]> git.proxmox.com Git - mirror_qemu.git/blob - hw/pc.c
pc: Generalize ISA IRQs to GSIs
[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 } QEMU_PACKED __attribute((__aligned__(4)));
82
83 struct e820_table {
84 uint32_t count;
85 struct e820_entry entry[E820_NR_ENTRIES];
86 } QEMU_PACKED __attribute((__aligned__(4)));
87
88 static struct e820_table e820_table;
89 struct hpet_fw_config hpet_cfg = {.count = UINT8_MAX};
90
91 void gsi_handler(void *opaque, int n, int level)
92 {
93 GSIState *s = opaque;
94
95 DPRINTF("pc: %s GSI %d\n", level ? "raising" : "lowering", n);
96 if (n < ISA_NUM_IRQS) {
97 qemu_set_irq(s->i8259_irq[n], level);
98 }
99 qemu_set_irq(s->ioapic_irq[n], level);
100 }
101
102 static void ioport80_write(void *opaque, uint32_t addr, uint32_t data)
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, uint32_t addr, uint32_t data)
121 {
122 qemu_irq_lower(ferr_irq);
123 }
124
125 /* TSC handling */
126 uint64_t cpu_get_tsc(CPUX86State *env)
127 {
128 return cpu_get_ticks();
129 }
130
131 /* SMM support */
132
133 static cpu_set_smm_t smm_set;
134 static void *smm_arg;
135
136 void cpu_smm_register(cpu_set_smm_t callback, void *arg)
137 {
138 assert(smm_set == NULL);
139 assert(smm_arg == NULL);
140 smm_set = callback;
141 smm_arg = arg;
142 }
143
144 void cpu_smm_update(CPUState *env)
145 {
146 if (smm_set && smm_arg && env == first_cpu)
147 smm_set(!!(env->hflags & HF_SMM_MASK), smm_arg);
148 }
149
150
151 /* IRQ handling */
152 int cpu_get_pic_interrupt(CPUState *env)
153 {
154 int intno;
155
156 intno = apic_get_interrupt(env->apic_state);
157 if (intno >= 0) {
158 /* set irq request if a PIC irq is still pending */
159 /* XXX: improve that */
160 pic_update_irq(isa_pic);
161 return intno;
162 }
163 /* read the irq from the PIC */
164 if (!apic_accept_pic_intr(env->apic_state)) {
165 return -1;
166 }
167
168 intno = pic_read_irq(isa_pic);
169 return intno;
170 }
171
172 static void pic_irq_request(void *opaque, int irq, int level)
173 {
174 CPUState *env = first_cpu;
175
176 DPRINTF("pic_irqs: %s irq %d\n", level? "raise" : "lower", irq);
177 if (env->apic_state) {
178 while (env) {
179 if (apic_accept_pic_intr(env->apic_state)) {
180 apic_deliver_pic_intr(env->apic_state, level);
181 }
182 env = env->next_cpu;
183 }
184 } else {
185 if (level)
186 cpu_interrupt(env, CPU_INTERRUPT_HARD);
187 else
188 cpu_reset_interrupt(env, CPU_INTERRUPT_HARD);
189 }
190 }
191
192 /* PC cmos mappings */
193
194 #define REG_EQUIPMENT_BYTE 0x14
195
196 static int cmos_get_fd_drive_type(FDriveType fd0)
197 {
198 int val;
199
200 switch (fd0) {
201 case FDRIVE_DRV_144:
202 /* 1.44 Mb 3"5 drive */
203 val = 4;
204 break;
205 case FDRIVE_DRV_288:
206 /* 2.88 Mb 3"5 drive */
207 val = 5;
208 break;
209 case FDRIVE_DRV_120:
210 /* 1.2 Mb 5"5 drive */
211 val = 2;
212 break;
213 case FDRIVE_DRV_NONE:
214 default:
215 val = 0;
216 break;
217 }
218 return val;
219 }
220
221 static void cmos_init_hd(int type_ofs, int info_ofs, BlockDriverState *hd,
222 ISADevice *s)
223 {
224 int cylinders, heads, sectors;
225 bdrv_get_geometry_hint(hd, &cylinders, &heads, &sectors);
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 *idebus0, *idebus1;
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 int val;
294 BlockDriverState *hd_table[4];
295 int i;
296
297 ide_get_bs(hd_table, arg->idebus0);
298 ide_get_bs(hd_table + 2, arg->idebus1);
299
300 rtc_set_memory(s, 0x12, (hd_table[0] ? 0xf0 : 0) | (hd_table[1] ? 0x0f : 0));
301 if (hd_table[0])
302 cmos_init_hd(0x19, 0x1b, hd_table[0], s);
303 if (hd_table[1])
304 cmos_init_hd(0x1a, 0x24, hd_table[1], s);
305
306 val = 0;
307 for (i = 0; i < 4; i++) {
308 if (hd_table[i]) {
309 int cylinders, heads, sectors, translation;
310 /* NOTE: bdrv_get_geometry_hint() returns the physical
311 geometry. It is always such that: 1 <= sects <= 63, 1
312 <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
313 geometry can be different if a translation is done. */
314 translation = bdrv_get_translation_hint(hd_table[i]);
315 if (translation == BIOS_ATA_TRANSLATION_AUTO) {
316 bdrv_get_geometry_hint(hd_table[i], &cylinders, &heads, &sectors);
317 if (cylinders <= 1024 && heads <= 16 && sectors <= 63) {
318 /* No translation. */
319 translation = 0;
320 } else {
321 /* LBA translation. */
322 translation = 1;
323 }
324 } else {
325 translation--;
326 }
327 val |= translation << (i * 2);
328 }
329 }
330 rtc_set_memory(s, 0x39, val);
331
332 qemu_unregister_reset(pc_cmos_init_late, opaque);
333 }
334
335 void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
336 const char *boot_device,
337 BusState *idebus0, BusState *idebus1,
338 ISADevice *s)
339 {
340 int val, nb, nb_heads, max_track, last_sect, i;
341 FDriveType fd_type[2];
342 DriveInfo *fd[2];
343 static pc_cmos_init_late_arg arg;
344
345 /* various important CMOS locations needed by PC/Bochs bios */
346
347 /* memory size */
348 val = 640; /* base memory in K */
349 rtc_set_memory(s, 0x15, val);
350 rtc_set_memory(s, 0x16, val >> 8);
351
352 val = (ram_size / 1024) - 1024;
353 if (val > 65535)
354 val = 65535;
355 rtc_set_memory(s, 0x17, val);
356 rtc_set_memory(s, 0x18, val >> 8);
357 rtc_set_memory(s, 0x30, val);
358 rtc_set_memory(s, 0x31, val >> 8);
359
360 if (above_4g_mem_size) {
361 rtc_set_memory(s, 0x5b, (unsigned int)above_4g_mem_size >> 16);
362 rtc_set_memory(s, 0x5c, (unsigned int)above_4g_mem_size >> 24);
363 rtc_set_memory(s, 0x5d, (uint64_t)above_4g_mem_size >> 32);
364 }
365
366 if (ram_size > (16 * 1024 * 1024))
367 val = (ram_size / 65536) - ((16 * 1024 * 1024) / 65536);
368 else
369 val = 0;
370 if (val > 65535)
371 val = 65535;
372 rtc_set_memory(s, 0x34, val);
373 rtc_set_memory(s, 0x35, val >> 8);
374
375 /* set the number of CPU */
376 rtc_set_memory(s, 0x5f, smp_cpus - 1);
377
378 /* set boot devices, and disable floppy signature check if requested */
379 if (set_boot_dev(s, boot_device, fd_bootchk)) {
380 exit(1);
381 }
382
383 /* floppy type */
384 for (i = 0; i < 2; i++) {
385 fd[i] = drive_get(IF_FLOPPY, 0, i);
386 if (fd[i] && bdrv_is_inserted(fd[i]->bdrv)) {
387 bdrv_get_floppy_geometry_hint(fd[i]->bdrv, &nb_heads, &max_track,
388 &last_sect, FDRIVE_DRV_NONE,
389 &fd_type[i]);
390 } else {
391 fd_type[i] = FDRIVE_DRV_NONE;
392 }
393 }
394 val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
395 cmos_get_fd_drive_type(fd_type[1]);
396 rtc_set_memory(s, 0x10, val);
397
398 val = 0;
399 nb = 0;
400 if (fd_type[0] < FDRIVE_DRV_NONE) {
401 nb++;
402 }
403 if (fd_type[1] < FDRIVE_DRV_NONE) {
404 nb++;
405 }
406 switch (nb) {
407 case 0:
408 break;
409 case 1:
410 val |= 0x01; /* 1 drive, ready for boot */
411 break;
412 case 2:
413 val |= 0x41; /* 2 drives, ready for boot */
414 break;
415 }
416 val |= 0x02; /* FPU is there */
417 val |= 0x04; /* PS/2 mouse installed */
418 rtc_set_memory(s, REG_EQUIPMENT_BYTE, val);
419
420 /* hard drives */
421 arg.rtc_state = s;
422 arg.idebus0 = idebus0;
423 arg.idebus1 = idebus1;
424 qemu_register_reset(pc_cmos_init_late, &arg);
425 }
426
427 /* port 92 stuff: could be split off */
428 typedef struct Port92State {
429 ISADevice dev;
430 MemoryRegion io;
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 const MemoryRegionPortio port92_portio[] = {
483 { 0, 1, 1, .read = port92_read, .write = port92_write },
484 PORTIO_END_OF_LIST(),
485 };
486
487 static const MemoryRegionOps port92_ops = {
488 .old_portio = port92_portio
489 };
490
491 static int port92_initfn(ISADevice *dev)
492 {
493 Port92State *s = DO_UPCAST(Port92State, dev, dev);
494
495 memory_region_init_io(&s->io, &port92_ops, s, "port92", 1);
496 isa_register_ioport(dev, &s->io, 0x92);
497
498 s->outport = 0;
499 return 0;
500 }
501
502 static ISADeviceInfo port92_info = {
503 .qdev.name = "port92",
504 .qdev.size = sizeof(Port92State),
505 .qdev.vmsd = &vmstate_port92_isa,
506 .qdev.no_user = 1,
507 .qdev.reset = port92_reset,
508 .init = port92_initfn,
509 };
510
511 static void port92_register(void)
512 {
513 isa_qdev_register(&port92_info);
514 }
515 device_init(port92_register)
516
517 static void handle_a20_line_change(void *opaque, int irq, int level)
518 {
519 CPUState *cpu = opaque;
520
521 /* XXX: send to all CPUs ? */
522 /* XXX: add logic to handle multiple A20 line sources */
523 cpu_x86_set_a20(cpu, level);
524 }
525
526 /***********************************************************/
527 /* Bochs BIOS debug ports */
528
529 static void bochs_bios_write(void *opaque, uint32_t addr, uint32_t val)
530 {
531 static const char shutdown_str[8] = "Shutdown";
532 static int shutdown_index = 0;
533
534 switch(addr) {
535 /* Bochs BIOS messages */
536 case 0x400:
537 case 0x401:
538 /* used to be panic, now unused */
539 break;
540 case 0x402:
541 case 0x403:
542 #ifdef DEBUG_BIOS
543 fprintf(stderr, "%c", val);
544 #endif
545 break;
546 case 0x8900:
547 /* same as Bochs power off */
548 if (val == shutdown_str[shutdown_index]) {
549 shutdown_index++;
550 if (shutdown_index == 8) {
551 shutdown_index = 0;
552 qemu_system_shutdown_request();
553 }
554 } else {
555 shutdown_index = 0;
556 }
557 break;
558
559 /* LGPL'ed VGA BIOS messages */
560 case 0x501:
561 case 0x502:
562 exit((val << 1) | 1);
563 case 0x500:
564 case 0x503:
565 #ifdef DEBUG_BIOS
566 fprintf(stderr, "%c", val);
567 #endif
568 break;
569 }
570 }
571
572 int e820_add_entry(uint64_t address, uint64_t length, uint32_t type)
573 {
574 int index = le32_to_cpu(e820_table.count);
575 struct e820_entry *entry;
576
577 if (index >= E820_NR_ENTRIES)
578 return -EBUSY;
579 entry = &e820_table.entry[index++];
580
581 entry->address = cpu_to_le64(address);
582 entry->length = cpu_to_le64(length);
583 entry->type = cpu_to_le32(type);
584
585 e820_table.count = cpu_to_le32(index);
586 return index;
587 }
588
589 static void *bochs_bios_init(void)
590 {
591 void *fw_cfg;
592 uint8_t *smbios_table;
593 size_t smbios_len;
594 uint64_t *numa_fw_cfg;
595 int i, j;
596
597 register_ioport_write(0x400, 1, 2, bochs_bios_write, NULL);
598 register_ioport_write(0x401, 1, 2, bochs_bios_write, NULL);
599 register_ioport_write(0x402, 1, 1, bochs_bios_write, NULL);
600 register_ioport_write(0x403, 1, 1, bochs_bios_write, NULL);
601 register_ioport_write(0x8900, 1, 1, bochs_bios_write, NULL);
602
603 register_ioport_write(0x501, 1, 1, bochs_bios_write, NULL);
604 register_ioport_write(0x501, 1, 2, bochs_bios_write, NULL);
605 register_ioport_write(0x502, 1, 2, bochs_bios_write, NULL);
606 register_ioport_write(0x500, 1, 1, bochs_bios_write, NULL);
607 register_ioport_write(0x503, 1, 1, bochs_bios_write, NULL);
608
609 fw_cfg = fw_cfg_init(BIOS_CFG_IOPORT, BIOS_CFG_IOPORT + 1, 0, 0);
610
611 fw_cfg_add_i32(fw_cfg, FW_CFG_ID, 1);
612 fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, (uint64_t)ram_size);
613 fw_cfg_add_bytes(fw_cfg, FW_CFG_ACPI_TABLES, (uint8_t *)acpi_tables,
614 acpi_tables_len);
615 fw_cfg_add_bytes(fw_cfg, FW_CFG_IRQ0_OVERRIDE, &irq0override, 1);
616
617 smbios_table = smbios_get_table(&smbios_len);
618 if (smbios_table)
619 fw_cfg_add_bytes(fw_cfg, FW_CFG_SMBIOS_ENTRIES,
620 smbios_table, smbios_len);
621 fw_cfg_add_bytes(fw_cfg, FW_CFG_E820_TABLE, (uint8_t *)&e820_table,
622 sizeof(struct e820_table));
623
624 fw_cfg_add_bytes(fw_cfg, FW_CFG_HPET, (uint8_t *)&hpet_cfg,
625 sizeof(struct hpet_fw_config));
626 /* allocate memory for the NUMA channel: one (64bit) word for the number
627 * of nodes, one word for each VCPU->node and one word for each node to
628 * hold the amount of memory.
629 */
630 numa_fw_cfg = g_malloc0((1 + smp_cpus + nb_numa_nodes) * 8);
631 numa_fw_cfg[0] = cpu_to_le64(nb_numa_nodes);
632 for (i = 0; i < smp_cpus; i++) {
633 for (j = 0; j < nb_numa_nodes; j++) {
634 if (node_cpumask[j] & (1 << i)) {
635 numa_fw_cfg[i + 1] = cpu_to_le64(j);
636 break;
637 }
638 }
639 }
640 for (i = 0; i < nb_numa_nodes; i++) {
641 numa_fw_cfg[smp_cpus + 1 + i] = cpu_to_le64(node_mem[i]);
642 }
643 fw_cfg_add_bytes(fw_cfg, FW_CFG_NUMA, (uint8_t *)numa_fw_cfg,
644 (1 + smp_cpus + nb_numa_nodes) * 8);
645
646 return fw_cfg;
647 }
648
649 static long get_file_size(FILE *f)
650 {
651 long where, size;
652
653 /* XXX: on Unix systems, using fstat() probably makes more sense */
654
655 where = ftell(f);
656 fseek(f, 0, SEEK_END);
657 size = ftell(f);
658 fseek(f, where, SEEK_SET);
659
660 return size;
661 }
662
663 static void load_linux(void *fw_cfg,
664 const char *kernel_filename,
665 const char *initrd_filename,
666 const char *kernel_cmdline,
667 target_phys_addr_t max_ram_size)
668 {
669 uint16_t protocol;
670 int setup_size, kernel_size, initrd_size = 0, cmdline_size;
671 uint32_t initrd_max;
672 uint8_t header[8192], *setup, *kernel, *initrd_data;
673 target_phys_addr_t real_addr, prot_addr, cmdline_addr, initrd_addr = 0;
674 FILE *f;
675 char *vmode;
676
677 /* Align to 16 bytes as a paranoia measure */
678 cmdline_size = (strlen(kernel_cmdline)+16) & ~15;
679
680 /* load the kernel header */
681 f = fopen(kernel_filename, "rb");
682 if (!f || !(kernel_size = get_file_size(f)) ||
683 fread(header, 1, MIN(ARRAY_SIZE(header), kernel_size), f) !=
684 MIN(ARRAY_SIZE(header), kernel_size)) {
685 fprintf(stderr, "qemu: could not load kernel '%s': %s\n",
686 kernel_filename, strerror(errno));
687 exit(1);
688 }
689
690 /* kernel protocol version */
691 #if 0
692 fprintf(stderr, "header magic: %#x\n", ldl_p(header+0x202));
693 #endif
694 if (ldl_p(header+0x202) == 0x53726448)
695 protocol = lduw_p(header+0x206);
696 else {
697 /* This looks like a multiboot kernel. If it is, let's stop
698 treating it like a Linux kernel. */
699 if (load_multiboot(fw_cfg, f, kernel_filename, initrd_filename,
700 kernel_cmdline, kernel_size, header))
701 return;
702 protocol = 0;
703 }
704
705 if (protocol < 0x200 || !(header[0x211] & 0x01)) {
706 /* Low kernel */
707 real_addr = 0x90000;
708 cmdline_addr = 0x9a000 - cmdline_size;
709 prot_addr = 0x10000;
710 } else if (protocol < 0x202) {
711 /* High but ancient kernel */
712 real_addr = 0x90000;
713 cmdline_addr = 0x9a000 - cmdline_size;
714 prot_addr = 0x100000;
715 } else {
716 /* High and recent kernel */
717 real_addr = 0x10000;
718 cmdline_addr = 0x20000;
719 prot_addr = 0x100000;
720 }
721
722 #if 0
723 fprintf(stderr,
724 "qemu: real_addr = 0x" TARGET_FMT_plx "\n"
725 "qemu: cmdline_addr = 0x" TARGET_FMT_plx "\n"
726 "qemu: prot_addr = 0x" TARGET_FMT_plx "\n",
727 real_addr,
728 cmdline_addr,
729 prot_addr);
730 #endif
731
732 /* highest address for loading the initrd */
733 if (protocol >= 0x203)
734 initrd_max = ldl_p(header+0x22c);
735 else
736 initrd_max = 0x37ffffff;
737
738 if (initrd_max >= max_ram_size-ACPI_DATA_SIZE)
739 initrd_max = max_ram_size-ACPI_DATA_SIZE-1;
740
741 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_ADDR, cmdline_addr);
742 fw_cfg_add_i32(fw_cfg, FW_CFG_CMDLINE_SIZE, strlen(kernel_cmdline)+1);
743 fw_cfg_add_bytes(fw_cfg, FW_CFG_CMDLINE_DATA,
744 (uint8_t*)strdup(kernel_cmdline),
745 strlen(kernel_cmdline)+1);
746
747 if (protocol >= 0x202) {
748 stl_p(header+0x228, cmdline_addr);
749 } else {
750 stw_p(header+0x20, 0xA33F);
751 stw_p(header+0x22, cmdline_addr-real_addr);
752 }
753
754 /* handle vga= parameter */
755 vmode = strstr(kernel_cmdline, "vga=");
756 if (vmode) {
757 unsigned int video_mode;
758 /* skip "vga=" */
759 vmode += 4;
760 if (!strncmp(vmode, "normal", 6)) {
761 video_mode = 0xffff;
762 } else if (!strncmp(vmode, "ext", 3)) {
763 video_mode = 0xfffe;
764 } else if (!strncmp(vmode, "ask", 3)) {
765 video_mode = 0xfffd;
766 } else {
767 video_mode = strtol(vmode, NULL, 0);
768 }
769 stw_p(header+0x1fa, video_mode);
770 }
771
772 /* loader type */
773 /* High nybble = B reserved for Qemu; low nybble is revision number.
774 If this code is substantially changed, you may want to consider
775 incrementing the revision. */
776 if (protocol >= 0x200)
777 header[0x210] = 0xB0;
778
779 /* heap */
780 if (protocol >= 0x201) {
781 header[0x211] |= 0x80; /* CAN_USE_HEAP */
782 stw_p(header+0x224, cmdline_addr-real_addr-0x200);
783 }
784
785 /* load initrd */
786 if (initrd_filename) {
787 if (protocol < 0x200) {
788 fprintf(stderr, "qemu: linux kernel too old to load a ram disk\n");
789 exit(1);
790 }
791
792 initrd_size = get_image_size(initrd_filename);
793 if (initrd_size < 0) {
794 fprintf(stderr, "qemu: error reading initrd %s\n",
795 initrd_filename);
796 exit(1);
797 }
798
799 initrd_addr = (initrd_max-initrd_size) & ~4095;
800
801 initrd_data = g_malloc(initrd_size);
802 load_image(initrd_filename, initrd_data);
803
804 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_ADDR, initrd_addr);
805 fw_cfg_add_i32(fw_cfg, FW_CFG_INITRD_SIZE, initrd_size);
806 fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, initrd_data, initrd_size);
807
808 stl_p(header+0x218, initrd_addr);
809 stl_p(header+0x21c, initrd_size);
810 }
811
812 /* load kernel and setup */
813 setup_size = header[0x1f1];
814 if (setup_size == 0)
815 setup_size = 4;
816 setup_size = (setup_size+1)*512;
817 kernel_size -= setup_size;
818
819 setup = g_malloc(setup_size);
820 kernel = g_malloc(kernel_size);
821 fseek(f, 0, SEEK_SET);
822 if (fread(setup, 1, setup_size, f) != setup_size) {
823 fprintf(stderr, "fread() failed\n");
824 exit(1);
825 }
826 if (fread(kernel, 1, kernel_size, f) != kernel_size) {
827 fprintf(stderr, "fread() failed\n");
828 exit(1);
829 }
830 fclose(f);
831 memcpy(setup, header, MIN(sizeof(header), setup_size));
832
833 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_ADDR, prot_addr);
834 fw_cfg_add_i32(fw_cfg, FW_CFG_KERNEL_SIZE, kernel_size);
835 fw_cfg_add_bytes(fw_cfg, FW_CFG_KERNEL_DATA, kernel, kernel_size);
836
837 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_ADDR, real_addr);
838 fw_cfg_add_i32(fw_cfg, FW_CFG_SETUP_SIZE, setup_size);
839 fw_cfg_add_bytes(fw_cfg, FW_CFG_SETUP_DATA, setup, setup_size);
840
841 option_rom[nb_option_roms].name = "linuxboot.bin";
842 option_rom[nb_option_roms].bootindex = 0;
843 nb_option_roms++;
844 }
845
846 #define NE2000_NB_MAX 6
847
848 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
849 0x280, 0x380 };
850 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
851
852 static const int parallel_io[MAX_PARALLEL_PORTS] = { 0x378, 0x278, 0x3bc };
853 static const int parallel_irq[MAX_PARALLEL_PORTS] = { 7, 7, 7 };
854
855 void pc_init_ne2k_isa(NICInfo *nd)
856 {
857 static int nb_ne2k = 0;
858
859 if (nb_ne2k == NE2000_NB_MAX)
860 return;
861 isa_ne2000_init(ne2000_io[nb_ne2k],
862 ne2000_irq[nb_ne2k], nd);
863 nb_ne2k++;
864 }
865
866 int cpu_is_bsp(CPUState *env)
867 {
868 /* We hard-wire the BSP to the first CPU. */
869 return env->cpu_index == 0;
870 }
871
872 DeviceState *cpu_get_current_apic(void)
873 {
874 if (cpu_single_env) {
875 return cpu_single_env->apic_state;
876 } else {
877 return NULL;
878 }
879 }
880
881 static DeviceState *apic_init(void *env, uint8_t apic_id)
882 {
883 DeviceState *dev;
884 SysBusDevice *d;
885 static int apic_mapped;
886
887 dev = qdev_create(NULL, "apic");
888 qdev_prop_set_uint8(dev, "id", apic_id);
889 qdev_prop_set_ptr(dev, "cpu_env", env);
890 qdev_init_nofail(dev);
891 d = sysbus_from_qdev(dev);
892
893 /* XXX: mapping more APICs at the same memory location */
894 if (apic_mapped == 0) {
895 /* NOTE: the APIC is directly connected to the CPU - it is not
896 on the global memory bus. */
897 /* XXX: what if the base changes? */
898 sysbus_mmio_map(d, 0, MSI_ADDR_BASE);
899 apic_mapped = 1;
900 }
901
902 msix_supported = 1;
903
904 return dev;
905 }
906
907 /* set CMOS shutdown status register (index 0xF) as S3_resume(0xFE)
908 BIOS will read it and start S3 resume at POST Entry */
909 void pc_cmos_set_s3_resume(void *opaque, int irq, int level)
910 {
911 ISADevice *s = opaque;
912
913 if (level) {
914 rtc_set_memory(s, 0xF, 0xFE);
915 }
916 }
917
918 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
919 {
920 CPUState *s = opaque;
921
922 if (level) {
923 cpu_interrupt(s, CPU_INTERRUPT_SMI);
924 }
925 }
926
927 static void pc_cpu_reset(void *opaque)
928 {
929 CPUState *env = opaque;
930
931 cpu_reset(env);
932 env->halted = !cpu_is_bsp(env);
933 }
934
935 static CPUState *pc_new_cpu(const char *cpu_model)
936 {
937 CPUState *env;
938
939 env = cpu_init(cpu_model);
940 if (!env) {
941 fprintf(stderr, "Unable to find x86 CPU definition\n");
942 exit(1);
943 }
944 if ((env->cpuid_features & CPUID_APIC) || smp_cpus > 1) {
945 env->cpuid_apic_id = env->cpu_index;
946 env->apic_state = apic_init(env, env->cpuid_apic_id);
947 }
948 qemu_register_reset(pc_cpu_reset, env);
949 pc_cpu_reset(env);
950 return env;
951 }
952
953 void pc_cpus_init(const char *cpu_model)
954 {
955 int i;
956
957 /* init CPUs */
958 if (cpu_model == NULL) {
959 #ifdef TARGET_X86_64
960 cpu_model = "qemu64";
961 #else
962 cpu_model = "qemu32";
963 #endif
964 }
965
966 for(i = 0; i < smp_cpus; i++) {
967 pc_new_cpu(cpu_model);
968 }
969 }
970
971 void pc_memory_init(MemoryRegion *system_memory,
972 const char *kernel_filename,
973 const char *kernel_cmdline,
974 const char *initrd_filename,
975 ram_addr_t below_4g_mem_size,
976 ram_addr_t above_4g_mem_size,
977 MemoryRegion *rom_memory,
978 MemoryRegion **ram_memory)
979 {
980 char *filename;
981 int ret, linux_boot, i;
982 MemoryRegion *ram, *bios, *isa_bios, *option_rom_mr;
983 MemoryRegion *ram_below_4g, *ram_above_4g;
984 int bios_size, isa_bios_size;
985 void *fw_cfg;
986
987 linux_boot = (kernel_filename != NULL);
988
989 /* Allocate RAM. We allocate it as a single memory region and use
990 * aliases to address portions of it, mostly for backwards compatiblity
991 * with older qemus that used qemu_ram_alloc().
992 */
993 ram = g_malloc(sizeof(*ram));
994 memory_region_init_ram(ram, NULL, "pc.ram",
995 below_4g_mem_size + above_4g_mem_size);
996 *ram_memory = ram;
997 ram_below_4g = g_malloc(sizeof(*ram_below_4g));
998 memory_region_init_alias(ram_below_4g, "ram-below-4g", ram,
999 0, below_4g_mem_size);
1000 memory_region_add_subregion(system_memory, 0, ram_below_4g);
1001 if (above_4g_mem_size > 0) {
1002 ram_above_4g = g_malloc(sizeof(*ram_above_4g));
1003 memory_region_init_alias(ram_above_4g, "ram-above-4g", ram,
1004 below_4g_mem_size, above_4g_mem_size);
1005 memory_region_add_subregion(system_memory, 0x100000000ULL,
1006 ram_above_4g);
1007 }
1008
1009 /* BIOS load */
1010 if (bios_name == NULL)
1011 bios_name = BIOS_FILENAME;
1012 filename = qemu_find_file(QEMU_FILE_TYPE_BIOS, bios_name);
1013 if (filename) {
1014 bios_size = get_image_size(filename);
1015 } else {
1016 bios_size = -1;
1017 }
1018 if (bios_size <= 0 ||
1019 (bios_size % 65536) != 0) {
1020 goto bios_error;
1021 }
1022 bios = g_malloc(sizeof(*bios));
1023 memory_region_init_ram(bios, NULL, "pc.bios", bios_size);
1024 memory_region_set_readonly(bios, true);
1025 ret = rom_add_file_fixed(bios_name, (uint32_t)(-bios_size), -1);
1026 if (ret != 0) {
1027 bios_error:
1028 fprintf(stderr, "qemu: could not load PC BIOS '%s'\n", bios_name);
1029 exit(1);
1030 }
1031 if (filename) {
1032 g_free(filename);
1033 }
1034 /* map the last 128KB of the BIOS in ISA space */
1035 isa_bios_size = bios_size;
1036 if (isa_bios_size > (128 * 1024))
1037 isa_bios_size = 128 * 1024;
1038 isa_bios = g_malloc(sizeof(*isa_bios));
1039 memory_region_init_alias(isa_bios, "isa-bios", bios,
1040 bios_size - isa_bios_size, isa_bios_size);
1041 memory_region_add_subregion_overlap(rom_memory,
1042 0x100000 - isa_bios_size,
1043 isa_bios,
1044 1);
1045 memory_region_set_readonly(isa_bios, true);
1046
1047 option_rom_mr = g_malloc(sizeof(*option_rom_mr));
1048 memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE);
1049 memory_region_add_subregion_overlap(rom_memory,
1050 PC_ROM_MIN_VGA,
1051 option_rom_mr,
1052 1);
1053
1054 /* map all the bios at the top of memory */
1055 memory_region_add_subregion(rom_memory,
1056 (uint32_t)(-bios_size),
1057 bios);
1058
1059 fw_cfg = bochs_bios_init();
1060 rom_set_fw(fw_cfg);
1061
1062 if (linux_boot) {
1063 load_linux(fw_cfg, kernel_filename, initrd_filename, kernel_cmdline, below_4g_mem_size);
1064 }
1065
1066 for (i = 0; i < nb_option_roms; i++) {
1067 rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1068 }
1069 }
1070
1071 qemu_irq *pc_allocate_cpu_irq(void)
1072 {
1073 return qemu_allocate_irqs(pic_irq_request, NULL, 1);
1074 }
1075
1076 void pc_vga_init(PCIBus *pci_bus)
1077 {
1078 if (cirrus_vga_enabled) {
1079 if (pci_bus) {
1080 pci_cirrus_vga_init(pci_bus);
1081 } else {
1082 isa_cirrus_vga_init(get_system_memory());
1083 }
1084 } else if (vmsvga_enabled) {
1085 if (pci_bus) {
1086 if (!pci_vmsvga_init(pci_bus)) {
1087 fprintf(stderr, "Warning: vmware_vga not available,"
1088 " using standard VGA instead\n");
1089 pci_vga_init(pci_bus);
1090 }
1091 } else {
1092 fprintf(stderr, "%s: vmware_vga: no PCI bus\n", __FUNCTION__);
1093 }
1094 #ifdef CONFIG_SPICE
1095 } else if (qxl_enabled) {
1096 if (pci_bus)
1097 pci_create_simple(pci_bus, -1, "qxl-vga");
1098 else
1099 fprintf(stderr, "%s: qxl: no PCI bus\n", __FUNCTION__);
1100 #endif
1101 } else if (std_vga_enabled) {
1102 if (pci_bus) {
1103 pci_vga_init(pci_bus);
1104 } else {
1105 isa_vga_init();
1106 }
1107 }
1108
1109 /*
1110 * sga does not suppress normal vga output. So a machine can have both a
1111 * vga card and sga manually enabled. Output will be seen on both.
1112 * For nographic case, sga is enabled at all times
1113 */
1114 if (display_type == DT_NOGRAPHIC) {
1115 isa_create_simple("sga");
1116 }
1117 }
1118
1119 static void cpu_request_exit(void *opaque, int irq, int level)
1120 {
1121 CPUState *env = cpu_single_env;
1122
1123 if (env && level) {
1124 cpu_exit(env);
1125 }
1126 }
1127
1128 void pc_basic_device_init(qemu_irq *gsi,
1129 ISADevice **rtc_state,
1130 bool no_vmport)
1131 {
1132 int i;
1133 DriveInfo *fd[MAX_FD];
1134 qemu_irq rtc_irq = NULL;
1135 qemu_irq *a20_line;
1136 ISADevice *i8042, *port92, *vmmouse, *pit;
1137 qemu_irq *cpu_exit_irq;
1138
1139 register_ioport_write(0x80, 1, 1, ioport80_write, NULL);
1140
1141 register_ioport_write(0xf0, 1, 1, ioportF0_write, NULL);
1142
1143 if (!no_hpet) {
1144 DeviceState *hpet = sysbus_try_create_simple("hpet", HPET_BASE, NULL);
1145
1146 if (hpet) {
1147 for (i = 0; i < GSI_NUM_PINS; i++) {
1148 sysbus_connect_irq(sysbus_from_qdev(hpet), i, gsi[i]);
1149 }
1150 rtc_irq = qdev_get_gpio_in(hpet, 0);
1151 }
1152 }
1153 *rtc_state = rtc_init(2000, rtc_irq);
1154
1155 qemu_register_boot_set(pc_boot_set, *rtc_state);
1156
1157 pit = pit_init(0x40, 0);
1158 pcspk_init(pit);
1159
1160 for(i = 0; i < MAX_SERIAL_PORTS; i++) {
1161 if (serial_hds[i]) {
1162 serial_isa_init(i, serial_hds[i]);
1163 }
1164 }
1165
1166 for(i = 0; i < MAX_PARALLEL_PORTS; i++) {
1167 if (parallel_hds[i]) {
1168 parallel_init(i, parallel_hds[i]);
1169 }
1170 }
1171
1172 a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1173 i8042 = isa_create_simple("i8042");
1174 i8042_setup_a20_line(i8042, &a20_line[0]);
1175 if (!no_vmport) {
1176 vmport_init();
1177 vmmouse = isa_try_create("vmmouse");
1178 } else {
1179 vmmouse = NULL;
1180 }
1181 if (vmmouse) {
1182 qdev_prop_set_ptr(&vmmouse->qdev, "ps2_mouse", i8042);
1183 qdev_init_nofail(&vmmouse->qdev);
1184 }
1185 port92 = isa_create_simple("port92");
1186 port92_init(port92, &a20_line[1]);
1187
1188 cpu_exit_irq = qemu_allocate_irqs(cpu_request_exit, NULL, 1);
1189 DMA_init(0, cpu_exit_irq);
1190
1191 for(i = 0; i < MAX_FD; i++) {
1192 fd[i] = drive_get(IF_FLOPPY, 0, i);
1193 }
1194 fdctrl_init_isa(fd);
1195 }
1196
1197 void pc_pci_device_init(PCIBus *pci_bus)
1198 {
1199 int max_bus;
1200 int bus;
1201
1202 max_bus = drive_get_max_bus(IF_SCSI);
1203 for (bus = 0; bus <= max_bus; bus++) {
1204 pci_create_simple(pci_bus, -1, "lsi53c895a");
1205 }
1206 }