]> git.proxmox.com Git - mirror_qemu.git/blob - hw/pc.c
Add new macro QEMU_PACKED for packed C structures
[mirror_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 "msix.h"
40 #include "sysbus.h"
41 #include "sysemu.h"
42 #include "blockdev.h"
43 #include "ui/qemu-spice.h"
44 #include "memory.h"
45 #include "exec-memory.h"
46
47 /* output Bochs bios info messages */
48 //#define DEBUG_BIOS
49
50 /* debug PC/ISA interrupts */
51 //#define DEBUG_IRQ
52
53 #ifdef DEBUG_IRQ
54 #define DPRINTF(fmt, ...) \
55 do { printf("CPUIRQ: " fmt , ## __VA_ARGS__); } while (0)
56 #else
57 #define DPRINTF(fmt, ...)
58 #endif
59
60 #define BIOS_FILENAME "bios.bin"
61
62 #define PC_MAX_BIOS_SIZE (4 * 1024 * 1024)
63
64 /* Leave a chunk of memory at the top of RAM for the BIOS ACPI tables. */
65 #define ACPI_DATA_SIZE 0x10000
66 #define BIOS_CFG_IOPORT 0x510
67 #define FW_CFG_ACPI_TABLES (FW_CFG_ARCH_LOCAL + 0)
68 #define FW_CFG_SMBIOS_ENTRIES (FW_CFG_ARCH_LOCAL + 1)
69 #define FW_CFG_IRQ0_OVERRIDE (FW_CFG_ARCH_LOCAL + 2)
70 #define FW_CFG_E820_TABLE (FW_CFG_ARCH_LOCAL + 3)
71 #define FW_CFG_HPET (FW_CFG_ARCH_LOCAL + 4)
72
73 #define MSI_ADDR_BASE 0xfee00000
74
75 #define E820_NR_ENTRIES 16
76
77 struct e820_entry {
78 uint64_t address;
79 uint64_t length;
80 uint32_t type;
81 } __attribute((__packed__, __aligned__(4)));
82
83 struct e820_table {
84 uint32_t count;
85 struct e820_entry entry[E820_NR_ENTRIES];
86 } __attribute((__packed__, __aligned__(4)));
87
88 static struct e820_table e820_table;
89 struct hpet_fw_config hpet_cfg = {.count = UINT8_MAX};
90
91 void isa_irq_handler(void *opaque, int n, int level)
92 {
93 IsaIrqState *isa = (IsaIrqState *)opaque;
94
95 DPRINTF("isa_irqs: %s irq %d\n", level? "raise" : "lower", n);
96 if (n < 16) {
97 qemu_set_irq(isa->i8259[n], level);
98 }
99 if (isa->ioapic)
100 qemu_set_irq(isa->ioapic[n], level);
101 };
102
103 static void ioport80_write(void *opaque, uint32_t addr, uint32_t data)
104 {
105 }
106
107 /* MSDOS compatibility mode FPU exception support */
108 static qemu_irq ferr_irq;
109
110 void pc_register_ferr_irq(qemu_irq irq)
111 {
112 ferr_irq = irq;
113 }
114
115 /* XXX: add IGNNE support */
116 void cpu_set_ferr(CPUX86State *s)
117 {
118 qemu_irq_raise(ferr_irq);
119 }
120
121 static void ioportF0_write(void *opaque, uint32_t addr, uint32_t data)
122 {
123 qemu_irq_lower(ferr_irq);
124 }
125
126 /* TSC handling */
127 uint64_t cpu_get_tsc(CPUX86State *env)
128 {
129 return cpu_get_ticks();
130 }
131
132 /* SMM support */
133
134 static cpu_set_smm_t smm_set;
135 static void *smm_arg;
136
137 void cpu_smm_register(cpu_set_smm_t callback, void *arg)
138 {
139 assert(smm_set == NULL);
140 assert(smm_arg == NULL);
141 smm_set = callback;
142 smm_arg = arg;
143 }
144
145 void cpu_smm_update(CPUState *env)
146 {
147 if (smm_set && smm_arg && env == first_cpu)
148 smm_set(!!(env->hflags & HF_SMM_MASK), smm_arg);
149 }
150
151
152 /* IRQ handling */
153 int cpu_get_pic_interrupt(CPUState *env)
154 {
155 int intno;
156
157 intno = apic_get_interrupt(env->apic_state);
158 if (intno >= 0) {
159 /* set irq request if a PIC irq is still pending */
160 /* XXX: improve that */
161 pic_update_irq(isa_pic);
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 CPUState *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(int type_ofs, int info_ofs, BlockDriverState *hd,
223 ISADevice *s)
224 {
225 int cylinders, heads, sectors;
226 bdrv_get_geometry_hint(hd, &cylinders, &heads, &sectors);
227 rtc_set_memory(s, type_ofs, 47);
228 rtc_set_memory(s, info_ofs, cylinders);
229 rtc_set_memory(s, info_ofs + 1, cylinders >> 8);
230 rtc_set_memory(s, info_ofs + 2, heads);
231 rtc_set_memory(s, info_ofs + 3, 0xff);
232 rtc_set_memory(s, info_ofs + 4, 0xff);
233 rtc_set_memory(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
234 rtc_set_memory(s, info_ofs + 6, cylinders);
235 rtc_set_memory(s, info_ofs + 7, cylinders >> 8);
236 rtc_set_memory(s, info_ofs + 8, sectors);
237 }
238
239 /* convert boot_device letter to something recognizable by the bios */
240 static int boot_device2nibble(char boot_device)
241 {
242 switch(boot_device) {
243 case 'a':
244 case 'b':
245 return 0x01; /* floppy boot */
246 case 'c':
247 return 0x02; /* hard drive boot */
248 case 'd':
249 return 0x03; /* CD-ROM boot */
250 case 'n':
251 return 0x04; /* Network boot */
252 }
253 return 0;
254 }
255
256 static int set_boot_dev(ISADevice *s, const char *boot_device, int fd_bootchk)
257 {
258 #define PC_MAX_BOOT_DEVICES 3
259 int nbds, bds[3] = { 0, };
260 int i;
261
262 nbds = strlen(boot_device);
263 if (nbds > PC_MAX_BOOT_DEVICES) {
264 error_report("Too many boot devices for PC");
265 return(1);
266 }
267 for (i = 0; i < nbds; i++) {
268 bds[i] = boot_device2nibble(boot_device[i]);
269 if (bds[i] == 0) {
270 error_report("Invalid boot device for PC: '%c'",
271 boot_device[i]);
272 return(1);
273 }
274 }
275 rtc_set_memory(s, 0x3d, (bds[1] << 4) | bds[0]);
276 rtc_set_memory(s, 0x38, (bds[2] << 4) | (fd_bootchk ? 0x0 : 0x1));
277 return(0);
278 }
279
280 static int pc_boot_set(void *opaque, const char *boot_device)
281 {
282 return set_boot_dev(opaque, boot_device, 0);
283 }
284
285 typedef struct pc_cmos_init_late_arg {
286 ISADevice *rtc_state;
287 BusState *idebus0, *idebus1;
288 } pc_cmos_init_late_arg;
289
290 static void pc_cmos_init_late(void *opaque)
291 {
292 pc_cmos_init_late_arg *arg = opaque;
293 ISADevice *s = arg->rtc_state;
294 int val;
295 BlockDriverState *hd_table[4];
296 int i;
297
298 ide_get_bs(hd_table, arg->idebus0);
299 ide_get_bs(hd_table + 2, arg->idebus1);
300
301 rtc_set_memory(s, 0x12, (hd_table[0] ? 0xf0 : 0) | (hd_table[1] ? 0x0f : 0));
302 if (hd_table[0])
303 cmos_init_hd(0x19, 0x1b, hd_table[0], s);
304 if (hd_table[1])
305 cmos_init_hd(0x1a, 0x24, hd_table[1], s);
306
307 val = 0;
308 for (i = 0; i < 4; i++) {
309 if (hd_table[i]) {
310 int cylinders, heads, sectors, translation;
311 /* NOTE: bdrv_get_geometry_hint() returns the physical
312 geometry. It is always such that: 1 <= sects <= 63, 1
313 <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
314 geometry can be different if a translation is done. */
315 translation = bdrv_get_translation_hint(hd_table[i]);
316 if (translation == BIOS_ATA_TRANSLATION_AUTO) {
317 bdrv_get_geometry_hint(hd_table[i], &cylinders, &heads, &sectors);
318 if (cylinders <= 1024 && heads <= 16 && sectors <= 63) {
319 /* No translation. */
320 translation = 0;
321 } else {
322 /* LBA translation. */
323 translation = 1;
324 }
325 } else {
326 translation--;
327 }
328 val |= translation << (i * 2);
329 }
330 }
331 rtc_set_memory(s, 0x39, val);
332
333 qemu_unregister_reset(pc_cmos_init_late, opaque);
334 }
335
336 void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
337 const char *boot_device,
338 BusState *idebus0, BusState *idebus1,
339 ISADevice *s)
340 {
341 int val, nb, nb_heads, max_track, last_sect, i;
342 FDriveType fd_type[2];
343 DriveInfo *fd[2];
344 static pc_cmos_init_late_arg arg;
345
346 /* various important CMOS locations needed by PC/Bochs bios */
347
348 /* memory size */
349 val = 640; /* base memory in K */
350 rtc_set_memory(s, 0x15, val);
351 rtc_set_memory(s, 0x16, val >> 8);
352
353 val = (ram_size / 1024) - 1024;
354 if (val > 65535)
355 val = 65535;
356 rtc_set_memory(s, 0x17, val);
357 rtc_set_memory(s, 0x18, val >> 8);
358 rtc_set_memory(s, 0x30, val);
359 rtc_set_memory(s, 0x31, val >> 8);
360
361 if (above_4g_mem_size) {
362 rtc_set_memory(s, 0x5b, (unsigned int)above_4g_mem_size >> 16);
363 rtc_set_memory(s, 0x5c, (unsigned int)above_4g_mem_size >> 24);
364 rtc_set_memory(s, 0x5d, (uint64_t)above_4g_mem_size >> 32);
365 }
366
367 if (ram_size > (16 * 1024 * 1024))
368 val = (ram_size / 65536) - ((16 * 1024 * 1024) / 65536);
369 else
370 val = 0;
371 if (val > 65535)
372 val = 65535;
373 rtc_set_memory(s, 0x34, val);
374 rtc_set_memory(s, 0x35, val >> 8);
375
376 /* set the number of CPU */
377 rtc_set_memory(s, 0x5f, smp_cpus - 1);
378
379 /* set boot devices, and disable floppy signature check if requested */
380 if (set_boot_dev(s, boot_device, fd_bootchk)) {
381 exit(1);
382 }
383
384 /* floppy type */
385 for (i = 0; i < 2; i++) {
386 fd[i] = drive_get(IF_FLOPPY, 0, i);
387 if (fd[i] && bdrv_is_inserted(fd[i]->bdrv)) {
388 bdrv_get_floppy_geometry_hint(fd[i]->bdrv, &nb_heads, &max_track,
389 &last_sect, FDRIVE_DRV_NONE,
390 &fd_type[i]);
391 } else {
392 fd_type[i] = FDRIVE_DRV_NONE;
393 }
394 }
395 val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
396 cmos_get_fd_drive_type(fd_type[1]);
397 rtc_set_memory(s, 0x10, val);
398
399 val = 0;
400 nb = 0;
401 if (fd_type[0] < FDRIVE_DRV_NONE) {
402 nb++;
403 }
404 if (fd_type[1] < FDRIVE_DRV_NONE) {
405 nb++;
406 }
407 switch (nb) {
408 case 0:
409 break;
410 case 1:
411 val |= 0x01; /* 1 drive, ready for boot */
412 break;
413 case 2:
414 val |= 0x41; /* 2 drives, ready for boot */
415 break;
416 }
417 val |= 0x02; /* FPU is there */
418 val |= 0x04; /* PS/2 mouse installed */
419 rtc_set_memory(s, REG_EQUIPMENT_BYTE, val);
420
421 /* hard drives */
422 arg.rtc_state = s;
423 arg.idebus0 = idebus0;
424 arg.idebus1 = idebus1;
425 qemu_register_reset(pc_cmos_init_late, &arg);
426 }
427
428 /* port 92 stuff: could be split off */
429 typedef struct Port92State {
430 ISADevice dev;
431 uint8_t outport;
432 qemu_irq *a20_out;
433 } Port92State;
434
435 static void port92_write(void *opaque, uint32_t addr, uint32_t val)
436 {
437 Port92State *s = opaque;
438
439 DPRINTF("port92: write 0x%02x\n", val);
440 s->outport = val;
441 qemu_set_irq(*s->a20_out, (val >> 1) & 1);
442 if (val & 1) {
443 qemu_system_reset_request();
444 }
445 }
446
447 static uint32_t port92_read(void *opaque, uint32_t addr)
448 {
449 Port92State *s = opaque;
450 uint32_t ret;
451
452 ret = s->outport;
453 DPRINTF("port92: read 0x%02x\n", ret);
454 return ret;
455 }
456
457 static void port92_init(ISADevice *dev, qemu_irq *a20_out)
458 {
459 Port92State *s = DO_UPCAST(Port92State, dev, dev);
460
461 s->a20_out = a20_out;
462 }
463
464 static const VMStateDescription vmstate_port92_isa = {
465 .name = "port92",
466 .version_id = 1,
467 .minimum_version_id = 1,
468 .minimum_version_id_old = 1,
469 .fields = (VMStateField []) {
470 VMSTATE_UINT8(outport, Port92State),
471 VMSTATE_END_OF_LIST()
472 }
473 };
474
475 static void port92_reset(DeviceState *d)
476 {
477 Port92State *s = container_of(d, Port92State, dev.qdev);
478
479 s->outport &= ~1;
480 }
481
482 static int port92_initfn(ISADevice *dev)
483 {
484 Port92State *s = DO_UPCAST(Port92State, dev, dev);
485
486 register_ioport_read(0x92, 1, 1, port92_read, s);
487 register_ioport_write(0x92, 1, 1, port92_write, s);
488 isa_init_ioport(dev, 0x92);
489 s->outport = 0;
490 return 0;
491 }
492
493 static ISADeviceInfo port92_info = {
494 .qdev.name = "port92",
495 .qdev.size = sizeof(Port92State),
496 .qdev.vmsd = &vmstate_port92_isa,
497 .qdev.no_user = 1,
498 .qdev.reset = port92_reset,
499 .init = port92_initfn,
500 };
501
502 static void port92_register(void)
503 {
504 isa_qdev_register(&port92_info);
505 }
506 device_init(port92_register)
507
508 static void handle_a20_line_change(void *opaque, int irq, int level)
509 {
510 CPUState *cpu = opaque;
511
512 /* XXX: send to all CPUs ? */
513 /* XXX: add logic to handle multiple A20 line sources */
514 cpu_x86_set_a20(cpu, level);
515 }
516
517 /***********************************************************/
518 /* Bochs BIOS debug ports */
519
520 static void bochs_bios_write(void *opaque, uint32_t addr, uint32_t val)
521 {
522 static const char shutdown_str[8] = "Shutdown";
523 static int shutdown_index = 0;
524
525 switch(addr) {
526 /* Bochs BIOS messages */
527 case 0x400:
528 case 0x401:
529 /* used to be panic, now unused */
530 break;
531 case 0x402:
532 case 0x403:
533 #ifdef DEBUG_BIOS
534 fprintf(stderr, "%c", val);
535 #endif
536 break;
537 case 0x8900:
538 /* same as Bochs power off */
539 if (val == shutdown_str[shutdown_index]) {
540 shutdown_index++;
541 if (shutdown_index == 8) {
542 shutdown_index = 0;
543 qemu_system_shutdown_request();
544 }
545 } else {
546 shutdown_index = 0;
547 }
548 break;
549
550 /* LGPL'ed VGA BIOS messages */
551 case 0x501:
552 case 0x502:
553 exit((val << 1) | 1);
554 case 0x500:
555 case 0x503:
556 #ifdef DEBUG_BIOS
557 fprintf(stderr, "%c", val);
558 #endif
559 break;
560 }
561 }
562
563 int e820_add_entry(uint64_t address, uint64_t length, uint32_t type)
564 {
565 int index = le32_to_cpu(e820_table.count);
566 struct e820_entry *entry;
567
568 if (index >= E820_NR_ENTRIES)
569 return -EBUSY;
570 entry = &e820_table.entry[index++];
571
572 entry->address = cpu_to_le64(address);
573 entry->length = cpu_to_le64(length);
574 entry->type = cpu_to_le32(type);
575
576 e820_table.count = cpu_to_le32(index);
577 return index;
578 }
579
580 static void *bochs_bios_init(void)
581 {
582 void *fw_cfg;
583 uint8_t *smbios_table;
584 size_t smbios_len;
585 uint64_t *numa_fw_cfg;
586 int i, j;
587
588 register_ioport_write(0x400, 1, 2, bochs_bios_write, NULL);
589 register_ioport_write(0x401, 1, 2, bochs_bios_write, NULL);
590 register_ioport_write(0x402, 1, 1, bochs_bios_write, NULL);
591 register_ioport_write(0x403, 1, 1, bochs_bios_write, NULL);
592 register_ioport_write(0x8900, 1, 1, bochs_bios_write, NULL);
593
594 register_ioport_write(0x501, 1, 1, bochs_bios_write, NULL);
595 register_ioport_write(0x501, 1, 2, bochs_bios_write, NULL);
596 register_ioport_write(0x502, 1, 2, bochs_bios_write, NULL);
597 register_ioport_write(0x500, 1, 1, bochs_bios_write, NULL);
598 register_ioport_write(0x503, 1, 1, bochs_bios_write, NULL);
599
600 fw_cfg = fw_cfg_init(BIOS_CFG_IOPORT, BIOS_CFG_IOPORT + 1, 0, 0);
601
602 fw_cfg_add_i32(fw_cfg, FW_CFG_ID, 1);
603 fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, (uint64_t)ram_size);
604 fw_cfg_add_bytes(fw_cfg, FW_CFG_ACPI_TABLES, (uint8_t *)acpi_tables,
605 acpi_tables_len);
606 fw_cfg_add_bytes(fw_cfg, FW_CFG_IRQ0_OVERRIDE, &irq0override, 1);
607
608 smbios_table = smbios_get_table(&smbios_len);
609 if (smbios_table)
610 fw_cfg_add_bytes(fw_cfg, FW_CFG_SMBIOS_ENTRIES,
611 smbios_table, smbios_len);
612 fw_cfg_add_bytes(fw_cfg, FW_CFG_E820_TABLE, (uint8_t *)&e820_table,
613 sizeof(struct e820_table));
614
615 fw_cfg_add_bytes(fw_cfg, FW_CFG_HPET, (uint8_t *)&hpet_cfg,
616 sizeof(struct hpet_fw_config));
617 /* allocate memory for the NUMA channel: one (64bit) word for the number
618 * of nodes, one word for each VCPU->node and one word for each node to
619 * hold the amount of memory.
620 */
621 numa_fw_cfg = g_malloc0((1 + smp_cpus + nb_numa_nodes) * 8);
622 numa_fw_cfg[0] = cpu_to_le64(nb_numa_nodes);
623 for (i = 0; i < smp_cpus; i++) {
624 for (j = 0; j < nb_numa_nodes; j++) {
625 if (node_cpumask[j] & (1 << i)) {
626 numa_fw_cfg[i + 1] = cpu_to_le64(j);
627 break;
628 }
629 }
630 }
631 for (i = 0; i < nb_numa_nodes; i++) {
632 numa_fw_cfg[smp_cpus + 1 + i] = cpu_to_le64(node_mem[i]);
633 }
634 fw_cfg_add_bytes(fw_cfg, FW_CFG_NUMA, (uint8_t *)numa_fw_cfg,
635 (1 + smp_cpus + nb_numa_nodes) * 8);
636
637 return fw_cfg;
638 }
639
640 static long get_file_size(FILE *f)
641 {
642 long where, size;
643
644 /* XXX: on Unix systems, using fstat() probably makes more sense */
645
646 where = ftell(f);
647 fseek(f, 0, SEEK_END);
648 size = ftell(f);
649 fseek(f, where, SEEK_SET);
650
651 return size;
652 }
653
654 static void load_linux(void *fw_cfg,
655 const char *kernel_filename,
656 const char *initrd_filename,
657 const char *kernel_cmdline,
658 target_phys_addr_t max_ram_size)
659 {
660 uint16_t protocol;
661 int setup_size, kernel_size, initrd_size = 0, cmdline_size;
662 uint32_t initrd_max;
663 uint8_t header[8192], *setup, *kernel, *initrd_data;
664 target_phys_addr_t real_addr, prot_addr, cmdline_addr, initrd_addr = 0;
665 FILE *f;
666 char *vmode;
667
668 /* Align to 16 bytes as a paranoia measure */
669 cmdline_size = (strlen(kernel_cmdline)+16) & ~15;
670
671 /* load the kernel header */
672 f = fopen(kernel_filename, "rb");
673 if (!f || !(kernel_size = get_file_size(f)) ||
674 fread(header, 1, MIN(ARRAY_SIZE(header), kernel_size), f) !=
675 MIN(ARRAY_SIZE(header), kernel_size)) {
676 fprintf(stderr, "qemu: could not load kernel '%s': %s\n",
677 kernel_filename, strerror(errno));
678 exit(1);
679 }
680
681 /* kernel protocol version */
682 #if 0
683 fprintf(stderr, "header magic: %#x\n", ldl_p(header+0x202));
684 #endif
685 if (ldl_p(header+0x202) == 0x53726448)
686 protocol = lduw_p(header+0x206);
687 else {
688 /* This looks like a multiboot kernel. If it is, let's stop
689 treating it like a Linux kernel. */
690 if (load_multiboot(fw_cfg, f, kernel_filename, initrd_filename,
691 kernel_cmdline, kernel_size, header))
692 return;
693 protocol = 0;
694 }
695
696 if (protocol < 0x200 || !(header[0x211] & 0x01)) {
697 /* Low kernel */
698 real_addr = 0x90000;
699 cmdline_addr = 0x9a000 - cmdline_size;
700 prot_addr = 0x10000;
701 } else if (protocol < 0x202) {
702 /* High but ancient kernel */
703 real_addr = 0x90000;
704 cmdline_addr = 0x9a000 - cmdline_size;
705 prot_addr = 0x100000;
706 } else {
707 /* High and recent kernel */
708 real_addr = 0x10000;
709 cmdline_addr = 0x20000;
710 prot_addr = 0x100000;
711 }
712
713 #if 0
714 fprintf(stderr,
715 "qemu: real_addr = 0x" TARGET_FMT_plx "\n"
716 "qemu: cmdline_addr = 0x" TARGET_FMT_plx "\n"
717 "qemu: prot_addr = 0x" TARGET_FMT_plx "\n",
718 real_addr,
719 cmdline_addr,
720 prot_addr);
721 #endif
722
723 /* highest address for loading the initrd */
724 if (protocol >= 0x203)
725 initrd_max = ldl_p(header+0x22c);
726 else
727 initrd_max = 0x37ffffff;
728
729 if (initrd_max >= max_ram_size-ACPI_DATA_SIZE)
730 initrd_max = max_ram_size-ACPI_DATA_SIZE-1;
731
732 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_ADDR, cmdline_addr);
733 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_SIZE, strlen(kernel_cmdline)+1);
734 fw_cfg_add_bytes(fw_cfg, FW_CFG_CMDLINE_DATA,
735 (uint8_t*)strdup(kernel_cmdline),
736 strlen(kernel_cmdline)+1);
737
738 if (protocol >= 0x202) {
739 stl_p(header+0x228, cmdline_addr);
740 } else {
741 stw_p(header+0x20, 0xA33F);
742 stw_p(header+0x22, cmdline_addr-real_addr);
743 }
744
745 /* handle vga= parameter */
746 vmode = strstr(kernel_cmdline, "vga=");
747 if (vmode) {
748 unsigned int video_mode;
749 /* skip "vga=" */
750 vmode += 4;
751 if (!strncmp(vmode, "normal", 6)) {
752 video_mode = 0xffff;
753 } else if (!strncmp(vmode, "ext", 3)) {
754 video_mode = 0xfffe;
755 } else if (!strncmp(vmode, "ask", 3)) {
756 video_mode = 0xfffd;
757 } else {
758 video_mode = strtol(vmode, NULL, 0);
759 }
760 stw_p(header+0x1fa, video_mode);
761 }
762
763 /* loader type */
764 /* High nybble = B reserved for Qemu; low nybble is revision number.
765 If this code is substantially changed, you may want to consider
766 incrementing the revision. */
767 if (protocol >= 0x200)
768 header[0x210] = 0xB0;
769
770 /* heap */
771 if (protocol >= 0x201) {
772 header[0x211] |= 0x80; /* CAN_USE_HEAP */
773 stw_p(header+0x224, cmdline_addr-real_addr-0x200);
774 }
775
776 /* load initrd */
777 if (initrd_filename) {
778 if (protocol < 0x200) {
779 fprintf(stderr, "qemu: linux kernel too old to load a ram disk\n");
780 exit(1);
781 }
782
783 initrd_size = get_image_size(initrd_filename);
784 if (initrd_size < 0) {
785 fprintf(stderr, "qemu: error reading initrd %s\n",
786 initrd_filename);
787 exit(1);
788 }
789
790 initrd_addr = (initrd_max-initrd_size) & ~4095;
791
792 initrd_data = g_malloc(initrd_size);
793 load_image(initrd_filename, initrd_data);
794
795 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_ADDR, initrd_addr);
796 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_SIZE, initrd_size);
797 fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, initrd_data, initrd_size);
798
799 stl_p(header+0x218, initrd_addr);
800 stl_p(header+0x21c, initrd_size);
801 }
802
803 /* load kernel and setup */
804 setup_size = header[0x1f1];
805 if (setup_size == 0)
806 setup_size = 4;
807 setup_size = (setup_size+1)*512;
808 kernel_size -= setup_size;
809
810 setup = g_malloc(setup_size);
811 kernel = g_malloc(kernel_size);
812 fseek(f, 0, SEEK_SET);
813 if (fread(setup, 1, setup_size, f) != setup_size) {
814 fprintf(stderr, "fread() failed\n");
815 exit(1);
816 }
817 if (fread(kernel, 1, kernel_size, f) != kernel_size) {
818 fprintf(stderr, "fread() failed\n");
819 exit(1);
820 }
821 fclose(f);
822 memcpy(setup, header, MIN(sizeof(header), setup_size));
823
824 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_ADDR, prot_addr);
825 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_SIZE, kernel_size);
826 fw_cfg_add_bytes(fw_cfg, FW_CFG_KERNEL_DATA, kernel, kernel_size);
827
828 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_ADDR, real_addr);
829 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_SIZE, setup_size);
830 fw_cfg_add_bytes(fw_cfg, FW_CFG_SETUP_DATA, setup, setup_size);
831
832 option_rom[nb_option_roms].name = "linuxboot.bin";
833 option_rom[nb_option_roms].bootindex = 0;
834 nb_option_roms++;
835 }
836
837 #define NE2000_NB_MAX 6
838
839 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
840 0x280, 0x380 };
841 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
842
843 static const int parallel_io[MAX_PARALLEL_PORTS] = { 0x378, 0x278, 0x3bc };
844 static const int parallel_irq[MAX_PARALLEL_PORTS] = { 7, 7, 7 };
845
846 void pc_init_ne2k_isa(NICInfo *nd)
847 {
848 static int nb_ne2k = 0;
849
850 if (nb_ne2k == NE2000_NB_MAX)
851 return;
852 isa_ne2000_init(ne2000_io[nb_ne2k],
853 ne2000_irq[nb_ne2k], nd);
854 nb_ne2k++;
855 }
856
857 int cpu_is_bsp(CPUState *env)
858 {
859 /* We hard-wire the BSP to the first CPU. */
860 return env->cpu_index == 0;
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 SysBusDevice *d;
876 static int apic_mapped;
877
878 dev = qdev_create(NULL, "apic");
879 qdev_prop_set_uint8(dev, "id", apic_id);
880 qdev_prop_set_ptr(dev, "cpu_env", env);
881 qdev_init_nofail(dev);
882 d = sysbus_from_qdev(dev);
883
884 /* XXX: mapping more APICs at the same memory location */
885 if (apic_mapped == 0) {
886 /* NOTE: the APIC is directly connected to the CPU - it is not
887 on the global memory bus. */
888 /* XXX: what if the base changes? */
889 sysbus_mmio_map(d, 0, MSI_ADDR_BASE);
890 apic_mapped = 1;
891 }
892
893 msix_supported = 1;
894
895 return dev;
896 }
897
898 /* set CMOS shutdown status register (index 0xF) as S3_resume(0xFE)
899 BIOS will read it and start S3 resume at POST Entry */
900 void pc_cmos_set_s3_resume(void *opaque, int irq, int level)
901 {
902 ISADevice *s = opaque;
903
904 if (level) {
905 rtc_set_memory(s, 0xF, 0xFE);
906 }
907 }
908
909 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
910 {
911 CPUState *s = opaque;
912
913 if (level) {
914 cpu_interrupt(s, CPU_INTERRUPT_SMI);
915 }
916 }
917
918 static void pc_cpu_reset(void *opaque)
919 {
920 CPUState *env = opaque;
921
922 cpu_reset(env);
923 env->halted = !cpu_is_bsp(env);
924 }
925
926 static CPUState *pc_new_cpu(const char *cpu_model)
927 {
928 CPUState *env;
929
930 env = cpu_init(cpu_model);
931 if (!env) {
932 fprintf(stderr, "Unable to find x86 CPU definition\n");
933 exit(1);
934 }
935 if ((env->cpuid_features & CPUID_APIC) || smp_cpus > 1) {
936 env->cpuid_apic_id = env->cpu_index;
937 env->apic_state = apic_init(env, env->cpuid_apic_id);
938 }
939 qemu_register_reset(pc_cpu_reset, env);
940 pc_cpu_reset(env);
941 return env;
942 }
943
944 void pc_cpus_init(const char *cpu_model)
945 {
946 int i;
947
948 /* init CPUs */
949 if (cpu_model == NULL) {
950 #ifdef TARGET_X86_64
951 cpu_model = "qemu64";
952 #else
953 cpu_model = "qemu32";
954 #endif
955 }
956
957 for(i = 0; i < smp_cpus; i++) {
958 pc_new_cpu(cpu_model);
959 }
960 }
961
962 void pc_memory_init(MemoryRegion *system_memory,
963 const char *kernel_filename,
964 const char *kernel_cmdline,
965 const char *initrd_filename,
966 ram_addr_t below_4g_mem_size,
967 ram_addr_t above_4g_mem_size,
968 MemoryRegion *pci_memory,
969 MemoryRegion **ram_memory)
970 {
971 char *filename;
972 int ret, linux_boot, i;
973 MemoryRegion *ram, *bios, *isa_bios, *option_rom_mr;
974 MemoryRegion *ram_below_4g, *ram_above_4g;
975 int bios_size, isa_bios_size;
976 void *fw_cfg;
977
978 linux_boot = (kernel_filename != NULL);
979
980 /* Allocate RAM. We allocate it as a single memory region and use
981 * aliases to address portions of it, mostly for backwards compatiblity
982 * with older qemus that used qemu_ram_alloc().
983 */
984 ram = g_malloc(sizeof(*ram));
985 memory_region_init_ram(ram, NULL, "pc.ram",
986 below_4g_mem_size + above_4g_mem_size);
987 *ram_memory = ram;
988 ram_below_4g = g_malloc(sizeof(*ram_below_4g));
989 memory_region_init_alias(ram_below_4g, "ram-below-4g", ram,
990 0, below_4g_mem_size);
991 memory_region_add_subregion(system_memory, 0, ram_below_4g);
992 if (above_4g_mem_size > 0) {
993 ram_above_4g = g_malloc(sizeof(*ram_above_4g));
994 memory_region_init_alias(ram_above_4g, "ram-above-4g", ram,
995 below_4g_mem_size, above_4g_mem_size);
996 memory_region_add_subregion(system_memory, 0x100000000ULL,
997 ram_above_4g);
998 }
999
1000 /* BIOS load */
1001 if (bios_name == NULL)
1002 bios_name = BIOS_FILENAME;
1003 filename = qemu_find_file(QEMU_FILE_TYPE_BIOS, bios_name);
1004 if (filename) {
1005 bios_size = get_image_size(filename);
1006 } else {
1007 bios_size = -1;
1008 }
1009 if (bios_size <= 0 ||
1010 (bios_size % 65536) != 0) {
1011 goto bios_error;
1012 }
1013 bios = g_malloc(sizeof(*bios));
1014 memory_region_init_ram(bios, NULL, "pc.bios", bios_size);
1015 memory_region_set_readonly(bios, true);
1016 ret = rom_add_file_fixed(bios_name, (uint32_t)(-bios_size), -1);
1017 if (ret != 0) {
1018 bios_error:
1019 fprintf(stderr, "qemu: could not load PC BIOS '%s'\n", bios_name);
1020 exit(1);
1021 }
1022 if (filename) {
1023 g_free(filename);
1024 }
1025 /* map the last 128KB of the BIOS in ISA space */
1026 isa_bios_size = bios_size;
1027 if (isa_bios_size > (128 * 1024))
1028 isa_bios_size = 128 * 1024;
1029 isa_bios = g_malloc(sizeof(*isa_bios));
1030 memory_region_init_alias(isa_bios, "isa-bios", bios,
1031 bios_size - isa_bios_size, isa_bios_size);
1032 memory_region_add_subregion_overlap(pci_memory,
1033 0x100000 - isa_bios_size,
1034 isa_bios,
1035 1);
1036 memory_region_set_readonly(isa_bios, true);
1037
1038 option_rom_mr = g_malloc(sizeof(*option_rom_mr));
1039 memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE);
1040 memory_region_add_subregion_overlap(pci_memory,
1041 PC_ROM_MIN_VGA,
1042 option_rom_mr,
1043 1);
1044
1045 /* map all the bios at the top of memory */
1046 memory_region_add_subregion(pci_memory,
1047 (uint32_t)(-bios_size),
1048 bios);
1049
1050 fw_cfg = bochs_bios_init();
1051 rom_set_fw(fw_cfg);
1052
1053 if (linux_boot) {
1054 load_linux(fw_cfg, kernel_filename, initrd_filename, kernel_cmdline, below_4g_mem_size);
1055 }
1056
1057 for (i = 0; i < nb_option_roms; i++) {
1058 rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1059 }
1060 }
1061
1062 qemu_irq *pc_allocate_cpu_irq(void)
1063 {
1064 return qemu_allocate_irqs(pic_irq_request, NULL, 1);
1065 }
1066
1067 void pc_vga_init(PCIBus *pci_bus)
1068 {
1069 if (cirrus_vga_enabled) {
1070 if (pci_bus) {
1071 pci_cirrus_vga_init(pci_bus);
1072 } else {
1073 isa_cirrus_vga_init(get_system_memory());
1074 }
1075 } else if (vmsvga_enabled) {
1076 if (pci_bus) {
1077 if (!pci_vmsvga_init(pci_bus)) {
1078 fprintf(stderr, "Warning: vmware_vga not available,"
1079 " using standard VGA instead\n");
1080 pci_vga_init(pci_bus);
1081 }
1082 } else {
1083 fprintf(stderr, "%s: vmware_vga: no PCI bus\n", __FUNCTION__);
1084 }
1085 #ifdef CONFIG_SPICE
1086 } else if (qxl_enabled) {
1087 if (pci_bus)
1088 pci_create_simple(pci_bus, -1, "qxl-vga");
1089 else
1090 fprintf(stderr, "%s: qxl: no PCI bus\n", __FUNCTION__);
1091 #endif
1092 } else if (std_vga_enabled) {
1093 if (pci_bus) {
1094 pci_vga_init(pci_bus);
1095 } else {
1096 isa_vga_init();
1097 }
1098 }
1099
1100 /*
1101 * sga does not suppress normal vga output. So a machine can have both a
1102 * vga card and sga manually enabled. Output will be seen on both.
1103 * For nographic case, sga is enabled at all times
1104 */
1105 if (display_type == DT_NOGRAPHIC) {
1106 isa_create_simple("sga");
1107 }
1108 }
1109
1110 static void cpu_request_exit(void *opaque, int irq, int level)
1111 {
1112 CPUState *env = cpu_single_env;
1113
1114 if (env && level) {
1115 cpu_exit(env);
1116 }
1117 }
1118
1119 void pc_basic_device_init(qemu_irq *isa_irq,
1120 ISADevice **rtc_state,
1121 bool no_vmport)
1122 {
1123 int i;
1124 DriveInfo *fd[MAX_FD];
1125 qemu_irq rtc_irq = NULL;
1126 qemu_irq *a20_line;
1127 ISADevice *i8042, *port92, *vmmouse, *pit;
1128 qemu_irq *cpu_exit_irq;
1129
1130 register_ioport_write(0x80, 1, 1, ioport80_write, NULL);
1131
1132 register_ioport_write(0xf0, 1, 1, ioportF0_write, NULL);
1133
1134 if (!no_hpet) {
1135 DeviceState *hpet = sysbus_try_create_simple("hpet", HPET_BASE, NULL);
1136
1137 if (hpet) {
1138 for (i = 0; i < 24; i++) {
1139 sysbus_connect_irq(sysbus_from_qdev(hpet), i, isa_irq[i]);
1140 }
1141 rtc_irq = qdev_get_gpio_in(hpet, 0);
1142 }
1143 }
1144 *rtc_state = rtc_init(2000, rtc_irq);
1145
1146 qemu_register_boot_set(pc_boot_set, *rtc_state);
1147
1148 pit = pit_init(0x40, 0);
1149 pcspk_init(pit);
1150
1151 for(i = 0; i < MAX_SERIAL_PORTS; i++) {
1152 if (serial_hds[i]) {
1153 serial_isa_init(i, serial_hds[i]);
1154 }
1155 }
1156
1157 for(i = 0; i < MAX_PARALLEL_PORTS; i++) {
1158 if (parallel_hds[i]) {
1159 parallel_init(i, parallel_hds[i]);
1160 }
1161 }
1162
1163 a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1164 i8042 = isa_create_simple("i8042");
1165 i8042_setup_a20_line(i8042, &a20_line[0]);
1166 if (!no_vmport) {
1167 vmport_init();
1168 vmmouse = isa_try_create("vmmouse");
1169 } else {
1170 vmmouse = NULL;
1171 }
1172 if (vmmouse) {
1173 qdev_prop_set_ptr(&vmmouse->qdev, "ps2_mouse", i8042);
1174 qdev_init_nofail(&vmmouse->qdev);
1175 }
1176 port92 = isa_create_simple("port92");
1177 port92_init(port92, &a20_line[1]);
1178
1179 cpu_exit_irq = qemu_allocate_irqs(cpu_request_exit, NULL, 1);
1180 DMA_init(0, cpu_exit_irq);
1181
1182 for(i = 0; i < MAX_FD; i++) {
1183 fd[i] = drive_get(IF_FLOPPY, 0, i);
1184 }
1185 fdctrl_init_isa(fd);
1186 }
1187
1188 void pc_pci_device_init(PCIBus *pci_bus)
1189 {
1190 int max_bus;
1191 int bus;
1192
1193 max_bus = drive_get_max_bus(IF_SCSI);
1194 for (bus = 0; bus <= max_bus; bus++) {
1195 pci_create_simple(pci_bus, -1, "lsi53c895a");
1196 }
1197 }