]> git.proxmox.com Git - qemu.git/blob - hw/pc.c
Add one new file vga-pci.h and cleanup on all platforms
[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 #include "vga-pci.h"
55
56 /* output Bochs bios info messages */
57 //#define DEBUG_BIOS
58
59 /* debug PC/ISA interrupts */
60 //#define DEBUG_IRQ
61
62 #ifdef DEBUG_IRQ
63 #define DPRINTF(fmt, ...) \
64 do { printf("CPUIRQ: " fmt , ## __VA_ARGS__); } while (0)
65 #else
66 #define DPRINTF(fmt, ...)
67 #endif
68
69 /* Leave a chunk of memory at the top of RAM for the BIOS ACPI tables. */
70 #define ACPI_DATA_SIZE 0x10000
71 #define BIOS_CFG_IOPORT 0x510
72 #define FW_CFG_ACPI_TABLES (FW_CFG_ARCH_LOCAL + 0)
73 #define FW_CFG_SMBIOS_ENTRIES (FW_CFG_ARCH_LOCAL + 1)
74 #define FW_CFG_IRQ0_OVERRIDE (FW_CFG_ARCH_LOCAL + 2)
75 #define FW_CFG_E820_TABLE (FW_CFG_ARCH_LOCAL + 3)
76 #define FW_CFG_HPET (FW_CFG_ARCH_LOCAL + 4)
77
78 #define MSI_ADDR_BASE 0xfee00000
79
80 #define E820_NR_ENTRIES 16
81
82 struct e820_entry {
83 uint64_t address;
84 uint64_t length;
85 uint32_t type;
86 } QEMU_PACKED __attribute((__aligned__(4)));
87
88 struct e820_table {
89 uint32_t count;
90 struct e820_entry entry[E820_NR_ENTRIES];
91 } QEMU_PACKED __attribute((__aligned__(4)));
92
93 static struct e820_table e820_table;
94 struct hpet_fw_config hpet_cfg = {.count = UINT8_MAX};
95
96 void gsi_handler(void *opaque, int n, int level)
97 {
98 GSIState *s = opaque;
99
100 DPRINTF("pc: %s GSI %d\n", level ? "raising" : "lowering", n);
101 if (n < ISA_NUM_IRQS) {
102 qemu_set_irq(s->i8259_irq[n], level);
103 }
104 qemu_set_irq(s->ioapic_irq[n], level);
105 }
106
107 static void ioport80_write(void *opaque, uint32_t addr, uint32_t data)
108 {
109 }
110
111 /* MSDOS compatibility mode FPU exception support */
112 static qemu_irq ferr_irq;
113
114 void pc_register_ferr_irq(qemu_irq irq)
115 {
116 ferr_irq = irq;
117 }
118
119 /* XXX: add IGNNE support */
120 void cpu_set_ferr(CPUX86State *s)
121 {
122 qemu_irq_raise(ferr_irq);
123 }
124
125 static void ioportF0_write(void *opaque, uint32_t addr, uint32_t data)
126 {
127 qemu_irq_lower(ferr_irq);
128 }
129
130 /* TSC handling */
131 uint64_t cpu_get_tsc(CPUX86State *env)
132 {
133 return cpu_get_ticks();
134 }
135
136 /* SMM support */
137
138 static cpu_set_smm_t smm_set;
139 static void *smm_arg;
140
141 void cpu_smm_register(cpu_set_smm_t callback, void *arg)
142 {
143 assert(smm_set == NULL);
144 assert(smm_arg == NULL);
145 smm_set = callback;
146 smm_arg = arg;
147 }
148
149 void cpu_smm_update(CPUX86State *env)
150 {
151 if (smm_set && smm_arg && env == first_cpu)
152 smm_set(!!(env->hflags & HF_SMM_MASK), smm_arg);
153 }
154
155
156 /* IRQ handling */
157 int cpu_get_pic_interrupt(CPUX86State *env)
158 {
159 int intno;
160
161 intno = apic_get_interrupt(env->apic_state);
162 if (intno >= 0) {
163 return intno;
164 }
165 /* read the irq from the PIC */
166 if (!apic_accept_pic_intr(env->apic_state)) {
167 return -1;
168 }
169
170 intno = pic_read_irq(isa_pic);
171 return intno;
172 }
173
174 static void pic_irq_request(void *opaque, int irq, int level)
175 {
176 CPUX86State *env = first_cpu;
177
178 DPRINTF("pic_irqs: %s irq %d\n", level? "raise" : "lower", irq);
179 if (env->apic_state) {
180 while (env) {
181 if (apic_accept_pic_intr(env->apic_state)) {
182 apic_deliver_pic_intr(env->apic_state, level);
183 }
184 env = env->next_cpu;
185 }
186 } else {
187 if (level)
188 cpu_interrupt(env, CPU_INTERRUPT_HARD);
189 else
190 cpu_reset_interrupt(env, CPU_INTERRUPT_HARD);
191 }
192 }
193
194 /* PC cmos mappings */
195
196 #define REG_EQUIPMENT_BYTE 0x14
197
198 static int cmos_get_fd_drive_type(FDriveType fd0)
199 {
200 int val;
201
202 switch (fd0) {
203 case FDRIVE_DRV_144:
204 /* 1.44 Mb 3"5 drive */
205 val = 4;
206 break;
207 case FDRIVE_DRV_288:
208 /* 2.88 Mb 3"5 drive */
209 val = 5;
210 break;
211 case FDRIVE_DRV_120:
212 /* 1.2 Mb 5"5 drive */
213 val = 2;
214 break;
215 case FDRIVE_DRV_NONE:
216 default:
217 val = 0;
218 break;
219 }
220 return val;
221 }
222
223 static void cmos_init_hd(ISADevice *s, int type_ofs, int info_ofs,
224 int16_t cylinders, int8_t heads, int8_t sectors)
225 {
226 rtc_set_memory(s, type_ofs, 47);
227 rtc_set_memory(s, info_ofs, cylinders);
228 rtc_set_memory(s, info_ofs + 1, cylinders >> 8);
229 rtc_set_memory(s, info_ofs + 2, heads);
230 rtc_set_memory(s, info_ofs + 3, 0xff);
231 rtc_set_memory(s, info_ofs + 4, 0xff);
232 rtc_set_memory(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
233 rtc_set_memory(s, info_ofs + 6, cylinders);
234 rtc_set_memory(s, info_ofs + 7, cylinders >> 8);
235 rtc_set_memory(s, info_ofs + 8, sectors);
236 }
237
238 /* convert boot_device letter to something recognizable by the bios */
239 static int boot_device2nibble(char boot_device)
240 {
241 switch(boot_device) {
242 case 'a':
243 case 'b':
244 return 0x01; /* floppy boot */
245 case 'c':
246 return 0x02; /* hard drive boot */
247 case 'd':
248 return 0x03; /* CD-ROM boot */
249 case 'n':
250 return 0x04; /* Network boot */
251 }
252 return 0;
253 }
254
255 static int set_boot_dev(ISADevice *s, const char *boot_device, int fd_bootchk)
256 {
257 #define PC_MAX_BOOT_DEVICES 3
258 int nbds, bds[3] = { 0, };
259 int i;
260
261 nbds = strlen(boot_device);
262 if (nbds > PC_MAX_BOOT_DEVICES) {
263 error_report("Too many boot devices for PC");
264 return(1);
265 }
266 for (i = 0; i < nbds; i++) {
267 bds[i] = boot_device2nibble(boot_device[i]);
268 if (bds[i] == 0) {
269 error_report("Invalid boot device for PC: '%c'",
270 boot_device[i]);
271 return(1);
272 }
273 }
274 rtc_set_memory(s, 0x3d, (bds[1] << 4) | bds[0]);
275 rtc_set_memory(s, 0x38, (bds[2] << 4) | (fd_bootchk ? 0x0 : 0x1));
276 return(0);
277 }
278
279 static int pc_boot_set(void *opaque, const char *boot_device)
280 {
281 return set_boot_dev(opaque, boot_device, 0);
282 }
283
284 typedef struct pc_cmos_init_late_arg {
285 ISADevice *rtc_state;
286 BusState *idebus[2];
287 } pc_cmos_init_late_arg;
288
289 static void pc_cmos_init_late(void *opaque)
290 {
291 pc_cmos_init_late_arg *arg = opaque;
292 ISADevice *s = arg->rtc_state;
293 int16_t cylinders;
294 int8_t heads, sectors;
295 int val;
296 int i, trans;
297
298 val = 0;
299 if (ide_get_geometry(arg->idebus[0], 0,
300 &cylinders, &heads, &sectors) >= 0) {
301 cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
302 val |= 0xf0;
303 }
304 if (ide_get_geometry(arg->idebus[0], 1,
305 &cylinders, &heads, &sectors) >= 0) {
306 cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
307 val |= 0x0f;
308 }
309 rtc_set_memory(s, 0x12, val);
310
311 val = 0;
312 for (i = 0; i < 4; i++) {
313 /* NOTE: ide_get_geometry() returns the physical
314 geometry. It is always such that: 1 <= sects <= 63, 1
315 <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
316 geometry can be different if a translation is done. */
317 if (ide_get_geometry(arg->idebus[i / 2], i % 2,
318 &cylinders, &heads, &sectors) >= 0) {
319 trans = ide_get_bios_chs_trans(arg->idebus[i / 2], i % 2) - 1;
320 assert((trans & ~3) == 0);
321 val |= trans << (i * 2);
322 }
323 }
324 rtc_set_memory(s, 0x39, val);
325
326 qemu_unregister_reset(pc_cmos_init_late, opaque);
327 }
328
329 void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
330 const char *boot_device,
331 ISADevice *floppy, BusState *idebus0, BusState *idebus1,
332 ISADevice *s)
333 {
334 int val, nb, i;
335 FDriveType fd_type[2] = { FDRIVE_DRV_NONE, FDRIVE_DRV_NONE };
336 static pc_cmos_init_late_arg arg;
337
338 /* various important CMOS locations needed by PC/Bochs bios */
339
340 /* memory size */
341 val = 640; /* base memory in K */
342 rtc_set_memory(s, 0x15, val);
343 rtc_set_memory(s, 0x16, val >> 8);
344
345 val = (ram_size / 1024) - 1024;
346 if (val > 65535)
347 val = 65535;
348 rtc_set_memory(s, 0x17, val);
349 rtc_set_memory(s, 0x18, val >> 8);
350 rtc_set_memory(s, 0x30, val);
351 rtc_set_memory(s, 0x31, val >> 8);
352
353 if (above_4g_mem_size) {
354 rtc_set_memory(s, 0x5b, (unsigned int)above_4g_mem_size >> 16);
355 rtc_set_memory(s, 0x5c, (unsigned int)above_4g_mem_size >> 24);
356 rtc_set_memory(s, 0x5d, (uint64_t)above_4g_mem_size >> 32);
357 }
358
359 if (ram_size > (16 * 1024 * 1024))
360 val = (ram_size / 65536) - ((16 * 1024 * 1024) / 65536);
361 else
362 val = 0;
363 if (val > 65535)
364 val = 65535;
365 rtc_set_memory(s, 0x34, val);
366 rtc_set_memory(s, 0x35, val >> 8);
367
368 /* set the number of CPU */
369 rtc_set_memory(s, 0x5f, smp_cpus - 1);
370
371 /* set boot devices, and disable floppy signature check if requested */
372 if (set_boot_dev(s, boot_device, fd_bootchk)) {
373 exit(1);
374 }
375
376 /* floppy type */
377 if (floppy) {
378 for (i = 0; i < 2; i++) {
379 fd_type[i] = isa_fdc_get_drive_type(floppy, i);
380 }
381 }
382 val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
383 cmos_get_fd_drive_type(fd_type[1]);
384 rtc_set_memory(s, 0x10, val);
385
386 val = 0;
387 nb = 0;
388 if (fd_type[0] < FDRIVE_DRV_NONE) {
389 nb++;
390 }
391 if (fd_type[1] < FDRIVE_DRV_NONE) {
392 nb++;
393 }
394 switch (nb) {
395 case 0:
396 break;
397 case 1:
398 val |= 0x01; /* 1 drive, ready for boot */
399 break;
400 case 2:
401 val |= 0x41; /* 2 drives, ready for boot */
402 break;
403 }
404 val |= 0x02; /* FPU is there */
405 val |= 0x04; /* PS/2 mouse installed */
406 rtc_set_memory(s, REG_EQUIPMENT_BYTE, val);
407
408 /* hard drives */
409 arg.rtc_state = s;
410 arg.idebus[0] = idebus0;
411 arg.idebus[1] = idebus1;
412 qemu_register_reset(pc_cmos_init_late, &arg);
413 }
414
415 /* port 92 stuff: could be split off */
416 typedef struct Port92State {
417 ISADevice dev;
418 MemoryRegion io;
419 uint8_t outport;
420 qemu_irq *a20_out;
421 } Port92State;
422
423 static void port92_write(void *opaque, uint32_t addr, uint32_t val)
424 {
425 Port92State *s = opaque;
426
427 DPRINTF("port92: write 0x%02x\n", val);
428 s->outport = val;
429 qemu_set_irq(*s->a20_out, (val >> 1) & 1);
430 if (val & 1) {
431 qemu_system_reset_request();
432 }
433 }
434
435 static uint32_t port92_read(void *opaque, uint32_t addr)
436 {
437 Port92State *s = opaque;
438 uint32_t ret;
439
440 ret = s->outport;
441 DPRINTF("port92: read 0x%02x\n", ret);
442 return ret;
443 }
444
445 static void port92_init(ISADevice *dev, qemu_irq *a20_out)
446 {
447 Port92State *s = DO_UPCAST(Port92State, dev, dev);
448
449 s->a20_out = a20_out;
450 }
451
452 static const VMStateDescription vmstate_port92_isa = {
453 .name = "port92",
454 .version_id = 1,
455 .minimum_version_id = 1,
456 .minimum_version_id_old = 1,
457 .fields = (VMStateField []) {
458 VMSTATE_UINT8(outport, Port92State),
459 VMSTATE_END_OF_LIST()
460 }
461 };
462
463 static void port92_reset(DeviceState *d)
464 {
465 Port92State *s = container_of(d, Port92State, dev.qdev);
466
467 s->outport &= ~1;
468 }
469
470 static const MemoryRegionPortio port92_portio[] = {
471 { 0, 1, 1, .read = port92_read, .write = port92_write },
472 PORTIO_END_OF_LIST(),
473 };
474
475 static const MemoryRegionOps port92_ops = {
476 .old_portio = port92_portio
477 };
478
479 static int port92_initfn(ISADevice *dev)
480 {
481 Port92State *s = DO_UPCAST(Port92State, dev, dev);
482
483 memory_region_init_io(&s->io, &port92_ops, s, "port92", 1);
484 isa_register_ioport(dev, &s->io, 0x92);
485
486 s->outport = 0;
487 return 0;
488 }
489
490 static void port92_class_initfn(ObjectClass *klass, void *data)
491 {
492 DeviceClass *dc = DEVICE_CLASS(klass);
493 ISADeviceClass *ic = ISA_DEVICE_CLASS(klass);
494 ic->init = port92_initfn;
495 dc->no_user = 1;
496 dc->reset = port92_reset;
497 dc->vmsd = &vmstate_port92_isa;
498 }
499
500 static TypeInfo port92_info = {
501 .name = "port92",
502 .parent = TYPE_ISA_DEVICE,
503 .instance_size = sizeof(Port92State),
504 .class_init = port92_class_initfn,
505 };
506
507 static void port92_register_types(void)
508 {
509 type_register_static(&port92_info);
510 }
511
512 type_init(port92_register_types)
513
514 static void handle_a20_line_change(void *opaque, int irq, int level)
515 {
516 CPUX86State *cpu = opaque;
517
518 /* XXX: send to all CPUs ? */
519 /* XXX: add logic to handle multiple A20 line sources */
520 cpu_x86_set_a20(cpu, level);
521 }
522
523 /***********************************************************/
524 /* Bochs BIOS debug ports */
525
526 static void bochs_bios_write(void *opaque, uint32_t addr, uint32_t val)
527 {
528 static const char shutdown_str[8] = "Shutdown";
529 static int shutdown_index = 0;
530
531 switch(addr) {
532 /* Bochs BIOS messages */
533 case 0x400:
534 case 0x401:
535 /* used to be panic, now unused */
536 break;
537 case 0x402:
538 case 0x403:
539 #ifdef DEBUG_BIOS
540 fprintf(stderr, "%c", val);
541 #endif
542 break;
543 case 0x8900:
544 /* same as Bochs power off */
545 if (val == shutdown_str[shutdown_index]) {
546 shutdown_index++;
547 if (shutdown_index == 8) {
548 shutdown_index = 0;
549 qemu_system_shutdown_request();
550 }
551 } else {
552 shutdown_index = 0;
553 }
554 break;
555
556 /* LGPL'ed VGA BIOS messages */
557 case 0x501:
558 case 0x502:
559 exit((val << 1) | 1);
560 case 0x500:
561 case 0x503:
562 #ifdef DEBUG_BIOS
563 fprintf(stderr, "%c", val);
564 #endif
565 break;
566 }
567 }
568
569 int e820_add_entry(uint64_t address, uint64_t length, uint32_t type)
570 {
571 int index = le32_to_cpu(e820_table.count);
572 struct e820_entry *entry;
573
574 if (index >= E820_NR_ENTRIES)
575 return -EBUSY;
576 entry = &e820_table.entry[index++];
577
578 entry->address = cpu_to_le64(address);
579 entry->length = cpu_to_le64(length);
580 entry->type = cpu_to_le32(type);
581
582 e820_table.count = cpu_to_le32(index);
583 return index;
584 }
585
586 static void *bochs_bios_init(void)
587 {
588 void *fw_cfg;
589 uint8_t *smbios_table;
590 size_t smbios_len;
591 uint64_t *numa_fw_cfg;
592 int i, j;
593
594 register_ioport_write(0x400, 1, 2, bochs_bios_write, NULL);
595 register_ioport_write(0x401, 1, 2, bochs_bios_write, NULL);
596 register_ioport_write(0x402, 1, 1, bochs_bios_write, NULL);
597 register_ioport_write(0x403, 1, 1, bochs_bios_write, NULL);
598 register_ioport_write(0x8900, 1, 1, bochs_bios_write, NULL);
599
600 register_ioport_write(0x501, 1, 1, bochs_bios_write, NULL);
601 register_ioport_write(0x501, 1, 2, bochs_bios_write, NULL);
602 register_ioport_write(0x502, 1, 2, bochs_bios_write, NULL);
603 register_ioport_write(0x500, 1, 1, bochs_bios_write, NULL);
604 register_ioport_write(0x503, 1, 1, bochs_bios_write, NULL);
605
606 fw_cfg = fw_cfg_init(BIOS_CFG_IOPORT, BIOS_CFG_IOPORT + 1, 0, 0);
607
608 fw_cfg_add_i32(fw_cfg, FW_CFG_ID, 1);
609 fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, (uint64_t)ram_size);
610 fw_cfg_add_bytes(fw_cfg, FW_CFG_ACPI_TABLES, (uint8_t *)acpi_tables,
611 acpi_tables_len);
612 fw_cfg_add_i32(fw_cfg, FW_CFG_IRQ0_OVERRIDE, kvm_allows_irq0_override());
613
614 smbios_table = smbios_get_table(&smbios_len);
615 if (smbios_table)
616 fw_cfg_add_bytes(fw_cfg, FW_CFG_SMBIOS_ENTRIES,
617 smbios_table, smbios_len);
618 fw_cfg_add_bytes(fw_cfg, FW_CFG_E820_TABLE, (uint8_t *)&e820_table,
619 sizeof(struct e820_table));
620
621 fw_cfg_add_bytes(fw_cfg, FW_CFG_HPET, (uint8_t *)&hpet_cfg,
622 sizeof(struct hpet_fw_config));
623 /* allocate memory for the NUMA channel: one (64bit) word for the number
624 * of nodes, one word for each VCPU->node and one word for each node to
625 * hold the amount of memory.
626 */
627 numa_fw_cfg = g_malloc0((1 + max_cpus + nb_numa_nodes) * 8);
628 numa_fw_cfg[0] = cpu_to_le64(nb_numa_nodes);
629 for (i = 0; i < max_cpus; i++) {
630 for (j = 0; j < nb_numa_nodes; j++) {
631 if (test_bit(i, node_cpumask[j])) {
632 numa_fw_cfg[i + 1] = cpu_to_le64(j);
633 break;
634 }
635 }
636 }
637 for (i = 0; i < nb_numa_nodes; i++) {
638 numa_fw_cfg[max_cpus + 1 + i] = cpu_to_le64(node_mem[i]);
639 }
640 fw_cfg_add_bytes(fw_cfg, FW_CFG_NUMA, (uint8_t *)numa_fw_cfg,
641 (1 + max_cpus + nb_numa_nodes) * 8);
642
643 return fw_cfg;
644 }
645
646 static long get_file_size(FILE *f)
647 {
648 long where, size;
649
650 /* XXX: on Unix systems, using fstat() probably makes more sense */
651
652 where = ftell(f);
653 fseek(f, 0, SEEK_END);
654 size = ftell(f);
655 fseek(f, where, SEEK_SET);
656
657 return size;
658 }
659
660 static void load_linux(void *fw_cfg,
661 const char *kernel_filename,
662 const char *initrd_filename,
663 const char *kernel_cmdline,
664 target_phys_addr_t max_ram_size)
665 {
666 uint16_t protocol;
667 int setup_size, kernel_size, initrd_size = 0, cmdline_size;
668 uint32_t initrd_max;
669 uint8_t header[8192], *setup, *kernel, *initrd_data;
670 target_phys_addr_t real_addr, prot_addr, cmdline_addr, initrd_addr = 0;
671 FILE *f;
672 char *vmode;
673
674 /* Align to 16 bytes as a paranoia measure */
675 cmdline_size = (strlen(kernel_cmdline)+16) & ~15;
676
677 /* load the kernel header */
678 f = fopen(kernel_filename, "rb");
679 if (!f || !(kernel_size = get_file_size(f)) ||
680 fread(header, 1, MIN(ARRAY_SIZE(header), kernel_size), f) !=
681 MIN(ARRAY_SIZE(header), kernel_size)) {
682 fprintf(stderr, "qemu: could not load kernel '%s': %s\n",
683 kernel_filename, strerror(errno));
684 exit(1);
685 }
686
687 /* kernel protocol version */
688 #if 0
689 fprintf(stderr, "header magic: %#x\n", ldl_p(header+0x202));
690 #endif
691 if (ldl_p(header+0x202) == 0x53726448)
692 protocol = lduw_p(header+0x206);
693 else {
694 /* This looks like a multiboot kernel. If it is, let's stop
695 treating it like a Linux kernel. */
696 if (load_multiboot(fw_cfg, f, kernel_filename, initrd_filename,
697 kernel_cmdline, kernel_size, header))
698 return;
699 protocol = 0;
700 }
701
702 if (protocol < 0x200 || !(header[0x211] & 0x01)) {
703 /* Low kernel */
704 real_addr = 0x90000;
705 cmdline_addr = 0x9a000 - cmdline_size;
706 prot_addr = 0x10000;
707 } else if (protocol < 0x202) {
708 /* High but ancient kernel */
709 real_addr = 0x90000;
710 cmdline_addr = 0x9a000 - cmdline_size;
711 prot_addr = 0x100000;
712 } else {
713 /* High and recent kernel */
714 real_addr = 0x10000;
715 cmdline_addr = 0x20000;
716 prot_addr = 0x100000;
717 }
718
719 #if 0
720 fprintf(stderr,
721 "qemu: real_addr = 0x" TARGET_FMT_plx "\n"
722 "qemu: cmdline_addr = 0x" TARGET_FMT_plx "\n"
723 "qemu: prot_addr = 0x" TARGET_FMT_plx "\n",
724 real_addr,
725 cmdline_addr,
726 prot_addr);
727 #endif
728
729 /* highest address for loading the initrd */
730 if (protocol >= 0x203)
731 initrd_max = ldl_p(header+0x22c);
732 else
733 initrd_max = 0x37ffffff;
734
735 if (initrd_max >= max_ram_size-ACPI_DATA_SIZE)
736 initrd_max = max_ram_size-ACPI_DATA_SIZE-1;
737
738 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_ADDR, cmdline_addr);
739 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_SIZE, strlen(kernel_cmdline)+1);
740 fw_cfg_add_bytes(fw_cfg, FW_CFG_CMDLINE_DATA,
741 (uint8_t*)strdup(kernel_cmdline),
742 strlen(kernel_cmdline)+1);
743
744 if (protocol >= 0x202) {
745 stl_p(header+0x228, cmdline_addr);
746 } else {
747 stw_p(header+0x20, 0xA33F);
748 stw_p(header+0x22, cmdline_addr-real_addr);
749 }
750
751 /* handle vga= parameter */
752 vmode = strstr(kernel_cmdline, "vga=");
753 if (vmode) {
754 unsigned int video_mode;
755 /* skip "vga=" */
756 vmode += 4;
757 if (!strncmp(vmode, "normal", 6)) {
758 video_mode = 0xffff;
759 } else if (!strncmp(vmode, "ext", 3)) {
760 video_mode = 0xfffe;
761 } else if (!strncmp(vmode, "ask", 3)) {
762 video_mode = 0xfffd;
763 } else {
764 video_mode = strtol(vmode, NULL, 0);
765 }
766 stw_p(header+0x1fa, video_mode);
767 }
768
769 /* loader type */
770 /* High nybble = B reserved for QEMU; low nybble is revision number.
771 If this code is substantially changed, you may want to consider
772 incrementing the revision. */
773 if (protocol >= 0x200)
774 header[0x210] = 0xB0;
775
776 /* heap */
777 if (protocol >= 0x201) {
778 header[0x211] |= 0x80; /* CAN_USE_HEAP */
779 stw_p(header+0x224, cmdline_addr-real_addr-0x200);
780 }
781
782 /* load initrd */
783 if (initrd_filename) {
784 if (protocol < 0x200) {
785 fprintf(stderr, "qemu: linux kernel too old to load a ram disk\n");
786 exit(1);
787 }
788
789 initrd_size = get_image_size(initrd_filename);
790 if (initrd_size < 0) {
791 fprintf(stderr, "qemu: error reading initrd %s\n",
792 initrd_filename);
793 exit(1);
794 }
795
796 initrd_addr = (initrd_max-initrd_size) & ~4095;
797
798 initrd_data = g_malloc(initrd_size);
799 load_image(initrd_filename, initrd_data);
800
801 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_ADDR, initrd_addr);
802 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_SIZE, initrd_size);
803 fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, initrd_data, initrd_size);
804
805 stl_p(header+0x218, initrd_addr);
806 stl_p(header+0x21c, initrd_size);
807 }
808
809 /* load kernel and setup */
810 setup_size = header[0x1f1];
811 if (setup_size == 0)
812 setup_size = 4;
813 setup_size = (setup_size+1)*512;
814 kernel_size -= setup_size;
815
816 setup = g_malloc(setup_size);
817 kernel = g_malloc(kernel_size);
818 fseek(f, 0, SEEK_SET);
819 if (fread(setup, 1, setup_size, f) != setup_size) {
820 fprintf(stderr, "fread() failed\n");
821 exit(1);
822 }
823 if (fread(kernel, 1, kernel_size, f) != kernel_size) {
824 fprintf(stderr, "fread() failed\n");
825 exit(1);
826 }
827 fclose(f);
828 memcpy(setup, header, MIN(sizeof(header), setup_size));
829
830 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_ADDR, prot_addr);
831 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_SIZE, kernel_size);
832 fw_cfg_add_bytes(fw_cfg, FW_CFG_KERNEL_DATA, kernel, kernel_size);
833
834 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_ADDR, real_addr);
835 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_SIZE, setup_size);
836 fw_cfg_add_bytes(fw_cfg, FW_CFG_SETUP_DATA, setup, setup_size);
837
838 option_rom[nb_option_roms].name = "linuxboot.bin";
839 option_rom[nb_option_roms].bootindex = 0;
840 nb_option_roms++;
841 }
842
843 #define NE2000_NB_MAX 6
844
845 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
846 0x280, 0x380 };
847 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
848
849 static const int parallel_io[MAX_PARALLEL_PORTS] = { 0x378, 0x278, 0x3bc };
850 static const int parallel_irq[MAX_PARALLEL_PORTS] = { 7, 7, 7 };
851
852 void pc_init_ne2k_isa(ISABus *bus, NICInfo *nd)
853 {
854 static int nb_ne2k = 0;
855
856 if (nb_ne2k == NE2000_NB_MAX)
857 return;
858 isa_ne2000_init(bus, ne2000_io[nb_ne2k],
859 ne2000_irq[nb_ne2k], nd);
860 nb_ne2k++;
861 }
862
863 DeviceState *cpu_get_current_apic(void)
864 {
865 if (cpu_single_env) {
866 return cpu_single_env->apic_state;
867 } else {
868 return NULL;
869 }
870 }
871
872 static DeviceState *apic_init(void *env, uint8_t apic_id)
873 {
874 DeviceState *dev;
875 static int apic_mapped;
876
877 if (kvm_irqchip_in_kernel()) {
878 dev = qdev_create(NULL, "kvm-apic");
879 } else if (xen_enabled()) {
880 dev = qdev_create(NULL, "xen-apic");
881 } else {
882 dev = qdev_create(NULL, "apic");
883 }
884
885 qdev_prop_set_uint8(dev, "id", apic_id);
886 qdev_prop_set_ptr(dev, "cpu_env", env);
887 qdev_init_nofail(dev);
888
889 /* XXX: mapping more APICs at the same memory location */
890 if (apic_mapped == 0) {
891 /* NOTE: the APIC is directly connected to the CPU - it is not
892 on the global memory bus. */
893 /* XXX: what if the base changes? */
894 sysbus_mmio_map(sysbus_from_qdev(dev), 0, MSI_ADDR_BASE);
895 apic_mapped = 1;
896 }
897
898 return dev;
899 }
900
901 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
902 {
903 CPUX86State *s = opaque;
904
905 if (level) {
906 cpu_interrupt(s, CPU_INTERRUPT_SMI);
907 }
908 }
909
910 static X86CPU *pc_new_cpu(const char *cpu_model)
911 {
912 X86CPU *cpu;
913 CPUX86State *env;
914
915 cpu = cpu_x86_init(cpu_model);
916 if (cpu == NULL) {
917 fprintf(stderr, "Unable to find x86 CPU definition\n");
918 exit(1);
919 }
920 env = &cpu->env;
921 if ((env->cpuid_features & CPUID_APIC) || smp_cpus > 1) {
922 env->apic_state = apic_init(env, env->cpuid_apic_id);
923 }
924 cpu_reset(CPU(cpu));
925 return cpu;
926 }
927
928 void pc_cpus_init(const char *cpu_model)
929 {
930 int i;
931
932 /* init CPUs */
933 if (cpu_model == NULL) {
934 #ifdef TARGET_X86_64
935 cpu_model = "qemu64";
936 #else
937 cpu_model = "qemu32";
938 #endif
939 }
940
941 for(i = 0; i < smp_cpus; i++) {
942 pc_new_cpu(cpu_model);
943 }
944 }
945
946 void *pc_memory_init(MemoryRegion *system_memory,
947 const char *kernel_filename,
948 const char *kernel_cmdline,
949 const char *initrd_filename,
950 ram_addr_t below_4g_mem_size,
951 ram_addr_t above_4g_mem_size,
952 MemoryRegion *rom_memory,
953 MemoryRegion **ram_memory)
954 {
955 int linux_boot, i;
956 MemoryRegion *ram, *option_rom_mr;
957 MemoryRegion *ram_below_4g, *ram_above_4g;
958 void *fw_cfg;
959
960 linux_boot = (kernel_filename != NULL);
961
962 /* Allocate RAM. We allocate it as a single memory region and use
963 * aliases to address portions of it, mostly for backwards compatibility
964 * with older qemus that used qemu_ram_alloc().
965 */
966 ram = g_malloc(sizeof(*ram));
967 memory_region_init_ram(ram, "pc.ram",
968 below_4g_mem_size + above_4g_mem_size);
969 vmstate_register_ram_global(ram);
970 *ram_memory = ram;
971 ram_below_4g = g_malloc(sizeof(*ram_below_4g));
972 memory_region_init_alias(ram_below_4g, "ram-below-4g", ram,
973 0, below_4g_mem_size);
974 memory_region_add_subregion(system_memory, 0, ram_below_4g);
975 if (above_4g_mem_size > 0) {
976 ram_above_4g = g_malloc(sizeof(*ram_above_4g));
977 memory_region_init_alias(ram_above_4g, "ram-above-4g", ram,
978 below_4g_mem_size, above_4g_mem_size);
979 memory_region_add_subregion(system_memory, 0x100000000ULL,
980 ram_above_4g);
981 }
982
983
984 /* Initialize PC system firmware */
985 pc_system_firmware_init(rom_memory);
986
987 option_rom_mr = g_malloc(sizeof(*option_rom_mr));
988 memory_region_init_ram(option_rom_mr, "pc.rom", PC_ROM_SIZE);
989 vmstate_register_ram_global(option_rom_mr);
990 memory_region_add_subregion_overlap(rom_memory,
991 PC_ROM_MIN_VGA,
992 option_rom_mr,
993 1);
994
995 fw_cfg = bochs_bios_init();
996 rom_set_fw(fw_cfg);
997
998 if (linux_boot) {
999 load_linux(fw_cfg, kernel_filename, initrd_filename, kernel_cmdline, below_4g_mem_size);
1000 }
1001
1002 for (i = 0; i < nb_option_roms; i++) {
1003 rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1004 }
1005 return fw_cfg;
1006 }
1007
1008 qemu_irq *pc_allocate_cpu_irq(void)
1009 {
1010 return qemu_allocate_irqs(pic_irq_request, NULL, 1);
1011 }
1012
1013 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1014 {
1015 DeviceState *dev = NULL;
1016
1017 if (cirrus_vga_enabled) {
1018 if (pci_bus) {
1019 dev = pci_cirrus_vga_init(pci_bus);
1020 } else {
1021 dev = &isa_create_simple(isa_bus, "isa-cirrus-vga")->qdev;
1022 }
1023 } else if (vmsvga_enabled) {
1024 if (pci_bus) {
1025 dev = pci_vmsvga_init(pci_bus);
1026 } else {
1027 fprintf(stderr, "%s: vmware_vga: no PCI bus\n", __FUNCTION__);
1028 }
1029 #ifdef CONFIG_SPICE
1030 } else if (qxl_enabled) {
1031 if (pci_bus) {
1032 dev = &pci_create_simple(pci_bus, -1, "qxl-vga")->qdev;
1033 } else {
1034 fprintf(stderr, "%s: qxl: no PCI bus\n", __FUNCTION__);
1035 }
1036 #endif
1037 } else if (std_vga_enabled) {
1038 if (pci_bus) {
1039 dev = pci_vga_init(pci_bus);
1040 } else {
1041 dev = isa_vga_init(isa_bus);
1042 }
1043 }
1044
1045 return dev;
1046 }
1047
1048 static void cpu_request_exit(void *opaque, int irq, int level)
1049 {
1050 CPUX86State *env = cpu_single_env;
1051
1052 if (env && level) {
1053 cpu_exit(env);
1054 }
1055 }
1056
1057 void pc_basic_device_init(ISABus *isa_bus, qemu_irq *gsi,
1058 ISADevice **rtc_state,
1059 ISADevice **floppy,
1060 bool no_vmport)
1061 {
1062 int i;
1063 DriveInfo *fd[MAX_FD];
1064 DeviceState *hpet = NULL;
1065 int pit_isa_irq = 0;
1066 qemu_irq pit_alt_irq = NULL;
1067 qemu_irq rtc_irq = NULL;
1068 qemu_irq *a20_line;
1069 ISADevice *i8042, *port92, *vmmouse, *pit = NULL;
1070 qemu_irq *cpu_exit_irq;
1071
1072 register_ioport_write(0x80, 1, 1, ioport80_write, NULL);
1073
1074 register_ioport_write(0xf0, 1, 1, ioportF0_write, NULL);
1075
1076 /*
1077 * Check if an HPET shall be created.
1078 *
1079 * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
1080 * when the HPET wants to take over. Thus we have to disable the latter.
1081 */
1082 if (!no_hpet && (!kvm_irqchip_in_kernel() || kvm_has_pit_state2())) {
1083 hpet = sysbus_try_create_simple("hpet", HPET_BASE, NULL);
1084
1085 if (hpet) {
1086 for (i = 0; i < GSI_NUM_PINS; i++) {
1087 sysbus_connect_irq(sysbus_from_qdev(hpet), i, gsi[i]);
1088 }
1089 pit_isa_irq = -1;
1090 pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1091 rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1092 }
1093 }
1094 *rtc_state = rtc_init(isa_bus, 2000, rtc_irq);
1095
1096 qemu_register_boot_set(pc_boot_set, *rtc_state);
1097
1098 if (!xen_enabled()) {
1099 if (kvm_irqchip_in_kernel()) {
1100 pit = kvm_pit_init(isa_bus, 0x40);
1101 } else {
1102 pit = pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1103 }
1104 if (hpet) {
1105 /* connect PIT to output control line of the HPET */
1106 qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(&pit->qdev, 0));
1107 }
1108 pcspk_init(isa_bus, pit);
1109 }
1110
1111 for(i = 0; i < MAX_SERIAL_PORTS; i++) {
1112 if (serial_hds[i]) {
1113 serial_isa_init(isa_bus, i, serial_hds[i]);
1114 }
1115 }
1116
1117 for(i = 0; i < MAX_PARALLEL_PORTS; i++) {
1118 if (parallel_hds[i]) {
1119 parallel_init(isa_bus, i, parallel_hds[i]);
1120 }
1121 }
1122
1123 a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1124 i8042 = isa_create_simple(isa_bus, "i8042");
1125 i8042_setup_a20_line(i8042, &a20_line[0]);
1126 if (!no_vmport) {
1127 vmport_init(isa_bus);
1128 vmmouse = isa_try_create(isa_bus, "vmmouse");
1129 } else {
1130 vmmouse = NULL;
1131 }
1132 if (vmmouse) {
1133 qdev_prop_set_ptr(&vmmouse->qdev, "ps2_mouse", i8042);
1134 qdev_init_nofail(&vmmouse->qdev);
1135 }
1136 port92 = isa_create_simple(isa_bus, "port92");
1137 port92_init(port92, &a20_line[1]);
1138
1139 cpu_exit_irq = qemu_allocate_irqs(cpu_request_exit, NULL, 1);
1140 DMA_init(0, cpu_exit_irq);
1141
1142 for(i = 0; i < MAX_FD; i++) {
1143 fd[i] = drive_get(IF_FLOPPY, 0, i);
1144 }
1145 *floppy = fdctrl_init_isa(isa_bus, fd);
1146 }
1147
1148 void pc_pci_device_init(PCIBus *pci_bus)
1149 {
1150 int max_bus;
1151 int bus;
1152
1153 max_bus = drive_get_max_bus(IF_SCSI);
1154 for (bus = 0; bus <= max_bus; bus++) {
1155 pci_create_simple(pci_bus, -1, "lsi53c895a");
1156 }
1157 }