]> git.proxmox.com Git - qemu.git/blob - hw/pc.c
Merge remote-tracking branch 'kraxel/usb.59' into staging
[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 /* base memory (first MiB) */
342 val = MIN(ram_size / 1024, 640);
343 rtc_set_memory(s, 0x15, val);
344 rtc_set_memory(s, 0x16, val >> 8);
345 /* extended memory (next 64MiB) */
346 if (ram_size > 1024 * 1024) {
347 val = (ram_size - 1024 * 1024) / 1024;
348 } else {
349 val = 0;
350 }
351 if (val > 65535)
352 val = 65535;
353 rtc_set_memory(s, 0x17, val);
354 rtc_set_memory(s, 0x18, val >> 8);
355 rtc_set_memory(s, 0x30, val);
356 rtc_set_memory(s, 0x31, val >> 8);
357 /* memory between 16MiB and 4GiB */
358 if (ram_size > 16 * 1024 * 1024) {
359 val = (ram_size - 16 * 1024 * 1024) / 65536;
360 } else {
361 val = 0;
362 }
363 if (val > 65535)
364 val = 65535;
365 rtc_set_memory(s, 0x34, val);
366 rtc_set_memory(s, 0x35, val >> 8);
367 /* memory above 4GiB */
368 val = above_4g_mem_size / 65536;
369 rtc_set_memory(s, 0x5b, val);
370 rtc_set_memory(s, 0x5c, val >> 8);
371 rtc_set_memory(s, 0x5d, val >> 16);
372
373 /* set the number of CPU */
374 rtc_set_memory(s, 0x5f, smp_cpus - 1);
375
376 /* set boot devices, and disable floppy signature check if requested */
377 if (set_boot_dev(s, boot_device, fd_bootchk)) {
378 exit(1);
379 }
380
381 /* floppy type */
382 if (floppy) {
383 for (i = 0; i < 2; i++) {
384 fd_type[i] = isa_fdc_get_drive_type(floppy, i);
385 }
386 }
387 val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
388 cmos_get_fd_drive_type(fd_type[1]);
389 rtc_set_memory(s, 0x10, val);
390
391 val = 0;
392 nb = 0;
393 if (fd_type[0] < FDRIVE_DRV_NONE) {
394 nb++;
395 }
396 if (fd_type[1] < FDRIVE_DRV_NONE) {
397 nb++;
398 }
399 switch (nb) {
400 case 0:
401 break;
402 case 1:
403 val |= 0x01; /* 1 drive, ready for boot */
404 break;
405 case 2:
406 val |= 0x41; /* 2 drives, ready for boot */
407 break;
408 }
409 val |= 0x02; /* FPU is there */
410 val |= 0x04; /* PS/2 mouse installed */
411 rtc_set_memory(s, REG_EQUIPMENT_BYTE, val);
412
413 /* hard drives */
414 arg.rtc_state = s;
415 arg.idebus[0] = idebus0;
416 arg.idebus[1] = idebus1;
417 qemu_register_reset(pc_cmos_init_late, &arg);
418 }
419
420 /* port 92 stuff: could be split off */
421 typedef struct Port92State {
422 ISADevice dev;
423 MemoryRegion io;
424 uint8_t outport;
425 qemu_irq *a20_out;
426 } Port92State;
427
428 static void port92_write(void *opaque, uint32_t addr, uint32_t val)
429 {
430 Port92State *s = opaque;
431
432 DPRINTF("port92: write 0x%02x\n", val);
433 s->outport = val;
434 qemu_set_irq(*s->a20_out, (val >> 1) & 1);
435 if (val & 1) {
436 qemu_system_reset_request();
437 }
438 }
439
440 static uint32_t port92_read(void *opaque, uint32_t addr)
441 {
442 Port92State *s = opaque;
443 uint32_t ret;
444
445 ret = s->outport;
446 DPRINTF("port92: read 0x%02x\n", ret);
447 return ret;
448 }
449
450 static void port92_init(ISADevice *dev, qemu_irq *a20_out)
451 {
452 Port92State *s = DO_UPCAST(Port92State, dev, dev);
453
454 s->a20_out = a20_out;
455 }
456
457 static const VMStateDescription vmstate_port92_isa = {
458 .name = "port92",
459 .version_id = 1,
460 .minimum_version_id = 1,
461 .minimum_version_id_old = 1,
462 .fields = (VMStateField []) {
463 VMSTATE_UINT8(outport, Port92State),
464 VMSTATE_END_OF_LIST()
465 }
466 };
467
468 static void port92_reset(DeviceState *d)
469 {
470 Port92State *s = container_of(d, Port92State, dev.qdev);
471
472 s->outport &= ~1;
473 }
474
475 static const MemoryRegionPortio port92_portio[] = {
476 { 0, 1, 1, .read = port92_read, .write = port92_write },
477 PORTIO_END_OF_LIST(),
478 };
479
480 static const MemoryRegionOps port92_ops = {
481 .old_portio = port92_portio
482 };
483
484 static int port92_initfn(ISADevice *dev)
485 {
486 Port92State *s = DO_UPCAST(Port92State, dev, dev);
487
488 memory_region_init_io(&s->io, &port92_ops, s, "port92", 1);
489 isa_register_ioport(dev, &s->io, 0x92);
490
491 s->outport = 0;
492 return 0;
493 }
494
495 static void port92_class_initfn(ObjectClass *klass, void *data)
496 {
497 DeviceClass *dc = DEVICE_CLASS(klass);
498 ISADeviceClass *ic = ISA_DEVICE_CLASS(klass);
499 ic->init = port92_initfn;
500 dc->no_user = 1;
501 dc->reset = port92_reset;
502 dc->vmsd = &vmstate_port92_isa;
503 }
504
505 static TypeInfo port92_info = {
506 .name = "port92",
507 .parent = TYPE_ISA_DEVICE,
508 .instance_size = sizeof(Port92State),
509 .class_init = port92_class_initfn,
510 };
511
512 static void port92_register_types(void)
513 {
514 type_register_static(&port92_info);
515 }
516
517 type_init(port92_register_types)
518
519 static void handle_a20_line_change(void *opaque, int irq, int level)
520 {
521 CPUX86State *cpu = opaque;
522
523 /* XXX: send to all CPUs ? */
524 /* XXX: add logic to handle multiple A20 line sources */
525 cpu_x86_set_a20(cpu, level);
526 }
527
528 /***********************************************************/
529 /* Bochs BIOS debug ports */
530
531 static void bochs_bios_write(void *opaque, uint32_t addr, uint32_t val)
532 {
533 static const char shutdown_str[8] = "Shutdown";
534 static int shutdown_index = 0;
535
536 switch(addr) {
537 /* Bochs BIOS messages */
538 case 0x400:
539 case 0x401:
540 /* used to be panic, now unused */
541 break;
542 case 0x402:
543 case 0x403:
544 #ifdef DEBUG_BIOS
545 fprintf(stderr, "%c", val);
546 #endif
547 break;
548 case 0x8900:
549 /* same as Bochs power off */
550 if (val == shutdown_str[shutdown_index]) {
551 shutdown_index++;
552 if (shutdown_index == 8) {
553 shutdown_index = 0;
554 qemu_system_shutdown_request();
555 }
556 } else {
557 shutdown_index = 0;
558 }
559 break;
560
561 /* LGPL'ed VGA BIOS messages */
562 case 0x501:
563 case 0x502:
564 exit((val << 1) | 1);
565 case 0x500:
566 case 0x503:
567 #ifdef DEBUG_BIOS
568 fprintf(stderr, "%c", val);
569 #endif
570 break;
571 }
572 }
573
574 int e820_add_entry(uint64_t address, uint64_t length, uint32_t type)
575 {
576 int index = le32_to_cpu(e820_table.count);
577 struct e820_entry *entry;
578
579 if (index >= E820_NR_ENTRIES)
580 return -EBUSY;
581 entry = &e820_table.entry[index++];
582
583 entry->address = cpu_to_le64(address);
584 entry->length = cpu_to_le64(length);
585 entry->type = cpu_to_le32(type);
586
587 e820_table.count = cpu_to_le32(index);
588 return index;
589 }
590
591 static void *bochs_bios_init(void)
592 {
593 void *fw_cfg;
594 uint8_t *smbios_table;
595 size_t smbios_len;
596 uint64_t *numa_fw_cfg;
597 int i, j;
598
599 register_ioport_write(0x400, 1, 2, bochs_bios_write, NULL);
600 register_ioport_write(0x401, 1, 2, bochs_bios_write, NULL);
601 register_ioport_write(0x402, 1, 1, bochs_bios_write, NULL);
602 register_ioport_write(0x403, 1, 1, bochs_bios_write, NULL);
603 register_ioport_write(0x8900, 1, 1, bochs_bios_write, NULL);
604
605 register_ioport_write(0x501, 1, 1, bochs_bios_write, NULL);
606 register_ioport_write(0x501, 1, 2, bochs_bios_write, NULL);
607 register_ioport_write(0x502, 1, 2, bochs_bios_write, NULL);
608 register_ioport_write(0x500, 1, 1, bochs_bios_write, NULL);
609 register_ioport_write(0x503, 1, 1, bochs_bios_write, NULL);
610
611 fw_cfg = fw_cfg_init(BIOS_CFG_IOPORT, BIOS_CFG_IOPORT + 1, 0, 0);
612
613 fw_cfg_add_i32(fw_cfg, FW_CFG_ID, 1);
614 fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, (uint64_t)ram_size);
615 fw_cfg_add_bytes(fw_cfg, FW_CFG_ACPI_TABLES, (uint8_t *)acpi_tables,
616 acpi_tables_len);
617 fw_cfg_add_i32(fw_cfg, FW_CFG_IRQ0_OVERRIDE, kvm_allows_irq0_override());
618
619 smbios_table = smbios_get_table(&smbios_len);
620 if (smbios_table)
621 fw_cfg_add_bytes(fw_cfg, FW_CFG_SMBIOS_ENTRIES,
622 smbios_table, smbios_len);
623 fw_cfg_add_bytes(fw_cfg, FW_CFG_E820_TABLE, (uint8_t *)&e820_table,
624 sizeof(struct e820_table));
625
626 fw_cfg_add_bytes(fw_cfg, FW_CFG_HPET, (uint8_t *)&hpet_cfg,
627 sizeof(struct hpet_fw_config));
628 /* allocate memory for the NUMA channel: one (64bit) word for the number
629 * of nodes, one word for each VCPU->node and one word for each node to
630 * hold the amount of memory.
631 */
632 numa_fw_cfg = g_malloc0((1 + max_cpus + nb_numa_nodes) * 8);
633 numa_fw_cfg[0] = cpu_to_le64(nb_numa_nodes);
634 for (i = 0; i < max_cpus; i++) {
635 for (j = 0; j < nb_numa_nodes; j++) {
636 if (test_bit(i, node_cpumask[j])) {
637 numa_fw_cfg[i + 1] = cpu_to_le64(j);
638 break;
639 }
640 }
641 }
642 for (i = 0; i < nb_numa_nodes; i++) {
643 numa_fw_cfg[max_cpus + 1 + i] = cpu_to_le64(node_mem[i]);
644 }
645 fw_cfg_add_bytes(fw_cfg, FW_CFG_NUMA, (uint8_t *)numa_fw_cfg,
646 (1 + max_cpus + nb_numa_nodes) * 8);
647
648 return fw_cfg;
649 }
650
651 static long get_file_size(FILE *f)
652 {
653 long where, size;
654
655 /* XXX: on Unix systems, using fstat() probably makes more sense */
656
657 where = ftell(f);
658 fseek(f, 0, SEEK_END);
659 size = ftell(f);
660 fseek(f, where, SEEK_SET);
661
662 return size;
663 }
664
665 static void load_linux(void *fw_cfg,
666 const char *kernel_filename,
667 const char *initrd_filename,
668 const char *kernel_cmdline,
669 target_phys_addr_t max_ram_size)
670 {
671 uint16_t protocol;
672 int setup_size, kernel_size, initrd_size = 0, cmdline_size;
673 uint32_t initrd_max;
674 uint8_t header[8192], *setup, *kernel, *initrd_data;
675 target_phys_addr_t real_addr, prot_addr, cmdline_addr, initrd_addr = 0;
676 FILE *f;
677 char *vmode;
678
679 /* Align to 16 bytes as a paranoia measure */
680 cmdline_size = (strlen(kernel_cmdline)+16) & ~15;
681
682 /* load the kernel header */
683 f = fopen(kernel_filename, "rb");
684 if (!f || !(kernel_size = get_file_size(f)) ||
685 fread(header, 1, MIN(ARRAY_SIZE(header), kernel_size), f) !=
686 MIN(ARRAY_SIZE(header), kernel_size)) {
687 fprintf(stderr, "qemu: could not load kernel '%s': %s\n",
688 kernel_filename, strerror(errno));
689 exit(1);
690 }
691
692 /* kernel protocol version */
693 #if 0
694 fprintf(stderr, "header magic: %#x\n", ldl_p(header+0x202));
695 #endif
696 if (ldl_p(header+0x202) == 0x53726448)
697 protocol = lduw_p(header+0x206);
698 else {
699 /* This looks like a multiboot kernel. If it is, let's stop
700 treating it like a Linux kernel. */
701 if (load_multiboot(fw_cfg, f, kernel_filename, initrd_filename,
702 kernel_cmdline, kernel_size, header))
703 return;
704 protocol = 0;
705 }
706
707 if (protocol < 0x200 || !(header[0x211] & 0x01)) {
708 /* Low kernel */
709 real_addr = 0x90000;
710 cmdline_addr = 0x9a000 - cmdline_size;
711 prot_addr = 0x10000;
712 } else if (protocol < 0x202) {
713 /* High but ancient kernel */
714 real_addr = 0x90000;
715 cmdline_addr = 0x9a000 - cmdline_size;
716 prot_addr = 0x100000;
717 } else {
718 /* High and recent kernel */
719 real_addr = 0x10000;
720 cmdline_addr = 0x20000;
721 prot_addr = 0x100000;
722 }
723
724 #if 0
725 fprintf(stderr,
726 "qemu: real_addr = 0x" TARGET_FMT_plx "\n"
727 "qemu: cmdline_addr = 0x" TARGET_FMT_plx "\n"
728 "qemu: prot_addr = 0x" TARGET_FMT_plx "\n",
729 real_addr,
730 cmdline_addr,
731 prot_addr);
732 #endif
733
734 /* highest address for loading the initrd */
735 if (protocol >= 0x203)
736 initrd_max = ldl_p(header+0x22c);
737 else
738 initrd_max = 0x37ffffff;
739
740 if (initrd_max >= max_ram_size-ACPI_DATA_SIZE)
741 initrd_max = max_ram_size-ACPI_DATA_SIZE-1;
742
743 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_ADDR, cmdline_addr);
744 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_SIZE, strlen(kernel_cmdline)+1);
745 fw_cfg_add_bytes(fw_cfg, FW_CFG_CMDLINE_DATA,
746 (uint8_t*)strdup(kernel_cmdline),
747 strlen(kernel_cmdline)+1);
748
749 if (protocol >= 0x202) {
750 stl_p(header+0x228, cmdline_addr);
751 } else {
752 stw_p(header+0x20, 0xA33F);
753 stw_p(header+0x22, cmdline_addr-real_addr);
754 }
755
756 /* handle vga= parameter */
757 vmode = strstr(kernel_cmdline, "vga=");
758 if (vmode) {
759 unsigned int video_mode;
760 /* skip "vga=" */
761 vmode += 4;
762 if (!strncmp(vmode, "normal", 6)) {
763 video_mode = 0xffff;
764 } else if (!strncmp(vmode, "ext", 3)) {
765 video_mode = 0xfffe;
766 } else if (!strncmp(vmode, "ask", 3)) {
767 video_mode = 0xfffd;
768 } else {
769 video_mode = strtol(vmode, NULL, 0);
770 }
771 stw_p(header+0x1fa, video_mode);
772 }
773
774 /* loader type */
775 /* High nybble = B reserved for QEMU; low nybble is revision number.
776 If this code is substantially changed, you may want to consider
777 incrementing the revision. */
778 if (protocol >= 0x200)
779 header[0x210] = 0xB0;
780
781 /* heap */
782 if (protocol >= 0x201) {
783 header[0x211] |= 0x80; /* CAN_USE_HEAP */
784 stw_p(header+0x224, cmdline_addr-real_addr-0x200);
785 }
786
787 /* load initrd */
788 if (initrd_filename) {
789 if (protocol < 0x200) {
790 fprintf(stderr, "qemu: linux kernel too old to load a ram disk\n");
791 exit(1);
792 }
793
794 initrd_size = get_image_size(initrd_filename);
795 if (initrd_size < 0) {
796 fprintf(stderr, "qemu: error reading initrd %s\n",
797 initrd_filename);
798 exit(1);
799 }
800
801 initrd_addr = (initrd_max-initrd_size) & ~4095;
802
803 initrd_data = g_malloc(initrd_size);
804 load_image(initrd_filename, initrd_data);
805
806 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_ADDR, initrd_addr);
807 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_SIZE, initrd_size);
808 fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, initrd_data, initrd_size);
809
810 stl_p(header+0x218, initrd_addr);
811 stl_p(header+0x21c, initrd_size);
812 }
813
814 /* load kernel and setup */
815 setup_size = header[0x1f1];
816 if (setup_size == 0)
817 setup_size = 4;
818 setup_size = (setup_size+1)*512;
819 kernel_size -= setup_size;
820
821 setup = g_malloc(setup_size);
822 kernel = g_malloc(kernel_size);
823 fseek(f, 0, SEEK_SET);
824 if (fread(setup, 1, setup_size, f) != setup_size) {
825 fprintf(stderr, "fread() failed\n");
826 exit(1);
827 }
828 if (fread(kernel, 1, kernel_size, f) != kernel_size) {
829 fprintf(stderr, "fread() failed\n");
830 exit(1);
831 }
832 fclose(f);
833 memcpy(setup, header, MIN(sizeof(header), setup_size));
834
835 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_ADDR, prot_addr);
836 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_SIZE, kernel_size);
837 fw_cfg_add_bytes(fw_cfg, FW_CFG_KERNEL_DATA, kernel, kernel_size);
838
839 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_ADDR, real_addr);
840 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_SIZE, setup_size);
841 fw_cfg_add_bytes(fw_cfg, FW_CFG_SETUP_DATA, setup, setup_size);
842
843 option_rom[nb_option_roms].name = "linuxboot.bin";
844 option_rom[nb_option_roms].bootindex = 0;
845 nb_option_roms++;
846 }
847
848 #define NE2000_NB_MAX 6
849
850 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
851 0x280, 0x380 };
852 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
853
854 static const int parallel_io[MAX_PARALLEL_PORTS] = { 0x378, 0x278, 0x3bc };
855 static const int parallel_irq[MAX_PARALLEL_PORTS] = { 7, 7, 7 };
856
857 void pc_init_ne2k_isa(ISABus *bus, NICInfo *nd)
858 {
859 static int nb_ne2k = 0;
860
861 if (nb_ne2k == NE2000_NB_MAX)
862 return;
863 isa_ne2000_init(bus, ne2000_io[nb_ne2k],
864 ne2000_irq[nb_ne2k], nd);
865 nb_ne2k++;
866 }
867
868 DeviceState *cpu_get_current_apic(void)
869 {
870 if (cpu_single_env) {
871 return cpu_single_env->apic_state;
872 } else {
873 return NULL;
874 }
875 }
876
877 static DeviceState *apic_init(void *env, uint8_t apic_id)
878 {
879 DeviceState *dev;
880 static int apic_mapped;
881
882 if (kvm_irqchip_in_kernel()) {
883 dev = qdev_create(NULL, "kvm-apic");
884 } else if (xen_enabled()) {
885 dev = qdev_create(NULL, "xen-apic");
886 } else {
887 dev = qdev_create(NULL, "apic");
888 }
889
890 qdev_prop_set_uint8(dev, "id", apic_id);
891 qdev_prop_set_ptr(dev, "cpu_env", env);
892 qdev_init_nofail(dev);
893
894 /* XXX: mapping more APICs at the same memory location */
895 if (apic_mapped == 0) {
896 /* NOTE: the APIC is directly connected to the CPU - it is not
897 on the global memory bus. */
898 /* XXX: what if the base changes? */
899 sysbus_mmio_map(sysbus_from_qdev(dev), 0, MSI_ADDR_BASE);
900 apic_mapped = 1;
901 }
902
903 return dev;
904 }
905
906 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
907 {
908 CPUX86State *s = opaque;
909
910 if (level) {
911 cpu_interrupt(s, CPU_INTERRUPT_SMI);
912 }
913 }
914
915 static X86CPU *pc_new_cpu(const char *cpu_model)
916 {
917 X86CPU *cpu;
918 CPUX86State *env;
919
920 cpu = cpu_x86_init(cpu_model);
921 if (cpu == NULL) {
922 fprintf(stderr, "Unable to find x86 CPU definition\n");
923 exit(1);
924 }
925 env = &cpu->env;
926 if ((env->cpuid_features & CPUID_APIC) || smp_cpus > 1) {
927 env->apic_state = apic_init(env, env->cpuid_apic_id);
928 }
929 cpu_reset(CPU(cpu));
930 return cpu;
931 }
932
933 void pc_cpus_init(const char *cpu_model)
934 {
935 int i;
936
937 /* init CPUs */
938 if (cpu_model == NULL) {
939 #ifdef TARGET_X86_64
940 cpu_model = "qemu64";
941 #else
942 cpu_model = "qemu32";
943 #endif
944 }
945
946 for(i = 0; i < smp_cpus; i++) {
947 pc_new_cpu(cpu_model);
948 }
949 }
950
951 void *pc_memory_init(MemoryRegion *system_memory,
952 const char *kernel_filename,
953 const char *kernel_cmdline,
954 const char *initrd_filename,
955 ram_addr_t below_4g_mem_size,
956 ram_addr_t above_4g_mem_size,
957 MemoryRegion *rom_memory,
958 MemoryRegion **ram_memory)
959 {
960 int linux_boot, i;
961 MemoryRegion *ram, *option_rom_mr;
962 MemoryRegion *ram_below_4g, *ram_above_4g;
963 void *fw_cfg;
964
965 linux_boot = (kernel_filename != NULL);
966
967 /* Allocate RAM. We allocate it as a single memory region and use
968 * aliases to address portions of it, mostly for backwards compatibility
969 * with older qemus that used qemu_ram_alloc().
970 */
971 ram = g_malloc(sizeof(*ram));
972 memory_region_init_ram(ram, "pc.ram",
973 below_4g_mem_size + above_4g_mem_size);
974 vmstate_register_ram_global(ram);
975 *ram_memory = ram;
976 ram_below_4g = g_malloc(sizeof(*ram_below_4g));
977 memory_region_init_alias(ram_below_4g, "ram-below-4g", ram,
978 0, below_4g_mem_size);
979 memory_region_add_subregion(system_memory, 0, ram_below_4g);
980 if (above_4g_mem_size > 0) {
981 ram_above_4g = g_malloc(sizeof(*ram_above_4g));
982 memory_region_init_alias(ram_above_4g, "ram-above-4g", ram,
983 below_4g_mem_size, above_4g_mem_size);
984 memory_region_add_subregion(system_memory, 0x100000000ULL,
985 ram_above_4g);
986 }
987
988
989 /* Initialize PC system firmware */
990 pc_system_firmware_init(rom_memory);
991
992 option_rom_mr = g_malloc(sizeof(*option_rom_mr));
993 memory_region_init_ram(option_rom_mr, "pc.rom", PC_ROM_SIZE);
994 vmstate_register_ram_global(option_rom_mr);
995 memory_region_add_subregion_overlap(rom_memory,
996 PC_ROM_MIN_VGA,
997 option_rom_mr,
998 1);
999
1000 fw_cfg = bochs_bios_init();
1001 rom_set_fw(fw_cfg);
1002
1003 if (linux_boot) {
1004 load_linux(fw_cfg, kernel_filename, initrd_filename, kernel_cmdline, below_4g_mem_size);
1005 }
1006
1007 for (i = 0; i < nb_option_roms; i++) {
1008 rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1009 }
1010 return fw_cfg;
1011 }
1012
1013 qemu_irq *pc_allocate_cpu_irq(void)
1014 {
1015 return qemu_allocate_irqs(pic_irq_request, NULL, 1);
1016 }
1017
1018 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1019 {
1020 DeviceState *dev = NULL;
1021
1022 if (cirrus_vga_enabled) {
1023 if (pci_bus) {
1024 dev = pci_cirrus_vga_init(pci_bus);
1025 } else {
1026 dev = &isa_create_simple(isa_bus, "isa-cirrus-vga")->qdev;
1027 }
1028 } else if (vmsvga_enabled) {
1029 if (pci_bus) {
1030 dev = pci_vmsvga_init(pci_bus);
1031 } else {
1032 fprintf(stderr, "%s: vmware_vga: no PCI bus\n", __FUNCTION__);
1033 }
1034 #ifdef CONFIG_SPICE
1035 } else if (qxl_enabled) {
1036 if (pci_bus) {
1037 dev = &pci_create_simple(pci_bus, -1, "qxl-vga")->qdev;
1038 } else {
1039 fprintf(stderr, "%s: qxl: no PCI bus\n", __FUNCTION__);
1040 }
1041 #endif
1042 } else if (std_vga_enabled) {
1043 if (pci_bus) {
1044 dev = pci_vga_init(pci_bus);
1045 } else {
1046 dev = isa_vga_init(isa_bus);
1047 }
1048 }
1049
1050 return dev;
1051 }
1052
1053 static void cpu_request_exit(void *opaque, int irq, int level)
1054 {
1055 CPUX86State *env = cpu_single_env;
1056
1057 if (env && level) {
1058 cpu_exit(env);
1059 }
1060 }
1061
1062 void pc_basic_device_init(ISABus *isa_bus, qemu_irq *gsi,
1063 ISADevice **rtc_state,
1064 ISADevice **floppy,
1065 bool no_vmport)
1066 {
1067 int i;
1068 DriveInfo *fd[MAX_FD];
1069 DeviceState *hpet = NULL;
1070 int pit_isa_irq = 0;
1071 qemu_irq pit_alt_irq = NULL;
1072 qemu_irq rtc_irq = NULL;
1073 qemu_irq *a20_line;
1074 ISADevice *i8042, *port92, *vmmouse, *pit = NULL;
1075 qemu_irq *cpu_exit_irq;
1076
1077 register_ioport_write(0x80, 1, 1, ioport80_write, NULL);
1078
1079 register_ioport_write(0xf0, 1, 1, ioportF0_write, NULL);
1080
1081 /*
1082 * Check if an HPET shall be created.
1083 *
1084 * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
1085 * when the HPET wants to take over. Thus we have to disable the latter.
1086 */
1087 if (!no_hpet && (!kvm_irqchip_in_kernel() || kvm_has_pit_state2())) {
1088 hpet = sysbus_try_create_simple("hpet", HPET_BASE, NULL);
1089
1090 if (hpet) {
1091 for (i = 0; i < GSI_NUM_PINS; i++) {
1092 sysbus_connect_irq(sysbus_from_qdev(hpet), i, gsi[i]);
1093 }
1094 pit_isa_irq = -1;
1095 pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1096 rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1097 }
1098 }
1099 *rtc_state = rtc_init(isa_bus, 2000, rtc_irq);
1100
1101 qemu_register_boot_set(pc_boot_set, *rtc_state);
1102
1103 if (!xen_enabled()) {
1104 if (kvm_irqchip_in_kernel()) {
1105 pit = kvm_pit_init(isa_bus, 0x40);
1106 } else {
1107 pit = pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1108 }
1109 if (hpet) {
1110 /* connect PIT to output control line of the HPET */
1111 qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(&pit->qdev, 0));
1112 }
1113 pcspk_init(isa_bus, pit);
1114 }
1115
1116 for(i = 0; i < MAX_SERIAL_PORTS; i++) {
1117 if (serial_hds[i]) {
1118 serial_isa_init(isa_bus, i, serial_hds[i]);
1119 }
1120 }
1121
1122 for(i = 0; i < MAX_PARALLEL_PORTS; i++) {
1123 if (parallel_hds[i]) {
1124 parallel_init(isa_bus, i, parallel_hds[i]);
1125 }
1126 }
1127
1128 a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1129 i8042 = isa_create_simple(isa_bus, "i8042");
1130 i8042_setup_a20_line(i8042, &a20_line[0]);
1131 if (!no_vmport) {
1132 vmport_init(isa_bus);
1133 vmmouse = isa_try_create(isa_bus, "vmmouse");
1134 } else {
1135 vmmouse = NULL;
1136 }
1137 if (vmmouse) {
1138 qdev_prop_set_ptr(&vmmouse->qdev, "ps2_mouse", i8042);
1139 qdev_init_nofail(&vmmouse->qdev);
1140 }
1141 port92 = isa_create_simple(isa_bus, "port92");
1142 port92_init(port92, &a20_line[1]);
1143
1144 cpu_exit_irq = qemu_allocate_irqs(cpu_request_exit, NULL, 1);
1145 DMA_init(0, cpu_exit_irq);
1146
1147 for(i = 0; i < MAX_FD; i++) {
1148 fd[i] = drive_get(IF_FLOPPY, 0, i);
1149 }
1150 *floppy = fdctrl_init_isa(isa_bus, fd);
1151 }
1152
1153 void pc_pci_device_init(PCIBus *pci_bus)
1154 {
1155 int max_bus;
1156 int bus;
1157
1158 max_bus = drive_get_max_bus(IF_SCSI);
1159 for (bus = 0; bus <= max_bus; bus++) {
1160 pci_create_simple(pci_bus, -1, "lsi53c895a");
1161 }
1162 }