]> git.proxmox.com Git - qemu.git/blob - hw/pc.c
pci: update all users to look in pci/
[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 "serial.h"
27 #include "apic.h"
28 #include "fdc.h"
29 #include "ide.h"
30 #include "pci/pci.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 "pci/msi.h"
42 #include "sysbus.h"
43 #include "sysemu.h"
44 #include "kvm.h"
45 #include "kvm_i386.h"
46 #include "xen.h"
47 #include "blockdev.h"
48 #include "hw/block-common.h"
49 #include "ui/qemu-spice.h"
50 #include "memory.h"
51 #include "exec-memory.h"
52 #include "arch_init.h"
53 #include "bitmap.h"
54
55 /* debug PC/ISA interrupts */
56 //#define DEBUG_IRQ
57
58 #ifdef DEBUG_IRQ
59 #define DPRINTF(fmt, ...) \
60 do { printf("CPUIRQ: " fmt , ## __VA_ARGS__); } while (0)
61 #else
62 #define DPRINTF(fmt, ...)
63 #endif
64
65 /* Leave a chunk of memory at the top of RAM for the BIOS ACPI tables. */
66 #define ACPI_DATA_SIZE 0x10000
67 #define BIOS_CFG_IOPORT 0x510
68 #define FW_CFG_ACPI_TABLES (FW_CFG_ARCH_LOCAL + 0)
69 #define FW_CFG_SMBIOS_ENTRIES (FW_CFG_ARCH_LOCAL + 1)
70 #define FW_CFG_IRQ0_OVERRIDE (FW_CFG_ARCH_LOCAL + 2)
71 #define FW_CFG_E820_TABLE (FW_CFG_ARCH_LOCAL + 3)
72 #define FW_CFG_HPET (FW_CFG_ARCH_LOCAL + 4)
73
74 #define E820_NR_ENTRIES 16
75
76 struct e820_entry {
77 uint64_t address;
78 uint64_t length;
79 uint32_t type;
80 } QEMU_PACKED __attribute((__aligned__(4)));
81
82 struct e820_table {
83 uint32_t count;
84 struct e820_entry entry[E820_NR_ENTRIES];
85 } QEMU_PACKED __attribute((__aligned__(4)));
86
87 static struct e820_table e820_table;
88 struct hpet_fw_config hpet_cfg = {.count = UINT8_MAX};
89
90 void gsi_handler(void *opaque, int n, int level)
91 {
92 GSIState *s = opaque;
93
94 DPRINTF("pc: %s GSI %d\n", level ? "raising" : "lowering", n);
95 if (n < ISA_NUM_IRQS) {
96 qemu_set_irq(s->i8259_irq[n], level);
97 }
98 qemu_set_irq(s->ioapic_irq[n], level);
99 }
100
101 static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
102 unsigned size)
103 {
104 }
105
106 /* MSDOS compatibility mode FPU exception support */
107 static qemu_irq ferr_irq;
108
109 void pc_register_ferr_irq(qemu_irq irq)
110 {
111 ferr_irq = irq;
112 }
113
114 /* XXX: add IGNNE support */
115 void cpu_set_ferr(CPUX86State *s)
116 {
117 qemu_irq_raise(ferr_irq);
118 }
119
120 static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
121 unsigned size)
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(CPUX86State *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(CPUX86State *env)
154 {
155 int intno;
156
157 intno = apic_get_interrupt(env->apic_state);
158 if (intno >= 0) {
159 return intno;
160 }
161 /* read the irq from the PIC */
162 if (!apic_accept_pic_intr(env->apic_state)) {
163 return -1;
164 }
165
166 intno = pic_read_irq(isa_pic);
167 return intno;
168 }
169
170 static void pic_irq_request(void *opaque, int irq, int level)
171 {
172 CPUX86State *env = first_cpu;
173
174 DPRINTF("pic_irqs: %s irq %d\n", level? "raise" : "lower", irq);
175 if (env->apic_state) {
176 while (env) {
177 if (apic_accept_pic_intr(env->apic_state)) {
178 apic_deliver_pic_intr(env->apic_state, level);
179 }
180 env = env->next_cpu;
181 }
182 } else {
183 if (level)
184 cpu_interrupt(env, CPU_INTERRUPT_HARD);
185 else
186 cpu_reset_interrupt(env, CPU_INTERRUPT_HARD);
187 }
188 }
189
190 /* PC cmos mappings */
191
192 #define REG_EQUIPMENT_BYTE 0x14
193
194 static int cmos_get_fd_drive_type(FDriveType fd0)
195 {
196 int val;
197
198 switch (fd0) {
199 case FDRIVE_DRV_144:
200 /* 1.44 Mb 3"5 drive */
201 val = 4;
202 break;
203 case FDRIVE_DRV_288:
204 /* 2.88 Mb 3"5 drive */
205 val = 5;
206 break;
207 case FDRIVE_DRV_120:
208 /* 1.2 Mb 5"5 drive */
209 val = 2;
210 break;
211 case FDRIVE_DRV_NONE:
212 default:
213 val = 0;
214 break;
215 }
216 return val;
217 }
218
219 static void cmos_init_hd(ISADevice *s, int type_ofs, int info_ofs,
220 int16_t cylinders, int8_t heads, int8_t sectors)
221 {
222 rtc_set_memory(s, type_ofs, 47);
223 rtc_set_memory(s, info_ofs, cylinders);
224 rtc_set_memory(s, info_ofs + 1, cylinders >> 8);
225 rtc_set_memory(s, info_ofs + 2, heads);
226 rtc_set_memory(s, info_ofs + 3, 0xff);
227 rtc_set_memory(s, info_ofs + 4, 0xff);
228 rtc_set_memory(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
229 rtc_set_memory(s, info_ofs + 6, cylinders);
230 rtc_set_memory(s, info_ofs + 7, cylinders >> 8);
231 rtc_set_memory(s, info_ofs + 8, sectors);
232 }
233
234 /* convert boot_device letter to something recognizable by the bios */
235 static int boot_device2nibble(char boot_device)
236 {
237 switch(boot_device) {
238 case 'a':
239 case 'b':
240 return 0x01; /* floppy boot */
241 case 'c':
242 return 0x02; /* hard drive boot */
243 case 'd':
244 return 0x03; /* CD-ROM boot */
245 case 'n':
246 return 0x04; /* Network boot */
247 }
248 return 0;
249 }
250
251 static int set_boot_dev(ISADevice *s, const char *boot_device, int fd_bootchk)
252 {
253 #define PC_MAX_BOOT_DEVICES 3
254 int nbds, bds[3] = { 0, };
255 int i;
256
257 nbds = strlen(boot_device);
258 if (nbds > PC_MAX_BOOT_DEVICES) {
259 error_report("Too many boot devices for PC");
260 return(1);
261 }
262 for (i = 0; i < nbds; i++) {
263 bds[i] = boot_device2nibble(boot_device[i]);
264 if (bds[i] == 0) {
265 error_report("Invalid boot device for PC: '%c'",
266 boot_device[i]);
267 return(1);
268 }
269 }
270 rtc_set_memory(s, 0x3d, (bds[1] << 4) | bds[0]);
271 rtc_set_memory(s, 0x38, (bds[2] << 4) | (fd_bootchk ? 0x0 : 0x1));
272 return(0);
273 }
274
275 static int pc_boot_set(void *opaque, const char *boot_device)
276 {
277 return set_boot_dev(opaque, boot_device, 0);
278 }
279
280 typedef struct pc_cmos_init_late_arg {
281 ISADevice *rtc_state;
282 BusState *idebus[2];
283 } pc_cmos_init_late_arg;
284
285 static void pc_cmos_init_late(void *opaque)
286 {
287 pc_cmos_init_late_arg *arg = opaque;
288 ISADevice *s = arg->rtc_state;
289 int16_t cylinders;
290 int8_t heads, sectors;
291 int val;
292 int i, trans;
293
294 val = 0;
295 if (ide_get_geometry(arg->idebus[0], 0,
296 &cylinders, &heads, &sectors) >= 0) {
297 cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
298 val |= 0xf0;
299 }
300 if (ide_get_geometry(arg->idebus[0], 1,
301 &cylinders, &heads, &sectors) >= 0) {
302 cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
303 val |= 0x0f;
304 }
305 rtc_set_memory(s, 0x12, val);
306
307 val = 0;
308 for (i = 0; i < 4; i++) {
309 /* NOTE: ide_get_geometry() returns the physical
310 geometry. It is always such that: 1 <= sects <= 63, 1
311 <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
312 geometry can be different if a translation is done. */
313 if (ide_get_geometry(arg->idebus[i / 2], i % 2,
314 &cylinders, &heads, &sectors) >= 0) {
315 trans = ide_get_bios_chs_trans(arg->idebus[i / 2], i % 2) - 1;
316 assert((trans & ~3) == 0);
317 val |= trans << (i * 2);
318 }
319 }
320 rtc_set_memory(s, 0x39, val);
321
322 qemu_unregister_reset(pc_cmos_init_late, opaque);
323 }
324
325 void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
326 const char *boot_device,
327 ISADevice *floppy, BusState *idebus0, BusState *idebus1,
328 ISADevice *s)
329 {
330 int val, nb, i;
331 FDriveType fd_type[2] = { FDRIVE_DRV_NONE, FDRIVE_DRV_NONE };
332 static pc_cmos_init_late_arg arg;
333
334 /* various important CMOS locations needed by PC/Bochs bios */
335
336 /* memory size */
337 /* base memory (first MiB) */
338 val = MIN(ram_size / 1024, 640);
339 rtc_set_memory(s, 0x15, val);
340 rtc_set_memory(s, 0x16, val >> 8);
341 /* extended memory (next 64MiB) */
342 if (ram_size > 1024 * 1024) {
343 val = (ram_size - 1024 * 1024) / 1024;
344 } else {
345 val = 0;
346 }
347 if (val > 65535)
348 val = 65535;
349 rtc_set_memory(s, 0x17, val);
350 rtc_set_memory(s, 0x18, val >> 8);
351 rtc_set_memory(s, 0x30, val);
352 rtc_set_memory(s, 0x31, val >> 8);
353 /* memory between 16MiB and 4GiB */
354 if (ram_size > 16 * 1024 * 1024) {
355 val = (ram_size - 16 * 1024 * 1024) / 65536;
356 } else {
357 val = 0;
358 }
359 if (val > 65535)
360 val = 65535;
361 rtc_set_memory(s, 0x34, val);
362 rtc_set_memory(s, 0x35, val >> 8);
363 /* memory above 4GiB */
364 val = above_4g_mem_size / 65536;
365 rtc_set_memory(s, 0x5b, val);
366 rtc_set_memory(s, 0x5c, val >> 8);
367 rtc_set_memory(s, 0x5d, val >> 16);
368
369 /* set the number of CPU */
370 rtc_set_memory(s, 0x5f, smp_cpus - 1);
371
372 /* set boot devices, and disable floppy signature check if requested */
373 if (set_boot_dev(s, boot_device, fd_bootchk)) {
374 exit(1);
375 }
376
377 /* floppy type */
378 if (floppy) {
379 for (i = 0; i < 2; i++) {
380 fd_type[i] = isa_fdc_get_drive_type(floppy, i);
381 }
382 }
383 val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
384 cmos_get_fd_drive_type(fd_type[1]);
385 rtc_set_memory(s, 0x10, val);
386
387 val = 0;
388 nb = 0;
389 if (fd_type[0] < FDRIVE_DRV_NONE) {
390 nb++;
391 }
392 if (fd_type[1] < FDRIVE_DRV_NONE) {
393 nb++;
394 }
395 switch (nb) {
396 case 0:
397 break;
398 case 1:
399 val |= 0x01; /* 1 drive, ready for boot */
400 break;
401 case 2:
402 val |= 0x41; /* 2 drives, ready for boot */
403 break;
404 }
405 val |= 0x02; /* FPU is there */
406 val |= 0x04; /* PS/2 mouse installed */
407 rtc_set_memory(s, REG_EQUIPMENT_BYTE, val);
408
409 /* hard drives */
410 arg.rtc_state = s;
411 arg.idebus[0] = idebus0;
412 arg.idebus[1] = idebus1;
413 qemu_register_reset(pc_cmos_init_late, &arg);
414 }
415
416 /* port 92 stuff: could be split off */
417 typedef struct Port92State {
418 ISADevice dev;
419 MemoryRegion io;
420 uint8_t outport;
421 qemu_irq *a20_out;
422 } Port92State;
423
424 static void port92_write(void *opaque, hwaddr addr, uint64_t val,
425 unsigned size)
426 {
427 Port92State *s = opaque;
428
429 DPRINTF("port92: write 0x%02x\n", val);
430 s->outport = val;
431 qemu_set_irq(*s->a20_out, (val >> 1) & 1);
432 if (val & 1) {
433 qemu_system_reset_request();
434 }
435 }
436
437 static uint64_t port92_read(void *opaque, hwaddr addr,
438 unsigned size)
439 {
440 Port92State *s = opaque;
441 uint32_t ret;
442
443 ret = s->outport;
444 DPRINTF("port92: read 0x%02x\n", ret);
445 return ret;
446 }
447
448 static void port92_init(ISADevice *dev, qemu_irq *a20_out)
449 {
450 Port92State *s = DO_UPCAST(Port92State, dev, dev);
451
452 s->a20_out = a20_out;
453 }
454
455 static const VMStateDescription vmstate_port92_isa = {
456 .name = "port92",
457 .version_id = 1,
458 .minimum_version_id = 1,
459 .minimum_version_id_old = 1,
460 .fields = (VMStateField []) {
461 VMSTATE_UINT8(outport, Port92State),
462 VMSTATE_END_OF_LIST()
463 }
464 };
465
466 static void port92_reset(DeviceState *d)
467 {
468 Port92State *s = container_of(d, Port92State, dev.qdev);
469
470 s->outport &= ~1;
471 }
472
473 static const MemoryRegionOps port92_ops = {
474 .read = port92_read,
475 .write = port92_write,
476 .impl = {
477 .min_access_size = 1,
478 .max_access_size = 1,
479 },
480 .endianness = DEVICE_LITTLE_ENDIAN,
481 };
482
483 static int port92_initfn(ISADevice *dev)
484 {
485 Port92State *s = DO_UPCAST(Port92State, dev, dev);
486
487 memory_region_init_io(&s->io, &port92_ops, s, "port92", 1);
488 isa_register_ioport(dev, &s->io, 0x92);
489
490 s->outport = 0;
491 return 0;
492 }
493
494 static void port92_class_initfn(ObjectClass *klass, void *data)
495 {
496 DeviceClass *dc = DEVICE_CLASS(klass);
497 ISADeviceClass *ic = ISA_DEVICE_CLASS(klass);
498 ic->init = port92_initfn;
499 dc->no_user = 1;
500 dc->reset = port92_reset;
501 dc->vmsd = &vmstate_port92_isa;
502 }
503
504 static TypeInfo port92_info = {
505 .name = "port92",
506 .parent = TYPE_ISA_DEVICE,
507 .instance_size = sizeof(Port92State),
508 .class_init = port92_class_initfn,
509 };
510
511 static void port92_register_types(void)
512 {
513 type_register_static(&port92_info);
514 }
515
516 type_init(port92_register_types)
517
518 static void handle_a20_line_change(void *opaque, int irq, int level)
519 {
520 CPUX86State *cpu = opaque;
521
522 /* XXX: send to all CPUs ? */
523 /* XXX: add logic to handle multiple A20 line sources */
524 cpu_x86_set_a20(cpu, level);
525 }
526
527 /***********************************************************/
528 /* Bochs BIOS debug ports */
529
530 static void bochs_bios_write(void *opaque, uint32_t addr, uint32_t val)
531 {
532 static const char shutdown_str[8] = "Shutdown";
533 static int shutdown_index = 0;
534
535 switch(addr) {
536 case 0x8900:
537 /* same as Bochs power off */
538 if (val == shutdown_str[shutdown_index]) {
539 shutdown_index++;
540 if (shutdown_index == 8) {
541 shutdown_index = 0;
542 qemu_system_shutdown_request();
543 }
544 } else {
545 shutdown_index = 0;
546 }
547 break;
548
549 case 0x501:
550 case 0x502:
551 exit((val << 1) | 1);
552 }
553 }
554
555 int e820_add_entry(uint64_t address, uint64_t length, uint32_t type)
556 {
557 int index = le32_to_cpu(e820_table.count);
558 struct e820_entry *entry;
559
560 if (index >= E820_NR_ENTRIES)
561 return -EBUSY;
562 entry = &e820_table.entry[index++];
563
564 entry->address = cpu_to_le64(address);
565 entry->length = cpu_to_le64(length);
566 entry->type = cpu_to_le32(type);
567
568 e820_table.count = cpu_to_le32(index);
569 return index;
570 }
571
572 static const MemoryRegionPortio bochs_bios_portio_list[] = {
573 { 0x500, 1, 1, .write = bochs_bios_write, }, /* 0x500 */
574 { 0x501, 1, 1, .write = bochs_bios_write, }, /* 0x501 */
575 { 0x501, 2, 2, .write = bochs_bios_write, }, /* 0x501 */
576 { 0x8900, 1, 1, .write = bochs_bios_write, }, /* 0x8900 */
577 PORTIO_END_OF_LIST(),
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 PortioList *bochs_bios_port_list = g_new(PortioList, 1);
588
589 portio_list_init(bochs_bios_port_list, bochs_bios_portio_list,
590 NULL, "bochs-bios");
591 portio_list_add(bochs_bios_port_list, get_system_io(), 0x0);
592
593 fw_cfg = fw_cfg_init(BIOS_CFG_IOPORT, BIOS_CFG_IOPORT + 1, 0, 0);
594
595 fw_cfg_add_i32(fw_cfg, FW_CFG_ID, 1);
596 fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, (uint64_t)ram_size);
597 fw_cfg_add_bytes(fw_cfg, FW_CFG_ACPI_TABLES, (uint8_t *)acpi_tables,
598 acpi_tables_len);
599 fw_cfg_add_i32(fw_cfg, FW_CFG_IRQ0_OVERRIDE, kvm_allows_irq0_override());
600
601 smbios_table = smbios_get_table(&smbios_len);
602 if (smbios_table)
603 fw_cfg_add_bytes(fw_cfg, FW_CFG_SMBIOS_ENTRIES,
604 smbios_table, smbios_len);
605 fw_cfg_add_bytes(fw_cfg, FW_CFG_E820_TABLE, (uint8_t *)&e820_table,
606 sizeof(struct e820_table));
607
608 fw_cfg_add_bytes(fw_cfg, FW_CFG_HPET, (uint8_t *)&hpet_cfg,
609 sizeof(struct hpet_fw_config));
610 /* allocate memory for the NUMA channel: one (64bit) word for the number
611 * of nodes, one word for each VCPU->node and one word for each node to
612 * hold the amount of memory.
613 */
614 numa_fw_cfg = g_malloc0((1 + max_cpus + nb_numa_nodes) * 8);
615 numa_fw_cfg[0] = cpu_to_le64(nb_numa_nodes);
616 for (i = 0; i < max_cpus; i++) {
617 for (j = 0; j < nb_numa_nodes; j++) {
618 if (test_bit(i, node_cpumask[j])) {
619 numa_fw_cfg[i + 1] = cpu_to_le64(j);
620 break;
621 }
622 }
623 }
624 for (i = 0; i < nb_numa_nodes; i++) {
625 numa_fw_cfg[max_cpus + 1 + i] = cpu_to_le64(node_mem[i]);
626 }
627 fw_cfg_add_bytes(fw_cfg, FW_CFG_NUMA, (uint8_t *)numa_fw_cfg,
628 (1 + max_cpus + nb_numa_nodes) * 8);
629
630 return fw_cfg;
631 }
632
633 static long get_file_size(FILE *f)
634 {
635 long where, size;
636
637 /* XXX: on Unix systems, using fstat() probably makes more sense */
638
639 where = ftell(f);
640 fseek(f, 0, SEEK_END);
641 size = ftell(f);
642 fseek(f, where, SEEK_SET);
643
644 return size;
645 }
646
647 static void load_linux(void *fw_cfg,
648 const char *kernel_filename,
649 const char *initrd_filename,
650 const char *kernel_cmdline,
651 hwaddr max_ram_size)
652 {
653 uint16_t protocol;
654 int setup_size, kernel_size, initrd_size = 0, cmdline_size;
655 uint32_t initrd_max;
656 uint8_t header[8192], *setup, *kernel, *initrd_data;
657 hwaddr real_addr, prot_addr, cmdline_addr, initrd_addr = 0;
658 FILE *f;
659 char *vmode;
660
661 /* Align to 16 bytes as a paranoia measure */
662 cmdline_size = (strlen(kernel_cmdline)+16) & ~15;
663
664 /* load the kernel header */
665 f = fopen(kernel_filename, "rb");
666 if (!f || !(kernel_size = get_file_size(f)) ||
667 fread(header, 1, MIN(ARRAY_SIZE(header), kernel_size), f) !=
668 MIN(ARRAY_SIZE(header), kernel_size)) {
669 fprintf(stderr, "qemu: could not load kernel '%s': %s\n",
670 kernel_filename, strerror(errno));
671 exit(1);
672 }
673
674 /* kernel protocol version */
675 #if 0
676 fprintf(stderr, "header magic: %#x\n", ldl_p(header+0x202));
677 #endif
678 if (ldl_p(header+0x202) == 0x53726448)
679 protocol = lduw_p(header+0x206);
680 else {
681 /* This looks like a multiboot kernel. If it is, let's stop
682 treating it like a Linux kernel. */
683 if (load_multiboot(fw_cfg, f, kernel_filename, initrd_filename,
684 kernel_cmdline, kernel_size, header))
685 return;
686 protocol = 0;
687 }
688
689 if (protocol < 0x200 || !(header[0x211] & 0x01)) {
690 /* Low kernel */
691 real_addr = 0x90000;
692 cmdline_addr = 0x9a000 - cmdline_size;
693 prot_addr = 0x10000;
694 } else if (protocol < 0x202) {
695 /* High but ancient kernel */
696 real_addr = 0x90000;
697 cmdline_addr = 0x9a000 - cmdline_size;
698 prot_addr = 0x100000;
699 } else {
700 /* High and recent kernel */
701 real_addr = 0x10000;
702 cmdline_addr = 0x20000;
703 prot_addr = 0x100000;
704 }
705
706 #if 0
707 fprintf(stderr,
708 "qemu: real_addr = 0x" TARGET_FMT_plx "\n"
709 "qemu: cmdline_addr = 0x" TARGET_FMT_plx "\n"
710 "qemu: prot_addr = 0x" TARGET_FMT_plx "\n",
711 real_addr,
712 cmdline_addr,
713 prot_addr);
714 #endif
715
716 /* highest address for loading the initrd */
717 if (protocol >= 0x203)
718 initrd_max = ldl_p(header+0x22c);
719 else
720 initrd_max = 0x37ffffff;
721
722 if (initrd_max >= max_ram_size-ACPI_DATA_SIZE)
723 initrd_max = max_ram_size-ACPI_DATA_SIZE-1;
724
725 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_ADDR, cmdline_addr);
726 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_SIZE, strlen(kernel_cmdline)+1);
727 fw_cfg_add_bytes(fw_cfg, FW_CFG_CMDLINE_DATA,
728 (uint8_t*)strdup(kernel_cmdline),
729 strlen(kernel_cmdline)+1);
730
731 if (protocol >= 0x202) {
732 stl_p(header+0x228, cmdline_addr);
733 } else {
734 stw_p(header+0x20, 0xA33F);
735 stw_p(header+0x22, cmdline_addr-real_addr);
736 }
737
738 /* handle vga= parameter */
739 vmode = strstr(kernel_cmdline, "vga=");
740 if (vmode) {
741 unsigned int video_mode;
742 /* skip "vga=" */
743 vmode += 4;
744 if (!strncmp(vmode, "normal", 6)) {
745 video_mode = 0xffff;
746 } else if (!strncmp(vmode, "ext", 3)) {
747 video_mode = 0xfffe;
748 } else if (!strncmp(vmode, "ask", 3)) {
749 video_mode = 0xfffd;
750 } else {
751 video_mode = strtol(vmode, NULL, 0);
752 }
753 stw_p(header+0x1fa, video_mode);
754 }
755
756 /* loader type */
757 /* High nybble = B reserved for QEMU; low nybble is revision number.
758 If this code is substantially changed, you may want to consider
759 incrementing the revision. */
760 if (protocol >= 0x200)
761 header[0x210] = 0xB0;
762
763 /* heap */
764 if (protocol >= 0x201) {
765 header[0x211] |= 0x80; /* CAN_USE_HEAP */
766 stw_p(header+0x224, cmdline_addr-real_addr-0x200);
767 }
768
769 /* load initrd */
770 if (initrd_filename) {
771 if (protocol < 0x200) {
772 fprintf(stderr, "qemu: linux kernel too old to load a ram disk\n");
773 exit(1);
774 }
775
776 initrd_size = get_image_size(initrd_filename);
777 if (initrd_size < 0) {
778 fprintf(stderr, "qemu: error reading initrd %s\n",
779 initrd_filename);
780 exit(1);
781 }
782
783 initrd_addr = (initrd_max-initrd_size) & ~4095;
784
785 initrd_data = g_malloc(initrd_size);
786 load_image(initrd_filename, initrd_data);
787
788 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_ADDR, initrd_addr);
789 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_SIZE, initrd_size);
790 fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, initrd_data, initrd_size);
791
792 stl_p(header+0x218, initrd_addr);
793 stl_p(header+0x21c, initrd_size);
794 }
795
796 /* load kernel and setup */
797 setup_size = header[0x1f1];
798 if (setup_size == 0)
799 setup_size = 4;
800 setup_size = (setup_size+1)*512;
801 kernel_size -= setup_size;
802
803 setup = g_malloc(setup_size);
804 kernel = g_malloc(kernel_size);
805 fseek(f, 0, SEEK_SET);
806 if (fread(setup, 1, setup_size, f) != setup_size) {
807 fprintf(stderr, "fread() failed\n");
808 exit(1);
809 }
810 if (fread(kernel, 1, kernel_size, f) != kernel_size) {
811 fprintf(stderr, "fread() failed\n");
812 exit(1);
813 }
814 fclose(f);
815 memcpy(setup, header, MIN(sizeof(header), setup_size));
816
817 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_ADDR, prot_addr);
818 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_SIZE, kernel_size);
819 fw_cfg_add_bytes(fw_cfg, FW_CFG_KERNEL_DATA, kernel, kernel_size);
820
821 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_ADDR, real_addr);
822 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_SIZE, setup_size);
823 fw_cfg_add_bytes(fw_cfg, FW_CFG_SETUP_DATA, setup, setup_size);
824
825 option_rom[nb_option_roms].name = "linuxboot.bin";
826 option_rom[nb_option_roms].bootindex = 0;
827 nb_option_roms++;
828 }
829
830 #define NE2000_NB_MAX 6
831
832 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
833 0x280, 0x380 };
834 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
835
836 static const int parallel_io[MAX_PARALLEL_PORTS] = { 0x378, 0x278, 0x3bc };
837 static const int parallel_irq[MAX_PARALLEL_PORTS] = { 7, 7, 7 };
838
839 void pc_init_ne2k_isa(ISABus *bus, NICInfo *nd)
840 {
841 static int nb_ne2k = 0;
842
843 if (nb_ne2k == NE2000_NB_MAX)
844 return;
845 isa_ne2000_init(bus, ne2000_io[nb_ne2k],
846 ne2000_irq[nb_ne2k], nd);
847 nb_ne2k++;
848 }
849
850 DeviceState *cpu_get_current_apic(void)
851 {
852 if (cpu_single_env) {
853 return cpu_single_env->apic_state;
854 } else {
855 return NULL;
856 }
857 }
858
859 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
860 {
861 CPUX86State *s = opaque;
862
863 if (level) {
864 cpu_interrupt(s, CPU_INTERRUPT_SMI);
865 }
866 }
867
868 void pc_cpus_init(const char *cpu_model)
869 {
870 int i;
871
872 /* init CPUs */
873 if (cpu_model == NULL) {
874 #ifdef TARGET_X86_64
875 cpu_model = "qemu64";
876 #else
877 cpu_model = "qemu32";
878 #endif
879 }
880
881 for (i = 0; i < smp_cpus; i++) {
882 if (!cpu_x86_init(cpu_model)) {
883 fprintf(stderr, "Unable to find x86 CPU definition\n");
884 exit(1);
885 }
886 }
887 }
888
889 void *pc_memory_init(MemoryRegion *system_memory,
890 const char *kernel_filename,
891 const char *kernel_cmdline,
892 const char *initrd_filename,
893 ram_addr_t below_4g_mem_size,
894 ram_addr_t above_4g_mem_size,
895 MemoryRegion *rom_memory,
896 MemoryRegion **ram_memory)
897 {
898 int linux_boot, i;
899 MemoryRegion *ram, *option_rom_mr;
900 MemoryRegion *ram_below_4g, *ram_above_4g;
901 void *fw_cfg;
902
903 linux_boot = (kernel_filename != NULL);
904
905 /* Allocate RAM. We allocate it as a single memory region and use
906 * aliases to address portions of it, mostly for backwards compatibility
907 * with older qemus that used qemu_ram_alloc().
908 */
909 ram = g_malloc(sizeof(*ram));
910 memory_region_init_ram(ram, "pc.ram",
911 below_4g_mem_size + above_4g_mem_size);
912 vmstate_register_ram_global(ram);
913 *ram_memory = ram;
914 ram_below_4g = g_malloc(sizeof(*ram_below_4g));
915 memory_region_init_alias(ram_below_4g, "ram-below-4g", ram,
916 0, below_4g_mem_size);
917 memory_region_add_subregion(system_memory, 0, ram_below_4g);
918 if (above_4g_mem_size > 0) {
919 ram_above_4g = g_malloc(sizeof(*ram_above_4g));
920 memory_region_init_alias(ram_above_4g, "ram-above-4g", ram,
921 below_4g_mem_size, above_4g_mem_size);
922 memory_region_add_subregion(system_memory, 0x100000000ULL,
923 ram_above_4g);
924 }
925
926
927 /* Initialize PC system firmware */
928 pc_system_firmware_init(rom_memory);
929
930 option_rom_mr = g_malloc(sizeof(*option_rom_mr));
931 memory_region_init_ram(option_rom_mr, "pc.rom", PC_ROM_SIZE);
932 vmstate_register_ram_global(option_rom_mr);
933 memory_region_add_subregion_overlap(rom_memory,
934 PC_ROM_MIN_VGA,
935 option_rom_mr,
936 1);
937
938 fw_cfg = bochs_bios_init();
939 rom_set_fw(fw_cfg);
940
941 if (linux_boot) {
942 load_linux(fw_cfg, kernel_filename, initrd_filename, kernel_cmdline, below_4g_mem_size);
943 }
944
945 for (i = 0; i < nb_option_roms; i++) {
946 rom_add_option(option_rom[i].name, option_rom[i].bootindex);
947 }
948 return fw_cfg;
949 }
950
951 qemu_irq *pc_allocate_cpu_irq(void)
952 {
953 return qemu_allocate_irqs(pic_irq_request, NULL, 1);
954 }
955
956 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
957 {
958 DeviceState *dev = NULL;
959
960 if (pci_bus) {
961 PCIDevice *pcidev = pci_vga_init(pci_bus);
962 dev = pcidev ? &pcidev->qdev : NULL;
963 } else if (isa_bus) {
964 ISADevice *isadev = isa_vga_init(isa_bus);
965 dev = isadev ? &isadev->qdev : NULL;
966 }
967 return dev;
968 }
969
970 static void cpu_request_exit(void *opaque, int irq, int level)
971 {
972 CPUX86State *env = cpu_single_env;
973
974 if (env && level) {
975 cpu_exit(env);
976 }
977 }
978
979 static const MemoryRegionOps ioport80_io_ops = {
980 .write = ioport80_write,
981 .endianness = DEVICE_NATIVE_ENDIAN,
982 .impl = {
983 .min_access_size = 1,
984 .max_access_size = 1,
985 },
986 };
987
988 static const MemoryRegionOps ioportF0_io_ops = {
989 .write = ioportF0_write,
990 .endianness = DEVICE_NATIVE_ENDIAN,
991 .impl = {
992 .min_access_size = 1,
993 .max_access_size = 1,
994 },
995 };
996
997 void pc_basic_device_init(ISABus *isa_bus, qemu_irq *gsi,
998 ISADevice **rtc_state,
999 ISADevice **floppy,
1000 bool no_vmport)
1001 {
1002 int i;
1003 DriveInfo *fd[MAX_FD];
1004 DeviceState *hpet = NULL;
1005 int pit_isa_irq = 0;
1006 qemu_irq pit_alt_irq = NULL;
1007 qemu_irq rtc_irq = NULL;
1008 qemu_irq *a20_line;
1009 ISADevice *i8042, *port92, *vmmouse, *pit = NULL;
1010 qemu_irq *cpu_exit_irq;
1011 MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1012 MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1013
1014 memory_region_init_io(ioport80_io, &ioport80_io_ops, NULL, "ioport80", 1);
1015 memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1016
1017 memory_region_init_io(ioportF0_io, &ioportF0_io_ops, NULL, "ioportF0", 1);
1018 memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1019
1020 /*
1021 * Check if an HPET shall be created.
1022 *
1023 * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
1024 * when the HPET wants to take over. Thus we have to disable the latter.
1025 */
1026 if (!no_hpet && (!kvm_irqchip_in_kernel() || kvm_has_pit_state2())) {
1027 hpet = sysbus_try_create_simple("hpet", HPET_BASE, NULL);
1028
1029 if (hpet) {
1030 for (i = 0; i < GSI_NUM_PINS; i++) {
1031 sysbus_connect_irq(sysbus_from_qdev(hpet), i, gsi[i]);
1032 }
1033 pit_isa_irq = -1;
1034 pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1035 rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1036 }
1037 }
1038 *rtc_state = rtc_init(isa_bus, 2000, rtc_irq);
1039
1040 qemu_register_boot_set(pc_boot_set, *rtc_state);
1041
1042 if (!xen_enabled()) {
1043 if (kvm_irqchip_in_kernel()) {
1044 pit = kvm_pit_init(isa_bus, 0x40);
1045 } else {
1046 pit = pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1047 }
1048 if (hpet) {
1049 /* connect PIT to output control line of the HPET */
1050 qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(&pit->qdev, 0));
1051 }
1052 pcspk_init(isa_bus, pit);
1053 }
1054
1055 for(i = 0; i < MAX_SERIAL_PORTS; i++) {
1056 if (serial_hds[i]) {
1057 serial_isa_init(isa_bus, i, serial_hds[i]);
1058 }
1059 }
1060
1061 for(i = 0; i < MAX_PARALLEL_PORTS; i++) {
1062 if (parallel_hds[i]) {
1063 parallel_init(isa_bus, i, parallel_hds[i]);
1064 }
1065 }
1066
1067 a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1068 i8042 = isa_create_simple(isa_bus, "i8042");
1069 i8042_setup_a20_line(i8042, &a20_line[0]);
1070 if (!no_vmport) {
1071 vmport_init(isa_bus);
1072 vmmouse = isa_try_create(isa_bus, "vmmouse");
1073 } else {
1074 vmmouse = NULL;
1075 }
1076 if (vmmouse) {
1077 qdev_prop_set_ptr(&vmmouse->qdev, "ps2_mouse", i8042);
1078 qdev_init_nofail(&vmmouse->qdev);
1079 }
1080 port92 = isa_create_simple(isa_bus, "port92");
1081 port92_init(port92, &a20_line[1]);
1082
1083 cpu_exit_irq = qemu_allocate_irqs(cpu_request_exit, NULL, 1);
1084 DMA_init(0, cpu_exit_irq);
1085
1086 for(i = 0; i < MAX_FD; i++) {
1087 fd[i] = drive_get(IF_FLOPPY, 0, i);
1088 }
1089 *floppy = fdctrl_init_isa(isa_bus, fd);
1090 }
1091
1092 void pc_nic_init(ISABus *isa_bus, PCIBus *pci_bus)
1093 {
1094 int i;
1095
1096 for (i = 0; i < nb_nics; i++) {
1097 NICInfo *nd = &nd_table[i];
1098
1099 if (!pci_bus || (nd->model && strcmp(nd->model, "ne2k_isa") == 0)) {
1100 pc_init_ne2k_isa(isa_bus, nd);
1101 } else {
1102 pci_nic_init_nofail(nd, "e1000", NULL);
1103 }
1104 }
1105 }
1106
1107 void pc_pci_device_init(PCIBus *pci_bus)
1108 {
1109 int max_bus;
1110 int bus;
1111
1112 max_bus = drive_get_max_bus(IF_SCSI);
1113 for (bus = 0; bus <= max_bus; bus++) {
1114 pci_create_simple(pci_bus, -1, "lsi53c895a");
1115 }
1116 }
1117
1118 void ioapic_init_gsi(GSIState *gsi_state, const char *parent_name)
1119 {
1120 DeviceState *dev;
1121 SysBusDevice *d;
1122 unsigned int i;
1123
1124 if (kvm_irqchip_in_kernel()) {
1125 dev = qdev_create(NULL, "kvm-ioapic");
1126 } else {
1127 dev = qdev_create(NULL, "ioapic");
1128 }
1129 if (parent_name) {
1130 object_property_add_child(object_resolve_path(parent_name, NULL),
1131 "ioapic", OBJECT(dev), NULL);
1132 }
1133 qdev_init_nofail(dev);
1134 d = sysbus_from_qdev(dev);
1135 sysbus_mmio_map(d, 0, 0xfec00000);
1136
1137 for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1138 gsi_state->ioapic_irq[i] = qdev_get_gpio_in(dev, i);
1139 }
1140 }