]> git.proxmox.com Git - qemu.git/blob - hw/pc.c
kvm: Move kvm_allows_irq0_override() to target-i386, fix return type
[qemu.git] / hw / pc.c
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.h"
25 #include "pc.h"
26 #include "apic.h"
27 #include "fdc.h"
28 #include "ide.h"
29 #include "pci.h"
30 #include "vmware_vga.h"
31 #include "monitor.h"
32 #include "fw_cfg.h"
33 #include "hpet_emul.h"
34 #include "smbios.h"
35 #include "loader.h"
36 #include "elf.h"
37 #include "multiboot.h"
38 #include "mc146818rtc.h"
39 #include "i8254.h"
40 #include "pcspk.h"
41 #include "msi.h"
42 #include "sysbus.h"
43 #include "sysemu.h"
44 #include "kvm.h"
45 #include "kvm_i386.h"
46 #include "xen.h"
47 #include "blockdev.h"
48 #include "hw/block-common.h"
49 #include "ui/qemu-spice.h"
50 #include "memory.h"
51 #include "exec-memory.h"
52 #include "arch_init.h"
53 #include "bitmap.h"
54
55 /* output Bochs bios info messages */
56 //#define DEBUG_BIOS
57
58 /* debug PC/ISA interrupts */
59 //#define DEBUG_IRQ
60
61 #ifdef DEBUG_IRQ
62 #define DPRINTF(fmt, ...) \
63 do { printf("CPUIRQ: " fmt , ## __VA_ARGS__); } while (0)
64 #else
65 #define DPRINTF(fmt, ...)
66 #endif
67
68 /* Leave a chunk of memory at the top of RAM for the BIOS ACPI tables. */
69 #define ACPI_DATA_SIZE 0x10000
70 #define BIOS_CFG_IOPORT 0x510
71 #define FW_CFG_ACPI_TABLES (FW_CFG_ARCH_LOCAL + 0)
72 #define FW_CFG_SMBIOS_ENTRIES (FW_CFG_ARCH_LOCAL + 1)
73 #define FW_CFG_IRQ0_OVERRIDE (FW_CFG_ARCH_LOCAL + 2)
74 #define FW_CFG_E820_TABLE (FW_CFG_ARCH_LOCAL + 3)
75 #define FW_CFG_HPET (FW_CFG_ARCH_LOCAL + 4)
76
77 #define MSI_ADDR_BASE 0xfee00000
78
79 #define E820_NR_ENTRIES 16
80
81 struct e820_entry {
82 uint64_t address;
83 uint64_t length;
84 uint32_t type;
85 } QEMU_PACKED __attribute((__aligned__(4)));
86
87 struct e820_table {
88 uint32_t count;
89 struct e820_entry entry[E820_NR_ENTRIES];
90 } QEMU_PACKED __attribute((__aligned__(4)));
91
92 static struct e820_table e820_table;
93 struct hpet_fw_config hpet_cfg = {.count = UINT8_MAX};
94
95 void gsi_handler(void *opaque, int n, int level)
96 {
97 GSIState *s = opaque;
98
99 DPRINTF("pc: %s GSI %d\n", level ? "raising" : "lowering", n);
100 if (n < ISA_NUM_IRQS) {
101 qemu_set_irq(s->i8259_irq[n], level);
102 }
103 qemu_set_irq(s->ioapic_irq[n], level);
104 }
105
106 static void ioport80_write(void *opaque, uint32_t addr, uint32_t data)
107 {
108 }
109
110 /* MSDOS compatibility mode FPU exception support */
111 static qemu_irq ferr_irq;
112
113 void pc_register_ferr_irq(qemu_irq irq)
114 {
115 ferr_irq = irq;
116 }
117
118 /* XXX: add IGNNE support */
119 void cpu_set_ferr(CPUX86State *s)
120 {
121 qemu_irq_raise(ferr_irq);
122 }
123
124 static void ioportF0_write(void *opaque, uint32_t addr, uint32_t data)
125 {
126 qemu_irq_lower(ferr_irq);
127 }
128
129 /* TSC handling */
130 uint64_t cpu_get_tsc(CPUX86State *env)
131 {
132 return cpu_get_ticks();
133 }
134
135 /* SMM support */
136
137 static cpu_set_smm_t smm_set;
138 static void *smm_arg;
139
140 void cpu_smm_register(cpu_set_smm_t callback, void *arg)
141 {
142 assert(smm_set == NULL);
143 assert(smm_arg == NULL);
144 smm_set = callback;
145 smm_arg = arg;
146 }
147
148 void cpu_smm_update(CPUX86State *env)
149 {
150 if (smm_set && smm_arg && env == first_cpu)
151 smm_set(!!(env->hflags & HF_SMM_MASK), smm_arg);
152 }
153
154
155 /* IRQ handling */
156 int cpu_get_pic_interrupt(CPUX86State *env)
157 {
158 int intno;
159
160 intno = apic_get_interrupt(env->apic_state);
161 if (intno >= 0) {
162 return intno;
163 }
164 /* read the irq from the PIC */
165 if (!apic_accept_pic_intr(env->apic_state)) {
166 return -1;
167 }
168
169 intno = pic_read_irq(isa_pic);
170 return intno;
171 }
172
173 static void pic_irq_request(void *opaque, int irq, int level)
174 {
175 CPUX86State *env = first_cpu;
176
177 DPRINTF("pic_irqs: %s irq %d\n", level? "raise" : "lower", irq);
178 if (env->apic_state) {
179 while (env) {
180 if (apic_accept_pic_intr(env->apic_state)) {
181 apic_deliver_pic_intr(env->apic_state, level);
182 }
183 env = env->next_cpu;
184 }
185 } else {
186 if (level)
187 cpu_interrupt(env, CPU_INTERRUPT_HARD);
188 else
189 cpu_reset_interrupt(env, CPU_INTERRUPT_HARD);
190 }
191 }
192
193 /* PC cmos mappings */
194
195 #define REG_EQUIPMENT_BYTE 0x14
196
197 static int cmos_get_fd_drive_type(FDriveType fd0)
198 {
199 int val;
200
201 switch (fd0) {
202 case FDRIVE_DRV_144:
203 /* 1.44 Mb 3"5 drive */
204 val = 4;
205 break;
206 case FDRIVE_DRV_288:
207 /* 2.88 Mb 3"5 drive */
208 val = 5;
209 break;
210 case FDRIVE_DRV_120:
211 /* 1.2 Mb 5"5 drive */
212 val = 2;
213 break;
214 case FDRIVE_DRV_NONE:
215 default:
216 val = 0;
217 break;
218 }
219 return val;
220 }
221
222 static void cmos_init_hd(ISADevice *s, int type_ofs, int info_ofs,
223 int16_t cylinders, int8_t heads, int8_t sectors)
224 {
225 rtc_set_memory(s, type_ofs, 47);
226 rtc_set_memory(s, info_ofs, cylinders);
227 rtc_set_memory(s, info_ofs + 1, cylinders >> 8);
228 rtc_set_memory(s, info_ofs + 2, heads);
229 rtc_set_memory(s, info_ofs + 3, 0xff);
230 rtc_set_memory(s, info_ofs + 4, 0xff);
231 rtc_set_memory(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
232 rtc_set_memory(s, info_ofs + 6, cylinders);
233 rtc_set_memory(s, info_ofs + 7, cylinders >> 8);
234 rtc_set_memory(s, info_ofs + 8, sectors);
235 }
236
237 /* convert boot_device letter to something recognizable by the bios */
238 static int boot_device2nibble(char boot_device)
239 {
240 switch(boot_device) {
241 case 'a':
242 case 'b':
243 return 0x01; /* floppy boot */
244 case 'c':
245 return 0x02; /* hard drive boot */
246 case 'd':
247 return 0x03; /* CD-ROM boot */
248 case 'n':
249 return 0x04; /* Network boot */
250 }
251 return 0;
252 }
253
254 static int set_boot_dev(ISADevice *s, const char *boot_device, int fd_bootchk)
255 {
256 #define PC_MAX_BOOT_DEVICES 3
257 int nbds, bds[3] = { 0, };
258 int i;
259
260 nbds = strlen(boot_device);
261 if (nbds > PC_MAX_BOOT_DEVICES) {
262 error_report("Too many boot devices for PC");
263 return(1);
264 }
265 for (i = 0; i < nbds; i++) {
266 bds[i] = boot_device2nibble(boot_device[i]);
267 if (bds[i] == 0) {
268 error_report("Invalid boot device for PC: '%c'",
269 boot_device[i]);
270 return(1);
271 }
272 }
273 rtc_set_memory(s, 0x3d, (bds[1] << 4) | bds[0]);
274 rtc_set_memory(s, 0x38, (bds[2] << 4) | (fd_bootchk ? 0x0 : 0x1));
275 return(0);
276 }
277
278 static int pc_boot_set(void *opaque, const char *boot_device)
279 {
280 return set_boot_dev(opaque, boot_device, 0);
281 }
282
283 typedef struct pc_cmos_init_late_arg {
284 ISADevice *rtc_state;
285 BusState *idebus[2];
286 } pc_cmos_init_late_arg;
287
288 static void pc_cmos_init_late(void *opaque)
289 {
290 pc_cmos_init_late_arg *arg = opaque;
291 ISADevice *s = arg->rtc_state;
292 int16_t cylinders;
293 int8_t heads, sectors;
294 int val;
295 int i, trans;
296
297 val = 0;
298 if (ide_get_geometry(arg->idebus[0], 0,
299 &cylinders, &heads, &sectors) >= 0) {
300 cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
301 val |= 0xf0;
302 }
303 if (ide_get_geometry(arg->idebus[0], 1,
304 &cylinders, &heads, &sectors) >= 0) {
305 cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
306 val |= 0x0f;
307 }
308 rtc_set_memory(s, 0x12, val);
309
310 val = 0;
311 for (i = 0; i < 4; i++) {
312 /* NOTE: ide_get_geometry() returns the physical
313 geometry. It is always such that: 1 <= sects <= 63, 1
314 <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
315 geometry can be different if a translation is done. */
316 if (ide_get_geometry(arg->idebus[i / 2], i % 2,
317 &cylinders, &heads, &sectors) >= 0) {
318 trans = ide_get_bios_chs_trans(arg->idebus[i / 2], i % 2) - 1;
319 assert((trans & ~3) == 0);
320 val |= trans << (i * 2);
321 }
322 }
323 rtc_set_memory(s, 0x39, val);
324
325 qemu_unregister_reset(pc_cmos_init_late, opaque);
326 }
327
328 void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
329 const char *boot_device,
330 ISADevice *floppy, BusState *idebus0, BusState *idebus1,
331 ISADevice *s)
332 {
333 int val, nb, i;
334 FDriveType fd_type[2] = { FDRIVE_DRV_NONE, FDRIVE_DRV_NONE };
335 static pc_cmos_init_late_arg arg;
336
337 /* various important CMOS locations needed by PC/Bochs bios */
338
339 /* memory size */
340 val = 640; /* base memory in K */
341 rtc_set_memory(s, 0x15, val);
342 rtc_set_memory(s, 0x16, val >> 8);
343
344 val = (ram_size / 1024) - 1024;
345 if (val > 65535)
346 val = 65535;
347 rtc_set_memory(s, 0x17, val);
348 rtc_set_memory(s, 0x18, val >> 8);
349 rtc_set_memory(s, 0x30, val);
350 rtc_set_memory(s, 0x31, val >> 8);
351
352 if (above_4g_mem_size) {
353 rtc_set_memory(s, 0x5b, (unsigned int)above_4g_mem_size >> 16);
354 rtc_set_memory(s, 0x5c, (unsigned int)above_4g_mem_size >> 24);
355 rtc_set_memory(s, 0x5d, (uint64_t)above_4g_mem_size >> 32);
356 }
357
358 if (ram_size > (16 * 1024 * 1024))
359 val = (ram_size / 65536) - ((16 * 1024 * 1024) / 65536);
360 else
361 val = 0;
362 if (val > 65535)
363 val = 65535;
364 rtc_set_memory(s, 0x34, val);
365 rtc_set_memory(s, 0x35, val >> 8);
366
367 /* set the number of CPU */
368 rtc_set_memory(s, 0x5f, smp_cpus - 1);
369
370 /* set boot devices, and disable floppy signature check if requested */
371 if (set_boot_dev(s, boot_device, fd_bootchk)) {
372 exit(1);
373 }
374
375 /* floppy type */
376 if (floppy) {
377 for (i = 0; i < 2; i++) {
378 fd_type[i] = isa_fdc_get_drive_type(floppy, i);
379 }
380 }
381 val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
382 cmos_get_fd_drive_type(fd_type[1]);
383 rtc_set_memory(s, 0x10, val);
384
385 val = 0;
386 nb = 0;
387 if (fd_type[0] < FDRIVE_DRV_NONE) {
388 nb++;
389 }
390 if (fd_type[1] < FDRIVE_DRV_NONE) {
391 nb++;
392 }
393 switch (nb) {
394 case 0:
395 break;
396 case 1:
397 val |= 0x01; /* 1 drive, ready for boot */
398 break;
399 case 2:
400 val |= 0x41; /* 2 drives, ready for boot */
401 break;
402 }
403 val |= 0x02; /* FPU is there */
404 val |= 0x04; /* PS/2 mouse installed */
405 rtc_set_memory(s, REG_EQUIPMENT_BYTE, val);
406
407 /* hard drives */
408 arg.rtc_state = s;
409 arg.idebus[0] = idebus0;
410 arg.idebus[1] = idebus1;
411 qemu_register_reset(pc_cmos_init_late, &arg);
412 }
413
414 /* port 92 stuff: could be split off */
415 typedef struct Port92State {
416 ISADevice dev;
417 MemoryRegion io;
418 uint8_t outport;
419 qemu_irq *a20_out;
420 } Port92State;
421
422 static void port92_write(void *opaque, uint32_t addr, uint32_t val)
423 {
424 Port92State *s = opaque;
425
426 DPRINTF("port92: write 0x%02x\n", val);
427 s->outport = val;
428 qemu_set_irq(*s->a20_out, (val >> 1) & 1);
429 if (val & 1) {
430 qemu_system_reset_request();
431 }
432 }
433
434 static uint32_t port92_read(void *opaque, uint32_t addr)
435 {
436 Port92State *s = opaque;
437 uint32_t ret;
438
439 ret = s->outport;
440 DPRINTF("port92: read 0x%02x\n", ret);
441 return ret;
442 }
443
444 static void port92_init(ISADevice *dev, qemu_irq *a20_out)
445 {
446 Port92State *s = DO_UPCAST(Port92State, dev, dev);
447
448 s->a20_out = a20_out;
449 }
450
451 static const VMStateDescription vmstate_port92_isa = {
452 .name = "port92",
453 .version_id = 1,
454 .minimum_version_id = 1,
455 .minimum_version_id_old = 1,
456 .fields = (VMStateField []) {
457 VMSTATE_UINT8(outport, Port92State),
458 VMSTATE_END_OF_LIST()
459 }
460 };
461
462 static void port92_reset(DeviceState *d)
463 {
464 Port92State *s = container_of(d, Port92State, dev.qdev);
465
466 s->outport &= ~1;
467 }
468
469 static const MemoryRegionPortio port92_portio[] = {
470 { 0, 1, 1, .read = port92_read, .write = port92_write },
471 PORTIO_END_OF_LIST(),
472 };
473
474 static const MemoryRegionOps port92_ops = {
475 .old_portio = port92_portio
476 };
477
478 static int port92_initfn(ISADevice *dev)
479 {
480 Port92State *s = DO_UPCAST(Port92State, dev, dev);
481
482 memory_region_init_io(&s->io, &port92_ops, s, "port92", 1);
483 isa_register_ioport(dev, &s->io, 0x92);
484
485 s->outport = 0;
486 return 0;
487 }
488
489 static void port92_class_initfn(ObjectClass *klass, void *data)
490 {
491 DeviceClass *dc = DEVICE_CLASS(klass);
492 ISADeviceClass *ic = ISA_DEVICE_CLASS(klass);
493 ic->init = port92_initfn;
494 dc->no_user = 1;
495 dc->reset = port92_reset;
496 dc->vmsd = &vmstate_port92_isa;
497 }
498
499 static TypeInfo port92_info = {
500 .name = "port92",
501 .parent = TYPE_ISA_DEVICE,
502 .instance_size = sizeof(Port92State),
503 .class_init = port92_class_initfn,
504 };
505
506 static void port92_register_types(void)
507 {
508 type_register_static(&port92_info);
509 }
510
511 type_init(port92_register_types)
512
513 static void handle_a20_line_change(void *opaque, int irq, int level)
514 {
515 CPUX86State *cpu = opaque;
516
517 /* XXX: send to all CPUs ? */
518 /* XXX: add logic to handle multiple A20 line sources */
519 cpu_x86_set_a20(cpu, level);
520 }
521
522 /***********************************************************/
523 /* Bochs BIOS debug ports */
524
525 static void bochs_bios_write(void *opaque, uint32_t addr, uint32_t val)
526 {
527 static const char shutdown_str[8] = "Shutdown";
528 static int shutdown_index = 0;
529
530 switch(addr) {
531 /* Bochs BIOS messages */
532 case 0x400:
533 case 0x401:
534 /* used to be panic, now unused */
535 break;
536 case 0x402:
537 case 0x403:
538 #ifdef DEBUG_BIOS
539 fprintf(stderr, "%c", val);
540 #endif
541 break;
542 case 0x8900:
543 /* same as Bochs power off */
544 if (val == shutdown_str[shutdown_index]) {
545 shutdown_index++;
546 if (shutdown_index == 8) {
547 shutdown_index = 0;
548 qemu_system_shutdown_request();
549 }
550 } else {
551 shutdown_index = 0;
552 }
553 break;
554
555 /* LGPL'ed VGA BIOS messages */
556 case 0x501:
557 case 0x502:
558 exit((val << 1) | 1);
559 case 0x500:
560 case 0x503:
561 #ifdef DEBUG_BIOS
562 fprintf(stderr, "%c", val);
563 #endif
564 break;
565 }
566 }
567
568 int e820_add_entry(uint64_t address, uint64_t length, uint32_t type)
569 {
570 int index = le32_to_cpu(e820_table.count);
571 struct e820_entry *entry;
572
573 if (index >= E820_NR_ENTRIES)
574 return -EBUSY;
575 entry = &e820_table.entry[index++];
576
577 entry->address = cpu_to_le64(address);
578 entry->length = cpu_to_le64(length);
579 entry->type = cpu_to_le32(type);
580
581 e820_table.count = cpu_to_le32(index);
582 return index;
583 }
584
585 static void *bochs_bios_init(void)
586 {
587 void *fw_cfg;
588 uint8_t *smbios_table;
589 size_t smbios_len;
590 uint64_t *numa_fw_cfg;
591 int i, j;
592
593 register_ioport_write(0x400, 1, 2, bochs_bios_write, NULL);
594 register_ioport_write(0x401, 1, 2, bochs_bios_write, NULL);
595 register_ioport_write(0x402, 1, 1, bochs_bios_write, NULL);
596 register_ioport_write(0x403, 1, 1, bochs_bios_write, NULL);
597 register_ioport_write(0x8900, 1, 1, bochs_bios_write, NULL);
598
599 register_ioport_write(0x501, 1, 1, bochs_bios_write, NULL);
600 register_ioport_write(0x501, 1, 2, bochs_bios_write, NULL);
601 register_ioport_write(0x502, 1, 2, bochs_bios_write, NULL);
602 register_ioport_write(0x500, 1, 1, bochs_bios_write, NULL);
603 register_ioport_write(0x503, 1, 1, bochs_bios_write, NULL);
604
605 fw_cfg = fw_cfg_init(BIOS_CFG_IOPORT, BIOS_CFG_IOPORT + 1, 0, 0);
606
607 fw_cfg_add_i32(fw_cfg, FW_CFG_ID, 1);
608 fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, (uint64_t)ram_size);
609 fw_cfg_add_bytes(fw_cfg, FW_CFG_ACPI_TABLES, (uint8_t *)acpi_tables,
610 acpi_tables_len);
611 fw_cfg_add_i32(fw_cfg, FW_CFG_IRQ0_OVERRIDE, kvm_allows_irq0_override());
612
613 smbios_table = smbios_get_table(&smbios_len);
614 if (smbios_table)
615 fw_cfg_add_bytes(fw_cfg, FW_CFG_SMBIOS_ENTRIES,
616 smbios_table, smbios_len);
617 fw_cfg_add_bytes(fw_cfg, FW_CFG_E820_TABLE, (uint8_t *)&e820_table,
618 sizeof(struct e820_table));
619
620 fw_cfg_add_bytes(fw_cfg, FW_CFG_HPET, (uint8_t *)&hpet_cfg,
621 sizeof(struct hpet_fw_config));
622 /* allocate memory for the NUMA channel: one (64bit) word for the number
623 * of nodes, one word for each VCPU->node and one word for each node to
624 * hold the amount of memory.
625 */
626 numa_fw_cfg = g_malloc0((1 + max_cpus + nb_numa_nodes) * 8);
627 numa_fw_cfg[0] = cpu_to_le64(nb_numa_nodes);
628 for (i = 0; i < max_cpus; i++) {
629 for (j = 0; j < nb_numa_nodes; j++) {
630 if (test_bit(i, node_cpumask[j])) {
631 numa_fw_cfg[i + 1] = cpu_to_le64(j);
632 break;
633 }
634 }
635 }
636 for (i = 0; i < nb_numa_nodes; i++) {
637 numa_fw_cfg[max_cpus + 1 + i] = cpu_to_le64(node_mem[i]);
638 }
639 fw_cfg_add_bytes(fw_cfg, FW_CFG_NUMA, (uint8_t *)numa_fw_cfg,
640 (1 + max_cpus + nb_numa_nodes) * 8);
641
642 return fw_cfg;
643 }
644
645 static long get_file_size(FILE *f)
646 {
647 long where, size;
648
649 /* XXX: on Unix systems, using fstat() probably makes more sense */
650
651 where = ftell(f);
652 fseek(f, 0, SEEK_END);
653 size = ftell(f);
654 fseek(f, where, SEEK_SET);
655
656 return size;
657 }
658
659 static void load_linux(void *fw_cfg,
660 const char *kernel_filename,
661 const char *initrd_filename,
662 const char *kernel_cmdline,
663 target_phys_addr_t max_ram_size)
664 {
665 uint16_t protocol;
666 int setup_size, kernel_size, initrd_size = 0, cmdline_size;
667 uint32_t initrd_max;
668 uint8_t header[8192], *setup, *kernel, *initrd_data;
669 target_phys_addr_t real_addr, prot_addr, cmdline_addr, initrd_addr = 0;
670 FILE *f;
671 char *vmode;
672
673 /* Align to 16 bytes as a paranoia measure */
674 cmdline_size = (strlen(kernel_cmdline)+16) & ~15;
675
676 /* load the kernel header */
677 f = fopen(kernel_filename, "rb");
678 if (!f || !(kernel_size = get_file_size(f)) ||
679 fread(header, 1, MIN(ARRAY_SIZE(header), kernel_size), f) !=
680 MIN(ARRAY_SIZE(header), kernel_size)) {
681 fprintf(stderr, "qemu: could not load kernel '%s': %s\n",
682 kernel_filename, strerror(errno));
683 exit(1);
684 }
685
686 /* kernel protocol version */
687 #if 0
688 fprintf(stderr, "header magic: %#x\n", ldl_p(header+0x202));
689 #endif
690 if (ldl_p(header+0x202) == 0x53726448)
691 protocol = lduw_p(header+0x206);
692 else {
693 /* This looks like a multiboot kernel. If it is, let's stop
694 treating it like a Linux kernel. */
695 if (load_multiboot(fw_cfg, f, kernel_filename, initrd_filename,
696 kernel_cmdline, kernel_size, header))
697 return;
698 protocol = 0;
699 }
700
701 if (protocol < 0x200 || !(header[0x211] & 0x01)) {
702 /* Low kernel */
703 real_addr = 0x90000;
704 cmdline_addr = 0x9a000 - cmdline_size;
705 prot_addr = 0x10000;
706 } else if (protocol < 0x202) {
707 /* High but ancient kernel */
708 real_addr = 0x90000;
709 cmdline_addr = 0x9a000 - cmdline_size;
710 prot_addr = 0x100000;
711 } else {
712 /* High and recent kernel */
713 real_addr = 0x10000;
714 cmdline_addr = 0x20000;
715 prot_addr = 0x100000;
716 }
717
718 #if 0
719 fprintf(stderr,
720 "qemu: real_addr = 0x" TARGET_FMT_plx "\n"
721 "qemu: cmdline_addr = 0x" TARGET_FMT_plx "\n"
722 "qemu: prot_addr = 0x" TARGET_FMT_plx "\n",
723 real_addr,
724 cmdline_addr,
725 prot_addr);
726 #endif
727
728 /* highest address for loading the initrd */
729 if (protocol >= 0x203)
730 initrd_max = ldl_p(header+0x22c);
731 else
732 initrd_max = 0x37ffffff;
733
734 if (initrd_max >= max_ram_size-ACPI_DATA_SIZE)
735 initrd_max = max_ram_size-ACPI_DATA_SIZE-1;
736
737 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_ADDR, cmdline_addr);
738 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_SIZE, strlen(kernel_cmdline)+1);
739 fw_cfg_add_bytes(fw_cfg, FW_CFG_CMDLINE_DATA,
740 (uint8_t*)strdup(kernel_cmdline),
741 strlen(kernel_cmdline)+1);
742
743 if (protocol >= 0x202) {
744 stl_p(header+0x228, cmdline_addr);
745 } else {
746 stw_p(header+0x20, 0xA33F);
747 stw_p(header+0x22, cmdline_addr-real_addr);
748 }
749
750 /* handle vga= parameter */
751 vmode = strstr(kernel_cmdline, "vga=");
752 if (vmode) {
753 unsigned int video_mode;
754 /* skip "vga=" */
755 vmode += 4;
756 if (!strncmp(vmode, "normal", 6)) {
757 video_mode = 0xffff;
758 } else if (!strncmp(vmode, "ext", 3)) {
759 video_mode = 0xfffe;
760 } else if (!strncmp(vmode, "ask", 3)) {
761 video_mode = 0xfffd;
762 } else {
763 video_mode = strtol(vmode, NULL, 0);
764 }
765 stw_p(header+0x1fa, video_mode);
766 }
767
768 /* loader type */
769 /* High nybble = B reserved for QEMU; low nybble is revision number.
770 If this code is substantially changed, you may want to consider
771 incrementing the revision. */
772 if (protocol >= 0x200)
773 header[0x210] = 0xB0;
774
775 /* heap */
776 if (protocol >= 0x201) {
777 header[0x211] |= 0x80; /* CAN_USE_HEAP */
778 stw_p(header+0x224, cmdline_addr-real_addr-0x200);
779 }
780
781 /* load initrd */
782 if (initrd_filename) {
783 if (protocol < 0x200) {
784 fprintf(stderr, "qemu: linux kernel too old to load a ram disk\n");
785 exit(1);
786 }
787
788 initrd_size = get_image_size(initrd_filename);
789 if (initrd_size < 0) {
790 fprintf(stderr, "qemu: error reading initrd %s\n",
791 initrd_filename);
792 exit(1);
793 }
794
795 initrd_addr = (initrd_max-initrd_size) & ~4095;
796
797 initrd_data = g_malloc(initrd_size);
798 load_image(initrd_filename, initrd_data);
799
800 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_ADDR, initrd_addr);
801 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_SIZE, initrd_size);
802 fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, initrd_data, initrd_size);
803
804 stl_p(header+0x218, initrd_addr);
805 stl_p(header+0x21c, initrd_size);
806 }
807
808 /* load kernel and setup */
809 setup_size = header[0x1f1];
810 if (setup_size == 0)
811 setup_size = 4;
812 setup_size = (setup_size+1)*512;
813 kernel_size -= setup_size;
814
815 setup = g_malloc(setup_size);
816 kernel = g_malloc(kernel_size);
817 fseek(f, 0, SEEK_SET);
818 if (fread(setup, 1, setup_size, f) != setup_size) {
819 fprintf(stderr, "fread() failed\n");
820 exit(1);
821 }
822 if (fread(kernel, 1, kernel_size, f) != kernel_size) {
823 fprintf(stderr, "fread() failed\n");
824 exit(1);
825 }
826 fclose(f);
827 memcpy(setup, header, MIN(sizeof(header), setup_size));
828
829 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_ADDR, prot_addr);
830 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_SIZE, kernel_size);
831 fw_cfg_add_bytes(fw_cfg, FW_CFG_KERNEL_DATA, kernel, kernel_size);
832
833 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_ADDR, real_addr);
834 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_SIZE, setup_size);
835 fw_cfg_add_bytes(fw_cfg, FW_CFG_SETUP_DATA, setup, setup_size);
836
837 option_rom[nb_option_roms].name = "linuxboot.bin";
838 option_rom[nb_option_roms].bootindex = 0;
839 nb_option_roms++;
840 }
841
842 #define NE2000_NB_MAX 6
843
844 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
845 0x280, 0x380 };
846 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
847
848 static const int parallel_io[MAX_PARALLEL_PORTS] = { 0x378, 0x278, 0x3bc };
849 static const int parallel_irq[MAX_PARALLEL_PORTS] = { 7, 7, 7 };
850
851 void pc_init_ne2k_isa(ISABus *bus, NICInfo *nd)
852 {
853 static int nb_ne2k = 0;
854
855 if (nb_ne2k == NE2000_NB_MAX)
856 return;
857 isa_ne2000_init(bus, ne2000_io[nb_ne2k],
858 ne2000_irq[nb_ne2k], nd);
859 nb_ne2k++;
860 }
861
862 DeviceState *cpu_get_current_apic(void)
863 {
864 if (cpu_single_env) {
865 return cpu_single_env->apic_state;
866 } else {
867 return NULL;
868 }
869 }
870
871 static DeviceState *apic_init(void *env, uint8_t apic_id)
872 {
873 DeviceState *dev;
874 static int apic_mapped;
875
876 if (kvm_irqchip_in_kernel()) {
877 dev = qdev_create(NULL, "kvm-apic");
878 } else if (xen_enabled()) {
879 dev = qdev_create(NULL, "xen-apic");
880 } else {
881 dev = qdev_create(NULL, "apic");
882 }
883
884 qdev_prop_set_uint8(dev, "id", apic_id);
885 qdev_prop_set_ptr(dev, "cpu_env", env);
886 qdev_init_nofail(dev);
887
888 /* XXX: mapping more APICs at the same memory location */
889 if (apic_mapped == 0) {
890 /* NOTE: the APIC is directly connected to the CPU - it is not
891 on the global memory bus. */
892 /* XXX: what if the base changes? */
893 sysbus_mmio_map(sysbus_from_qdev(dev), 0, MSI_ADDR_BASE);
894 apic_mapped = 1;
895 }
896
897 return dev;
898 }
899
900 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
901 {
902 CPUX86State *s = opaque;
903
904 if (level) {
905 cpu_interrupt(s, CPU_INTERRUPT_SMI);
906 }
907 }
908
909 static X86CPU *pc_new_cpu(const char *cpu_model)
910 {
911 X86CPU *cpu;
912 CPUX86State *env;
913
914 cpu = cpu_x86_init(cpu_model);
915 if (cpu == NULL) {
916 fprintf(stderr, "Unable to find x86 CPU definition\n");
917 exit(1);
918 }
919 env = &cpu->env;
920 if ((env->cpuid_features & CPUID_APIC) || smp_cpus > 1) {
921 env->apic_state = apic_init(env, env->cpuid_apic_id);
922 }
923 cpu_reset(CPU(cpu));
924 return cpu;
925 }
926
927 void pc_cpus_init(const char *cpu_model)
928 {
929 int i;
930
931 /* init CPUs */
932 if (cpu_model == NULL) {
933 #ifdef TARGET_X86_64
934 cpu_model = "qemu64";
935 #else
936 cpu_model = "qemu32";
937 #endif
938 }
939
940 for(i = 0; i < smp_cpus; i++) {
941 pc_new_cpu(cpu_model);
942 }
943 }
944
945 void *pc_memory_init(MemoryRegion *system_memory,
946 const char *kernel_filename,
947 const char *kernel_cmdline,
948 const char *initrd_filename,
949 ram_addr_t below_4g_mem_size,
950 ram_addr_t above_4g_mem_size,
951 MemoryRegion *rom_memory,
952 MemoryRegion **ram_memory)
953 {
954 int linux_boot, i;
955 MemoryRegion *ram, *option_rom_mr;
956 MemoryRegion *ram_below_4g, *ram_above_4g;
957 void *fw_cfg;
958
959 linux_boot = (kernel_filename != NULL);
960
961 /* Allocate RAM. We allocate it as a single memory region and use
962 * aliases to address portions of it, mostly for backwards compatibility
963 * with older qemus that used qemu_ram_alloc().
964 */
965 ram = g_malloc(sizeof(*ram));
966 memory_region_init_ram(ram, "pc.ram",
967 below_4g_mem_size + above_4g_mem_size);
968 vmstate_register_ram_global(ram);
969 *ram_memory = ram;
970 ram_below_4g = g_malloc(sizeof(*ram_below_4g));
971 memory_region_init_alias(ram_below_4g, "ram-below-4g", ram,
972 0, below_4g_mem_size);
973 memory_region_add_subregion(system_memory, 0, ram_below_4g);
974 if (above_4g_mem_size > 0) {
975 ram_above_4g = g_malloc(sizeof(*ram_above_4g));
976 memory_region_init_alias(ram_above_4g, "ram-above-4g", ram,
977 below_4g_mem_size, above_4g_mem_size);
978 memory_region_add_subregion(system_memory, 0x100000000ULL,
979 ram_above_4g);
980 }
981
982
983 /* Initialize PC system firmware */
984 pc_system_firmware_init(rom_memory);
985
986 option_rom_mr = g_malloc(sizeof(*option_rom_mr));
987 memory_region_init_ram(option_rom_mr, "pc.rom", PC_ROM_SIZE);
988 vmstate_register_ram_global(option_rom_mr);
989 memory_region_add_subregion_overlap(rom_memory,
990 PC_ROM_MIN_VGA,
991 option_rom_mr,
992 1);
993
994 fw_cfg = bochs_bios_init();
995 rom_set_fw(fw_cfg);
996
997 if (linux_boot) {
998 load_linux(fw_cfg, kernel_filename, initrd_filename, kernel_cmdline, below_4g_mem_size);
999 }
1000
1001 for (i = 0; i < nb_option_roms; i++) {
1002 rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1003 }
1004 return fw_cfg;
1005 }
1006
1007 qemu_irq *pc_allocate_cpu_irq(void)
1008 {
1009 return qemu_allocate_irqs(pic_irq_request, NULL, 1);
1010 }
1011
1012 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1013 {
1014 DeviceState *dev = NULL;
1015
1016 if (cirrus_vga_enabled) {
1017 if (pci_bus) {
1018 dev = pci_cirrus_vga_init(pci_bus);
1019 } else {
1020 dev = &isa_create_simple(isa_bus, "isa-cirrus-vga")->qdev;
1021 }
1022 } else if (vmsvga_enabled) {
1023 if (pci_bus) {
1024 dev = pci_vmsvga_init(pci_bus);
1025 } else {
1026 fprintf(stderr, "%s: vmware_vga: no PCI bus\n", __FUNCTION__);
1027 }
1028 #ifdef CONFIG_SPICE
1029 } else if (qxl_enabled) {
1030 if (pci_bus) {
1031 dev = &pci_create_simple(pci_bus, -1, "qxl-vga")->qdev;
1032 } else {
1033 fprintf(stderr, "%s: qxl: no PCI bus\n", __FUNCTION__);
1034 }
1035 #endif
1036 } else if (std_vga_enabled) {
1037 if (pci_bus) {
1038 dev = pci_vga_init(pci_bus);
1039 } else {
1040 dev = isa_vga_init(isa_bus);
1041 }
1042 }
1043
1044 return dev;
1045 }
1046
1047 static void cpu_request_exit(void *opaque, int irq, int level)
1048 {
1049 CPUX86State *env = cpu_single_env;
1050
1051 if (env && level) {
1052 cpu_exit(env);
1053 }
1054 }
1055
1056 void pc_basic_device_init(ISABus *isa_bus, qemu_irq *gsi,
1057 ISADevice **rtc_state,
1058 ISADevice **floppy,
1059 bool no_vmport)
1060 {
1061 int i;
1062 DriveInfo *fd[MAX_FD];
1063 DeviceState *hpet = NULL;
1064 int pit_isa_irq = 0;
1065 qemu_irq pit_alt_irq = NULL;
1066 qemu_irq rtc_irq = NULL;
1067 qemu_irq *a20_line;
1068 ISADevice *i8042, *port92, *vmmouse, *pit = NULL;
1069 qemu_irq *cpu_exit_irq;
1070
1071 register_ioport_write(0x80, 1, 1, ioport80_write, NULL);
1072
1073 register_ioport_write(0xf0, 1, 1, ioportF0_write, NULL);
1074
1075 /*
1076 * Check if an HPET shall be created.
1077 *
1078 * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
1079 * when the HPET wants to take over. Thus we have to disable the latter.
1080 */
1081 if (!no_hpet && (!kvm_irqchip_in_kernel() || kvm_has_pit_state2())) {
1082 hpet = sysbus_try_create_simple("hpet", HPET_BASE, NULL);
1083
1084 if (hpet) {
1085 for (i = 0; i < GSI_NUM_PINS; i++) {
1086 sysbus_connect_irq(sysbus_from_qdev(hpet), i, gsi[i]);
1087 }
1088 pit_isa_irq = -1;
1089 pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1090 rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1091 }
1092 }
1093 *rtc_state = rtc_init(isa_bus, 2000, rtc_irq);
1094
1095 qemu_register_boot_set(pc_boot_set, *rtc_state);
1096
1097 if (!xen_enabled()) {
1098 if (kvm_irqchip_in_kernel()) {
1099 pit = kvm_pit_init(isa_bus, 0x40);
1100 } else {
1101 pit = pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1102 }
1103 if (hpet) {
1104 /* connect PIT to output control line of the HPET */
1105 qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(&pit->qdev, 0));
1106 }
1107 pcspk_init(isa_bus, pit);
1108 }
1109
1110 for(i = 0; i < MAX_SERIAL_PORTS; i++) {
1111 if (serial_hds[i]) {
1112 serial_isa_init(isa_bus, i, serial_hds[i]);
1113 }
1114 }
1115
1116 for(i = 0; i < MAX_PARALLEL_PORTS; i++) {
1117 if (parallel_hds[i]) {
1118 parallel_init(isa_bus, i, parallel_hds[i]);
1119 }
1120 }
1121
1122 a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1123 i8042 = isa_create_simple(isa_bus, "i8042");
1124 i8042_setup_a20_line(i8042, &a20_line[0]);
1125 if (!no_vmport) {
1126 vmport_init(isa_bus);
1127 vmmouse = isa_try_create(isa_bus, "vmmouse");
1128 } else {
1129 vmmouse = NULL;
1130 }
1131 if (vmmouse) {
1132 qdev_prop_set_ptr(&vmmouse->qdev, "ps2_mouse", i8042);
1133 qdev_init_nofail(&vmmouse->qdev);
1134 }
1135 port92 = isa_create_simple(isa_bus, "port92");
1136 port92_init(port92, &a20_line[1]);
1137
1138 cpu_exit_irq = qemu_allocate_irqs(cpu_request_exit, NULL, 1);
1139 DMA_init(0, cpu_exit_irq);
1140
1141 for(i = 0; i < MAX_FD; i++) {
1142 fd[i] = drive_get(IF_FLOPPY, 0, i);
1143 }
1144 *floppy = fdctrl_init_isa(isa_bus, fd);
1145 }
1146
1147 void pc_pci_device_init(PCIBus *pci_bus)
1148 {
1149 int max_bus;
1150 int bus;
1151
1152 max_bus = drive_get_max_bus(IF_SCSI);
1153 for (bus = 0; bus <= max_bus; bus++) {
1154 pci_create_simple(pci_bus, -1, "lsi53c895a");
1155 }
1156 }