]> git.proxmox.com Git - mirror_qemu.git/blame - hw/i386/pc.c
Merge remote-tracking branch 'remotes/bonzini/memory' into staging
[mirror_qemu.git] / hw / i386 / pc.c
CommitLineData
80cabfad
FB
1/*
2 * QEMU PC System Emulator
5fafdf24 3 *
80cabfad 4 * Copyright (c) 2003-2004 Fabrice Bellard
5fafdf24 5 *
80cabfad
FB
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 */
83c9f4ca 24#include "hw/hw.h"
0d09e41a
PB
25#include "hw/i386/pc.h"
26#include "hw/char/serial.h"
27#include "hw/i386/apic.h"
28#include "hw/block/fdc.h"
83c9f4ca
PB
29#include "hw/ide.h"
30#include "hw/pci/pci.h"
83c9089e 31#include "monitor/monitor.h"
0d09e41a
PB
32#include "hw/nvram/fw_cfg.h"
33#include "hw/timer/hpet.h"
34#include "hw/i386/smbios.h"
83c9f4ca 35#include "hw/loader.h"
ca20cf32 36#include "elf.h"
47b43a1f 37#include "multiboot.h"
0d09e41a
PB
38#include "hw/timer/mc146818rtc.h"
39#include "hw/timer/i8254.h"
40#include "hw/audio/pcspk.h"
83c9f4ca
PB
41#include "hw/pci/msi.h"
42#include "hw/sysbus.h"
9c17d615
PB
43#include "sysemu/sysemu.h"
44#include "sysemu/kvm.h"
1d31f66b 45#include "kvm_i386.h"
0d09e41a 46#include "hw/xen/xen.h"
9c17d615 47#include "sysemu/blockdev.h"
0d09e41a 48#include "hw/block/block.h"
a19cbfb3 49#include "ui/qemu-spice.h"
022c62cb
PB
50#include "exec/memory.h"
51#include "exec/address-spaces.h"
9c17d615 52#include "sysemu/arch_init.h"
1de7afc9 53#include "qemu/bitmap.h"
0c764a9d 54#include "qemu/config-file.h"
0445259b 55#include "hw/acpi/acpi.h"
5ff020b7 56#include "hw/acpi/cpu_hotplug.h"
53a89e26 57#include "hw/cpu/icc_bus.h"
c649983b 58#include "hw/boards.h"
39848901 59#include "hw/pci/pci_host.h"
72c194f7 60#include "acpi-build.h"
80cabfad 61
471fd342
BS
62/* debug PC/ISA interrupts */
63//#define DEBUG_IRQ
64
65#ifdef DEBUG_IRQ
66#define DPRINTF(fmt, ...) \
67 do { printf("CPUIRQ: " fmt , ## __VA_ARGS__); } while (0)
68#else
69#define DPRINTF(fmt, ...)
70#endif
71
a80274c3
PB
72/* Leave a chunk of memory at the top of RAM for the BIOS ACPI tables. */
73#define ACPI_DATA_SIZE 0x10000
3cce6243 74#define BIOS_CFG_IOPORT 0x510
8a92ea2f 75#define FW_CFG_ACPI_TABLES (FW_CFG_ARCH_LOCAL + 0)
b6f6e3d3 76#define FW_CFG_SMBIOS_ENTRIES (FW_CFG_ARCH_LOCAL + 1)
6b35e7bf 77#define FW_CFG_IRQ0_OVERRIDE (FW_CFG_ARCH_LOCAL + 2)
4c5b10b7 78#define FW_CFG_E820_TABLE (FW_CFG_ARCH_LOCAL + 3)
40ac17cd 79#define FW_CFG_HPET (FW_CFG_ARCH_LOCAL + 4)
80cabfad 80
4c5b10b7
JS
81#define E820_NR_ENTRIES 16
82
83struct e820_entry {
84 uint64_t address;
85 uint64_t length;
86 uint32_t type;
541dc0d4 87} QEMU_PACKED __attribute((__aligned__(4)));
4c5b10b7
JS
88
89struct e820_table {
90 uint32_t count;
91 struct e820_entry entry[E820_NR_ENTRIES];
541dc0d4 92} QEMU_PACKED __attribute((__aligned__(4)));
4c5b10b7 93
7d67110f
GH
94static struct e820_table e820_reserve;
95static struct e820_entry *e820_table;
96static unsigned e820_entries;
dd703b99 97struct hpet_fw_config hpet_cfg = {.count = UINT8_MAX};
4c5b10b7 98
b881fbe9 99void gsi_handler(void *opaque, int n, int level)
1452411b 100{
b881fbe9 101 GSIState *s = opaque;
1452411b 102
b881fbe9
JK
103 DPRINTF("pc: %s GSI %d\n", level ? "raising" : "lowering", n);
104 if (n < ISA_NUM_IRQS) {
105 qemu_set_irq(s->i8259_irq[n], level);
1632dc6a 106 }
b881fbe9 107 qemu_set_irq(s->ioapic_irq[n], level);
2e9947d2 108}
1452411b 109
258711c6
JG
110static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
111 unsigned size)
80cabfad
FB
112{
113}
114
c02e1eac
JG
115static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
116{
a6fc23e5 117 return 0xffffffffffffffffULL;
c02e1eac
JG
118}
119
f929aad6 120/* MSDOS compatibility mode FPU exception support */
d537cf6c 121static qemu_irq ferr_irq;
8e78eb28
IY
122
123void pc_register_ferr_irq(qemu_irq irq)
124{
125 ferr_irq = irq;
126}
127
f929aad6
FB
128/* XXX: add IGNNE support */
129void cpu_set_ferr(CPUX86State *s)
130{
d537cf6c 131 qemu_irq_raise(ferr_irq);
f929aad6
FB
132}
133
258711c6
JG
134static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
135 unsigned size)
f929aad6 136{
d537cf6c 137 qemu_irq_lower(ferr_irq);
f929aad6
FB
138}
139
c02e1eac
JG
140static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
141{
a6fc23e5 142 return 0xffffffffffffffffULL;
c02e1eac
JG
143}
144
28ab0e2e 145/* TSC handling */
28ab0e2e
FB
146uint64_t cpu_get_tsc(CPUX86State *env)
147{
4a1418e0 148 return cpu_get_ticks();
28ab0e2e
FB
149}
150
a5954d5c 151/* SMM support */
f885f1ea
IY
152
153static cpu_set_smm_t smm_set;
154static void *smm_arg;
155
156void cpu_smm_register(cpu_set_smm_t callback, void *arg)
157{
158 assert(smm_set == NULL);
159 assert(smm_arg == NULL);
160 smm_set = callback;
161 smm_arg = arg;
162}
163
4a8fa5dc 164void cpu_smm_update(CPUX86State *env)
a5954d5c 165{
182735ef 166 if (smm_set && smm_arg && CPU(x86_env_get_cpu(env)) == first_cpu) {
f885f1ea 167 smm_set(!!(env->hflags & HF_SMM_MASK), smm_arg);
182735ef 168 }
a5954d5c
FB
169}
170
171
3de388f6 172/* IRQ handling */
4a8fa5dc 173int cpu_get_pic_interrupt(CPUX86State *env)
3de388f6 174{
02e51483 175 X86CPU *cpu = x86_env_get_cpu(env);
3de388f6
FB
176 int intno;
177
02e51483 178 intno = apic_get_interrupt(cpu->apic_state);
3de388f6 179 if (intno >= 0) {
3de388f6
FB
180 return intno;
181 }
3de388f6 182 /* read the irq from the PIC */
02e51483 183 if (!apic_accept_pic_intr(cpu->apic_state)) {
0e21e12b 184 return -1;
cf6d64bf 185 }
0e21e12b 186
3de388f6
FB
187 intno = pic_read_irq(isa_pic);
188 return intno;
189}
190
d537cf6c 191static void pic_irq_request(void *opaque, int irq, int level)
3de388f6 192{
182735ef
AF
193 CPUState *cs = first_cpu;
194 X86CPU *cpu = X86_CPU(cs);
a5b38b51 195
471fd342 196 DPRINTF("pic_irqs: %s irq %d\n", level? "raise" : "lower", irq);
02e51483 197 if (cpu->apic_state) {
bdc44640 198 CPU_FOREACH(cs) {
182735ef 199 cpu = X86_CPU(cs);
02e51483
CF
200 if (apic_accept_pic_intr(cpu->apic_state)) {
201 apic_deliver_pic_intr(cpu->apic_state, level);
cf6d64bf 202 }
d5529471
AJ
203 }
204 } else {
d8ed887b 205 if (level) {
c3affe56 206 cpu_interrupt(cs, CPU_INTERRUPT_HARD);
d8ed887b
AF
207 } else {
208 cpu_reset_interrupt(cs, CPU_INTERRUPT_HARD);
209 }
a5b38b51 210 }
3de388f6
FB
211}
212
b0a21b53
FB
213/* PC cmos mappings */
214
80cabfad
FB
215#define REG_EQUIPMENT_BYTE 0x14
216
d288c7ba 217static int cmos_get_fd_drive_type(FDriveType fd0)
777428f2
FB
218{
219 int val;
220
221 switch (fd0) {
d288c7ba 222 case FDRIVE_DRV_144:
777428f2
FB
223 /* 1.44 Mb 3"5 drive */
224 val = 4;
225 break;
d288c7ba 226 case FDRIVE_DRV_288:
777428f2
FB
227 /* 2.88 Mb 3"5 drive */
228 val = 5;
229 break;
d288c7ba 230 case FDRIVE_DRV_120:
777428f2
FB
231 /* 1.2 Mb 5"5 drive */
232 val = 2;
233 break;
d288c7ba 234 case FDRIVE_DRV_NONE:
777428f2
FB
235 default:
236 val = 0;
237 break;
238 }
239 return val;
240}
241
9139046c
MA
242static void cmos_init_hd(ISADevice *s, int type_ofs, int info_ofs,
243 int16_t cylinders, int8_t heads, int8_t sectors)
ba6c2377 244{
ba6c2377
FB
245 rtc_set_memory(s, type_ofs, 47);
246 rtc_set_memory(s, info_ofs, cylinders);
247 rtc_set_memory(s, info_ofs + 1, cylinders >> 8);
248 rtc_set_memory(s, info_ofs + 2, heads);
249 rtc_set_memory(s, info_ofs + 3, 0xff);
250 rtc_set_memory(s, info_ofs + 4, 0xff);
251 rtc_set_memory(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
252 rtc_set_memory(s, info_ofs + 6, cylinders);
253 rtc_set_memory(s, info_ofs + 7, cylinders >> 8);
254 rtc_set_memory(s, info_ofs + 8, sectors);
255}
256
6ac0e82d
AZ
257/* convert boot_device letter to something recognizable by the bios */
258static int boot_device2nibble(char boot_device)
259{
260 switch(boot_device) {
261 case 'a':
262 case 'b':
263 return 0x01; /* floppy boot */
264 case 'c':
265 return 0x02; /* hard drive boot */
266 case 'd':
267 return 0x03; /* CD-ROM boot */
268 case 'n':
269 return 0x04; /* Network boot */
270 }
271 return 0;
272}
273
e1123015 274static int set_boot_dev(ISADevice *s, const char *boot_device)
0ecdffbb
AJ
275{
276#define PC_MAX_BOOT_DEVICES 3
0ecdffbb
AJ
277 int nbds, bds[3] = { 0, };
278 int i;
279
280 nbds = strlen(boot_device);
281 if (nbds > PC_MAX_BOOT_DEVICES) {
1ecda02b 282 error_report("Too many boot devices for PC");
0ecdffbb
AJ
283 return(1);
284 }
285 for (i = 0; i < nbds; i++) {
286 bds[i] = boot_device2nibble(boot_device[i]);
287 if (bds[i] == 0) {
1ecda02b
MA
288 error_report("Invalid boot device for PC: '%c'",
289 boot_device[i]);
0ecdffbb
AJ
290 return(1);
291 }
292 }
293 rtc_set_memory(s, 0x3d, (bds[1] << 4) | bds[0]);
d9346e81 294 rtc_set_memory(s, 0x38, (bds[2] << 4) | (fd_bootchk ? 0x0 : 0x1));
0ecdffbb
AJ
295 return(0);
296}
297
d9346e81
MA
298static int pc_boot_set(void *opaque, const char *boot_device)
299{
e1123015 300 return set_boot_dev(opaque, boot_device);
d9346e81
MA
301}
302
c0897e0c
MA
303typedef struct pc_cmos_init_late_arg {
304 ISADevice *rtc_state;
9139046c 305 BusState *idebus[2];
c0897e0c
MA
306} pc_cmos_init_late_arg;
307
308static void pc_cmos_init_late(void *opaque)
309{
310 pc_cmos_init_late_arg *arg = opaque;
311 ISADevice *s = arg->rtc_state;
9139046c
MA
312 int16_t cylinders;
313 int8_t heads, sectors;
c0897e0c 314 int val;
2adc99b2 315 int i, trans;
c0897e0c 316
9139046c
MA
317 val = 0;
318 if (ide_get_geometry(arg->idebus[0], 0,
319 &cylinders, &heads, &sectors) >= 0) {
320 cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
321 val |= 0xf0;
322 }
323 if (ide_get_geometry(arg->idebus[0], 1,
324 &cylinders, &heads, &sectors) >= 0) {
325 cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
326 val |= 0x0f;
327 }
328 rtc_set_memory(s, 0x12, val);
c0897e0c
MA
329
330 val = 0;
331 for (i = 0; i < 4; i++) {
9139046c
MA
332 /* NOTE: ide_get_geometry() returns the physical
333 geometry. It is always such that: 1 <= sects <= 63, 1
334 <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
335 geometry can be different if a translation is done. */
336 if (ide_get_geometry(arg->idebus[i / 2], i % 2,
337 &cylinders, &heads, &sectors) >= 0) {
2adc99b2
MA
338 trans = ide_get_bios_chs_trans(arg->idebus[i / 2], i % 2) - 1;
339 assert((trans & ~3) == 0);
340 val |= trans << (i * 2);
c0897e0c
MA
341 }
342 }
343 rtc_set_memory(s, 0x39, val);
344
345 qemu_unregister_reset(pc_cmos_init_late, opaque);
346}
347
b8b7456d
IM
348typedef struct RTCCPUHotplugArg {
349 Notifier cpu_added_notifier;
350 ISADevice *rtc_state;
351} RTCCPUHotplugArg;
352
353static void rtc_notify_cpu_added(Notifier *notifier, void *data)
354{
355 RTCCPUHotplugArg *arg = container_of(notifier, RTCCPUHotplugArg,
356 cpu_added_notifier);
357 ISADevice *s = arg->rtc_state;
358
359 /* increment the number of CPUs */
360 rtc_set_memory(s, 0x5f, rtc_get_memory(s, 0x5f) + 1);
361}
362
845773ab 363void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
c0897e0c 364 const char *boot_device,
34d4260e 365 ISADevice *floppy, BusState *idebus0, BusState *idebus1,
63ffb564 366 ISADevice *s)
80cabfad 367{
61a8d649 368 int val, nb, i;
980bda8b 369 FDriveType fd_type[2] = { FDRIVE_DRV_NONE, FDRIVE_DRV_NONE };
c0897e0c 370 static pc_cmos_init_late_arg arg;
b8b7456d 371 static RTCCPUHotplugArg cpu_hotplug_cb;
b0a21b53 372
b0a21b53 373 /* various important CMOS locations needed by PC/Bochs bios */
80cabfad
FB
374
375 /* memory size */
e89001f7
MA
376 /* base memory (first MiB) */
377 val = MIN(ram_size / 1024, 640);
333190eb
FB
378 rtc_set_memory(s, 0x15, val);
379 rtc_set_memory(s, 0x16, val >> 8);
e89001f7
MA
380 /* extended memory (next 64MiB) */
381 if (ram_size > 1024 * 1024) {
382 val = (ram_size - 1024 * 1024) / 1024;
383 } else {
384 val = 0;
385 }
80cabfad
FB
386 if (val > 65535)
387 val = 65535;
b0a21b53
FB
388 rtc_set_memory(s, 0x17, val);
389 rtc_set_memory(s, 0x18, val >> 8);
390 rtc_set_memory(s, 0x30, val);
391 rtc_set_memory(s, 0x31, val >> 8);
e89001f7
MA
392 /* memory between 16MiB and 4GiB */
393 if (ram_size > 16 * 1024 * 1024) {
394 val = (ram_size - 16 * 1024 * 1024) / 65536;
395 } else {
9da98861 396 val = 0;
e89001f7 397 }
80cabfad
FB
398 if (val > 65535)
399 val = 65535;
b0a21b53
FB
400 rtc_set_memory(s, 0x34, val);
401 rtc_set_memory(s, 0x35, val >> 8);
e89001f7
MA
402 /* memory above 4GiB */
403 val = above_4g_mem_size / 65536;
404 rtc_set_memory(s, 0x5b, val);
405 rtc_set_memory(s, 0x5c, val >> 8);
406 rtc_set_memory(s, 0x5d, val >> 16);
3b46e624 407
298e01b6
AJ
408 /* set the number of CPU */
409 rtc_set_memory(s, 0x5f, smp_cpus - 1);
b8b7456d
IM
410 /* init CPU hotplug notifier */
411 cpu_hotplug_cb.rtc_state = s;
412 cpu_hotplug_cb.cpu_added_notifier.notify = rtc_notify_cpu_added;
413 qemu_register_cpu_added_notifier(&cpu_hotplug_cb.cpu_added_notifier);
298e01b6 414
e1123015 415 if (set_boot_dev(s, boot_device)) {
28c5af54
JM
416 exit(1);
417 }
80cabfad 418
b41a2cd1 419 /* floppy type */
34d4260e 420 if (floppy) {
34d4260e 421 for (i = 0; i < 2; i++) {
61a8d649 422 fd_type[i] = isa_fdc_get_drive_type(floppy, i);
63ffb564
BS
423 }
424 }
425 val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
426 cmos_get_fd_drive_type(fd_type[1]);
b0a21b53 427 rtc_set_memory(s, 0x10, val);
3b46e624 428
b0a21b53 429 val = 0;
b41a2cd1 430 nb = 0;
63ffb564 431 if (fd_type[0] < FDRIVE_DRV_NONE) {
80cabfad 432 nb++;
d288c7ba 433 }
63ffb564 434 if (fd_type[1] < FDRIVE_DRV_NONE) {
80cabfad 435 nb++;
d288c7ba 436 }
80cabfad
FB
437 switch (nb) {
438 case 0:
439 break;
440 case 1:
b0a21b53 441 val |= 0x01; /* 1 drive, ready for boot */
80cabfad
FB
442 break;
443 case 2:
b0a21b53 444 val |= 0x41; /* 2 drives, ready for boot */
80cabfad
FB
445 break;
446 }
b0a21b53
FB
447 val |= 0x02; /* FPU is there */
448 val |= 0x04; /* PS/2 mouse installed */
449 rtc_set_memory(s, REG_EQUIPMENT_BYTE, val);
450
ba6c2377 451 /* hard drives */
c0897e0c 452 arg.rtc_state = s;
9139046c
MA
453 arg.idebus[0] = idebus0;
454 arg.idebus[1] = idebus1;
c0897e0c 455 qemu_register_reset(pc_cmos_init_late, &arg);
80cabfad
FB
456}
457
a0881c64
AF
458#define TYPE_PORT92 "port92"
459#define PORT92(obj) OBJECT_CHECK(Port92State, (obj), TYPE_PORT92)
460
4b78a802
BS
461/* port 92 stuff: could be split off */
462typedef struct Port92State {
a0881c64
AF
463 ISADevice parent_obj;
464
23af670e 465 MemoryRegion io;
4b78a802
BS
466 uint8_t outport;
467 qemu_irq *a20_out;
468} Port92State;
469
93ef4192
AG
470static void port92_write(void *opaque, hwaddr addr, uint64_t val,
471 unsigned size)
4b78a802
BS
472{
473 Port92State *s = opaque;
4700a316 474 int oldval = s->outport;
4b78a802
BS
475
476 DPRINTF("port92: write 0x%02x\n", val);
477 s->outport = val;
478 qemu_set_irq(*s->a20_out, (val >> 1) & 1);
4700a316 479 if ((val & 1) && !(oldval & 1)) {
4b78a802
BS
480 qemu_system_reset_request();
481 }
482}
483
93ef4192
AG
484static uint64_t port92_read(void *opaque, hwaddr addr,
485 unsigned size)
4b78a802
BS
486{
487 Port92State *s = opaque;
488 uint32_t ret;
489
490 ret = s->outport;
491 DPRINTF("port92: read 0x%02x\n", ret);
492 return ret;
493}
494
495static void port92_init(ISADevice *dev, qemu_irq *a20_out)
496{
a0881c64 497 Port92State *s = PORT92(dev);
4b78a802
BS
498
499 s->a20_out = a20_out;
500}
501
502static const VMStateDescription vmstate_port92_isa = {
503 .name = "port92",
504 .version_id = 1,
505 .minimum_version_id = 1,
d49805ae 506 .fields = (VMStateField[]) {
4b78a802
BS
507 VMSTATE_UINT8(outport, Port92State),
508 VMSTATE_END_OF_LIST()
509 }
510};
511
512static void port92_reset(DeviceState *d)
513{
a0881c64 514 Port92State *s = PORT92(d);
4b78a802
BS
515
516 s->outport &= ~1;
517}
518
23af670e 519static const MemoryRegionOps port92_ops = {
93ef4192
AG
520 .read = port92_read,
521 .write = port92_write,
522 .impl = {
523 .min_access_size = 1,
524 .max_access_size = 1,
525 },
526 .endianness = DEVICE_LITTLE_ENDIAN,
23af670e
RH
527};
528
db895a1e 529static void port92_initfn(Object *obj)
4b78a802 530{
db895a1e 531 Port92State *s = PORT92(obj);
4b78a802 532
1437c94b 533 memory_region_init_io(&s->io, OBJECT(s), &port92_ops, s, "port92", 1);
23af670e 534
4b78a802 535 s->outport = 0;
db895a1e
AF
536}
537
538static void port92_realizefn(DeviceState *dev, Error **errp)
539{
540 ISADevice *isadev = ISA_DEVICE(dev);
541 Port92State *s = PORT92(dev);
542
543 isa_register_ioport(isadev, &s->io, 0x92);
4b78a802
BS
544}
545
8f04ee08
AL
546static void port92_class_initfn(ObjectClass *klass, void *data)
547{
39bffca2 548 DeviceClass *dc = DEVICE_CLASS(klass);
db895a1e 549
db895a1e 550 dc->realize = port92_realizefn;
39bffca2
AL
551 dc->reset = port92_reset;
552 dc->vmsd = &vmstate_port92_isa;
f3b17640
MA
553 /*
554 * Reason: unlike ordinary ISA devices, this one needs additional
555 * wiring: its A20 output line needs to be wired up by
556 * port92_init().
557 */
558 dc->cannot_instantiate_with_device_add_yet = true;
8f04ee08
AL
559}
560
8c43a6f0 561static const TypeInfo port92_info = {
a0881c64 562 .name = TYPE_PORT92,
39bffca2
AL
563 .parent = TYPE_ISA_DEVICE,
564 .instance_size = sizeof(Port92State),
db895a1e 565 .instance_init = port92_initfn,
39bffca2 566 .class_init = port92_class_initfn,
4b78a802
BS
567};
568
83f7d43a 569static void port92_register_types(void)
4b78a802 570{
39bffca2 571 type_register_static(&port92_info);
4b78a802 572}
83f7d43a
AF
573
574type_init(port92_register_types)
4b78a802 575
956a3e6b 576static void handle_a20_line_change(void *opaque, int irq, int level)
59b8ad81 577{
cc36a7a2 578 X86CPU *cpu = opaque;
e1a23744 579
956a3e6b 580 /* XXX: send to all CPUs ? */
4b78a802 581 /* XXX: add logic to handle multiple A20 line sources */
cc36a7a2 582 x86_cpu_set_a20(cpu, level);
e1a23744
FB
583}
584
4c5b10b7
JS
585int e820_add_entry(uint64_t address, uint64_t length, uint32_t type)
586{
7d67110f 587 int index = le32_to_cpu(e820_reserve.count);
4c5b10b7
JS
588 struct e820_entry *entry;
589
7d67110f
GH
590 if (type != E820_RAM) {
591 /* old FW_CFG_E820_TABLE entry -- reservations only */
592 if (index >= E820_NR_ENTRIES) {
593 return -EBUSY;
594 }
595 entry = &e820_reserve.entry[index++];
596
597 entry->address = cpu_to_le64(address);
598 entry->length = cpu_to_le64(length);
599 entry->type = cpu_to_le32(type);
600
601 e820_reserve.count = cpu_to_le32(index);
602 }
4c5b10b7 603
7d67110f
GH
604 /* new "etc/e820" file -- include ram too */
605 e820_table = g_realloc(e820_table,
606 sizeof(struct e820_entry) * (e820_entries+1));
607 e820_table[e820_entries].address = cpu_to_le64(address);
608 e820_table[e820_entries].length = cpu_to_le64(length);
609 e820_table[e820_entries].type = cpu_to_le32(type);
610 e820_entries++;
4c5b10b7 611
7d67110f 612 return e820_entries;
4c5b10b7
JS
613}
614
7bf8ef19
GS
615int e820_get_num_entries(void)
616{
617 return e820_entries;
618}
619
620bool e820_get_entry(int idx, uint32_t type, uint64_t *address, uint64_t *length)
621{
622 if (idx < e820_entries && e820_table[idx].type == cpu_to_le32(type)) {
623 *address = le64_to_cpu(e820_table[idx].address);
624 *length = le64_to_cpu(e820_table[idx].length);
625 return true;
626 }
627 return false;
628}
629
1d934e89
EH
630/* Calculates the limit to CPU APIC ID values
631 *
632 * This function returns the limit for the APIC ID value, so that all
633 * CPU APIC IDs are < pc_apic_id_limit().
634 *
635 * This is used for FW_CFG_MAX_CPUS. See comments on bochs_bios_init().
636 */
637static unsigned int pc_apic_id_limit(unsigned int max_cpus)
638{
639 return x86_cpu_apic_id_from_index(max_cpus - 1) + 1;
640}
641
a88b362c 642static FWCfgState *bochs_bios_init(void)
80cabfad 643{
a88b362c 644 FWCfgState *fw_cfg;
c97294ec
GS
645 uint8_t *smbios_tables, *smbios_anchor;
646 size_t smbios_tables_len, smbios_anchor_len;
11c2fd3e
AL
647 uint64_t *numa_fw_cfg;
648 int i, j;
1d934e89 649 unsigned int apic_id_limit = pc_apic_id_limit(max_cpus);
3cce6243
BS
650
651 fw_cfg = fw_cfg_init(BIOS_CFG_IOPORT, BIOS_CFG_IOPORT + 1, 0, 0);
1d934e89
EH
652 /* FW_CFG_MAX_CPUS is a bit confusing/problematic on x86:
653 *
654 * SeaBIOS needs FW_CFG_MAX_CPUS for CPU hotplug, but the CPU hotplug
655 * QEMU<->SeaBIOS interface is not based on the "CPU index", but on the APIC
656 * ID of hotplugged CPUs[1]. This means that FW_CFG_MAX_CPUS is not the
657 * "maximum number of CPUs", but the "limit to the APIC ID values SeaBIOS
658 * may see".
659 *
660 * So, this means we must not use max_cpus, here, but the maximum possible
661 * APIC ID value, plus one.
662 *
663 * [1] The only kind of "CPU identifier" used between SeaBIOS and QEMU is
664 * the APIC ID, not the "CPU index"
665 */
666 fw_cfg_add_i16(fw_cfg, FW_CFG_MAX_CPUS, (uint16_t)apic_id_limit);
3cce6243 667 fw_cfg_add_i32(fw_cfg, FW_CFG_ID, 1);
905fdcb5 668 fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, (uint64_t)ram_size);
089da572
MA
669 fw_cfg_add_bytes(fw_cfg, FW_CFG_ACPI_TABLES,
670 acpi_tables, acpi_tables_len);
9b5b76d4 671 fw_cfg_add_i32(fw_cfg, FW_CFG_IRQ0_OVERRIDE, kvm_allows_irq0_override());
b6f6e3d3 672
c97294ec
GS
673 smbios_tables = smbios_get_table_legacy(&smbios_tables_len);
674 if (smbios_tables) {
b6f6e3d3 675 fw_cfg_add_bytes(fw_cfg, FW_CFG_SMBIOS_ENTRIES,
c97294ec
GS
676 smbios_tables, smbios_tables_len);
677 }
678
679 smbios_get_tables(&smbios_tables, &smbios_tables_len,
680 &smbios_anchor, &smbios_anchor_len);
681 if (smbios_anchor) {
682 fw_cfg_add_file(fw_cfg, "etc/smbios/smbios-tables",
683 smbios_tables, smbios_tables_len);
684 fw_cfg_add_file(fw_cfg, "etc/smbios/smbios-anchor",
685 smbios_anchor, smbios_anchor_len);
686 }
687
089da572 688 fw_cfg_add_bytes(fw_cfg, FW_CFG_E820_TABLE,
7d67110f
GH
689 &e820_reserve, sizeof(e820_reserve));
690 fw_cfg_add_file(fw_cfg, "etc/e820", e820_table,
691 sizeof(struct e820_entry) * e820_entries);
11c2fd3e 692
089da572 693 fw_cfg_add_bytes(fw_cfg, FW_CFG_HPET, &hpet_cfg, sizeof(hpet_cfg));
11c2fd3e
AL
694 /* allocate memory for the NUMA channel: one (64bit) word for the number
695 * of nodes, one word for each VCPU->node and one word for each node to
696 * hold the amount of memory.
697 */
1d934e89 698 numa_fw_cfg = g_new0(uint64_t, 1 + apic_id_limit + nb_numa_nodes);
11c2fd3e 699 numa_fw_cfg[0] = cpu_to_le64(nb_numa_nodes);
991dfefd 700 for (i = 0; i < max_cpus; i++) {
1d934e89
EH
701 unsigned int apic_id = x86_cpu_apic_id_from_index(i);
702 assert(apic_id < apic_id_limit);
11c2fd3e 703 for (j = 0; j < nb_numa_nodes; j++) {
ee785fed 704 if (test_bit(i, node_cpumask[j])) {
1d934e89 705 numa_fw_cfg[apic_id + 1] = cpu_to_le64(j);
11c2fd3e
AL
706 break;
707 }
708 }
709 }
710 for (i = 0; i < nb_numa_nodes; i++) {
1d934e89 711 numa_fw_cfg[apic_id_limit + 1 + i] = cpu_to_le64(node_mem[i]);
11c2fd3e 712 }
089da572 713 fw_cfg_add_bytes(fw_cfg, FW_CFG_NUMA, numa_fw_cfg,
1d934e89
EH
714 (1 + apic_id_limit + nb_numa_nodes) *
715 sizeof(*numa_fw_cfg));
bf483392
AG
716
717 return fw_cfg;
80cabfad
FB
718}
719
642a4f96
TS
720static long get_file_size(FILE *f)
721{
722 long where, size;
723
724 /* XXX: on Unix systems, using fstat() probably makes more sense */
725
726 where = ftell(f);
727 fseek(f, 0, SEEK_END);
728 size = ftell(f);
729 fseek(f, where, SEEK_SET);
730
731 return size;
732}
733
a88b362c 734static void load_linux(FWCfgState *fw_cfg,
4fc9af53 735 const char *kernel_filename,
0f9d76e5
LG
736 const char *initrd_filename,
737 const char *kernel_cmdline,
a8170e5e 738 hwaddr max_ram_size)
642a4f96
TS
739{
740 uint16_t protocol;
5cea8590 741 int setup_size, kernel_size, initrd_size = 0, cmdline_size;
642a4f96 742 uint32_t initrd_max;
57a46d05 743 uint8_t header[8192], *setup, *kernel, *initrd_data;
a8170e5e 744 hwaddr real_addr, prot_addr, cmdline_addr, initrd_addr = 0;
45a50b16 745 FILE *f;
bf4e5d92 746 char *vmode;
642a4f96
TS
747
748 /* Align to 16 bytes as a paranoia measure */
749 cmdline_size = (strlen(kernel_cmdline)+16) & ~15;
750
751 /* load the kernel header */
752 f = fopen(kernel_filename, "rb");
753 if (!f || !(kernel_size = get_file_size(f)) ||
0f9d76e5
LG
754 fread(header, 1, MIN(ARRAY_SIZE(header), kernel_size), f) !=
755 MIN(ARRAY_SIZE(header), kernel_size)) {
756 fprintf(stderr, "qemu: could not load kernel '%s': %s\n",
757 kernel_filename, strerror(errno));
758 exit(1);
642a4f96
TS
759 }
760
761 /* kernel protocol version */
bc4edd79 762#if 0
642a4f96 763 fprintf(stderr, "header magic: %#x\n", ldl_p(header+0x202));
bc4edd79 764#endif
0f9d76e5
LG
765 if (ldl_p(header+0x202) == 0x53726448) {
766 protocol = lduw_p(header+0x206);
767 } else {
768 /* This looks like a multiboot kernel. If it is, let's stop
769 treating it like a Linux kernel. */
52001445 770 if (load_multiboot(fw_cfg, f, kernel_filename, initrd_filename,
0f9d76e5 771 kernel_cmdline, kernel_size, header)) {
82663ee2 772 return;
0f9d76e5
LG
773 }
774 protocol = 0;
f16408df 775 }
642a4f96
TS
776
777 if (protocol < 0x200 || !(header[0x211] & 0x01)) {
0f9d76e5
LG
778 /* Low kernel */
779 real_addr = 0x90000;
780 cmdline_addr = 0x9a000 - cmdline_size;
781 prot_addr = 0x10000;
642a4f96 782 } else if (protocol < 0x202) {
0f9d76e5
LG
783 /* High but ancient kernel */
784 real_addr = 0x90000;
785 cmdline_addr = 0x9a000 - cmdline_size;
786 prot_addr = 0x100000;
642a4f96 787 } else {
0f9d76e5
LG
788 /* High and recent kernel */
789 real_addr = 0x10000;
790 cmdline_addr = 0x20000;
791 prot_addr = 0x100000;
642a4f96
TS
792 }
793
bc4edd79 794#if 0
642a4f96 795 fprintf(stderr,
0f9d76e5
LG
796 "qemu: real_addr = 0x" TARGET_FMT_plx "\n"
797 "qemu: cmdline_addr = 0x" TARGET_FMT_plx "\n"
798 "qemu: prot_addr = 0x" TARGET_FMT_plx "\n",
799 real_addr,
800 cmdline_addr,
801 prot_addr);
bc4edd79 802#endif
642a4f96
TS
803
804 /* highest address for loading the initrd */
0f9d76e5
LG
805 if (protocol >= 0x203) {
806 initrd_max = ldl_p(header+0x22c);
807 } else {
808 initrd_max = 0x37ffffff;
809 }
642a4f96 810
e6ade764
GC
811 if (initrd_max >= max_ram_size-ACPI_DATA_SIZE)
812 initrd_max = max_ram_size-ACPI_DATA_SIZE-1;
642a4f96 813
57a46d05
AG
814 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_ADDR, cmdline_addr);
815 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_SIZE, strlen(kernel_cmdline)+1);
96f80586 816 fw_cfg_add_string(fw_cfg, FW_CFG_CMDLINE_DATA, kernel_cmdline);
642a4f96
TS
817
818 if (protocol >= 0x202) {
0f9d76e5 819 stl_p(header+0x228, cmdline_addr);
642a4f96 820 } else {
0f9d76e5
LG
821 stw_p(header+0x20, 0xA33F);
822 stw_p(header+0x22, cmdline_addr-real_addr);
642a4f96
TS
823 }
824
bf4e5d92
PT
825 /* handle vga= parameter */
826 vmode = strstr(kernel_cmdline, "vga=");
827 if (vmode) {
828 unsigned int video_mode;
829 /* skip "vga=" */
830 vmode += 4;
831 if (!strncmp(vmode, "normal", 6)) {
832 video_mode = 0xffff;
833 } else if (!strncmp(vmode, "ext", 3)) {
834 video_mode = 0xfffe;
835 } else if (!strncmp(vmode, "ask", 3)) {
836 video_mode = 0xfffd;
837 } else {
838 video_mode = strtol(vmode, NULL, 0);
839 }
840 stw_p(header+0x1fa, video_mode);
841 }
842
642a4f96 843 /* loader type */
5cbdb3a3 844 /* High nybble = B reserved for QEMU; low nybble is revision number.
642a4f96
TS
845 If this code is substantially changed, you may want to consider
846 incrementing the revision. */
0f9d76e5
LG
847 if (protocol >= 0x200) {
848 header[0x210] = 0xB0;
849 }
642a4f96
TS
850 /* heap */
851 if (protocol >= 0x201) {
0f9d76e5
LG
852 header[0x211] |= 0x80; /* CAN_USE_HEAP */
853 stw_p(header+0x224, cmdline_addr-real_addr-0x200);
642a4f96
TS
854 }
855
856 /* load initrd */
857 if (initrd_filename) {
0f9d76e5
LG
858 if (protocol < 0x200) {
859 fprintf(stderr, "qemu: linux kernel too old to load a ram disk\n");
860 exit(1);
861 }
642a4f96 862
0f9d76e5 863 initrd_size = get_image_size(initrd_filename);
d6fa4b77 864 if (initrd_size < 0) {
7454e51d
MT
865 fprintf(stderr, "qemu: error reading initrd %s: %s\n",
866 initrd_filename, strerror(errno));
d6fa4b77
MK
867 exit(1);
868 }
869
45a50b16 870 initrd_addr = (initrd_max-initrd_size) & ~4095;
57a46d05 871
7267c094 872 initrd_data = g_malloc(initrd_size);
57a46d05
AG
873 load_image(initrd_filename, initrd_data);
874
875 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_ADDR, initrd_addr);
876 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_SIZE, initrd_size);
877 fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, initrd_data, initrd_size);
642a4f96 878
0f9d76e5
LG
879 stl_p(header+0x218, initrd_addr);
880 stl_p(header+0x21c, initrd_size);
642a4f96
TS
881 }
882
45a50b16 883 /* load kernel and setup */
642a4f96 884 setup_size = header[0x1f1];
0f9d76e5
LG
885 if (setup_size == 0) {
886 setup_size = 4;
887 }
642a4f96 888 setup_size = (setup_size+1)*512;
45a50b16 889 kernel_size -= setup_size;
642a4f96 890
7267c094
AL
891 setup = g_malloc(setup_size);
892 kernel = g_malloc(kernel_size);
45a50b16 893 fseek(f, 0, SEEK_SET);
5a41ecc5
KS
894 if (fread(setup, 1, setup_size, f) != setup_size) {
895 fprintf(stderr, "fread() failed\n");
896 exit(1);
897 }
898 if (fread(kernel, 1, kernel_size, f) != kernel_size) {
899 fprintf(stderr, "fread() failed\n");
900 exit(1);
901 }
642a4f96 902 fclose(f);
45a50b16 903 memcpy(setup, header, MIN(sizeof(header), setup_size));
57a46d05
AG
904
905 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_ADDR, prot_addr);
906 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_SIZE, kernel_size);
907 fw_cfg_add_bytes(fw_cfg, FW_CFG_KERNEL_DATA, kernel, kernel_size);
908
909 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_ADDR, real_addr);
910 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_SIZE, setup_size);
911 fw_cfg_add_bytes(fw_cfg, FW_CFG_SETUP_DATA, setup, setup_size);
912
2e55e842
GN
913 option_rom[nb_option_roms].name = "linuxboot.bin";
914 option_rom[nb_option_roms].bootindex = 0;
57a46d05 915 nb_option_roms++;
642a4f96
TS
916}
917
b41a2cd1
FB
918#define NE2000_NB_MAX 6
919
675d6f82
BS
920static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
921 0x280, 0x380 };
922static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
b41a2cd1 923
48a18b3c 924void pc_init_ne2k_isa(ISABus *bus, NICInfo *nd)
a41b2ff2
PB
925{
926 static int nb_ne2k = 0;
927
928 if (nb_ne2k == NE2000_NB_MAX)
929 return;
48a18b3c 930 isa_ne2000_init(bus, ne2000_io[nb_ne2k],
9453c5bc 931 ne2000_irq[nb_ne2k], nd);
a41b2ff2
PB
932 nb_ne2k++;
933}
934
92a16d7a 935DeviceState *cpu_get_current_apic(void)
0e26b7b8 936{
4917cf44
AF
937 if (current_cpu) {
938 X86CPU *cpu = X86_CPU(current_cpu);
02e51483 939 return cpu->apic_state;
0e26b7b8
BS
940 } else {
941 return NULL;
942 }
943}
944
845773ab 945void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
53b67b30 946{
c3affe56 947 X86CPU *cpu = opaque;
53b67b30
BS
948
949 if (level) {
c3affe56 950 cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
53b67b30
BS
951 }
952}
953
62fc403f
IM
954static X86CPU *pc_new_cpu(const char *cpu_model, int64_t apic_id,
955 DeviceState *icc_bridge, Error **errp)
31050930
IM
956{
957 X86CPU *cpu;
958 Error *local_err = NULL;
959
cd7b87ff
AF
960 cpu = cpu_x86_create(cpu_model, icc_bridge, &local_err);
961 if (local_err != NULL) {
962 error_propagate(errp, local_err);
963 return NULL;
31050930
IM
964 }
965
966 object_property_set_int(OBJECT(cpu), apic_id, "apic-id", &local_err);
967 object_property_set_bool(OBJECT(cpu), true, "realized", &local_err);
968
969 if (local_err) {
31050930 970 error_propagate(errp, local_err);
cd7b87ff
AF
971 object_unref(OBJECT(cpu));
972 cpu = NULL;
31050930
IM
973 }
974 return cpu;
975}
976
c649983b
IM
977static const char *current_cpu_model;
978
979void pc_hot_add_cpu(const int64_t id, Error **errp)
980{
981 DeviceState *icc_bridge;
982 int64_t apic_id = x86_cpu_apic_id_from_index(id);
983
8de433cb
IM
984 if (id < 0) {
985 error_setg(errp, "Invalid CPU id: %" PRIi64, id);
986 return;
987 }
988
c649983b
IM
989 if (cpu_exists(apic_id)) {
990 error_setg(errp, "Unable to add CPU: %" PRIi64
991 ", it already exists", id);
992 return;
993 }
994
995 if (id >= max_cpus) {
996 error_setg(errp, "Unable to add CPU: %" PRIi64
997 ", max allowed: %d", id, max_cpus - 1);
998 return;
999 }
1000
5ff020b7
EH
1001 if (apic_id >= ACPI_CPU_HOTPLUG_ID_LIMIT) {
1002 error_setg(errp, "Unable to add CPU: %" PRIi64
1003 ", resulting APIC ID (%" PRIi64 ") is too large",
1004 id, apic_id);
1005 return;
1006 }
1007
c649983b
IM
1008 icc_bridge = DEVICE(object_resolve_path_type("icc-bridge",
1009 TYPE_ICC_BRIDGE, NULL));
1010 pc_new_cpu(current_cpu_model, apic_id, icc_bridge, errp);
1011}
1012
62fc403f 1013void pc_cpus_init(const char *cpu_model, DeviceState *icc_bridge)
70166477
IY
1014{
1015 int i;
53a89e26 1016 X86CPU *cpu = NULL;
31050930 1017 Error *error = NULL;
f03bd716 1018 unsigned long apic_id_limit;
70166477
IY
1019
1020 /* init CPUs */
1021 if (cpu_model == NULL) {
1022#ifdef TARGET_X86_64
1023 cpu_model = "qemu64";
1024#else
1025 cpu_model = "qemu32";
1026#endif
1027 }
c649983b 1028 current_cpu_model = cpu_model;
70166477 1029
f03bd716
EH
1030 apic_id_limit = pc_apic_id_limit(max_cpus);
1031 if (apic_id_limit > ACPI_CPU_HOTPLUG_ID_LIMIT) {
1032 error_report("max_cpus is too large. APIC ID of last CPU is %lu",
1033 apic_id_limit - 1);
1034 exit(1);
1035 }
1036
bdeec802 1037 for (i = 0; i < smp_cpus; i++) {
53a89e26
IM
1038 cpu = pc_new_cpu(cpu_model, x86_cpu_apic_id_from_index(i),
1039 icc_bridge, &error);
31050930 1040 if (error) {
4a44d85e 1041 error_report("%s", error_get_pretty(error));
31050930 1042 error_free(error);
bdeec802
IM
1043 exit(1);
1044 }
70166477 1045 }
53a89e26
IM
1046
1047 /* map APIC MMIO area if CPU has APIC */
02e51483 1048 if (cpu && cpu->apic_state) {
53a89e26
IM
1049 /* XXX: what if the base changes? */
1050 sysbus_mmio_map_overlap(SYS_BUS_DEVICE(icc_bridge), 0,
1051 APIC_DEFAULT_ADDRESS, 0x1000);
1052 }
c97294ec
GS
1053
1054 /* tell smbios about cpuid version and features */
1055 smbios_set_cpuid(cpu->env.cpuid_version, cpu->env.features[FEAT_1_EDX]);
70166477
IY
1056}
1057
f8c457b8
MT
1058/* pci-info ROM file. Little endian format */
1059typedef struct PcRomPciInfo {
1060 uint64_t w32_min;
1061 uint64_t w32_max;
1062 uint64_t w64_min;
1063 uint64_t w64_max;
1064} PcRomPciInfo;
1065
1066static void pc_fw_cfg_guest_info(PcGuestInfo *guest_info)
1067{
1068 PcRomPciInfo *info;
39848901
IM
1069 Object *pci_info;
1070 bool ambiguous = false;
1071
d26d9e14 1072 if (!guest_info->has_pci_info || !guest_info->fw_cfg) {
f8c457b8
MT
1073 return;
1074 }
39848901
IM
1075 pci_info = object_resolve_path_type("", TYPE_PCI_HOST_BRIDGE, &ambiguous);
1076 g_assert(!ambiguous);
1077 if (!pci_info) {
1078 return;
1079 }
f8c457b8
MT
1080
1081 info = g_malloc(sizeof *info);
39848901
IM
1082 info->w32_min = cpu_to_le64(object_property_get_int(pci_info,
1083 PCI_HOST_PROP_PCI_HOLE_START, NULL));
1084 info->w32_max = cpu_to_le64(object_property_get_int(pci_info,
1085 PCI_HOST_PROP_PCI_HOLE_END, NULL));
1086 info->w64_min = cpu_to_le64(object_property_get_int(pci_info,
1087 PCI_HOST_PROP_PCI_HOLE64_START, NULL));
1088 info->w64_max = cpu_to_le64(object_property_get_int(pci_info,
1089 PCI_HOST_PROP_PCI_HOLE64_END, NULL));
f8c457b8
MT
1090 /* Pass PCI hole info to guest via a side channel.
1091 * Required so guest PCI enumeration does the right thing. */
1092 fw_cfg_add_file(guest_info->fw_cfg, "etc/pci-info", info, sizeof *info);
1093}
1094
3459a625
MT
1095typedef struct PcGuestInfoState {
1096 PcGuestInfo info;
1097 Notifier machine_done;
1098} PcGuestInfoState;
1099
1100static
1101void pc_guest_info_machine_done(Notifier *notifier, void *data)
1102{
1103 PcGuestInfoState *guest_info_state = container_of(notifier,
1104 PcGuestInfoState,
1105 machine_done);
f8c457b8 1106 pc_fw_cfg_guest_info(&guest_info_state->info);
72c194f7 1107 acpi_setup(&guest_info_state->info);
3459a625
MT
1108}
1109
1110PcGuestInfo *pc_guest_info_init(ram_addr_t below_4g_mem_size,
1111 ram_addr_t above_4g_mem_size)
1112{
1113 PcGuestInfoState *guest_info_state = g_malloc0(sizeof *guest_info_state);
1114 PcGuestInfo *guest_info = &guest_info_state->info;
b20c9bd5
MT
1115 int i, j;
1116
f30ee8a9 1117 guest_info->ram_size_below_4g = below_4g_mem_size;
b20c9bd5
MT
1118 guest_info->ram_size = below_4g_mem_size + above_4g_mem_size;
1119 guest_info->apic_id_limit = pc_apic_id_limit(max_cpus);
1120 guest_info->apic_xrupt_override = kvm_allows_irq0_override();
1121 guest_info->numa_nodes = nb_numa_nodes;
1122 guest_info->node_mem = g_memdup(node_mem, guest_info->numa_nodes *
1123 sizeof *guest_info->node_mem);
1124 guest_info->node_cpu = g_malloc0(guest_info->apic_id_limit *
1125 sizeof *guest_info->node_cpu);
1126
1127 for (i = 0; i < max_cpus; i++) {
1128 unsigned int apic_id = x86_cpu_apic_id_from_index(i);
1129 assert(apic_id < guest_info->apic_id_limit);
1130 for (j = 0; j < nb_numa_nodes; j++) {
1131 if (test_bit(i, node_cpumask[j])) {
1132 guest_info->node_cpu[apic_id] = j;
1133 break;
1134 }
1135 }
1136 }
3459a625 1137
3459a625
MT
1138 guest_info_state->machine_done.notify = pc_guest_info_machine_done;
1139 qemu_add_machine_init_done_notifier(&guest_info_state->machine_done);
1140 return guest_info;
1141}
1142
83d08f26
MT
1143/* setup pci memory address space mapping into system address space */
1144void pc_pci_as_mapping_init(Object *owner, MemoryRegion *system_memory,
1145 MemoryRegion *pci_address_space)
39848901 1146{
83d08f26
MT
1147 /* Set to lower priority than RAM */
1148 memory_region_add_subregion_overlap(system_memory, 0x0,
1149 pci_address_space, -1);
39848901
IM
1150}
1151
f7e4dd6c
GH
1152void pc_acpi_init(const char *default_dsdt)
1153{
c5a98cf3 1154 char *filename;
f7e4dd6c
GH
1155
1156 if (acpi_tables != NULL) {
1157 /* manually set via -acpitable, leave it alone */
1158 return;
1159 }
1160
1161 filename = qemu_find_file(QEMU_FILE_TYPE_BIOS, default_dsdt);
1162 if (filename == NULL) {
1163 fprintf(stderr, "WARNING: failed to find %s\n", default_dsdt);
c5a98cf3
LE
1164 } else {
1165 char *arg;
1166 QemuOpts *opts;
1167 Error *err = NULL;
f7e4dd6c 1168
c5a98cf3 1169 arg = g_strdup_printf("file=%s", filename);
0c764a9d 1170
c5a98cf3
LE
1171 /* creates a deep copy of "arg" */
1172 opts = qemu_opts_parse(qemu_find_opts("acpi"), arg, 0);
1173 g_assert(opts != NULL);
0c764a9d 1174
1a4b2666 1175 acpi_table_add_builtin(opts, &err);
c5a98cf3 1176 if (err) {
4a44d85e
SA
1177 error_report("WARNING: failed to load %s: %s", filename,
1178 error_get_pretty(err));
c5a98cf3
LE
1179 error_free(err);
1180 }
1181 g_free(arg);
1182 g_free(filename);
f7e4dd6c 1183 }
f7e4dd6c
GH
1184}
1185
a88b362c
LE
1186FWCfgState *pc_memory_init(MemoryRegion *system_memory,
1187 const char *kernel_filename,
1188 const char *kernel_cmdline,
1189 const char *initrd_filename,
1190 ram_addr_t below_4g_mem_size,
1191 ram_addr_t above_4g_mem_size,
1192 MemoryRegion *rom_memory,
3459a625
MT
1193 MemoryRegion **ram_memory,
1194 PcGuestInfo *guest_info)
80cabfad 1195{
cbc5b5f3
JJ
1196 int linux_boot, i;
1197 MemoryRegion *ram, *option_rom_mr;
00cb2a99 1198 MemoryRegion *ram_below_4g, *ram_above_4g;
a88b362c 1199 FWCfgState *fw_cfg;
d592d303 1200
80cabfad
FB
1201 linux_boot = (kernel_filename != NULL);
1202
00cb2a99 1203 /* Allocate RAM. We allocate it as a single memory region and use
66a0a2cb 1204 * aliases to address portions of it, mostly for backwards compatibility
00cb2a99
AK
1205 * with older qemus that used qemu_ram_alloc().
1206 */
7267c094 1207 ram = g_malloc(sizeof(*ram));
2c9b15ca 1208 memory_region_init_ram(ram, NULL, "pc.ram",
00cb2a99 1209 below_4g_mem_size + above_4g_mem_size);
c5705a77 1210 vmstate_register_ram_global(ram);
ae0a5466 1211 *ram_memory = ram;
7267c094 1212 ram_below_4g = g_malloc(sizeof(*ram_below_4g));
2c9b15ca 1213 memory_region_init_alias(ram_below_4g, NULL, "ram-below-4g", ram,
00cb2a99
AK
1214 0, below_4g_mem_size);
1215 memory_region_add_subregion(system_memory, 0, ram_below_4g);
7db16f24 1216 e820_add_entry(0, below_4g_mem_size, E820_RAM);
bbe80adf 1217 if (above_4g_mem_size > 0) {
7267c094 1218 ram_above_4g = g_malloc(sizeof(*ram_above_4g));
2c9b15ca 1219 memory_region_init_alias(ram_above_4g, NULL, "ram-above-4g", ram,
00cb2a99
AK
1220 below_4g_mem_size, above_4g_mem_size);
1221 memory_region_add_subregion(system_memory, 0x100000000ULL,
1222 ram_above_4g);
0624c7f9 1223 e820_add_entry(0x100000000ULL, above_4g_mem_size, E820_RAM);
bbe80adf 1224 }
82b36dc3 1225
cbc5b5f3
JJ
1226
1227 /* Initialize PC system firmware */
6dd2a5c9 1228 pc_system_firmware_init(rom_memory, guest_info->isapc_ram_fw);
00cb2a99 1229
7267c094 1230 option_rom_mr = g_malloc(sizeof(*option_rom_mr));
2c9b15ca 1231 memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE);
c5705a77 1232 vmstate_register_ram_global(option_rom_mr);
4463aee6 1233 memory_region_add_subregion_overlap(rom_memory,
00cb2a99
AK
1234 PC_ROM_MIN_VGA,
1235 option_rom_mr,
1236 1);
f753ff16 1237
bf483392 1238 fw_cfg = bochs_bios_init();
8832cb80 1239 rom_set_fw(fw_cfg);
1d108d97 1240
f753ff16 1241 if (linux_boot) {
81a204e4 1242 load_linux(fw_cfg, kernel_filename, initrd_filename, kernel_cmdline, below_4g_mem_size);
f753ff16
PB
1243 }
1244
1245 for (i = 0; i < nb_option_roms; i++) {
2e55e842 1246 rom_add_option(option_rom[i].name, option_rom[i].bootindex);
406c8df3 1247 }
3459a625 1248 guest_info->fw_cfg = fw_cfg;
459ae5ea 1249 return fw_cfg;
3d53f5c3
IY
1250}
1251
845773ab
IY
1252qemu_irq *pc_allocate_cpu_irq(void)
1253{
1254 return qemu_allocate_irqs(pic_irq_request, NULL, 1);
1255}
1256
48a18b3c 1257DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
765d7908 1258{
ad6d45fa
AL
1259 DeviceState *dev = NULL;
1260
16094b75
AJ
1261 if (pci_bus) {
1262 PCIDevice *pcidev = pci_vga_init(pci_bus);
1263 dev = pcidev ? &pcidev->qdev : NULL;
1264 } else if (isa_bus) {
1265 ISADevice *isadev = isa_vga_init(isa_bus);
4a17cc4f 1266 dev = isadev ? DEVICE(isadev) : NULL;
765d7908 1267 }
ad6d45fa 1268 return dev;
765d7908
IY
1269}
1270
4556bd8b
BS
1271static void cpu_request_exit(void *opaque, int irq, int level)
1272{
4917cf44 1273 CPUState *cpu = current_cpu;
4556bd8b 1274
4917cf44
AF
1275 if (cpu && level) {
1276 cpu_exit(cpu);
4556bd8b
BS
1277 }
1278}
1279
258711c6
JG
1280static const MemoryRegionOps ioport80_io_ops = {
1281 .write = ioport80_write,
c02e1eac 1282 .read = ioport80_read,
258711c6
JG
1283 .endianness = DEVICE_NATIVE_ENDIAN,
1284 .impl = {
1285 .min_access_size = 1,
1286 .max_access_size = 1,
1287 },
1288};
1289
1290static const MemoryRegionOps ioportF0_io_ops = {
1291 .write = ioportF0_write,
c02e1eac 1292 .read = ioportF0_read,
258711c6
JG
1293 .endianness = DEVICE_NATIVE_ENDIAN,
1294 .impl = {
1295 .min_access_size = 1,
1296 .max_access_size = 1,
1297 },
1298};
1299
48a18b3c 1300void pc_basic_device_init(ISABus *isa_bus, qemu_irq *gsi,
1611977c 1301 ISADevice **rtc_state,
34d4260e 1302 ISADevice **floppy,
7a10ef51
LPF
1303 bool no_vmport,
1304 uint32 hpet_irqs)
ffe513da
IY
1305{
1306 int i;
1307 DriveInfo *fd[MAX_FD];
ce967e2f
JK
1308 DeviceState *hpet = NULL;
1309 int pit_isa_irq = 0;
1310 qemu_irq pit_alt_irq = NULL;
7d932dfd 1311 qemu_irq rtc_irq = NULL;
956a3e6b 1312 qemu_irq *a20_line;
c2d8d311 1313 ISADevice *i8042, *port92, *vmmouse, *pit = NULL;
4556bd8b 1314 qemu_irq *cpu_exit_irq;
258711c6
JG
1315 MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1316 MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
ffe513da 1317
2c9b15ca 1318 memory_region_init_io(ioport80_io, NULL, &ioport80_io_ops, NULL, "ioport80", 1);
258711c6 1319 memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
ffe513da 1320
2c9b15ca 1321 memory_region_init_io(ioportF0_io, NULL, &ioportF0_io_ops, NULL, "ioportF0", 1);
258711c6 1322 memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
ffe513da 1323
5d17c0d2
JK
1324 /*
1325 * Check if an HPET shall be created.
1326 *
1327 * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
1328 * when the HPET wants to take over. Thus we have to disable the latter.
1329 */
1330 if (!no_hpet && (!kvm_irqchip_in_kernel() || kvm_has_pit_state2())) {
7a10ef51 1331 /* In order to set property, here not using sysbus_try_create_simple */
51116102 1332 hpet = qdev_try_create(NULL, TYPE_HPET);
dd703b99 1333 if (hpet) {
7a10ef51
LPF
1334 /* For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-1.7
1335 * and earlier, use IRQ2 for compat. Otherwise, use IRQ16~23,
1336 * IRQ8 and IRQ2.
1337 */
1338 uint8_t compat = object_property_get_int(OBJECT(hpet),
1339 HPET_INTCAP, NULL);
1340 if (!compat) {
1341 qdev_prop_set_uint32(hpet, HPET_INTCAP, hpet_irqs);
1342 }
1343 qdev_init_nofail(hpet);
1344 sysbus_mmio_map(SYS_BUS_DEVICE(hpet), 0, HPET_BASE);
1345
b881fbe9 1346 for (i = 0; i < GSI_NUM_PINS; i++) {
1356b98d 1347 sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
dd703b99 1348 }
ce967e2f
JK
1349 pit_isa_irq = -1;
1350 pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1351 rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
822557eb 1352 }
ffe513da 1353 }
48a18b3c 1354 *rtc_state = rtc_init(isa_bus, 2000, rtc_irq);
7d932dfd
JK
1355
1356 qemu_register_boot_set(pc_boot_set, *rtc_state);
1357
c2d8d311
SS
1358 if (!xen_enabled()) {
1359 if (kvm_irqchip_in_kernel()) {
1360 pit = kvm_pit_init(isa_bus, 0x40);
1361 } else {
1362 pit = pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1363 }
1364 if (hpet) {
1365 /* connect PIT to output control line of the HPET */
4a17cc4f 1366 qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(DEVICE(pit), 0));
c2d8d311
SS
1367 }
1368 pcspk_init(isa_bus, pit);
ce967e2f 1369 }
ffe513da
IY
1370
1371 for(i = 0; i < MAX_SERIAL_PORTS; i++) {
1372 if (serial_hds[i]) {
48a18b3c 1373 serial_isa_init(isa_bus, i, serial_hds[i]);
ffe513da
IY
1374 }
1375 }
1376
1377 for(i = 0; i < MAX_PARALLEL_PORTS; i++) {
1378 if (parallel_hds[i]) {
48a18b3c 1379 parallel_init(isa_bus, i, parallel_hds[i]);
ffe513da
IY
1380 }
1381 }
1382
182735ef 1383 a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
48a18b3c 1384 i8042 = isa_create_simple(isa_bus, "i8042");
4b78a802 1385 i8042_setup_a20_line(i8042, &a20_line[0]);
1611977c 1386 if (!no_vmport) {
48a18b3c
HP
1387 vmport_init(isa_bus);
1388 vmmouse = isa_try_create(isa_bus, "vmmouse");
1611977c
AP
1389 } else {
1390 vmmouse = NULL;
1391 }
86d86414 1392 if (vmmouse) {
4a17cc4f
AF
1393 DeviceState *dev = DEVICE(vmmouse);
1394 qdev_prop_set_ptr(dev, "ps2_mouse", i8042);
1395 qdev_init_nofail(dev);
86d86414 1396 }
48a18b3c 1397 port92 = isa_create_simple(isa_bus, "port92");
4b78a802 1398 port92_init(port92, &a20_line[1]);
956a3e6b 1399
4556bd8b
BS
1400 cpu_exit_irq = qemu_allocate_irqs(cpu_request_exit, NULL, 1);
1401 DMA_init(0, cpu_exit_irq);
ffe513da
IY
1402
1403 for(i = 0; i < MAX_FD; i++) {
1404 fd[i] = drive_get(IF_FLOPPY, 0, i);
1405 }
48a18b3c 1406 *floppy = fdctrl_init_isa(isa_bus, fd);
ffe513da
IY
1407}
1408
9011a1a7
IY
1409void pc_nic_init(ISABus *isa_bus, PCIBus *pci_bus)
1410{
1411 int i;
1412
1413 for (i = 0; i < nb_nics; i++) {
1414 NICInfo *nd = &nd_table[i];
1415
1416 if (!pci_bus || (nd->model && strcmp(nd->model, "ne2k_isa") == 0)) {
1417 pc_init_ne2k_isa(isa_bus, nd);
1418 } else {
29b358f9 1419 pci_nic_init_nofail(nd, pci_bus, "e1000", NULL);
9011a1a7
IY
1420 }
1421 }
1422}
1423
845773ab 1424void pc_pci_device_init(PCIBus *pci_bus)
e3a5cf42
IY
1425{
1426 int max_bus;
1427 int bus;
1428
1429 max_bus = drive_get_max_bus(IF_SCSI);
1430 for (bus = 0; bus <= max_bus; bus++) {
1431 pci_create_simple(pci_bus, -1, "lsi53c895a");
1432 }
1433}
a39e3564
JB
1434
1435void ioapic_init_gsi(GSIState *gsi_state, const char *parent_name)
1436{
1437 DeviceState *dev;
1438 SysBusDevice *d;
1439 unsigned int i;
1440
1441 if (kvm_irqchip_in_kernel()) {
1442 dev = qdev_create(NULL, "kvm-ioapic");
1443 } else {
1444 dev = qdev_create(NULL, "ioapic");
1445 }
1446 if (parent_name) {
1447 object_property_add_child(object_resolve_path(parent_name, NULL),
1448 "ioapic", OBJECT(dev), NULL);
1449 }
1450 qdev_init_nofail(dev);
1356b98d 1451 d = SYS_BUS_DEVICE(dev);
3a4a4697 1452 sysbus_mmio_map(d, 0, IO_APIC_DEFAULT_ADDRESS);
a39e3564
JB
1453
1454 for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1455 gsi_state->ioapic_irq[i] = qdev_get_gpio_in(dev, i);
1456 }
1457}