xref: /openbmc/qemu/hw/i386/pc.c (revision 5ba03e2dd785362026917e4cc8a1fd2c64e8e62c)
1 /*
2  * QEMU PC System Emulator
3  *
4  * Copyright (c) 2003-2004 Fabrice Bellard
5  *
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:
12  *
13  * The above copyright notice and this permission notice shall be included in
14  * all copies or substantial portions of the Software.
15  *
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
22  * THE SOFTWARE.
23  */
24 #include "hw/hw.h"
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"
31 #include "hw/ide.h"
32 #include "hw/pci/pci.h"
33 #include "hw/pci/pci_bus.h"
34 #include "monitor/monitor.h"
35 #include "hw/nvram/fw_cfg.h"
36 #include "hw/timer/hpet.h"
37 #include "hw/i386/smbios.h"
38 #include "hw/loader.h"
39 #include "elf.h"
40 #include "multiboot.h"
41 #include "hw/timer/mc146818rtc.h"
42 #include "hw/timer/i8254.h"
43 #include "hw/audio/pcspk.h"
44 #include "hw/pci/msi.h"
45 #include "hw/sysbus.h"
46 #include "sysemu/sysemu.h"
47 #include "sysemu/numa.h"
48 #include "sysemu/kvm.h"
49 #include "sysemu/qtest.h"
50 #include "kvm_i386.h"
51 #include "hw/xen/xen.h"
52 #include "sysemu/block-backend.h"
53 #include "hw/block/block.h"
54 #include "ui/qemu-spice.h"
55 #include "exec/memory.h"
56 #include "exec/address-spaces.h"
57 #include "sysemu/arch_init.h"
58 #include "qemu/bitmap.h"
59 #include "qemu/config-file.h"
60 #include "hw/acpi/acpi.h"
61 #include "hw/acpi/cpu_hotplug.h"
62 #include "hw/cpu/icc_bus.h"
63 #include "hw/boards.h"
64 #include "hw/pci/pci_host.h"
65 #include "acpi-build.h"
66 #include "hw/mem/pc-dimm.h"
67 #include "trace.h"
68 #include "qapi/visitor.h"
69 #include "qapi-visit.h"
70 
71 /* debug PC/ISA interrupts */
72 //#define DEBUG_IRQ
73 
74 #ifdef DEBUG_IRQ
75 #define DPRINTF(fmt, ...)                                       \
76     do { printf("CPUIRQ: " fmt , ## __VA_ARGS__); } while (0)
77 #else
78 #define DPRINTF(fmt, ...)
79 #endif
80 
81 /* Leave a chunk of memory at the top of RAM for the BIOS ACPI tables
82  * (128K) and other BIOS datastructures (less than 4K reported to be used at
83  * the moment, 32K should be enough for a while).  */
84 static unsigned acpi_data_size = 0x20000 + 0x8000;
85 void pc_set_legacy_acpi_data_size(void)
86 {
87     acpi_data_size = 0x10000;
88 }
89 
90 #define BIOS_CFG_IOPORT 0x510
91 #define FW_CFG_ACPI_TABLES (FW_CFG_ARCH_LOCAL + 0)
92 #define FW_CFG_SMBIOS_ENTRIES (FW_CFG_ARCH_LOCAL + 1)
93 #define FW_CFG_IRQ0_OVERRIDE (FW_CFG_ARCH_LOCAL + 2)
94 #define FW_CFG_E820_TABLE (FW_CFG_ARCH_LOCAL + 3)
95 #define FW_CFG_HPET (FW_CFG_ARCH_LOCAL + 4)
96 
97 #define E820_NR_ENTRIES		16
98 
99 struct e820_entry {
100     uint64_t address;
101     uint64_t length;
102     uint32_t type;
103 } QEMU_PACKED __attribute((__aligned__(4)));
104 
105 struct e820_table {
106     uint32_t count;
107     struct e820_entry entry[E820_NR_ENTRIES];
108 } QEMU_PACKED __attribute((__aligned__(4)));
109 
110 static struct e820_table e820_reserve;
111 static struct e820_entry *e820_table;
112 static unsigned e820_entries;
113 struct hpet_fw_config hpet_cfg = {.count = UINT8_MAX};
114 
115 void gsi_handler(void *opaque, int n, int level)
116 {
117     GSIState *s = opaque;
118 
119     DPRINTF("pc: %s GSI %d\n", level ? "raising" : "lowering", n);
120     if (n < ISA_NUM_IRQS) {
121         qemu_set_irq(s->i8259_irq[n], level);
122     }
123     qemu_set_irq(s->ioapic_irq[n], level);
124 }
125 
126 static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
127                            unsigned size)
128 {
129 }
130 
131 static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
132 {
133     return 0xffffffffffffffffULL;
134 }
135 
136 /* MSDOS compatibility mode FPU exception support */
137 static qemu_irq ferr_irq;
138 
139 void pc_register_ferr_irq(qemu_irq irq)
140 {
141     ferr_irq = irq;
142 }
143 
144 /* XXX: add IGNNE support */
145 void cpu_set_ferr(CPUX86State *s)
146 {
147     qemu_irq_raise(ferr_irq);
148 }
149 
150 static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
151                            unsigned size)
152 {
153     qemu_irq_lower(ferr_irq);
154 }
155 
156 static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
157 {
158     return 0xffffffffffffffffULL;
159 }
160 
161 /* TSC handling */
162 uint64_t cpu_get_tsc(CPUX86State *env)
163 {
164     return cpu_get_ticks();
165 }
166 
167 /* SMM support */
168 
169 static cpu_set_smm_t smm_set;
170 static void *smm_arg;
171 
172 void cpu_smm_register(cpu_set_smm_t callback, void *arg)
173 {
174     assert(smm_set == NULL);
175     assert(smm_arg == NULL);
176     smm_set = callback;
177     smm_arg = arg;
178 }
179 
180 void cpu_smm_update(CPUX86State *env)
181 {
182     if (smm_set && smm_arg && CPU(x86_env_get_cpu(env)) == first_cpu) {
183         smm_set(!!(env->hflags & HF_SMM_MASK), smm_arg);
184     }
185 }
186 
187 
188 /* IRQ handling */
189 int cpu_get_pic_interrupt(CPUX86State *env)
190 {
191     X86CPU *cpu = x86_env_get_cpu(env);
192     int intno;
193 
194     intno = apic_get_interrupt(cpu->apic_state);
195     if (intno >= 0) {
196         return intno;
197     }
198     /* read the irq from the PIC */
199     if (!apic_accept_pic_intr(cpu->apic_state)) {
200         return -1;
201     }
202 
203     intno = pic_read_irq(isa_pic);
204     return intno;
205 }
206 
207 static void pic_irq_request(void *opaque, int irq, int level)
208 {
209     CPUState *cs = first_cpu;
210     X86CPU *cpu = X86_CPU(cs);
211 
212     DPRINTF("pic_irqs: %s irq %d\n", level? "raise" : "lower", irq);
213     if (cpu->apic_state) {
214         CPU_FOREACH(cs) {
215             cpu = X86_CPU(cs);
216             if (apic_accept_pic_intr(cpu->apic_state)) {
217                 apic_deliver_pic_intr(cpu->apic_state, level);
218             }
219         }
220     } else {
221         if (level) {
222             cpu_interrupt(cs, CPU_INTERRUPT_HARD);
223         } else {
224             cpu_reset_interrupt(cs, CPU_INTERRUPT_HARD);
225         }
226     }
227 }
228 
229 /* PC cmos mappings */
230 
231 #define REG_EQUIPMENT_BYTE          0x14
232 
233 static int cmos_get_fd_drive_type(FDriveType fd0)
234 {
235     int val;
236 
237     switch (fd0) {
238     case FDRIVE_DRV_144:
239         /* 1.44 Mb 3"5 drive */
240         val = 4;
241         break;
242     case FDRIVE_DRV_288:
243         /* 2.88 Mb 3"5 drive */
244         val = 5;
245         break;
246     case FDRIVE_DRV_120:
247         /* 1.2 Mb 5"5 drive */
248         val = 2;
249         break;
250     case FDRIVE_DRV_NONE:
251     default:
252         val = 0;
253         break;
254     }
255     return val;
256 }
257 
258 static void cmos_init_hd(ISADevice *s, int type_ofs, int info_ofs,
259                          int16_t cylinders, int8_t heads, int8_t sectors)
260 {
261     rtc_set_memory(s, type_ofs, 47);
262     rtc_set_memory(s, info_ofs, cylinders);
263     rtc_set_memory(s, info_ofs + 1, cylinders >> 8);
264     rtc_set_memory(s, info_ofs + 2, heads);
265     rtc_set_memory(s, info_ofs + 3, 0xff);
266     rtc_set_memory(s, info_ofs + 4, 0xff);
267     rtc_set_memory(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
268     rtc_set_memory(s, info_ofs + 6, cylinders);
269     rtc_set_memory(s, info_ofs + 7, cylinders >> 8);
270     rtc_set_memory(s, info_ofs + 8, sectors);
271 }
272 
273 /* convert boot_device letter to something recognizable by the bios */
274 static int boot_device2nibble(char boot_device)
275 {
276     switch(boot_device) {
277     case 'a':
278     case 'b':
279         return 0x01; /* floppy boot */
280     case 'c':
281         return 0x02; /* hard drive boot */
282     case 'd':
283         return 0x03; /* CD-ROM boot */
284     case 'n':
285         return 0x04; /* Network boot */
286     }
287     return 0;
288 }
289 
290 static void set_boot_dev(ISADevice *s, const char *boot_device, Error **errp)
291 {
292 #define PC_MAX_BOOT_DEVICES 3
293     int nbds, bds[3] = { 0, };
294     int i;
295 
296     nbds = strlen(boot_device);
297     if (nbds > PC_MAX_BOOT_DEVICES) {
298         error_setg(errp, "Too many boot devices for PC");
299         return;
300     }
301     for (i = 0; i < nbds; i++) {
302         bds[i] = boot_device2nibble(boot_device[i]);
303         if (bds[i] == 0) {
304             error_setg(errp, "Invalid boot device for PC: '%c'",
305                        boot_device[i]);
306             return;
307         }
308     }
309     rtc_set_memory(s, 0x3d, (bds[1] << 4) | bds[0]);
310     rtc_set_memory(s, 0x38, (bds[2] << 4) | (fd_bootchk ? 0x0 : 0x1));
311 }
312 
313 static void pc_boot_set(void *opaque, const char *boot_device, Error **errp)
314 {
315     set_boot_dev(opaque, boot_device, errp);
316 }
317 
318 typedef struct pc_cmos_init_late_arg {
319     ISADevice *rtc_state;
320     BusState *idebus[2];
321 } pc_cmos_init_late_arg;
322 
323 static void pc_cmos_init_late(void *opaque)
324 {
325     pc_cmos_init_late_arg *arg = opaque;
326     ISADevice *s = arg->rtc_state;
327     int16_t cylinders;
328     int8_t heads, sectors;
329     int val;
330     int i, trans;
331 
332     val = 0;
333     if (ide_get_geometry(arg->idebus[0], 0,
334                          &cylinders, &heads, &sectors) >= 0) {
335         cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
336         val |= 0xf0;
337     }
338     if (ide_get_geometry(arg->idebus[0], 1,
339                          &cylinders, &heads, &sectors) >= 0) {
340         cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
341         val |= 0x0f;
342     }
343     rtc_set_memory(s, 0x12, val);
344 
345     val = 0;
346     for (i = 0; i < 4; i++) {
347         /* NOTE: ide_get_geometry() returns the physical
348            geometry.  It is always such that: 1 <= sects <= 63, 1
349            <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
350            geometry can be different if a translation is done. */
351         if (ide_get_geometry(arg->idebus[i / 2], i % 2,
352                              &cylinders, &heads, &sectors) >= 0) {
353             trans = ide_get_bios_chs_trans(arg->idebus[i / 2], i % 2) - 1;
354             assert((trans & ~3) == 0);
355             val |= trans << (i * 2);
356         }
357     }
358     rtc_set_memory(s, 0x39, val);
359 
360     qemu_unregister_reset(pc_cmos_init_late, opaque);
361 }
362 
363 void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
364                   const char *boot_device, MachineState *machine,
365                   ISADevice *floppy, BusState *idebus0, BusState *idebus1,
366                   ISADevice *s)
367 {
368     int val, nb, i;
369     FDriveType fd_type[2] = { FDRIVE_DRV_NONE, FDRIVE_DRV_NONE };
370     static pc_cmos_init_late_arg arg;
371     PCMachineState *pc_machine = PC_MACHINE(machine);
372     Error *local_err = NULL;
373 
374     /* various important CMOS locations needed by PC/Bochs bios */
375 
376     /* memory size */
377     /* base memory (first MiB) */
378     val = MIN(ram_size / 1024, 640);
379     rtc_set_memory(s, 0x15, val);
380     rtc_set_memory(s, 0x16, val >> 8);
381     /* extended memory (next 64MiB) */
382     if (ram_size > 1024 * 1024) {
383         val = (ram_size - 1024 * 1024) / 1024;
384     } else {
385         val = 0;
386     }
387     if (val > 65535)
388         val = 65535;
389     rtc_set_memory(s, 0x17, val);
390     rtc_set_memory(s, 0x18, val >> 8);
391     rtc_set_memory(s, 0x30, val);
392     rtc_set_memory(s, 0x31, val >> 8);
393     /* memory between 16MiB and 4GiB */
394     if (ram_size > 16 * 1024 * 1024) {
395         val = (ram_size - 16 * 1024 * 1024) / 65536;
396     } else {
397         val = 0;
398     }
399     if (val > 65535)
400         val = 65535;
401     rtc_set_memory(s, 0x34, val);
402     rtc_set_memory(s, 0x35, val >> 8);
403     /* memory above 4GiB */
404     val = above_4g_mem_size / 65536;
405     rtc_set_memory(s, 0x5b, val);
406     rtc_set_memory(s, 0x5c, val >> 8);
407     rtc_set_memory(s, 0x5d, val >> 16);
408 
409     /* set the number of CPU */
410     rtc_set_memory(s, 0x5f, smp_cpus - 1);
411 
412     object_property_add_link(OBJECT(machine), "rtc_state",
413                              TYPE_ISA_DEVICE,
414                              (Object **)&pc_machine->rtc,
415                              object_property_allow_set_link,
416                              OBJ_PROP_LINK_UNREF_ON_RELEASE, &error_abort);
417     object_property_set_link(OBJECT(machine), OBJECT(s),
418                              "rtc_state", &error_abort);
419 
420     set_boot_dev(s, boot_device, &local_err);
421     if (local_err) {
422         error_report_err(local_err);
423         exit(1);
424     }
425 
426     /* floppy type */
427     if (floppy) {
428         for (i = 0; i < 2; i++) {
429             fd_type[i] = isa_fdc_get_drive_type(floppy, i);
430         }
431     }
432     val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
433         cmos_get_fd_drive_type(fd_type[1]);
434     rtc_set_memory(s, 0x10, val);
435 
436     val = 0;
437     nb = 0;
438     if (fd_type[0] < FDRIVE_DRV_NONE) {
439         nb++;
440     }
441     if (fd_type[1] < FDRIVE_DRV_NONE) {
442         nb++;
443     }
444     switch (nb) {
445     case 0:
446         break;
447     case 1:
448         val |= 0x01; /* 1 drive, ready for boot */
449         break;
450     case 2:
451         val |= 0x41; /* 2 drives, ready for boot */
452         break;
453     }
454     val |= 0x02; /* FPU is there */
455     val |= 0x04; /* PS/2 mouse installed */
456     rtc_set_memory(s, REG_EQUIPMENT_BYTE, val);
457 
458     /* hard drives */
459     arg.rtc_state = s;
460     arg.idebus[0] = idebus0;
461     arg.idebus[1] = idebus1;
462     qemu_register_reset(pc_cmos_init_late, &arg);
463 }
464 
465 #define TYPE_PORT92 "port92"
466 #define PORT92(obj) OBJECT_CHECK(Port92State, (obj), TYPE_PORT92)
467 
468 /* port 92 stuff: could be split off */
469 typedef struct Port92State {
470     ISADevice parent_obj;
471 
472     MemoryRegion io;
473     uint8_t outport;
474     qemu_irq *a20_out;
475 } Port92State;
476 
477 static void port92_write(void *opaque, hwaddr addr, uint64_t val,
478                          unsigned size)
479 {
480     Port92State *s = opaque;
481     int oldval = s->outport;
482 
483     DPRINTF("port92: write 0x%02" PRIx64 "\n", val);
484     s->outport = val;
485     qemu_set_irq(*s->a20_out, (val >> 1) & 1);
486     if ((val & 1) && !(oldval & 1)) {
487         qemu_system_reset_request();
488     }
489 }
490 
491 static uint64_t port92_read(void *opaque, hwaddr addr,
492                             unsigned size)
493 {
494     Port92State *s = opaque;
495     uint32_t ret;
496 
497     ret = s->outport;
498     DPRINTF("port92: read 0x%02x\n", ret);
499     return ret;
500 }
501 
502 static void port92_init(ISADevice *dev, qemu_irq *a20_out)
503 {
504     Port92State *s = PORT92(dev);
505 
506     s->a20_out = a20_out;
507 }
508 
509 static const VMStateDescription vmstate_port92_isa = {
510     .name = "port92",
511     .version_id = 1,
512     .minimum_version_id = 1,
513     .fields = (VMStateField[]) {
514         VMSTATE_UINT8(outport, Port92State),
515         VMSTATE_END_OF_LIST()
516     }
517 };
518 
519 static void port92_reset(DeviceState *d)
520 {
521     Port92State *s = PORT92(d);
522 
523     s->outport &= ~1;
524 }
525 
526 static const MemoryRegionOps port92_ops = {
527     .read = port92_read,
528     .write = port92_write,
529     .impl = {
530         .min_access_size = 1,
531         .max_access_size = 1,
532     },
533     .endianness = DEVICE_LITTLE_ENDIAN,
534 };
535 
536 static void port92_initfn(Object *obj)
537 {
538     Port92State *s = PORT92(obj);
539 
540     memory_region_init_io(&s->io, OBJECT(s), &port92_ops, s, "port92", 1);
541 
542     s->outport = 0;
543 }
544 
545 static void port92_realizefn(DeviceState *dev, Error **errp)
546 {
547     ISADevice *isadev = ISA_DEVICE(dev);
548     Port92State *s = PORT92(dev);
549 
550     isa_register_ioport(isadev, &s->io, 0x92);
551 }
552 
553 static void port92_class_initfn(ObjectClass *klass, void *data)
554 {
555     DeviceClass *dc = DEVICE_CLASS(klass);
556 
557     dc->realize = port92_realizefn;
558     dc->reset = port92_reset;
559     dc->vmsd = &vmstate_port92_isa;
560     /*
561      * Reason: unlike ordinary ISA devices, this one needs additional
562      * wiring: its A20 output line needs to be wired up by
563      * port92_init().
564      */
565     dc->cannot_instantiate_with_device_add_yet = true;
566 }
567 
568 static const TypeInfo port92_info = {
569     .name          = TYPE_PORT92,
570     .parent        = TYPE_ISA_DEVICE,
571     .instance_size = sizeof(Port92State),
572     .instance_init = port92_initfn,
573     .class_init    = port92_class_initfn,
574 };
575 
576 static void port92_register_types(void)
577 {
578     type_register_static(&port92_info);
579 }
580 
581 type_init(port92_register_types)
582 
583 static void handle_a20_line_change(void *opaque, int irq, int level)
584 {
585     X86CPU *cpu = opaque;
586 
587     /* XXX: send to all CPUs ? */
588     /* XXX: add logic to handle multiple A20 line sources */
589     x86_cpu_set_a20(cpu, level);
590 }
591 
592 int e820_add_entry(uint64_t address, uint64_t length, uint32_t type)
593 {
594     int index = le32_to_cpu(e820_reserve.count);
595     struct e820_entry *entry;
596 
597     if (type != E820_RAM) {
598         /* old FW_CFG_E820_TABLE entry -- reservations only */
599         if (index >= E820_NR_ENTRIES) {
600             return -EBUSY;
601         }
602         entry = &e820_reserve.entry[index++];
603 
604         entry->address = cpu_to_le64(address);
605         entry->length = cpu_to_le64(length);
606         entry->type = cpu_to_le32(type);
607 
608         e820_reserve.count = cpu_to_le32(index);
609     }
610 
611     /* new "etc/e820" file -- include ram too */
612     e820_table = g_renew(struct e820_entry, e820_table, e820_entries + 1);
613     e820_table[e820_entries].address = cpu_to_le64(address);
614     e820_table[e820_entries].length = cpu_to_le64(length);
615     e820_table[e820_entries].type = cpu_to_le32(type);
616     e820_entries++;
617 
618     return e820_entries;
619 }
620 
621 int e820_get_num_entries(void)
622 {
623     return e820_entries;
624 }
625 
626 bool e820_get_entry(int idx, uint32_t type, uint64_t *address, uint64_t *length)
627 {
628     if (idx < e820_entries && e820_table[idx].type == cpu_to_le32(type)) {
629         *address = le64_to_cpu(e820_table[idx].address);
630         *length = le64_to_cpu(e820_table[idx].length);
631         return true;
632     }
633     return false;
634 }
635 
636 /* Enables contiguous-apic-ID mode, for compatibility */
637 static bool compat_apic_id_mode;
638 
639 void enable_compat_apic_id_mode(void)
640 {
641     compat_apic_id_mode = true;
642 }
643 
644 /* Calculates initial APIC ID for a specific CPU index
645  *
646  * Currently we need to be able to calculate the APIC ID from the CPU index
647  * alone (without requiring a CPU object), as the QEMU<->Seabios interfaces have
648  * no concept of "CPU index", and the NUMA tables on fw_cfg need the APIC ID of
649  * all CPUs up to max_cpus.
650  */
651 static uint32_t x86_cpu_apic_id_from_index(unsigned int cpu_index)
652 {
653     uint32_t correct_id;
654     static bool warned;
655 
656     correct_id = x86_apicid_from_cpu_idx(smp_cores, smp_threads, cpu_index);
657     if (compat_apic_id_mode) {
658         if (cpu_index != correct_id && !warned && !qtest_enabled()) {
659             error_report("APIC IDs set in compatibility mode, "
660                          "CPU topology won't match the configuration");
661             warned = true;
662         }
663         return cpu_index;
664     } else {
665         return correct_id;
666     }
667 }
668 
669 /* Calculates the limit to CPU APIC ID values
670  *
671  * This function returns the limit for the APIC ID value, so that all
672  * CPU APIC IDs are < pc_apic_id_limit().
673  *
674  * This is used for FW_CFG_MAX_CPUS. See comments on bochs_bios_init().
675  */
676 static unsigned int pc_apic_id_limit(unsigned int max_cpus)
677 {
678     return x86_cpu_apic_id_from_index(max_cpus - 1) + 1;
679 }
680 
681 static FWCfgState *bochs_bios_init(void)
682 {
683     FWCfgState *fw_cfg;
684     uint8_t *smbios_tables, *smbios_anchor;
685     size_t smbios_tables_len, smbios_anchor_len;
686     uint64_t *numa_fw_cfg;
687     int i, j;
688     unsigned int apic_id_limit = pc_apic_id_limit(max_cpus);
689 
690     fw_cfg = fw_cfg_init_io(BIOS_CFG_IOPORT);
691     /* FW_CFG_MAX_CPUS is a bit confusing/problematic on x86:
692      *
693      * SeaBIOS needs FW_CFG_MAX_CPUS for CPU hotplug, but the CPU hotplug
694      * QEMU<->SeaBIOS interface is not based on the "CPU index", but on the APIC
695      * ID of hotplugged CPUs[1]. This means that FW_CFG_MAX_CPUS is not the
696      * "maximum number of CPUs", but the "limit to the APIC ID values SeaBIOS
697      * may see".
698      *
699      * So, this means we must not use max_cpus, here, but the maximum possible
700      * APIC ID value, plus one.
701      *
702      * [1] The only kind of "CPU identifier" used between SeaBIOS and QEMU is
703      *     the APIC ID, not the "CPU index"
704      */
705     fw_cfg_add_i16(fw_cfg, FW_CFG_MAX_CPUS, (uint16_t)apic_id_limit);
706     fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, (uint64_t)ram_size);
707     fw_cfg_add_bytes(fw_cfg, FW_CFG_ACPI_TABLES,
708                      acpi_tables, acpi_tables_len);
709     fw_cfg_add_i32(fw_cfg, FW_CFG_IRQ0_OVERRIDE, kvm_allows_irq0_override());
710 
711     smbios_tables = smbios_get_table_legacy(&smbios_tables_len);
712     if (smbios_tables) {
713         fw_cfg_add_bytes(fw_cfg, FW_CFG_SMBIOS_ENTRIES,
714                          smbios_tables, smbios_tables_len);
715     }
716 
717     smbios_get_tables(&smbios_tables, &smbios_tables_len,
718                       &smbios_anchor, &smbios_anchor_len);
719     if (smbios_anchor) {
720         fw_cfg_add_file(fw_cfg, "etc/smbios/smbios-tables",
721                         smbios_tables, smbios_tables_len);
722         fw_cfg_add_file(fw_cfg, "etc/smbios/smbios-anchor",
723                         smbios_anchor, smbios_anchor_len);
724     }
725 
726     fw_cfg_add_bytes(fw_cfg, FW_CFG_E820_TABLE,
727                      &e820_reserve, sizeof(e820_reserve));
728     fw_cfg_add_file(fw_cfg, "etc/e820", e820_table,
729                     sizeof(struct e820_entry) * e820_entries);
730 
731     fw_cfg_add_bytes(fw_cfg, FW_CFG_HPET, &hpet_cfg, sizeof(hpet_cfg));
732     /* allocate memory for the NUMA channel: one (64bit) word for the number
733      * of nodes, one word for each VCPU->node and one word for each node to
734      * hold the amount of memory.
735      */
736     numa_fw_cfg = g_new0(uint64_t, 1 + apic_id_limit + nb_numa_nodes);
737     numa_fw_cfg[0] = cpu_to_le64(nb_numa_nodes);
738     for (i = 0; i < max_cpus; i++) {
739         unsigned int apic_id = x86_cpu_apic_id_from_index(i);
740         assert(apic_id < apic_id_limit);
741         for (j = 0; j < nb_numa_nodes; j++) {
742             if (test_bit(i, numa_info[j].node_cpu)) {
743                 numa_fw_cfg[apic_id + 1] = cpu_to_le64(j);
744                 break;
745             }
746         }
747     }
748     for (i = 0; i < nb_numa_nodes; i++) {
749         numa_fw_cfg[apic_id_limit + 1 + i] = cpu_to_le64(numa_info[i].node_mem);
750     }
751     fw_cfg_add_bytes(fw_cfg, FW_CFG_NUMA, numa_fw_cfg,
752                      (1 + apic_id_limit + nb_numa_nodes) *
753                      sizeof(*numa_fw_cfg));
754 
755     return fw_cfg;
756 }
757 
758 static long get_file_size(FILE *f)
759 {
760     long where, size;
761 
762     /* XXX: on Unix systems, using fstat() probably makes more sense */
763 
764     where = ftell(f);
765     fseek(f, 0, SEEK_END);
766     size = ftell(f);
767     fseek(f, where, SEEK_SET);
768 
769     return size;
770 }
771 
772 static void load_linux(FWCfgState *fw_cfg,
773                        const char *kernel_filename,
774                        const char *initrd_filename,
775                        const char *kernel_cmdline,
776                        hwaddr max_ram_size)
777 {
778     uint16_t protocol;
779     int setup_size, kernel_size, initrd_size = 0, cmdline_size;
780     uint32_t initrd_max;
781     uint8_t header[8192], *setup, *kernel, *initrd_data;
782     hwaddr real_addr, prot_addr, cmdline_addr, initrd_addr = 0;
783     FILE *f;
784     char *vmode;
785 
786     /* Align to 16 bytes as a paranoia measure */
787     cmdline_size = (strlen(kernel_cmdline)+16) & ~15;
788 
789     /* load the kernel header */
790     f = fopen(kernel_filename, "rb");
791     if (!f || !(kernel_size = get_file_size(f)) ||
792         fread(header, 1, MIN(ARRAY_SIZE(header), kernel_size), f) !=
793         MIN(ARRAY_SIZE(header), kernel_size)) {
794         fprintf(stderr, "qemu: could not load kernel '%s': %s\n",
795                 kernel_filename, strerror(errno));
796         exit(1);
797     }
798 
799     /* kernel protocol version */
800 #if 0
801     fprintf(stderr, "header magic: %#x\n", ldl_p(header+0x202));
802 #endif
803     if (ldl_p(header+0x202) == 0x53726448) {
804         protocol = lduw_p(header+0x206);
805     } else {
806         /* This looks like a multiboot kernel. If it is, let's stop
807            treating it like a Linux kernel. */
808         if (load_multiboot(fw_cfg, f, kernel_filename, initrd_filename,
809                            kernel_cmdline, kernel_size, header)) {
810             return;
811         }
812         protocol = 0;
813     }
814 
815     if (protocol < 0x200 || !(header[0x211] & 0x01)) {
816         /* Low kernel */
817         real_addr    = 0x90000;
818         cmdline_addr = 0x9a000 - cmdline_size;
819         prot_addr    = 0x10000;
820     } else if (protocol < 0x202) {
821         /* High but ancient kernel */
822         real_addr    = 0x90000;
823         cmdline_addr = 0x9a000 - cmdline_size;
824         prot_addr    = 0x100000;
825     } else {
826         /* High and recent kernel */
827         real_addr    = 0x10000;
828         cmdline_addr = 0x20000;
829         prot_addr    = 0x100000;
830     }
831 
832 #if 0
833     fprintf(stderr,
834             "qemu: real_addr     = 0x" TARGET_FMT_plx "\n"
835             "qemu: cmdline_addr  = 0x" TARGET_FMT_plx "\n"
836             "qemu: prot_addr     = 0x" TARGET_FMT_plx "\n",
837             real_addr,
838             cmdline_addr,
839             prot_addr);
840 #endif
841 
842     /* highest address for loading the initrd */
843     if (protocol >= 0x203) {
844         initrd_max = ldl_p(header+0x22c);
845     } else {
846         initrd_max = 0x37ffffff;
847     }
848 
849     if (initrd_max >= max_ram_size - acpi_data_size) {
850         initrd_max = max_ram_size - acpi_data_size - 1;
851     }
852 
853     fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_ADDR, cmdline_addr);
854     fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_SIZE, strlen(kernel_cmdline)+1);
855     fw_cfg_add_string(fw_cfg, FW_CFG_CMDLINE_DATA, kernel_cmdline);
856 
857     if (protocol >= 0x202) {
858         stl_p(header+0x228, cmdline_addr);
859     } else {
860         stw_p(header+0x20, 0xA33F);
861         stw_p(header+0x22, cmdline_addr-real_addr);
862     }
863 
864     /* handle vga= parameter */
865     vmode = strstr(kernel_cmdline, "vga=");
866     if (vmode) {
867         unsigned int video_mode;
868         /* skip "vga=" */
869         vmode += 4;
870         if (!strncmp(vmode, "normal", 6)) {
871             video_mode = 0xffff;
872         } else if (!strncmp(vmode, "ext", 3)) {
873             video_mode = 0xfffe;
874         } else if (!strncmp(vmode, "ask", 3)) {
875             video_mode = 0xfffd;
876         } else {
877             video_mode = strtol(vmode, NULL, 0);
878         }
879         stw_p(header+0x1fa, video_mode);
880     }
881 
882     /* loader type */
883     /* High nybble = B reserved for QEMU; low nybble is revision number.
884        If this code is substantially changed, you may want to consider
885        incrementing the revision. */
886     if (protocol >= 0x200) {
887         header[0x210] = 0xB0;
888     }
889     /* heap */
890     if (protocol >= 0x201) {
891         header[0x211] |= 0x80;	/* CAN_USE_HEAP */
892         stw_p(header+0x224, cmdline_addr-real_addr-0x200);
893     }
894 
895     /* load initrd */
896     if (initrd_filename) {
897         if (protocol < 0x200) {
898             fprintf(stderr, "qemu: linux kernel too old to load a ram disk\n");
899             exit(1);
900         }
901 
902         initrd_size = get_image_size(initrd_filename);
903         if (initrd_size < 0) {
904             fprintf(stderr, "qemu: error reading initrd %s: %s\n",
905                     initrd_filename, strerror(errno));
906             exit(1);
907         }
908 
909         initrd_addr = (initrd_max-initrd_size) & ~4095;
910 
911         initrd_data = g_malloc(initrd_size);
912         load_image(initrd_filename, initrd_data);
913 
914         fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_ADDR, initrd_addr);
915         fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_SIZE, initrd_size);
916         fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, initrd_data, initrd_size);
917 
918         stl_p(header+0x218, initrd_addr);
919         stl_p(header+0x21c, initrd_size);
920     }
921 
922     /* load kernel and setup */
923     setup_size = header[0x1f1];
924     if (setup_size == 0) {
925         setup_size = 4;
926     }
927     setup_size = (setup_size+1)*512;
928     kernel_size -= setup_size;
929 
930     setup  = g_malloc(setup_size);
931     kernel = g_malloc(kernel_size);
932     fseek(f, 0, SEEK_SET);
933     if (fread(setup, 1, setup_size, f) != setup_size) {
934         fprintf(stderr, "fread() failed\n");
935         exit(1);
936     }
937     if (fread(kernel, 1, kernel_size, f) != kernel_size) {
938         fprintf(stderr, "fread() failed\n");
939         exit(1);
940     }
941     fclose(f);
942     memcpy(setup, header, MIN(sizeof(header), setup_size));
943 
944     fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_ADDR, prot_addr);
945     fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_SIZE, kernel_size);
946     fw_cfg_add_bytes(fw_cfg, FW_CFG_KERNEL_DATA, kernel, kernel_size);
947 
948     fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_ADDR, real_addr);
949     fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_SIZE, setup_size);
950     fw_cfg_add_bytes(fw_cfg, FW_CFG_SETUP_DATA, setup, setup_size);
951 
952     option_rom[nb_option_roms].name = "linuxboot.bin";
953     option_rom[nb_option_roms].bootindex = 0;
954     nb_option_roms++;
955 }
956 
957 #define NE2000_NB_MAX 6
958 
959 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
960                                               0x280, 0x380 };
961 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
962 
963 void pc_init_ne2k_isa(ISABus *bus, NICInfo *nd)
964 {
965     static int nb_ne2k = 0;
966 
967     if (nb_ne2k == NE2000_NB_MAX)
968         return;
969     isa_ne2000_init(bus, ne2000_io[nb_ne2k],
970                     ne2000_irq[nb_ne2k], nd);
971     nb_ne2k++;
972 }
973 
974 DeviceState *cpu_get_current_apic(void)
975 {
976     if (current_cpu) {
977         X86CPU *cpu = X86_CPU(current_cpu);
978         return cpu->apic_state;
979     } else {
980         return NULL;
981     }
982 }
983 
984 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
985 {
986     X86CPU *cpu = opaque;
987 
988     if (level) {
989         cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
990     }
991 }
992 
993 static X86CPU *pc_new_cpu(const char *cpu_model, int64_t apic_id,
994                           DeviceState *icc_bridge, Error **errp)
995 {
996     X86CPU *cpu = NULL;
997     Error *local_err = NULL;
998 
999     if (icc_bridge == NULL) {
1000         error_setg(&local_err, "Invalid icc-bridge value");
1001         goto out;
1002     }
1003 
1004     cpu = cpu_x86_create(cpu_model, &local_err);
1005     if (local_err != NULL) {
1006         goto out;
1007     }
1008 
1009     qdev_set_parent_bus(DEVICE(cpu), qdev_get_child_bus(icc_bridge, "icc"));
1010     object_unref(OBJECT(cpu));
1011 
1012     object_property_set_int(OBJECT(cpu), apic_id, "apic-id", &local_err);
1013     object_property_set_bool(OBJECT(cpu), true, "realized", &local_err);
1014 
1015 out:
1016     if (local_err) {
1017         error_propagate(errp, local_err);
1018         object_unref(OBJECT(cpu));
1019         cpu = NULL;
1020     }
1021     return cpu;
1022 }
1023 
1024 static const char *current_cpu_model;
1025 
1026 void pc_hot_add_cpu(const int64_t id, Error **errp)
1027 {
1028     DeviceState *icc_bridge;
1029     int64_t apic_id = x86_cpu_apic_id_from_index(id);
1030 
1031     if (id < 0) {
1032         error_setg(errp, "Invalid CPU id: %" PRIi64, id);
1033         return;
1034     }
1035 
1036     if (cpu_exists(apic_id)) {
1037         error_setg(errp, "Unable to add CPU: %" PRIi64
1038                    ", it already exists", id);
1039         return;
1040     }
1041 
1042     if (id >= max_cpus) {
1043         error_setg(errp, "Unable to add CPU: %" PRIi64
1044                    ", max allowed: %d", id, max_cpus - 1);
1045         return;
1046     }
1047 
1048     if (apic_id >= ACPI_CPU_HOTPLUG_ID_LIMIT) {
1049         error_setg(errp, "Unable to add CPU: %" PRIi64
1050                    ", resulting APIC ID (%" PRIi64 ") is too large",
1051                    id, apic_id);
1052         return;
1053     }
1054 
1055     icc_bridge = DEVICE(object_resolve_path_type("icc-bridge",
1056                                                  TYPE_ICC_BRIDGE, NULL));
1057     pc_new_cpu(current_cpu_model, apic_id, icc_bridge, errp);
1058 }
1059 
1060 void pc_cpus_init(const char *cpu_model, DeviceState *icc_bridge)
1061 {
1062     int i;
1063     X86CPU *cpu = NULL;
1064     Error *error = NULL;
1065     unsigned long apic_id_limit;
1066 
1067     /* init CPUs */
1068     if (cpu_model == NULL) {
1069 #ifdef TARGET_X86_64
1070         cpu_model = "qemu64";
1071 #else
1072         cpu_model = "qemu32";
1073 #endif
1074     }
1075     current_cpu_model = cpu_model;
1076 
1077     apic_id_limit = pc_apic_id_limit(max_cpus);
1078     if (apic_id_limit > ACPI_CPU_HOTPLUG_ID_LIMIT) {
1079         error_report("max_cpus is too large. APIC ID of last CPU is %lu",
1080                      apic_id_limit - 1);
1081         exit(1);
1082     }
1083 
1084     for (i = 0; i < smp_cpus; i++) {
1085         cpu = pc_new_cpu(cpu_model, x86_cpu_apic_id_from_index(i),
1086                          icc_bridge, &error);
1087         if (error) {
1088             error_report_err(error);
1089             exit(1);
1090         }
1091     }
1092 
1093     /* map APIC MMIO area if CPU has APIC */
1094     if (cpu && cpu->apic_state) {
1095         /* XXX: what if the base changes? */
1096         sysbus_mmio_map_overlap(SYS_BUS_DEVICE(icc_bridge), 0,
1097                                 APIC_DEFAULT_ADDRESS, 0x1000);
1098     }
1099 
1100     /* tell smbios about cpuid version and features */
1101     smbios_set_cpuid(cpu->env.cpuid_version, cpu->env.features[FEAT_1_EDX]);
1102 }
1103 
1104 /* pci-info ROM file. Little endian format */
1105 typedef struct PcRomPciInfo {
1106     uint64_t w32_min;
1107     uint64_t w32_max;
1108     uint64_t w64_min;
1109     uint64_t w64_max;
1110 } PcRomPciInfo;
1111 
1112 typedef struct PcGuestInfoState {
1113     PcGuestInfo info;
1114     Notifier machine_done;
1115 } PcGuestInfoState;
1116 
1117 static
1118 void pc_guest_info_machine_done(Notifier *notifier, void *data)
1119 {
1120     PcGuestInfoState *guest_info_state = container_of(notifier,
1121                                                       PcGuestInfoState,
1122                                                       machine_done);
1123     PCIBus *bus = find_i440fx();
1124 
1125     if (bus) {
1126         int extra_hosts = 0;
1127 
1128         QLIST_FOREACH(bus, &bus->child, sibling) {
1129             /* look for expander root buses */
1130             if (pci_bus_is_root(bus)) {
1131                 extra_hosts++;
1132             }
1133         }
1134         if (extra_hosts && guest_info_state->info.fw_cfg) {
1135             uint64_t *val = g_malloc(sizeof(*val));
1136             *val = cpu_to_le64(extra_hosts);
1137             fw_cfg_add_file(guest_info_state->info.fw_cfg,
1138                     "etc/extra-pci-roots", val, sizeof(*val));
1139         }
1140     }
1141 
1142     acpi_setup(&guest_info_state->info);
1143 }
1144 
1145 PcGuestInfo *pc_guest_info_init(ram_addr_t below_4g_mem_size,
1146                                 ram_addr_t above_4g_mem_size)
1147 {
1148     PcGuestInfoState *guest_info_state = g_malloc0(sizeof *guest_info_state);
1149     PcGuestInfo *guest_info = &guest_info_state->info;
1150     int i, j;
1151 
1152     guest_info->ram_size_below_4g = below_4g_mem_size;
1153     guest_info->ram_size = below_4g_mem_size + above_4g_mem_size;
1154     guest_info->apic_id_limit = pc_apic_id_limit(max_cpus);
1155     guest_info->apic_xrupt_override = kvm_allows_irq0_override();
1156     guest_info->numa_nodes = nb_numa_nodes;
1157     guest_info->node_mem = g_malloc0(guest_info->numa_nodes *
1158                                     sizeof *guest_info->node_mem);
1159     for (i = 0; i < nb_numa_nodes; i++) {
1160         guest_info->node_mem[i] = numa_info[i].node_mem;
1161     }
1162 
1163     guest_info->node_cpu = g_malloc0(guest_info->apic_id_limit *
1164                                      sizeof *guest_info->node_cpu);
1165 
1166     for (i = 0; i < max_cpus; i++) {
1167         unsigned int apic_id = x86_cpu_apic_id_from_index(i);
1168         assert(apic_id < guest_info->apic_id_limit);
1169         for (j = 0; j < nb_numa_nodes; j++) {
1170             if (test_bit(i, numa_info[j].node_cpu)) {
1171                 guest_info->node_cpu[apic_id] = j;
1172                 break;
1173             }
1174         }
1175     }
1176 
1177     guest_info_state->machine_done.notify = pc_guest_info_machine_done;
1178     qemu_add_machine_init_done_notifier(&guest_info_state->machine_done);
1179     return guest_info;
1180 }
1181 
1182 /* setup pci memory address space mapping into system address space */
1183 void pc_pci_as_mapping_init(Object *owner, MemoryRegion *system_memory,
1184                             MemoryRegion *pci_address_space)
1185 {
1186     /* Set to lower priority than RAM */
1187     memory_region_add_subregion_overlap(system_memory, 0x0,
1188                                         pci_address_space, -1);
1189 }
1190 
1191 void pc_acpi_init(const char *default_dsdt)
1192 {
1193     char *filename;
1194 
1195     if (acpi_tables != NULL) {
1196         /* manually set via -acpitable, leave it alone */
1197         return;
1198     }
1199 
1200     filename = qemu_find_file(QEMU_FILE_TYPE_BIOS, default_dsdt);
1201     if (filename == NULL) {
1202         fprintf(stderr, "WARNING: failed to find %s\n", default_dsdt);
1203     } else {
1204         QemuOpts *opts = qemu_opts_create(qemu_find_opts("acpi"), NULL, 0,
1205                                           &error_abort);
1206         Error *err = NULL;
1207 
1208         qemu_opt_set(opts, "file", filename, &error_abort);
1209 
1210         acpi_table_add_builtin(opts, &err);
1211         if (err) {
1212             error_report("WARNING: failed to load %s: %s", filename,
1213                          error_get_pretty(err));
1214             error_free(err);
1215         }
1216         g_free(filename);
1217     }
1218 }
1219 
1220 FWCfgState *xen_load_linux(const char *kernel_filename,
1221                            const char *kernel_cmdline,
1222                            const char *initrd_filename,
1223                            ram_addr_t below_4g_mem_size,
1224                            PcGuestInfo *guest_info)
1225 {
1226     int i;
1227     FWCfgState *fw_cfg;
1228 
1229     assert(kernel_filename != NULL);
1230 
1231     fw_cfg = fw_cfg_init_io(BIOS_CFG_IOPORT);
1232     rom_set_fw(fw_cfg);
1233 
1234     load_linux(fw_cfg, kernel_filename, initrd_filename,
1235                kernel_cmdline, below_4g_mem_size);
1236     for (i = 0; i < nb_option_roms; i++) {
1237         assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
1238                !strcmp(option_rom[i].name, "multiboot.bin"));
1239         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1240     }
1241     guest_info->fw_cfg = fw_cfg;
1242     return fw_cfg;
1243 }
1244 
1245 FWCfgState *pc_memory_init(MachineState *machine,
1246                            MemoryRegion *system_memory,
1247                            ram_addr_t below_4g_mem_size,
1248                            ram_addr_t above_4g_mem_size,
1249                            MemoryRegion *rom_memory,
1250                            MemoryRegion **ram_memory,
1251                            PcGuestInfo *guest_info)
1252 {
1253     int linux_boot, i;
1254     MemoryRegion *ram, *option_rom_mr;
1255     MemoryRegion *ram_below_4g, *ram_above_4g;
1256     FWCfgState *fw_cfg;
1257     PCMachineState *pcms = PC_MACHINE(machine);
1258 
1259     assert(machine->ram_size == below_4g_mem_size + above_4g_mem_size);
1260 
1261     linux_boot = (machine->kernel_filename != NULL);
1262 
1263     /* Allocate RAM.  We allocate it as a single memory region and use
1264      * aliases to address portions of it, mostly for backwards compatibility
1265      * with older qemus that used qemu_ram_alloc().
1266      */
1267     ram = g_malloc(sizeof(*ram));
1268     memory_region_allocate_system_memory(ram, NULL, "pc.ram",
1269                                          machine->ram_size);
1270     *ram_memory = ram;
1271     ram_below_4g = g_malloc(sizeof(*ram_below_4g));
1272     memory_region_init_alias(ram_below_4g, NULL, "ram-below-4g", ram,
1273                              0, below_4g_mem_size);
1274     memory_region_add_subregion(system_memory, 0, ram_below_4g);
1275     e820_add_entry(0, below_4g_mem_size, E820_RAM);
1276     if (above_4g_mem_size > 0) {
1277         ram_above_4g = g_malloc(sizeof(*ram_above_4g));
1278         memory_region_init_alias(ram_above_4g, NULL, "ram-above-4g", ram,
1279                                  below_4g_mem_size, above_4g_mem_size);
1280         memory_region_add_subregion(system_memory, 0x100000000ULL,
1281                                     ram_above_4g);
1282         e820_add_entry(0x100000000ULL, above_4g_mem_size, E820_RAM);
1283     }
1284 
1285     if (!guest_info->has_reserved_memory &&
1286         (machine->ram_slots ||
1287          (machine->maxram_size > machine->ram_size))) {
1288         MachineClass *mc = MACHINE_GET_CLASS(machine);
1289 
1290         error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
1291                      mc->name);
1292         exit(EXIT_FAILURE);
1293     }
1294 
1295     /* initialize hotplug memory address space */
1296     if (guest_info->has_reserved_memory &&
1297         (machine->ram_size < machine->maxram_size)) {
1298         ram_addr_t hotplug_mem_size =
1299             machine->maxram_size - machine->ram_size;
1300 
1301         if (machine->ram_slots > ACPI_MAX_RAM_SLOTS) {
1302             error_report("unsupported amount of memory slots: %"PRIu64,
1303                          machine->ram_slots);
1304             exit(EXIT_FAILURE);
1305         }
1306 
1307         if (QEMU_ALIGN_UP(machine->maxram_size,
1308                           TARGET_PAGE_SIZE) != machine->maxram_size) {
1309             error_report("maximum memory size must by aligned to multiple of "
1310                          "%d bytes", TARGET_PAGE_SIZE);
1311             exit(EXIT_FAILURE);
1312         }
1313 
1314         pcms->hotplug_memory_base =
1315             ROUND_UP(0x100000000ULL + above_4g_mem_size, 1ULL << 30);
1316 
1317         if (pcms->enforce_aligned_dimm) {
1318             /* size hotplug region assuming 1G page max alignment per slot */
1319             hotplug_mem_size += (1ULL << 30) * machine->ram_slots;
1320         }
1321 
1322         if ((pcms->hotplug_memory_base + hotplug_mem_size) <
1323             hotplug_mem_size) {
1324             error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT,
1325                          machine->maxram_size);
1326             exit(EXIT_FAILURE);
1327         }
1328 
1329         memory_region_init(&pcms->hotplug_memory, OBJECT(pcms),
1330                            "hotplug-memory", hotplug_mem_size);
1331         memory_region_add_subregion(system_memory, pcms->hotplug_memory_base,
1332                                     &pcms->hotplug_memory);
1333     }
1334 
1335     /* Initialize PC system firmware */
1336     pc_system_firmware_init(rom_memory, guest_info->isapc_ram_fw);
1337 
1338     option_rom_mr = g_malloc(sizeof(*option_rom_mr));
1339     memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE,
1340                            &error_abort);
1341     vmstate_register_ram_global(option_rom_mr);
1342     memory_region_add_subregion_overlap(rom_memory,
1343                                         PC_ROM_MIN_VGA,
1344                                         option_rom_mr,
1345                                         1);
1346 
1347     fw_cfg = bochs_bios_init();
1348     rom_set_fw(fw_cfg);
1349 
1350     if (guest_info->has_reserved_memory && pcms->hotplug_memory_base) {
1351         uint64_t *val = g_malloc(sizeof(*val));
1352         *val = cpu_to_le64(ROUND_UP(pcms->hotplug_memory_base, 0x1ULL << 30));
1353         fw_cfg_add_file(fw_cfg, "etc/reserved-memory-end", val, sizeof(*val));
1354     }
1355 
1356     if (linux_boot) {
1357         load_linux(fw_cfg, machine->kernel_filename, machine->initrd_filename,
1358                    machine->kernel_cmdline, below_4g_mem_size);
1359     }
1360 
1361     for (i = 0; i < nb_option_roms; i++) {
1362         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1363     }
1364     guest_info->fw_cfg = fw_cfg;
1365     return fw_cfg;
1366 }
1367 
1368 qemu_irq *pc_allocate_cpu_irq(void)
1369 {
1370     return qemu_allocate_irqs(pic_irq_request, NULL, 1);
1371 }
1372 
1373 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1374 {
1375     DeviceState *dev = NULL;
1376 
1377     if (pci_bus) {
1378         PCIDevice *pcidev = pci_vga_init(pci_bus);
1379         dev = pcidev ? &pcidev->qdev : NULL;
1380     } else if (isa_bus) {
1381         ISADevice *isadev = isa_vga_init(isa_bus);
1382         dev = isadev ? DEVICE(isadev) : NULL;
1383     }
1384     return dev;
1385 }
1386 
1387 static void cpu_request_exit(void *opaque, int irq, int level)
1388 {
1389     CPUState *cpu = current_cpu;
1390 
1391     if (cpu && level) {
1392         cpu_exit(cpu);
1393     }
1394 }
1395 
1396 static const MemoryRegionOps ioport80_io_ops = {
1397     .write = ioport80_write,
1398     .read = ioport80_read,
1399     .endianness = DEVICE_NATIVE_ENDIAN,
1400     .impl = {
1401         .min_access_size = 1,
1402         .max_access_size = 1,
1403     },
1404 };
1405 
1406 static const MemoryRegionOps ioportF0_io_ops = {
1407     .write = ioportF0_write,
1408     .read = ioportF0_read,
1409     .endianness = DEVICE_NATIVE_ENDIAN,
1410     .impl = {
1411         .min_access_size = 1,
1412         .max_access_size = 1,
1413     },
1414 };
1415 
1416 void pc_basic_device_init(ISABus *isa_bus, qemu_irq *gsi,
1417                           ISADevice **rtc_state,
1418                           bool create_fdctrl,
1419                           ISADevice **floppy,
1420                           bool no_vmport,
1421                           uint32 hpet_irqs)
1422 {
1423     int i;
1424     DriveInfo *fd[MAX_FD];
1425     DeviceState *hpet = NULL;
1426     int pit_isa_irq = 0;
1427     qemu_irq pit_alt_irq = NULL;
1428     qemu_irq rtc_irq = NULL;
1429     qemu_irq *a20_line;
1430     ISADevice *i8042, *port92, *vmmouse, *pit = NULL;
1431     qemu_irq *cpu_exit_irq;
1432     MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1433     MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1434 
1435     memory_region_init_io(ioport80_io, NULL, &ioport80_io_ops, NULL, "ioport80", 1);
1436     memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1437 
1438     memory_region_init_io(ioportF0_io, NULL, &ioportF0_io_ops, NULL, "ioportF0", 1);
1439     memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1440 
1441     /*
1442      * Check if an HPET shall be created.
1443      *
1444      * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
1445      * when the HPET wants to take over. Thus we have to disable the latter.
1446      */
1447     if (!no_hpet && (!kvm_irqchip_in_kernel() || kvm_has_pit_state2())) {
1448         /* In order to set property, here not using sysbus_try_create_simple */
1449         hpet = qdev_try_create(NULL, TYPE_HPET);
1450         if (hpet) {
1451             /* For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-1.7
1452              * and earlier, use IRQ2 for compat. Otherwise, use IRQ16~23,
1453              * IRQ8 and IRQ2.
1454              */
1455             uint8_t compat = object_property_get_int(OBJECT(hpet),
1456                     HPET_INTCAP, NULL);
1457             if (!compat) {
1458                 qdev_prop_set_uint32(hpet, HPET_INTCAP, hpet_irqs);
1459             }
1460             qdev_init_nofail(hpet);
1461             sysbus_mmio_map(SYS_BUS_DEVICE(hpet), 0, HPET_BASE);
1462 
1463             for (i = 0; i < GSI_NUM_PINS; i++) {
1464                 sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
1465             }
1466             pit_isa_irq = -1;
1467             pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1468             rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1469         }
1470     }
1471     *rtc_state = rtc_init(isa_bus, 2000, rtc_irq);
1472 
1473     qemu_register_boot_set(pc_boot_set, *rtc_state);
1474 
1475     if (!xen_enabled()) {
1476         if (kvm_irqchip_in_kernel()) {
1477             pit = kvm_pit_init(isa_bus, 0x40);
1478         } else {
1479             pit = pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1480         }
1481         if (hpet) {
1482             /* connect PIT to output control line of the HPET */
1483             qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(DEVICE(pit), 0));
1484         }
1485         pcspk_init(isa_bus, pit);
1486     }
1487 
1488     serial_hds_isa_init(isa_bus, MAX_SERIAL_PORTS);
1489     parallel_hds_isa_init(isa_bus, MAX_PARALLEL_PORTS);
1490 
1491     a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1492     i8042 = isa_create_simple(isa_bus, "i8042");
1493     i8042_setup_a20_line(i8042, &a20_line[0]);
1494     if (!no_vmport) {
1495         vmport_init(isa_bus);
1496         vmmouse = isa_try_create(isa_bus, "vmmouse");
1497     } else {
1498         vmmouse = NULL;
1499     }
1500     if (vmmouse) {
1501         DeviceState *dev = DEVICE(vmmouse);
1502         qdev_prop_set_ptr(dev, "ps2_mouse", i8042);
1503         qdev_init_nofail(dev);
1504     }
1505     port92 = isa_create_simple(isa_bus, "port92");
1506     port92_init(port92, &a20_line[1]);
1507 
1508     cpu_exit_irq = qemu_allocate_irqs(cpu_request_exit, NULL, 1);
1509     DMA_init(0, cpu_exit_irq);
1510 
1511     for(i = 0; i < MAX_FD; i++) {
1512         fd[i] = drive_get(IF_FLOPPY, 0, i);
1513         create_fdctrl |= !!fd[i];
1514     }
1515     *floppy = create_fdctrl ? fdctrl_init_isa(isa_bus, fd) : NULL;
1516 }
1517 
1518 void pc_nic_init(ISABus *isa_bus, PCIBus *pci_bus)
1519 {
1520     int i;
1521 
1522     for (i = 0; i < nb_nics; i++) {
1523         NICInfo *nd = &nd_table[i];
1524 
1525         if (!pci_bus || (nd->model && strcmp(nd->model, "ne2k_isa") == 0)) {
1526             pc_init_ne2k_isa(isa_bus, nd);
1527         } else {
1528             pci_nic_init_nofail(nd, pci_bus, "e1000", NULL);
1529         }
1530     }
1531 }
1532 
1533 void pc_pci_device_init(PCIBus *pci_bus)
1534 {
1535     int max_bus;
1536     int bus;
1537 
1538     max_bus = drive_get_max_bus(IF_SCSI);
1539     for (bus = 0; bus <= max_bus; bus++) {
1540         pci_create_simple(pci_bus, -1, "lsi53c895a");
1541     }
1542 }
1543 
1544 void ioapic_init_gsi(GSIState *gsi_state, const char *parent_name)
1545 {
1546     DeviceState *dev;
1547     SysBusDevice *d;
1548     unsigned int i;
1549 
1550     if (kvm_irqchip_in_kernel()) {
1551         dev = qdev_create(NULL, "kvm-ioapic");
1552     } else {
1553         dev = qdev_create(NULL, "ioapic");
1554     }
1555     if (parent_name) {
1556         object_property_add_child(object_resolve_path(parent_name, NULL),
1557                                   "ioapic", OBJECT(dev), NULL);
1558     }
1559     qdev_init_nofail(dev);
1560     d = SYS_BUS_DEVICE(dev);
1561     sysbus_mmio_map(d, 0, IO_APIC_DEFAULT_ADDRESS);
1562 
1563     for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1564         gsi_state->ioapic_irq[i] = qdev_get_gpio_in(dev, i);
1565     }
1566 }
1567 
1568 static void pc_dimm_plug(HotplugHandler *hotplug_dev,
1569                          DeviceState *dev, Error **errp)
1570 {
1571     int slot;
1572     HotplugHandlerClass *hhc;
1573     Error *local_err = NULL;
1574     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1575     MachineState *machine = MACHINE(hotplug_dev);
1576     PCDIMMDevice *dimm = PC_DIMM(dev);
1577     PCDIMMDeviceClass *ddc = PC_DIMM_GET_CLASS(dimm);
1578     MemoryRegion *mr = ddc->get_memory_region(dimm);
1579     uint64_t existing_dimms_capacity = 0;
1580     uint64_t align = TARGET_PAGE_SIZE;
1581     uint64_t addr;
1582 
1583     addr = object_property_get_int(OBJECT(dimm), PC_DIMM_ADDR_PROP, &local_err);
1584     if (local_err) {
1585         goto out;
1586     }
1587 
1588     if (memory_region_get_alignment(mr) && pcms->enforce_aligned_dimm) {
1589         align = memory_region_get_alignment(mr);
1590     }
1591 
1592     addr = pc_dimm_get_free_addr(pcms->hotplug_memory_base,
1593                                  memory_region_size(&pcms->hotplug_memory),
1594                                  !addr ? NULL : &addr, align,
1595                                  memory_region_size(mr), &local_err);
1596     if (local_err) {
1597         goto out;
1598     }
1599 
1600     existing_dimms_capacity = pc_existing_dimms_capacity(&local_err);
1601     if (local_err) {
1602         goto out;
1603     }
1604 
1605     if (existing_dimms_capacity + memory_region_size(mr) >
1606         machine->maxram_size - machine->ram_size) {
1607         error_setg(&local_err, "not enough space, currently 0x%" PRIx64
1608                    " in use of total hot pluggable 0x" RAM_ADDR_FMT,
1609                    existing_dimms_capacity,
1610                    machine->maxram_size - machine->ram_size);
1611         goto out;
1612     }
1613 
1614     object_property_set_int(OBJECT(dev), addr, PC_DIMM_ADDR_PROP, &local_err);
1615     if (local_err) {
1616         goto out;
1617     }
1618     trace_mhp_pc_dimm_assigned_address(addr);
1619 
1620     slot = object_property_get_int(OBJECT(dev), PC_DIMM_SLOT_PROP, &local_err);
1621     if (local_err) {
1622         goto out;
1623     }
1624 
1625     slot = pc_dimm_get_free_slot(slot == PC_DIMM_UNASSIGNED_SLOT ? NULL : &slot,
1626                                  machine->ram_slots, &local_err);
1627     if (local_err) {
1628         goto out;
1629     }
1630     object_property_set_int(OBJECT(dev), slot, PC_DIMM_SLOT_PROP, &local_err);
1631     if (local_err) {
1632         goto out;
1633     }
1634     trace_mhp_pc_dimm_assigned_slot(slot);
1635 
1636     if (!pcms->acpi_dev) {
1637         error_setg(&local_err,
1638                    "memory hotplug is not enabled: missing acpi device");
1639         goto out;
1640     }
1641 
1642     if (kvm_enabled() && !kvm_has_free_slot(machine)) {
1643         error_setg(&local_err, "hypervisor has no free memory slots left");
1644         goto out;
1645     }
1646 
1647     memory_region_add_subregion(&pcms->hotplug_memory,
1648                                 addr - pcms->hotplug_memory_base, mr);
1649     vmstate_register_ram(mr, dev);
1650 
1651     hhc = HOTPLUG_HANDLER_GET_CLASS(pcms->acpi_dev);
1652     hhc->plug(HOTPLUG_HANDLER(pcms->acpi_dev), dev, &local_err);
1653 out:
1654     error_propagate(errp, local_err);
1655 }
1656 
1657 static void pc_dimm_unplug_request(HotplugHandler *hotplug_dev,
1658                                    DeviceState *dev, Error **errp)
1659 {
1660     HotplugHandlerClass *hhc;
1661     Error *local_err = NULL;
1662     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1663 
1664     if (!pcms->acpi_dev) {
1665         error_setg(&local_err,
1666                    "memory hotplug is not enabled: missing acpi device");
1667         goto out;
1668     }
1669 
1670     hhc = HOTPLUG_HANDLER_GET_CLASS(pcms->acpi_dev);
1671     hhc->unplug_request(HOTPLUG_HANDLER(pcms->acpi_dev), dev, &local_err);
1672 
1673 out:
1674     error_propagate(errp, local_err);
1675 }
1676 
1677 static void pc_dimm_unplug(HotplugHandler *hotplug_dev,
1678                            DeviceState *dev, Error **errp)
1679 {
1680     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1681     PCDIMMDevice *dimm = PC_DIMM(dev);
1682     PCDIMMDeviceClass *ddc = PC_DIMM_GET_CLASS(dimm);
1683     MemoryRegion *mr = ddc->get_memory_region(dimm);
1684     HotplugHandlerClass *hhc;
1685     Error *local_err = NULL;
1686 
1687     hhc = HOTPLUG_HANDLER_GET_CLASS(pcms->acpi_dev);
1688     hhc->unplug(HOTPLUG_HANDLER(pcms->acpi_dev), dev, &local_err);
1689 
1690     if (local_err) {
1691         goto out;
1692     }
1693 
1694     memory_region_del_subregion(&pcms->hotplug_memory, mr);
1695     vmstate_unregister_ram(mr, dev);
1696 
1697     object_unparent(OBJECT(dev));
1698 
1699  out:
1700     error_propagate(errp, local_err);
1701 }
1702 
1703 static void pc_cpu_plug(HotplugHandler *hotplug_dev,
1704                         DeviceState *dev, Error **errp)
1705 {
1706     HotplugHandlerClass *hhc;
1707     Error *local_err = NULL;
1708     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1709 
1710     if (!dev->hotplugged) {
1711         goto out;
1712     }
1713 
1714     if (!pcms->acpi_dev) {
1715         error_setg(&local_err,
1716                    "cpu hotplug is not enabled: missing acpi device");
1717         goto out;
1718     }
1719 
1720     hhc = HOTPLUG_HANDLER_GET_CLASS(pcms->acpi_dev);
1721     hhc->plug(HOTPLUG_HANDLER(pcms->acpi_dev), dev, &local_err);
1722     if (local_err) {
1723         goto out;
1724     }
1725 
1726     /* increment the number of CPUs */
1727     rtc_set_memory(pcms->rtc, 0x5f, rtc_get_memory(pcms->rtc, 0x5f) + 1);
1728 out:
1729     error_propagate(errp, local_err);
1730 }
1731 
1732 static void pc_machine_device_plug_cb(HotplugHandler *hotplug_dev,
1733                                       DeviceState *dev, Error **errp)
1734 {
1735     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1736         pc_dimm_plug(hotplug_dev, dev, errp);
1737     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1738         pc_cpu_plug(hotplug_dev, dev, errp);
1739     }
1740 }
1741 
1742 static void pc_machine_device_unplug_request_cb(HotplugHandler *hotplug_dev,
1743                                                 DeviceState *dev, Error **errp)
1744 {
1745     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1746         pc_dimm_unplug_request(hotplug_dev, dev, errp);
1747     } else {
1748         error_setg(errp, "acpi: device unplug request for not supported device"
1749                    " type: %s", object_get_typename(OBJECT(dev)));
1750     }
1751 }
1752 
1753 static void pc_machine_device_unplug_cb(HotplugHandler *hotplug_dev,
1754                                         DeviceState *dev, Error **errp)
1755 {
1756     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1757         pc_dimm_unplug(hotplug_dev, dev, errp);
1758     } else {
1759         error_setg(errp, "acpi: device unplug for not supported device"
1760                    " type: %s", object_get_typename(OBJECT(dev)));
1761     }
1762 }
1763 
1764 static HotplugHandler *pc_get_hotpug_handler(MachineState *machine,
1765                                              DeviceState *dev)
1766 {
1767     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(machine);
1768 
1769     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM) ||
1770         object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1771         return HOTPLUG_HANDLER(machine);
1772     }
1773 
1774     return pcmc->get_hotplug_handler ?
1775         pcmc->get_hotplug_handler(machine, dev) : NULL;
1776 }
1777 
1778 static void
1779 pc_machine_get_hotplug_memory_region_size(Object *obj, Visitor *v, void *opaque,
1780                                           const char *name, Error **errp)
1781 {
1782     PCMachineState *pcms = PC_MACHINE(obj);
1783     int64_t value = memory_region_size(&pcms->hotplug_memory);
1784 
1785     visit_type_int(v, &value, name, errp);
1786 }
1787 
1788 static void pc_machine_get_max_ram_below_4g(Object *obj, Visitor *v,
1789                                          void *opaque, const char *name,
1790                                          Error **errp)
1791 {
1792     PCMachineState *pcms = PC_MACHINE(obj);
1793     uint64_t value = pcms->max_ram_below_4g;
1794 
1795     visit_type_size(v, &value, name, errp);
1796 }
1797 
1798 static void pc_machine_set_max_ram_below_4g(Object *obj, Visitor *v,
1799                                          void *opaque, const char *name,
1800                                          Error **errp)
1801 {
1802     PCMachineState *pcms = PC_MACHINE(obj);
1803     Error *error = NULL;
1804     uint64_t value;
1805 
1806     visit_type_size(v, &value, name, &error);
1807     if (error) {
1808         error_propagate(errp, error);
1809         return;
1810     }
1811     if (value > (1ULL << 32)) {
1812         error_set(&error, ERROR_CLASS_GENERIC_ERROR,
1813                   "Machine option 'max-ram-below-4g=%"PRIu64
1814                   "' expects size less than or equal to 4G", value);
1815         error_propagate(errp, error);
1816         return;
1817     }
1818 
1819     if (value < (1ULL << 20)) {
1820         error_report("Warning: small max_ram_below_4g(%"PRIu64
1821                      ") less than 1M.  BIOS may not work..",
1822                      value);
1823     }
1824 
1825     pcms->max_ram_below_4g = value;
1826 }
1827 
1828 static void pc_machine_get_vmport(Object *obj, Visitor *v, void *opaque,
1829                                   const char *name, Error **errp)
1830 {
1831     PCMachineState *pcms = PC_MACHINE(obj);
1832     OnOffAuto vmport = pcms->vmport;
1833 
1834     visit_type_OnOffAuto(v, &vmport, name, errp);
1835 }
1836 
1837 static void pc_machine_set_vmport(Object *obj, Visitor *v, void *opaque,
1838                                   const char *name, Error **errp)
1839 {
1840     PCMachineState *pcms = PC_MACHINE(obj);
1841 
1842     visit_type_OnOffAuto(v, &pcms->vmport, name, errp);
1843 }
1844 
1845 static bool pc_machine_get_aligned_dimm(Object *obj, Error **errp)
1846 {
1847     PCMachineState *pcms = PC_MACHINE(obj);
1848 
1849     return pcms->enforce_aligned_dimm;
1850 }
1851 
1852 static void pc_machine_initfn(Object *obj)
1853 {
1854     PCMachineState *pcms = PC_MACHINE(obj);
1855 
1856     object_property_add(obj, PC_MACHINE_MEMHP_REGION_SIZE, "int",
1857                         pc_machine_get_hotplug_memory_region_size,
1858                         NULL, NULL, NULL, NULL);
1859 
1860     pcms->max_ram_below_4g = 1ULL << 32; /* 4G */
1861     object_property_add(obj, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
1862                         pc_machine_get_max_ram_below_4g,
1863                         pc_machine_set_max_ram_below_4g,
1864                         NULL, NULL, NULL);
1865     object_property_set_description(obj, PC_MACHINE_MAX_RAM_BELOW_4G,
1866                                     "Maximum ram below the 4G boundary (32bit boundary)",
1867                                     NULL);
1868 
1869     pcms->vmport = ON_OFF_AUTO_AUTO;
1870     object_property_add(obj, PC_MACHINE_VMPORT, "OnOffAuto",
1871                         pc_machine_get_vmport,
1872                         pc_machine_set_vmport,
1873                         NULL, NULL, NULL);
1874     object_property_set_description(obj, PC_MACHINE_VMPORT,
1875                                     "Enable vmport (pc & q35)",
1876                                     NULL);
1877 
1878     pcms->enforce_aligned_dimm = true;
1879     object_property_add_bool(obj, PC_MACHINE_ENFORCE_ALIGNED_DIMM,
1880                              pc_machine_get_aligned_dimm,
1881                              NULL, NULL);
1882 }
1883 
1884 static unsigned pc_cpu_index_to_socket_id(unsigned cpu_index)
1885 {
1886     unsigned pkg_id, core_id, smt_id;
1887     x86_topo_ids_from_idx(smp_cores, smp_threads, cpu_index,
1888                           &pkg_id, &core_id, &smt_id);
1889     return pkg_id;
1890 }
1891 
1892 static void pc_machine_class_init(ObjectClass *oc, void *data)
1893 {
1894     MachineClass *mc = MACHINE_CLASS(oc);
1895     PCMachineClass *pcmc = PC_MACHINE_CLASS(oc);
1896     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
1897 
1898     pcmc->get_hotplug_handler = mc->get_hotplug_handler;
1899     mc->get_hotplug_handler = pc_get_hotpug_handler;
1900     mc->cpu_index_to_socket_id = pc_cpu_index_to_socket_id;
1901     hc->plug = pc_machine_device_plug_cb;
1902     hc->unplug_request = pc_machine_device_unplug_request_cb;
1903     hc->unplug = pc_machine_device_unplug_cb;
1904 }
1905 
1906 static const TypeInfo pc_machine_info = {
1907     .name = TYPE_PC_MACHINE,
1908     .parent = TYPE_MACHINE,
1909     .abstract = true,
1910     .instance_size = sizeof(PCMachineState),
1911     .instance_init = pc_machine_initfn,
1912     .class_size = sizeof(PCMachineClass),
1913     .class_init = pc_machine_class_init,
1914     .interfaces = (InterfaceInfo[]) {
1915          { TYPE_HOTPLUG_HANDLER },
1916          { }
1917     },
1918 };
1919 
1920 static void pc_machine_register_types(void)
1921 {
1922     type_register_static(&pc_machine_info);
1923 }
1924 
1925 type_init(pc_machine_register_types)
1926