1);
1468 MemoryRegion
*ioportF0_io
= g_new(MemoryRegion
, 1);
1470 memory_region_init_io(ioport80_io
, NULL
, &ioport80_io_ops
, NULL
, "ioport80", 1);
1471 memory_region_add_subregion(isa_bus
->address_space_io
, 0x80, ioport80_io
);
1473 memory_region_init_io(ioportF0_io
, NULL
, &ioportF0_io_ops
, NULL
, "ioportF0", 1);
1474 memory_region_add_subregion(isa_bus
->address_space_io
, 0xf0, ioportF0_io
);
1477 * Check if an HPET shall be created.
1479 * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
1480 * when the HPET wants to take over. Thus we have to disable the latter.
1482 if (!no_hpet
&& (!kvm_irqchip_in_kernel() || kvm_has_pit_state2())) {
1483 /* In order to set property, here not using sysbus_try_create_simple */
1484 hpet
= qdev_try_create(NULL
, TYPE_HPET
);
1486 /* For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-1.7
1487 * and earlier, use IRQ2 for compat. Otherwise, use IRQ16~23,
1490 uint8_t compat
= object_property_get_int(OBJECT(hpet
),
1493 qdev_prop_set_uint32(hpet
, HPET_INTCAP
, hpet_irqs
);
1495 qdev_init_nofail(hpet
);
1496 sysbus_mmio_map(SYS_BUS_DEVICE(hpet
), 0, HPET_BASE
);
1498 for (i
= 0; i
< GSI_NUM_PINS
; i
++) {
1499 sysbus_connect_irq(SYS_BUS_DEVICE(hpet
), i
, gsi
[i
]);
1502 pit_alt_irq
= qdev_get_gpio_in(hpet
, HPET_LEGACY_PIT_INT
);
1503 rtc_irq
= qdev_get_gpio_in(hpet
, HPET_LEGACY_RTC_INT
);
1506 *rtc_state
= rtc_init(isa_bus
, 2000, rtc_irq
);
1508 qemu_register_boot_set(pc_boot_set
, *rtc_state
);
1510 if (!xen_enabled()) {
1511 if (kvm_pit_in_kernel()) {
1512 pit
= kvm_pit_init(isa_bus
, 0x40);
1514 pit
= pit_init(isa_bus
, 0x40, pit_isa_irq
, pit_alt_irq
);
1517 /* connect PIT to output control line of the HPET */
1518 qdev_connect_gpio_out(hpet
, 0, qdev_get_gpio_in(DEVICE(pit
), 0));
1520 pcspk_init(isa_bus
, pit
);
1523 serial_hds_isa_init(isa_bus
, MAX_SERIAL_PORTS
);
1524 parallel_hds_isa_init(isa_bus
, MAX_PARALLEL_PORTS
);
1526 a20_line
= qemu_allocate_irqs(handle_a20_line_change
, first_cpu
, 2);
1527 i8042
= isa_create_simple(isa_bus
, "i8042");
1528 i8042_setup_a20_line(i8042
, &a20_line
[0]);
1530 vmport_init(isa_bus
);
1531 vmmouse
= isa_try_create(isa_bus
, "vmmouse");
1536 DeviceState
*dev
= DEVICE(vmmouse
);
1537 qdev_prop_set_ptr(dev
, "ps2_mouse", i8042
);
1538 qdev_init_nofail(dev
);
1540 port92
= isa_create_simple(isa_bus
, "port92");
1541 port92_init(port92
, &a20_line
[1]);
1545 for(i
= 0; i
< MAX_FD
; i
++) {
1546 fd
[i
] = drive_get(IF_FLOPPY
, 0, i
);
1547 create_fdctrl
|= !!fd
[i
];
1549 if (create_fdctrl
) {
1550 fdctrl_init_isa(isa_bus
, fd
);
1554 void pc_nic_init(ISABus
*isa_bus
, PCIBus
*pci_bus
)
1558 for (i
= 0; i
< nb_nics
; i
++) {
1559 NICInfo
*nd
= &nd_table
[i
];
1561 if (!pci_bus
|| (nd
->model
&& strcmp(nd
->model
, "ne2k_isa") == 0)) {
1562 pc_init_ne2k_isa(isa_bus
, nd
);
1564 pci_nic_init_nofail(nd
, pci_bus
, "e1000", NULL
);
1569 void pc_pci_device_init(PCIBus
*pci_bus
)
1574 max_bus
= drive_get_max_bus(IF_SCSI
);
1575 for (bus
= 0; bus
<= max_bus
; bus
++) {
1576 pci_create_simple(pci_bus
, -1, "lsi53c895a");
1580 void ioapic_init_gsi(GSIState
*gsi_state
, const char *parent_name
)
1586 if (kvm_ioapic_in_kernel()) {
1587 dev
= qdev_create(NULL
, "kvm-ioapic");
1589 dev
= qdev_create(NULL
, "ioapic");
1592 object_property_add_child(object_resolve_path(parent_name
, NULL
),
1593 "ioapic", OBJECT(dev
), NULL
);
1595 qdev_init_nofail(dev
);
1596 d
= SYS_BUS_DEVICE(dev
);
1597 sysbus_mmio_map(d
, 0, IO_APIC_DEFAULT_ADDRESS
);
1599 for (i
= 0; i
< IOAPIC_NUM_PINS
; i
++) {
1600 gsi_state
->ioapic_irq
[i
] = qdev_get_gpio_in(dev
, i
);
1604 static void pc_dimm_plug(HotplugHandler
*hotplug_dev
,
1605 DeviceState
*dev
, Error
**errp
)
1607 HotplugHandlerClass
*hhc
;
1608 Error
*local_err
= NULL
;
1609 PCMachineState
*pcms
= PC_MACHINE(hotplug_dev
);
1610 PCMachineClass
*pcmc
= PC_MACHINE_GET_CLASS(pcms
);
1611 PCDIMMDevice
*dimm
= PC_DIMM(dev
);
1612 PCDIMMDeviceClass
*ddc
= PC_DIMM_GET_CLASS(dimm
);
1613 MemoryRegion
*mr
= ddc
->get_memory_region(dimm
);
1614 uint64_t align
= TARGET_PAGE_SIZE
;
1616 if (memory_region_get_alignment(mr
) && pcmc
->enforce_aligned_dimm
) {
1617 align
= memory_region_get_alignment(mr
);
1620 if (!pcms
->acpi_dev
) {
1621 error_setg(&local_err
,
1622 "memory hotplug is not enabled: missing acpi device");
1626 pc_dimm_memory_plug(dev
, &pcms
->hotplug_memory
, mr
, align
, &local_err
);
1631 hhc
= HOTPLUG_HANDLER_GET_CLASS(pcms
->acpi_dev
);
1632 hhc
->plug(HOTPLUG_HANDLER(pcms
->acpi_dev
), dev
, &error_abort
);
1634 error_propagate(errp
, local_err
);
1637 static void pc_dimm_unplug_request(HotplugHandler
*hotplug_dev
,
1638 DeviceState
*dev
, Error
**errp
)
1640 HotplugHandlerClass
*hhc
;
1641 Error
*local_err
= NULL
;
1642 PCMachineState
*pcms
= PC_MACHINE(hotplug_dev
);
1644 if (!pcms
->acpi_dev
) {
1645 error_setg(&local_err
,
1646 "memory hotplug is not enabled: missing acpi device");
1650 hhc
= HOTPLUG_HANDLER_GET_CLASS(pcms
->acpi_dev
);
1651 hhc
->unplug_request(HOTPLUG_HANDLER(pcms
->acpi_dev
), dev
, &local_err
);
1654 error_propagate(errp
, local_err
);
1657 static void pc_dimm_unplug(HotplugHandler
*hotplug_dev
,
1658 DeviceState
*dev
, Error
**errp
)
1660 PCMachineState
*pcms
= PC_MACHINE(hotplug_dev
);
1661 PCDIMMDevice
*dimm
= PC_DIMM(dev
);
1662 PCDIMMDeviceClass
*ddc
= PC_DIMM_GET_CLASS(dimm
);
1663 MemoryRegion
*mr
= ddc
->get_memory_region(dimm
);
1664 HotplugHandlerClass
*hhc
;
1665 Error
*local_err
= NULL
;
1667 hhc
= HOTPLUG_HANDLER_GET_CLASS(pcms
->acpi_dev
);
1668 hhc
->unplug(HOTPLUG_HANDLER(pcms
->acpi_dev
), dev
, &local_err
);
1674 pc_dimm_memory_unplug(dev
, &pcms
->hotplug_memory
, mr
);
1675 object_unparent(OBJECT(dev
));
1678 error_propagate(errp
, local_err
);
1681 static void pc_cpu_plug(HotplugHandler
*hotplug_dev
,
1682 DeviceState
*dev
, Error
**errp
)
1684 HotplugHandlerClass
*hhc
;
1685 Error
*local_err
= NULL
;
1686 PCMachineState
*pcms
= PC_MACHINE(hotplug_dev
);
1688 if (!dev
->hotplugged
) {
1692 if (!pcms
->acpi_dev
) {
1693 error_setg(&local_err
,
1694 "cpu hotplug is not enabled: missing acpi device");
1698 hhc
= HOTPLUG_HANDLER_GET_CLASS(pcms
->acpi_dev
);
1699 hhc
->plug(HOTPLUG_HANDLER(pcms
->acpi_dev
), dev
, &local_err
);
1704 /* increment the number of CPUs */
1705 rtc_set_memory(pcms
->rtc
, 0x5f, rtc_get_memory(pcms
->rtc
, 0x5f) + 1);
1707 error_propagate(errp
, local_err
);
1710 static void pc_machine_device_plug_cb(HotplugHandler
*hotplug_dev
,
1711 DeviceState
*dev
, Error
**errp
)
1713 if (object_dynamic_cast(OBJECT(dev
), TYPE_PC_DIMM
)) {
1714 pc_dimm_plug(hotplug_dev
, dev
, errp
);
1715 } else if (object_dynamic_cast(OBJECT(dev
), TYPE_CPU
)) {
1716 pc_cpu_plug(hotplug_dev
, dev
, errp
);
1720 static void pc_machine_device_unplug_request_cb(HotplugHandler
*hotplug_dev
,
1721 DeviceState
*dev
, Error
**errp
)
1723 if (object_dynamic_cast(OBJECT(dev
), TYPE_PC_DIMM
)) {
1724 pc_dimm_unplug_request(hotplug_dev
, dev
, errp
);
1726 error_setg(errp
, "acpi: device unplug request for not supported device"
1727 " type: %s", object_get_typename(OBJECT(dev
)));
1731 static void pc_machine_device_unplug_cb(HotplugHandler
*hotplug_dev
,
1732 DeviceState
*dev
, Error
**errp
)
1734 if (object_dynamic_cast(OBJECT(dev
), TYPE_PC_DIMM
)) {
1735 pc_dimm_unplug(hotplug_dev
, dev
, errp
);
1737 error_setg(errp
, "acpi: device unplug for not supported device"
1738 " type: %s", object_get_typename(OBJECT(dev
)));
1742 static HotplugHandler
*pc_get_hotpug_handler(MachineState
*machine
,
1745 PCMachineClass
*pcmc
= PC_MACHINE_GET_CLASS(machine
);
1747 if (object_dynamic_cast(OBJECT(dev
), TYPE_PC_DIMM
) ||
1748 object_dynamic_cast(OBJECT(dev
), TYPE_CPU
)) {
1749 return HOTPLUG_HANDLER(machine
);
1752 return pcmc
->get_hotplug_handler
?
1753 pcmc
->get_hotplug_handler(machine
, dev
) : NULL
;
1757 pc_machine_get_hotplug_memory_region_size(Object
*obj
, Visitor
*v
, void *opaque
,
1758 const char *name
, Error
**errp
)
1760 PCMachineState
*pcms
= PC_MACHINE(obj
);
1761 int64_t value
= memory_region_size(&pcms
->hotplug_memory
.mr
);
1763 visit_type_int(v
, &value
, name
, errp
);
1766 static void pc_machine_get_max_ram_below_4g(Object
*obj
, Visitor
*v
,
1767 void *opaque
, const char *name
,
1770 PCMachineState
*pcms
= PC_MACHINE(obj
);
1771 uint64_t value
= pcms
->max_ram_below_4g
;
1773 visit_type_size(v
, &value
, name
, errp
);
1776 static void pc_machine_set_max_ram_below_4g(Object
*obj
, Visitor
*v
,
1777 void *opaque
, const char *name
,
1780 PCMachineState
*pcms
= PC_MACHINE(obj
);
1781 Error
*error
= NULL
;
1784 visit_type_size(v
, &value
, name
, &error
);
1786 error_propagate(errp
, error
);
1789 if (value
> (1ULL << 32)) {
1791 "Machine option 'max-ram-below-4g=%"PRIu64
1792 "' expects size less than or equal to 4G", value
);
1793 error_propagate(errp
, error
);
1797 if (value
< (1ULL << 20)) {
1798 error_report("Warning: small max_ram_below_4g(%"PRIu64
1799 ") less than 1M. BIOS may not work..",
1803 pcms
->max_ram_below_4g
= value
;
1806 static void pc_machine_get_vmport(Object
*obj
, Visitor
*v
, void *opaque
,
1807 const char *name
, Error
**errp
)
1809 PCMachineState
*pcms
= PC_MACHINE(obj
);
1810 OnOffAuto vmport
= pcms
->vmport
;
1812 visit_type_OnOffAuto(v
, &vmport
, name
, errp
);
1815 static void pc_machine_set_vmport(Object
*obj
, Visitor
*v
, void *opaque
,
1816 const char *name
, Error
**errp
)
1818 PCMachineState
*pcms
= PC_MACHINE(obj
);
1820 visit_type_OnOffAuto(v
, &pcms
->vmport
, name
, errp
);
1823 bool pc_machine_is_smm_enabled(PCMachineState
*pcms
)
1825 bool smm_available
= false;
1827 if (pcms
->smm
== ON_OFF_AUTO_OFF
) {
1831 if (tcg_enabled() || qtest_enabled()) {
1832 smm_available
= true;
1833 } else if (kvm_enabled()) {
1834 smm_available
= kvm_has_smm();
1837 if (smm_available
) {
1841 if (pcms
->smm
== ON_OFF_AUTO_ON
) {
1842 error_report("System Management Mode not supported by this hypervisor.");
1848 static void pc_machine_get_smm(Object
*obj
, Visitor
*v
, void *opaque
,
1849 const char *name
, Error
**errp
)
1851 PCMachineState
*pcms
= PC_MACHINE(obj
);
1852 OnOffAuto smm
= pcms
->smm
;
1854 visit_type_OnOffAuto(v
, &smm
, name
, errp
);
1857 static void pc_machine_set_smm(Object
*obj
, Visitor
*v
, void *opaque
,
1858 const char *name
, Error
**errp
)
1860 PCMachineState
*pcms
= PC_MACHINE(obj
);
1862 visit_type_OnOffAuto(v
, &pcms
->smm
, name
, errp
);
1865 static bool pc_machine_get_nvdimm(Object
*obj
, Error
**errp
)
1867 PCMachineState
*pcms
= PC_MACHINE(obj
);
1869 return pcms
->nvdimm
;
1872 static void pc_machine_set_nvdimm(Object
*obj
, bool value
, Error
**errp
)
1874 PCMachineState
*pcms
= PC_MACHINE(obj
);
1876 pcms
->nvdimm
= value
;
1879 static void pc_machine_initfn(Object
*obj
)
1881 PCMachineState
*pcms
= PC_MACHINE(obj
);
1883 object_property_add(obj
, PC_MACHINE_MEMHP_REGION_SIZE
, "int",
1884 pc_machine_get_hotplug_memory_region_size
,
1885 NULL
, NULL
, NULL
, &error_abort
);
1887 pcms
->max_ram_below_4g
= 1ULL << 32; /* 4G */
1888 object_property_add(obj
, PC_MACHINE_MAX_RAM_BELOW_4G
, "size",
1889 pc_machine_get_max_ram_below_4g
,
1890 pc_machine_set_max_ram_below_4g
,
1891 NULL
, NULL
, &error_abort
);
1892 object_property_set_description(obj
, PC_MACHINE_MAX_RAM_BELOW_4G
,
1893 "Maximum ram below the 4G boundary (32bit boundary)",
1896 pcms
->smm
= ON_OFF_AUTO_AUTO
;
1897 object_property_add(obj
, PC_MACHINE_SMM
, "OnOffAuto",
1900 NULL
, NULL
, &error_abort
);
1901 object_property_set_description(obj
, PC_MACHINE_SMM
,
1902 "Enable SMM (pc & q35)",
1905 pcms
->vmport
= ON_OFF_AUTO_AUTO
;
1906 object_property_add(obj
, PC_MACHINE_VMPORT
, "OnOffAuto",
1907 pc_machine_get_vmport
,
1908 pc_machine_set_vmport
,
1909 NULL
, NULL
, &error_abort
);
1910 object_property_set_description(obj
, PC_MACHINE_VMPORT
,
1911 "Enable vmport (pc & q35)",
1914 /* nvdimm is disabled on default. */
1915 pcms
->nvdimm
= false;
1916 object_property_add_bool(obj
, PC_MACHINE_NVDIMM
, pc_machine_get_nvdimm
,
1917 pc_machine_set_nvdimm
, &error_abort
);
1920 static void pc_machine_reset(void)
1925 qemu_devices_reset();
1927 /* Reset APIC after devices have been reset to cancel
1928 * any changes that qemu_devices_reset() might have done.
1933 if (cpu
->apic_state
) {
1934 device_reset(cpu
->apic_state
);
1939 static unsigned pc_cpu_index_to_socket_id(unsigned cpu_index
)
1941 X86CPUTopoInfo topo
;
1942 x86_topo_ids_from_idx(smp_cores
, smp_threads
, cpu_index
,
1947 static void pc_machine_class_init(ObjectClass
*oc
, void *data
)
1949 MachineClass
*mc
= MACHINE_CLASS(oc
);
1950 PCMachineClass
*pcmc
= PC_MACHINE_CLASS(oc
);
1951 HotplugHandlerClass
*hc
= HOTPLUG_HANDLER_CLASS(oc
);
1953 pcmc
->get_hotplug_handler
= mc
->get_hotplug_handler
;
1954 pcmc
->pci_enabled
= true;
1955 pcmc
->has_acpi_build
= true;
1956 pcmc
->rsdp_in_ram
= true;
1957 pcmc
->smbios_defaults
= true;
1958 pcmc
->smbios_uuid_encoded
= true;
1959 pcmc
->gigabyte_align
= true;
1960 pcmc
->has_reserved_memory
= true;
1961 pcmc
->kvmclock_enabled
= true;
1962 pcmc
->enforce_aligned_dimm
= true;
1963 /* BIOS ACPI tables: 128K. Other BIOS datastructures: less than 4K reported
1964 * to be used at the moment, 32K should be enough for a while. */
1965 pcmc
->acpi_data_size
= 0x20000 + 0x8000;
1966 mc
->get_hotplug_handler
= pc_get_hotpug_handler
;
1967 mc
->cpu_index_to_socket_id
= pc_cpu_index_to_socket_id
;
1968 mc
->default_boot_order
= "cad";
1969 mc
->hot_add_cpu
= pc_hot_add_cpu
;
1971 mc
->reset
= pc_machine_reset
;
1972 hc
->plug
= pc_machine_device_plug_cb
;
1973 hc
->unplug_request
= pc_machine_device_unplug_request_cb
;
1974 hc
->unplug
= pc_machine_device_unplug_cb
;
1977 static const TypeInfo pc_machine_info
= {
1978 .name
= TYPE_PC_MACHINE
,
1979 .parent
= TYPE_MACHINE
,
1981 .instance_size
= sizeof(PCMachineState
),
1982 .instance_init
= pc_machine_initfn
,
1983 .class_size
= sizeof(PCMachineClass
),
1984 .class_init
= pc_machine_class_init
,
1985 .interfaces
= (InterfaceInfo
[]) {
1986 { TYPE_HOTPLUG_HANDLER
},
1991 static void pc_machine_register_types(void)
1993 type_register_static(&pc_machine_info
);
1996 type_init(pc_machine_register_types
)