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/block/fdc.h"
30 #include "hw/pci/pci.h"
31 #include "monitor/monitor.h"
32 #include "hw/nvram/fw_cfg.h"
33 #include "hw/timer/hpet.h"
34 #include "hw/i386/smbios.h"
35 #include "hw/loader.h"
37 #include "multiboot.h"
38 #include "hw/timer/mc146818rtc.h"
39 #include "hw/timer/i8254.h"
40 #include "hw/audio/pcspk.h"
41 #include "hw/pci/msi.h"
42 #include "hw/sysbus.h"
43 #include "sysemu/sysemu.h"
44 #include "sysemu/kvm.h"
46 #include "hw/xen/xen.h"
47 #include "sysemu/blockdev.h"
48 #include "hw/block/block.h"
49 #include "ui/qemu-spice.h"
50 #include "exec/memory.h"
51 #include "exec/address-spaces.h"
52 #include "sysemu/arch_init.h"
53 #include "qemu/bitmap.h"
54 #include "qemu/config-file.h"
55 #include "hw/acpi/acpi.h"
56 #include "hw/cpu/icc_bus.h"
57 #include "hw/boards.h"
59 /* debug PC/ISA interrupts */
63 #define DPRINTF(fmt, ...) \
64 do { printf("CPUIRQ: " fmt , ## __VA_ARGS__); } while (0)
66 #define DPRINTF(fmt, ...)
69 /* Leave a chunk of memory at the top of RAM for the BIOS ACPI tables. */
70 #define ACPI_DATA_SIZE 0x10000
71 #define BIOS_CFG_IOPORT 0x510
72 #define FW_CFG_ACPI_TABLES (FW_CFG_ARCH_LOCAL + 0)
73 #define FW_CFG_SMBIOS_ENTRIES (FW_CFG_ARCH_LOCAL + 1)
74 #define FW_CFG_IRQ0_OVERRIDE (FW_CFG_ARCH_LOCAL + 2)
75 #define FW_CFG_E820_TABLE (FW_CFG_ARCH_LOCAL + 3)
76 #define FW_CFG_HPET (FW_CFG_ARCH_LOCAL + 4)
78 #define IO_APIC_DEFAULT_ADDRESS 0xfec00000
80 #define E820_NR_ENTRIES 16
86 } QEMU_PACKED __attribute((__aligned__(4)));
90 struct e820_entry entry[E820_NR_ENTRIES];
91 } QEMU_PACKED __attribute((__aligned__(4)));
93 static struct e820_table e820_table;
94 struct hpet_fw_config hpet_cfg = {.count = UINT8_MAX};
96 void gsi_handler(void *opaque, int n, int level)
100 DPRINTF("pc: %s GSI %d\n", level ? "raising" : "lowering", n);
101 if (n < ISA_NUM_IRQS) {
102 qemu_set_irq(s->i8259_irq[n], level);
104 qemu_set_irq(s->ioapic_irq[n], level);
107 static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
112 static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
114 return 0xffffffffffffffffULL;
117 /* MSDOS compatibility mode FPU exception support */
118 static qemu_irq ferr_irq;
120 void pc_register_ferr_irq(qemu_irq irq)
125 /* XXX: add IGNNE support */
126 void cpu_set_ferr(CPUX86State *s)
128 qemu_irq_raise(ferr_irq);
131 static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
134 qemu_irq_lower(ferr_irq);
137 static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
139 return 0xffffffffffffffffULL;
143 uint64_t cpu_get_tsc(CPUX86State *env)
145 return cpu_get_ticks();
150 static cpu_set_smm_t smm_set;
151 static void *smm_arg;
153 void cpu_smm_register(cpu_set_smm_t callback, void *arg)
155 assert(smm_set == NULL);
156 assert(smm_arg == NULL);
161 void cpu_smm_update(CPUX86State *env)
163 if (smm_set && smm_arg && env == first_cpu)
164 smm_set(!!(env->hflags & HF_SMM_MASK), smm_arg);
169 int cpu_get_pic_interrupt(CPUX86State *env)
173 intno = apic_get_interrupt(env->apic_state);
177 /* read the irq from the PIC */
178 if (!apic_accept_pic_intr(env->apic_state)) {
182 intno = pic_read_irq(isa_pic);
186 static void pic_irq_request(void *opaque, int irq, int level)
188 CPUX86State *env = first_cpu;
190 DPRINTF("pic_irqs: %s irq %d\n", level? "raise" : "lower", irq);
191 if (env->apic_state) {
193 if (apic_accept_pic_intr(env->apic_state)) {
194 apic_deliver_pic_intr(env->apic_state, level);
199 CPUState *cs = CPU(x86_env_get_cpu(env));
201 cpu_interrupt(cs, CPU_INTERRUPT_HARD);
203 cpu_reset_interrupt(cs, CPU_INTERRUPT_HARD);
208 /* PC cmos mappings */
210 #define REG_EQUIPMENT_BYTE 0x14
212 static int cmos_get_fd_drive_type(FDriveType fd0)
218 /* 1.44 Mb 3"5 drive */
222 /* 2.88 Mb 3"5 drive */
226 /* 1.2 Mb 5"5 drive */
229 case FDRIVE_DRV_NONE:
237 static void cmos_init_hd(ISADevice *s, int type_ofs, int info_ofs,
238 int16_t cylinders, int8_t heads, int8_t sectors)
240 rtc_set_memory(s, type_ofs, 47);
241 rtc_set_memory(s, info_ofs, cylinders);
242 rtc_set_memory(s, info_ofs + 1, cylinders >> 8);
243 rtc_set_memory(s, info_ofs + 2, heads);
244 rtc_set_memory(s, info_ofs + 3, 0xff);
245 rtc_set_memory(s, info_ofs + 4, 0xff);
246 rtc_set_memory(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
247 rtc_set_memory(s, info_ofs + 6, cylinders);
248 rtc_set_memory(s, info_ofs + 7, cylinders >> 8);
249 rtc_set_memory(s, info_ofs + 8, sectors);
252 /* convert boot_device letter to something recognizable by the bios */
253 static int boot_device2nibble(char boot_device)
255 switch(boot_device) {
258 return 0x01; /* floppy boot */
260 return 0x02; /* hard drive boot */
262 return 0x03; /* CD-ROM boot */
264 return 0x04; /* Network boot */
269 static int set_boot_dev(ISADevice *s, const char *boot_device, int fd_bootchk)
271 #define PC_MAX_BOOT_DEVICES 3
272 int nbds, bds[3] = { 0, };
275 nbds = strlen(boot_device);
276 if (nbds > PC_MAX_BOOT_DEVICES) {
277 error_report("Too many boot devices for PC");
280 for (i = 0; i < nbds; i++) {
281 bds[i] = boot_device2nibble(boot_device[i]);
283 error_report("Invalid boot device for PC: '%c'",
288 rtc_set_memory(s, 0x3d, (bds[1] << 4) | bds[0]);
289 rtc_set_memory(s, 0x38, (bds[2] << 4) | (fd_bootchk ? 0x0 : 0x1));
293 static int pc_boot_set(void *opaque, const char *boot_device)
295 return set_boot_dev(opaque, boot_device, 0);
298 typedef struct pc_cmos_init_late_arg {
299 ISADevice *rtc_state;
301 } pc_cmos_init_late_arg;
303 static void pc_cmos_init_late(void *opaque)
305 pc_cmos_init_late_arg *arg = opaque;
306 ISADevice *s = arg->rtc_state;
308 int8_t heads, sectors;
313 if (ide_get_geometry(arg->idebus[0], 0,
314 &cylinders, &heads, §ors) >= 0) {
315 cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
318 if (ide_get_geometry(arg->idebus[0], 1,
319 &cylinders, &heads, §ors) >= 0) {
320 cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
323 rtc_set_memory(s, 0x12, val);
326 for (i = 0; i < 4; i++) {
327 /* NOTE: ide_get_geometry() returns the physical
328 geometry. It is always such that: 1 <= sects <= 63, 1
329 <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
330 geometry can be different if a translation is done. */
331 if (ide_get_geometry(arg->idebus[i / 2], i % 2,
332 &cylinders, &heads, §ors) >= 0) {
333 trans = ide_get_bios_chs_trans(arg->idebus[i / 2], i % 2) - 1;
334 assert((trans & ~3) == 0);
335 val |= trans << (i * 2);
338 rtc_set_memory(s, 0x39, val);
340 qemu_unregister_reset(pc_cmos_init_late, opaque);
343 typedef struct RTCCPUHotplugArg {
344 Notifier cpu_added_notifier;
345 ISADevice *rtc_state;
348 static void rtc_notify_cpu_added(Notifier *notifier, void *data)
350 RTCCPUHotplugArg *arg = container_of(notifier, RTCCPUHotplugArg,
352 ISADevice *s = arg->rtc_state;
354 /* increment the number of CPUs */
355 rtc_set_memory(s, 0x5f, rtc_get_memory(s, 0x5f) + 1);
358 void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
359 const char *boot_device,
360 ISADevice *floppy, BusState *idebus0, BusState *idebus1,
364 FDriveType fd_type[2] = { FDRIVE_DRV_NONE, FDRIVE_DRV_NONE };
365 static pc_cmos_init_late_arg arg;
366 static RTCCPUHotplugArg cpu_hotplug_cb;
368 /* various important CMOS locations needed by PC/Bochs bios */
371 /* base memory (first MiB) */
372 val = MIN(ram_size / 1024, 640);
373 rtc_set_memory(s, 0x15, val);
374 rtc_set_memory(s, 0x16, val >> 8);
375 /* extended memory (next 64MiB) */
376 if (ram_size > 1024 * 1024) {
377 val = (ram_size - 1024 * 1024) / 1024;
383 rtc_set_memory(s, 0x17, val);
384 rtc_set_memory(s, 0x18, val >> 8);
385 rtc_set_memory(s, 0x30, val);
386 rtc_set_memory(s, 0x31, val >> 8);
387 /* memory between 16MiB and 4GiB */
388 if (ram_size > 16 * 1024 * 1024) {
389 val = (ram_size - 16 * 1024 * 1024) / 65536;
395 rtc_set_memory(s, 0x34, val);
396 rtc_set_memory(s, 0x35, val >> 8);
397 /* memory above 4GiB */
398 val = above_4g_mem_size / 65536;
399 rtc_set_memory(s, 0x5b, val);
400 rtc_set_memory(s, 0x5c, val >> 8);
401 rtc_set_memory(s, 0x5d, val >> 16);
403 /* set the number of CPU */
404 rtc_set_memory(s, 0x5f, smp_cpus - 1);
405 /* init CPU hotplug notifier */
406 cpu_hotplug_cb.rtc_state = s;
407 cpu_hotplug_cb.cpu_added_notifier.notify = rtc_notify_cpu_added;
408 qemu_register_cpu_added_notifier(&cpu_hotplug_cb.cpu_added_notifier);
410 /* set boot devices, and disable floppy signature check if requested */
411 if (set_boot_dev(s, boot_device, fd_bootchk)) {
417 for (i = 0; i < 2; i++) {
418 fd_type[i] = isa_fdc_get_drive_type(floppy, i);
421 val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
422 cmos_get_fd_drive_type(fd_type[1]);
423 rtc_set_memory(s, 0x10, val);
427 if (fd_type[0] < FDRIVE_DRV_NONE) {
430 if (fd_type[1] < FDRIVE_DRV_NONE) {
437 val |= 0x01; /* 1 drive, ready for boot */
440 val |= 0x41; /* 2 drives, ready for boot */
443 val |= 0x02; /* FPU is there */
444 val |= 0x04; /* PS/2 mouse installed */
445 rtc_set_memory(s, REG_EQUIPMENT_BYTE, val);
449 arg.idebus[0] = idebus0;
450 arg.idebus[1] = idebus1;
451 qemu_register_reset(pc_cmos_init_late, &arg);
454 #define TYPE_PORT92 "port92"
455 #define PORT92(obj) OBJECT_CHECK(Port92State, (obj), TYPE_PORT92)
457 /* port 92 stuff: could be split off */
458 typedef struct Port92State {
459 ISADevice parent_obj;
466 static void port92_write(void *opaque, hwaddr addr, uint64_t val,
469 Port92State *s = opaque;
471 DPRINTF("port92: write 0x%02x\n", val);
473 qemu_set_irq(*s->a20_out, (val >> 1) & 1);
475 qemu_system_reset_request();
479 static uint64_t port92_read(void *opaque, hwaddr addr,
482 Port92State *s = opaque;
486 DPRINTF("port92: read 0x%02x\n", ret);
490 static void port92_init(ISADevice *dev, qemu_irq *a20_out)
492 Port92State *s = PORT92(dev);
494 s->a20_out = a20_out;
497 static const VMStateDescription vmstate_port92_isa = {
500 .minimum_version_id = 1,
501 .minimum_version_id_old = 1,
502 .fields = (VMStateField []) {
503 VMSTATE_UINT8(outport, Port92State),
504 VMSTATE_END_OF_LIST()
508 static void port92_reset(DeviceState *d)
510 Port92State *s = PORT92(d);
515 static const MemoryRegionOps port92_ops = {
517 .write = port92_write,
519 .min_access_size = 1,
520 .max_access_size = 1,
522 .endianness = DEVICE_LITTLE_ENDIAN,
525 static void port92_initfn(Object *obj)
527 Port92State *s = PORT92(obj);
529 memory_region_init_io(&s->io, &port92_ops, s, "port92", 1);
534 static void port92_realizefn(DeviceState *dev, Error **errp)
536 ISADevice *isadev = ISA_DEVICE(dev);
537 Port92State *s = PORT92(dev);
539 isa_register_ioport(isadev, &s->io, 0x92);
542 static void port92_class_initfn(ObjectClass *klass, void *data)
544 DeviceClass *dc = DEVICE_CLASS(klass);
547 dc->realize = port92_realizefn;
548 dc->reset = port92_reset;
549 dc->vmsd = &vmstate_port92_isa;
552 static const TypeInfo port92_info = {
554 .parent = TYPE_ISA_DEVICE,
555 .instance_size = sizeof(Port92State),
556 .instance_init = port92_initfn,
557 .class_init = port92_class_initfn,
560 static void port92_register_types(void)
562 type_register_static(&port92_info);
565 type_init(port92_register_types)
567 static void handle_a20_line_change(void *opaque, int irq, int level)
569 X86CPU *cpu = opaque;
571 /* XXX: send to all CPUs ? */
572 /* XXX: add logic to handle multiple A20 line sources */
573 x86_cpu_set_a20(cpu, level);
576 int e820_add_entry(uint64_t address, uint64_t length, uint32_t type)
578 int index = le32_to_cpu(e820_table.count);
579 struct e820_entry *entry;
581 if (index >= E820_NR_ENTRIES)
583 entry = &e820_table.entry[index++];
585 entry->address = cpu_to_le64(address);
586 entry->length = cpu_to_le64(length);
587 entry->type = cpu_to_le32(type);
589 e820_table.count = cpu_to_le32(index);
593 /* Calculates the limit to CPU APIC ID values
595 * This function returns the limit for the APIC ID value, so that all
596 * CPU APIC IDs are < pc_apic_id_limit().
598 * This is used for FW_CFG_MAX_CPUS. See comments on bochs_bios_init().
600 static unsigned int pc_apic_id_limit(unsigned int max_cpus)
602 return x86_cpu_apic_id_from_index(max_cpus - 1) + 1;
605 static FWCfgState *bochs_bios_init(void)
608 uint8_t *smbios_table;
610 uint64_t *numa_fw_cfg;
612 unsigned int apic_id_limit = pc_apic_id_limit(max_cpus);
614 fw_cfg = fw_cfg_init(BIOS_CFG_IOPORT, BIOS_CFG_IOPORT + 1, 0, 0);
615 /* FW_CFG_MAX_CPUS is a bit confusing/problematic on x86:
617 * SeaBIOS needs FW_CFG_MAX_CPUS for CPU hotplug, but the CPU hotplug
618 * QEMU<->SeaBIOS interface is not based on the "CPU index", but on the APIC
619 * ID of hotplugged CPUs[1]. This means that FW_CFG_MAX_CPUS is not the
620 * "maximum number of CPUs", but the "limit to the APIC ID values SeaBIOS
623 * So, this means we must not use max_cpus, here, but the maximum possible
624 * APIC ID value, plus one.
626 * [1] The only kind of "CPU identifier" used between SeaBIOS and QEMU is
627 * the APIC ID, not the "CPU index"
629 fw_cfg_add_i16(fw_cfg, FW_CFG_MAX_CPUS, (uint16_t)apic_id_limit);
630 fw_cfg_add_i32(fw_cfg, FW_CFG_ID, 1);
631 fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, (uint64_t)ram_size);
632 fw_cfg_add_bytes(fw_cfg, FW_CFG_ACPI_TABLES,
633 acpi_tables, acpi_tables_len);
634 fw_cfg_add_i32(fw_cfg, FW_CFG_IRQ0_OVERRIDE, kvm_allows_irq0_override());
636 smbios_table = smbios_get_table(&smbios_len);
638 fw_cfg_add_bytes(fw_cfg, FW_CFG_SMBIOS_ENTRIES,
639 smbios_table, smbios_len);
640 fw_cfg_add_bytes(fw_cfg, FW_CFG_E820_TABLE,
641 &e820_table, sizeof(e820_table));
643 fw_cfg_add_bytes(fw_cfg, FW_CFG_HPET, &hpet_cfg, sizeof(hpet_cfg));
644 /* allocate memory for the NUMA channel: one (64bit) word for the number
645 * of nodes, one word for each VCPU->node and one word for each node to
646 * hold the amount of memory.
648 numa_fw_cfg = g_new0(uint64_t, 1 + apic_id_limit + nb_numa_nodes);
649 numa_fw_cfg[0] = cpu_to_le64(nb_numa_nodes);
650 for (i = 0; i < max_cpus; i++) {
651 unsigned int apic_id = x86_cpu_apic_id_from_index(i);
652 assert(apic_id < apic_id_limit);
653 for (j = 0; j < nb_numa_nodes; j++) {
654 if (test_bit(i, node_cpumask[j])) {
655 numa_fw_cfg[apic_id + 1] = cpu_to_le64(j);
660 for (i = 0; i < nb_numa_nodes; i++) {
661 numa_fw_cfg[apic_id_limit + 1 + i] = cpu_to_le64(node_mem[i]);
663 fw_cfg_add_bytes(fw_cfg, FW_CFG_NUMA, numa_fw_cfg,
664 (1 + apic_id_limit + nb_numa_nodes) *
665 sizeof(*numa_fw_cfg));
670 static long get_file_size(FILE *f)
674 /* XXX: on Unix systems, using fstat() probably makes more sense */
677 fseek(f, 0, SEEK_END);
679 fseek(f, where, SEEK_SET);
684 static void load_linux(FWCfgState *fw_cfg,
685 const char *kernel_filename,
686 const char *initrd_filename,
687 const char *kernel_cmdline,
691 int setup_size, kernel_size, initrd_size = 0, cmdline_size;
693 uint8_t header[8192], *setup, *kernel, *initrd_data;
694 hwaddr real_addr, prot_addr, cmdline_addr, initrd_addr = 0;
698 /* Align to 16 bytes as a paranoia measure */
699 cmdline_size = (strlen(kernel_cmdline)+16) & ~15;
701 /* load the kernel header */
702 f = fopen(kernel_filename, "rb");
703 if (!f || !(kernel_size = get_file_size(f)) ||
704 fread(header, 1, MIN(ARRAY_SIZE(header), kernel_size), f) !=
705 MIN(ARRAY_SIZE(header), kernel_size)) {
706 fprintf(stderr, "qemu: could not load kernel '%s': %s\n",
707 kernel_filename, strerror(errno));
711 /* kernel protocol version */
713 fprintf(stderr, "header magic: %#x\n", ldl_p(header+0x202));
715 if (ldl_p(header+0x202) == 0x53726448) {
716 protocol = lduw_p(header+0x206);
718 /* This looks like a multiboot kernel. If it is, let's stop
719 treating it like a Linux kernel. */
720 if (load_multiboot(fw_cfg, f, kernel_filename, initrd_filename,
721 kernel_cmdline, kernel_size, header)) {
727 if (protocol < 0x200 || !(header[0x211] & 0x01)) {
730 cmdline_addr = 0x9a000 - cmdline_size;
732 } else if (protocol < 0x202) {
733 /* High but ancient kernel */
735 cmdline_addr = 0x9a000 - cmdline_size;
736 prot_addr = 0x100000;
738 /* High and recent kernel */
740 cmdline_addr = 0x20000;
741 prot_addr = 0x100000;
746 "qemu: real_addr = 0x" TARGET_FMT_plx "\n"
747 "qemu: cmdline_addr = 0x" TARGET_FMT_plx "\n"
748 "qemu: prot_addr = 0x" TARGET_FMT_plx "\n",
754 /* highest address for loading the initrd */
755 if (protocol >= 0x203) {
756 initrd_max = ldl_p(header+0x22c);
758 initrd_max = 0x37ffffff;
761 if (initrd_max >= max_ram_size-ACPI_DATA_SIZE)
762 initrd_max = max_ram_size-ACPI_DATA_SIZE-1;
764 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_ADDR, cmdline_addr);
765 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_SIZE, strlen(kernel_cmdline)+1);
766 fw_cfg_add_string(fw_cfg, FW_CFG_CMDLINE_DATA, kernel_cmdline);
768 if (protocol >= 0x202) {
769 stl_p(header+0x228, cmdline_addr);
771 stw_p(header+0x20, 0xA33F);
772 stw_p(header+0x22, cmdline_addr-real_addr);
775 /* handle vga= parameter */
776 vmode = strstr(kernel_cmdline, "vga=");
778 unsigned int video_mode;
781 if (!strncmp(vmode, "normal", 6)) {
783 } else if (!strncmp(vmode, "ext", 3)) {
785 } else if (!strncmp(vmode, "ask", 3)) {
788 video_mode = strtol(vmode, NULL, 0);
790 stw_p(header+0x1fa, video_mode);
794 /* High nybble = B reserved for QEMU; low nybble is revision number.
795 If this code is substantially changed, you may want to consider
796 incrementing the revision. */
797 if (protocol >= 0x200) {
798 header[0x210] = 0xB0;
801 if (protocol >= 0x201) {
802 header[0x211] |= 0x80; /* CAN_USE_HEAP */
803 stw_p(header+0x224, cmdline_addr-real_addr-0x200);
807 if (initrd_filename) {
808 if (protocol < 0x200) {
809 fprintf(stderr, "qemu: linux kernel too old to load a ram disk\n");
813 initrd_size = get_image_size(initrd_filename);
814 if (initrd_size < 0) {
815 fprintf(stderr, "qemu: error reading initrd %s\n",
820 initrd_addr = (initrd_max-initrd_size) & ~4095;
822 initrd_data = g_malloc(initrd_size);
823 load_image(initrd_filename, initrd_data);
825 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_ADDR, initrd_addr);
826 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_SIZE, initrd_size);
827 fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, initrd_data, initrd_size);
829 stl_p(header+0x218, initrd_addr);
830 stl_p(header+0x21c, initrd_size);
833 /* load kernel and setup */
834 setup_size = header[0x1f1];
835 if (setup_size == 0) {
838 setup_size = (setup_size+1)*512;
839 kernel_size -= setup_size;
841 setup = g_malloc(setup_size);
842 kernel = g_malloc(kernel_size);
843 fseek(f, 0, SEEK_SET);
844 if (fread(setup, 1, setup_size, f) != setup_size) {
845 fprintf(stderr, "fread() failed\n");
848 if (fread(kernel, 1, kernel_size, f) != kernel_size) {
849 fprintf(stderr, "fread() failed\n");
853 memcpy(setup, header, MIN(sizeof(header), setup_size));
855 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_ADDR, prot_addr);
856 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_SIZE, kernel_size);
857 fw_cfg_add_bytes(fw_cfg, FW_CFG_KERNEL_DATA, kernel, kernel_size);
859 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_ADDR, real_addr);
860 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_SIZE, setup_size);
861 fw_cfg_add_bytes(fw_cfg, FW_CFG_SETUP_DATA, setup, setup_size);
863 option_rom[nb_option_roms].name = "linuxboot.bin";
864 option_rom[nb_option_roms].bootindex = 0;
868 #define NE2000_NB_MAX 6
870 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
872 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
874 static const int parallel_io[MAX_PARALLEL_PORTS] = { 0x378, 0x278, 0x3bc };
875 static const int parallel_irq[MAX_PARALLEL_PORTS] = { 7, 7, 7 };
877 void pc_init_ne2k_isa(ISABus *bus, NICInfo *nd)
879 static int nb_ne2k = 0;
881 if (nb_ne2k == NE2000_NB_MAX)
883 isa_ne2000_init(bus, ne2000_io[nb_ne2k],
884 ne2000_irq[nb_ne2k], nd);
888 DeviceState *cpu_get_current_apic(void)
890 if (cpu_single_env) {
891 return cpu_single_env->apic_state;
897 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
899 X86CPU *cpu = opaque;
902 cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
906 static X86CPU *pc_new_cpu(const char *cpu_model, int64_t apic_id,
907 DeviceState *icc_bridge, Error **errp)
910 Error *local_err = NULL;
912 cpu = cpu_x86_create(cpu_model, icc_bridge, errp);
917 object_property_set_int(OBJECT(cpu), apic_id, "apic-id", &local_err);
918 object_property_set_bool(OBJECT(cpu), true, "realized", &local_err);
922 object_unref(OBJECT(cpu));
925 error_propagate(errp, local_err);
930 static const char *current_cpu_model;
932 void pc_hot_add_cpu(const int64_t id, Error **errp)
934 DeviceState *icc_bridge;
935 int64_t apic_id = x86_cpu_apic_id_from_index(id);
937 if (cpu_exists(apic_id)) {
938 error_setg(errp, "Unable to add CPU: %" PRIi64
939 ", it already exists", id);
943 if (id >= max_cpus) {
944 error_setg(errp, "Unable to add CPU: %" PRIi64
945 ", max allowed: %d", id, max_cpus - 1);
949 icc_bridge = DEVICE(object_resolve_path_type("icc-bridge",
950 TYPE_ICC_BRIDGE, NULL));
951 pc_new_cpu(current_cpu_model, apic_id, icc_bridge, errp);
954 void pc_cpus_init(const char *cpu_model, DeviceState *icc_bridge)
961 if (cpu_model == NULL) {
963 cpu_model = "qemu64";
965 cpu_model = "qemu32";
968 current_cpu_model = cpu_model;
970 for (i = 0; i < smp_cpus; i++) {
971 cpu = pc_new_cpu(cpu_model, x86_cpu_apic_id_from_index(i),
974 fprintf(stderr, "%s\n", error_get_pretty(error));
980 /* map APIC MMIO area if CPU has APIC */
981 if (cpu && cpu->env.apic_state) {
982 /* XXX: what if the base changes? */
983 sysbus_mmio_map_overlap(SYS_BUS_DEVICE(icc_bridge), 0,
984 APIC_DEFAULT_ADDRESS, 0x1000);
988 void pc_acpi_init(const char *default_dsdt)
992 if (acpi_tables != NULL) {
993 /* manually set via -acpitable, leave it alone */
997 filename = qemu_find_file(QEMU_FILE_TYPE_BIOS, default_dsdt);
998 if (filename == NULL) {
999 fprintf(stderr, "WARNING: failed to find %s\n", default_dsdt);
1005 arg = g_strdup_printf("file=%s", filename);
1007 /* creates a deep copy of "arg" */
1008 opts = qemu_opts_parse(qemu_find_opts("acpi"), arg, 0);
1009 g_assert(opts != NULL);
1011 acpi_table_add(opts, &err);
1013 fprintf(stderr, "WARNING: failed to load %s: %s\n", filename,
1014 error_get_pretty(err));
1022 FWCfgState *pc_memory_init(MemoryRegion *system_memory,
1023 const char *kernel_filename,
1024 const char *kernel_cmdline,
1025 const char *initrd_filename,
1026 ram_addr_t below_4g_mem_size,
1027 ram_addr_t above_4g_mem_size,
1028 MemoryRegion *rom_memory,
1029 MemoryRegion **ram_memory)
1032 MemoryRegion *ram, *option_rom_mr;
1033 MemoryRegion *ram_below_4g, *ram_above_4g;
1036 linux_boot = (kernel_filename != NULL);
1038 /* Allocate RAM. We allocate it as a single memory region and use
1039 * aliases to address portions of it, mostly for backwards compatibility
1040 * with older qemus that used qemu_ram_alloc().
1042 ram = g_malloc(sizeof(*ram));
1043 memory_region_init_ram(ram, "pc.ram",
1044 below_4g_mem_size + above_4g_mem_size);
1045 vmstate_register_ram_global(ram);
1047 ram_below_4g = g_malloc(sizeof(*ram_below_4g));
1048 memory_region_init_alias(ram_below_4g, "ram-below-4g", ram,
1049 0, below_4g_mem_size);
1050 memory_region_add_subregion(system_memory, 0, ram_below_4g);
1051 if (above_4g_mem_size > 0) {
1052 ram_above_4g = g_malloc(sizeof(*ram_above_4g));
1053 memory_region_init_alias(ram_above_4g, "ram-above-4g", ram,
1054 below_4g_mem_size, above_4g_mem_size);
1055 memory_region_add_subregion(system_memory, 0x100000000ULL,
1060 /* Initialize PC system firmware */
1061 pc_system_firmware_init(rom_memory);
1063 option_rom_mr = g_malloc(sizeof(*option_rom_mr));
1064 memory_region_init_ram(option_rom_mr, "pc.rom", PC_ROM_SIZE);
1065 vmstate_register_ram_global(option_rom_mr);
1066 memory_region_add_subregion_overlap(rom_memory,
1071 fw_cfg = bochs_bios_init();
1075 load_linux(fw_cfg, kernel_filename, initrd_filename, kernel_cmdline, below_4g_mem_size);
1078 for (i = 0; i < nb_option_roms; i++) {
1079 rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1084 qemu_irq *pc_allocate_cpu_irq(void)
1086 return qemu_allocate_irqs(pic_irq_request, NULL, 1);
1089 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1091 DeviceState *dev = NULL;
1094 PCIDevice *pcidev = pci_vga_init(pci_bus);
1095 dev = pcidev ? &pcidev->qdev : NULL;
1096 } else if (isa_bus) {
1097 ISADevice *isadev = isa_vga_init(isa_bus);
1098 dev = isadev ? &isadev->qdev : NULL;
1103 static void cpu_request_exit(void *opaque, int irq, int level)
1105 CPUX86State *env = cpu_single_env;
1112 static const MemoryRegionOps ioport80_io_ops = {
1113 .write = ioport80_write,
1114 .read = ioport80_read,
1115 .endianness = DEVICE_NATIVE_ENDIAN,
1117 .min_access_size = 1,
1118 .max_access_size = 1,
1122 static const MemoryRegionOps ioportF0_io_ops = {
1123 .write = ioportF0_write,
1124 .read = ioportF0_read,
1125 .endianness = DEVICE_NATIVE_ENDIAN,
1127 .min_access_size = 1,
1128 .max_access_size = 1,
1132 void pc_basic_device_init(ISABus *isa_bus, qemu_irq *gsi,
1133 ISADevice **rtc_state,
1138 DriveInfo *fd[MAX_FD];
1139 DeviceState *hpet = NULL;
1140 int pit_isa_irq = 0;
1141 qemu_irq pit_alt_irq = NULL;
1142 qemu_irq rtc_irq = NULL;
1144 ISADevice *i8042, *port92, *vmmouse, *pit = NULL;
1145 qemu_irq *cpu_exit_irq;
1146 MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1147 MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1149 memory_region_init_io(ioport80_io, &ioport80_io_ops, NULL, "ioport80", 1);
1150 memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1152 memory_region_init_io(ioportF0_io, &ioportF0_io_ops, NULL, "ioportF0", 1);
1153 memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1156 * Check if an HPET shall be created.
1158 * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
1159 * when the HPET wants to take over. Thus we have to disable the latter.
1161 if (!no_hpet && (!kvm_irqchip_in_kernel() || kvm_has_pit_state2())) {
1162 hpet = sysbus_try_create_simple("hpet", HPET_BASE, NULL);
1165 for (i = 0; i < GSI_NUM_PINS; i++) {
1166 sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
1169 pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1170 rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1173 *rtc_state = rtc_init(isa_bus, 2000, rtc_irq);
1175 qemu_register_boot_set(pc_boot_set, *rtc_state);
1177 if (!xen_enabled()) {
1178 if (kvm_irqchip_in_kernel()) {
1179 pit = kvm_pit_init(isa_bus, 0x40);
1181 pit = pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1184 /* connect PIT to output control line of the HPET */
1185 qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(&pit->qdev, 0));
1187 pcspk_init(isa_bus, pit);
1190 for(i = 0; i < MAX_SERIAL_PORTS; i++) {
1191 if (serial_hds[i]) {
1192 serial_isa_init(isa_bus, i, serial_hds[i]);
1196 for(i = 0; i < MAX_PARALLEL_PORTS; i++) {
1197 if (parallel_hds[i]) {
1198 parallel_init(isa_bus, i, parallel_hds[i]);
1202 a20_line = qemu_allocate_irqs(handle_a20_line_change,
1203 x86_env_get_cpu(first_cpu), 2);
1204 i8042 = isa_create_simple(isa_bus, "i8042");
1205 i8042_setup_a20_line(i8042, &a20_line[0]);
1207 vmport_init(isa_bus);
1208 vmmouse = isa_try_create(isa_bus, "vmmouse");
1213 qdev_prop_set_ptr(&vmmouse->qdev, "ps2_mouse", i8042);
1214 qdev_init_nofail(&vmmouse->qdev);
1216 port92 = isa_create_simple(isa_bus, "port92");
1217 port92_init(port92, &a20_line[1]);
1219 cpu_exit_irq = qemu_allocate_irqs(cpu_request_exit, NULL, 1);
1220 DMA_init(0, cpu_exit_irq);
1222 for(i = 0; i < MAX_FD; i++) {
1223 fd[i] = drive_get(IF_FLOPPY, 0, i);
1225 *floppy = fdctrl_init_isa(isa_bus, fd);
1228 void pc_nic_init(ISABus *isa_bus, PCIBus *pci_bus)
1232 for (i = 0; i < nb_nics; i++) {
1233 NICInfo *nd = &nd_table[i];
1235 if (!pci_bus || (nd->model && strcmp(nd->model, "ne2k_isa") == 0)) {
1236 pc_init_ne2k_isa(isa_bus, nd);
1238 pci_nic_init_nofail(nd, "e1000", NULL);
1243 void pc_pci_device_init(PCIBus *pci_bus)
1248 max_bus = drive_get_max_bus(IF_SCSI);
1249 for (bus = 0; bus <= max_bus; bus++) {
1250 pci_create_simple(pci_bus, -1, "lsi53c895a");
1254 void ioapic_init_gsi(GSIState *gsi_state, const char *parent_name)
1260 if (kvm_irqchip_in_kernel()) {
1261 dev = qdev_create(NULL, "kvm-ioapic");
1263 dev = qdev_create(NULL, "ioapic");
1266 object_property_add_child(object_resolve_path(parent_name, NULL),
1267 "ioapic", OBJECT(dev), NULL);
1269 qdev_init_nofail(dev);
1270 d = SYS_BUS_DEVICE(dev);
1271 sysbus_mmio_map(d, 0, IO_APIC_DEFAULT_ADDRESS);
1273 for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1274 gsi_state->ioapic_irq[i] = qdev_get_gpio_in(dev, i);