vkernel - Add usage()
[dragonfly.git] / sys / platform / vkernel / platform / init.c
CommitLineData
9c059ae3
MD
1/*
2 * Copyright (c) 2006 The DragonFly Project. All rights reserved.
3 *
4 * This code is derived from software contributed to The DragonFly Project
5 * by Matthew Dillon <dillon@backplane.com>
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 *
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in
15 * the documentation and/or other materials provided with the
16 * distribution.
17 * 3. Neither the name of The DragonFly Project nor the names of its
18 * contributors may be used to endorse or promote products derived
19 * from this software without specific, prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
24 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
25 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
26 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
27 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
28 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
29 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
30 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
31 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 * SUCH DAMAGE.
33 *
2c4a81d7 34 * $DragonFly: src/sys/platform/vkernel/platform/init.c,v 1.56 2008/05/27 07:48:00 dillon Exp $
9c059ae3
MD
35 */
36
44a09dd6
MD
37#include <sys/types.h>
38#include <sys/systm.h>
e4a473f1 39#include <sys/kernel.h>
44a09dd6
MD
40#include <sys/stat.h>
41#include <sys/mman.h>
e4a473f1
MD
42#include <sys/cons.h>
43#include <sys/random.h>
44#include <sys/vkernel.h>
8a9d8995 45#include <sys/tls.h>
884d6eea 46#include <sys/reboot.h>
e4a473f1 47#include <sys/proc.h>
6f7b98e0 48#include <sys/msgbuf.h>
6efe4829 49#include <sys/vmspace.h>
d53b52ae
SZ
50#include <sys/socket.h>
51#include <sys/sockio.h>
702acf06 52#include <sys/sysctl.h>
91be174d 53#include <sys/un.h>
44a09dd6
MD
54#include <vm/vm_page.h>
55
702acf06 56#include <machine/cpu.h>
8a9d8995 57#include <machine/globaldata.h>
e4a473f1
MD
58#include <machine/tls.h>
59#include <machine/md_var.h>
60#include <machine/vmparam.h>
870b0161 61#include <cpu/specialreg.h>
8a9d8995 62
d53b52ae
SZ
63#include <net/if.h>
64#include <net/if_arp.h>
65#include <net/ethernet.h>
66#include <net/bridge/if_bridgevar.h>
67#include <netinet/in.h>
68#include <arpa/inet.h>
69
9c059ae3 70#include <stdio.h>
44a09dd6 71#include <stdlib.h>
057f0718 72#include <stdarg.h>
278c4e67 73#include <stdbool.h>
44a09dd6
MD
74#include <unistd.h>
75#include <fcntl.h>
76#include <string.h>
77#include <err.h>
78#include <errno.h>
8a9d8995 79#include <assert.h>
278c4e67 80#include <sysexits.h>
9c059ae3 81
e4a473f1 82vm_paddr_t phys_avail[16];
9c059ae3 83vm_paddr_t Maxmem;
71152ac6 84vm_paddr_t Maxmem_bytes;
9f90716f 85int physmem;
9c059ae3 86int MemImageFd = -1;
ee63ee00
SW
87struct vkdisk_info DiskInfo[VKDISK_MAX];
88int DiskNum;
d53b52ae
SZ
89struct vknetif_info NetifInfo[VKNETIF_MAX];
90int NetifNum;
b02b4170 91char *pid_file;
e4a473f1
MD
92vm_offset_t KvaStart;
93vm_offset_t KvaEnd;
9c059ae3 94vm_offset_t KvaSize;
e4a473f1
MD
95vm_offset_t virtual_start;
96vm_offset_t virtual_end;
791c6551
MD
97vm_offset_t virtual2_start;
98vm_offset_t virtual2_end;
e4a473f1
MD
99vm_offset_t kernel_vm_end;
100vm_offset_t crashdumpmap;
6f7b98e0
MD
101vm_offset_t clean_sva;
102vm_offset_t clean_eva;
103struct msgbuf *msgbufp;
104caddr_t ptvmmap;
e4a473f1 105vpte_t *KernelPTD;
71152ac6 106vpte_t *KernelPTA; /* Warning: Offset for direct VA translation */
6f7b98e0 107u_int cpu_feature; /* XXX */
870b0161
MD
108int tsc_present;
109int64_t tsc_frequency;
c5b0b0ba 110int optcpus; /* number of cpus - see mp_start() */
702acf06
JT
111int lwp_cpu_lock; /* if/how to lock virtual CPUs to real CPUs */
112int real_ncpus; /* number of real CPUs */
113int next_cpu; /* next real CPU to lock a virtual CPU to */
9c059ae3 114
1a8fe9c5
MD
115int via_feature_xcrypt = 0; /* XXX */
116int via_feature_rng = 0; /* XXX */
117
8a9d8995
MD
118struct privatespace *CPU_prvspace;
119
252b2227 120static struct trapframe proc0_tf;
e4a473f1
MD
121static void *proc0paddr;
122
44a09dd6
MD
123static void init_sys_memory(char *imageFile);
124static void init_kern_memory(void);
8a9d8995 125static void init_globaldata(void);
e4a473f1 126static void init_vkernel(void);
a72d8a9f 127static void init_disk(char *diskExp[], int diskFileNum, enum vkdisk_type type);
ee63ee00 128static void init_netif(char *netifExp[], int netifFileNum);
278c4e67
SG
129static void writepid(void);
130static void cleanpid(void);
91be174d 131static int unix_connect(const char *path);
278c4e67
SG
132static void usage_err(const char *ctl, ...);
133static void usage_help(_Bool);
9c059ae3 134
40d0276c
MD
135static int save_ac;
136static char **save_av;
137
9c059ae3
MD
138/*
139 * Kernel startup for virtual kernels - standard main()
140 */
141int
142main(int ac, char **av)
143{
44a09dd6 144 char *memImageFile = NULL;
d53b52ae 145 char *netifFile[VKNETIF_MAX];
ee63ee00 146 char *diskFile[VKDISK_MAX];
a72d8a9f 147 char *cdFile[VKDISK_MAX];
44a09dd6 148 char *suffix;
702acf06 149 char *endp;
d53b52ae 150 int netifFileNum = 0;
ee63ee00 151 int diskFileNum = 0;
a72d8a9f 152 int cdFileNum = 0;
41556a7a 153 int bootOnDisk = -1; /* set below to vcd (0) or vkd (1) */
44a09dd6 154 int c;
aaf8b91f 155 int i;
a6d77ab5 156 int j;
aaf8b91f 157 int n;
a6d77ab5 158 int isq;
dc7503af 159 int real_vkernel_enable;
aecc8383 160 int supports_sse;
870b0161 161 size_t vsize;
b02b4170 162
40d0276c
MD
163 save_ac = ac;
164 save_av = av;
165
9c059ae3
MD
166 /*
167 * Process options
168 */
7e42c007 169 kernel_mem_readonly = 1;
c5b0b0ba
MD
170#ifdef SMP
171 optcpus = 2;
172#endif
702acf06 173 lwp_cpu_lock = LCL_NONE;
7e42c007 174
27c3aec1 175 real_vkernel_enable = 0;
870b0161
MD
176 vsize = sizeof(real_vkernel_enable);
177 sysctlbyname("vm.vkernel_enable", &real_vkernel_enable, &vsize, NULL,0);
dc7503af 178
27c3aec1
MD
179 if (real_vkernel_enable == 0) {
180 errx(1, "vm.vkernel_enable is 0, must be set "
181 "to 1 to execute a vkernel!");
182 }
dc7503af 183
27c3aec1 184 real_ncpus = 1;
870b0161
MD
185 vsize = sizeof(real_ncpus);
186 sysctlbyname("hw.ncpu", &real_ncpus, &vsize, NULL, 0);
702acf06 187
278c4e67
SG
188 if (ac < 2)
189 usage_help(false);
190
191 while ((c = getopt(ac, av, "c:hsvl:m:n:r:e:i:p:I:U")) != -1) {
9c059ae3 192 switch(c) {
aaf8b91f
MD
193 case 'e':
194 /*
195 * name=value:name=value:name=value...
a6d77ab5
MD
196 * name="value"...
197 *
198 * Allow values to be quoted but note that shells
199 * may remove the quotes, so using this feature
200 * to embed colons may require a backslash.
aaf8b91f
MD
201 */
202 n = strlen(optarg);
a6d77ab5 203 isq = 0;
aaf8b91f 204 kern_envp = malloc(n + 2);
a6d77ab5
MD
205 for (i = j = 0; i < n; ++i) {
206 if (optarg[i] == '"')
207 isq ^= 1;
208 else if (optarg[i] == '\'')
209 isq ^= 2;
210 else if (isq == 0 && optarg[i] == ':')
211 kern_envp[j++] = 0;
aaf8b91f 212 else
a6d77ab5 213 kern_envp[j++] = optarg[i];
aaf8b91f 214 }
a6d77ab5
MD
215 kern_envp[j++] = 0;
216 kern_envp[j++] = 0;
aaf8b91f 217 break;
884d6eea
MD
218 case 's':
219 boothowto |= RB_SINGLE;
220 break;
9c059ae3
MD
221 case 'v':
222 bootverbose = 1;
223 break;
224 case 'i':
225 memImageFile = optarg;
226 break;
d869938c 227 case 'I':
d53b52ae 228 if (netifFileNum < VKNETIF_MAX)
40d0276c 229 netifFile[netifFileNum++] = strdup(optarg);
d869938c 230 break;
9c059ae3 231 case 'r':
41556a7a
MD
232 if (bootOnDisk < 0)
233 bootOnDisk = 1;
a72d8a9f 234 if (diskFileNum + cdFileNum < VKDISK_MAX)
40d0276c 235 diskFile[diskFileNum++] = strdup(optarg);
9c059ae3 236 break;
a72d8a9f 237 case 'c':
41556a7a
MD
238 if (bootOnDisk < 0)
239 bootOnDisk = 0;
a72d8a9f 240 if (diskFileNum + cdFileNum < VKDISK_MAX)
40d0276c 241 cdFile[cdFileNum++] = strdup(optarg);
a72d8a9f 242 break;
9c059ae3 243 case 'm':
71152ac6 244 Maxmem_bytes = strtoull(optarg, &suffix, 0);
9c059ae3
MD
245 if (suffix) {
246 switch(*suffix) {
247 case 'g':
248 case 'G':
71152ac6 249 Maxmem_bytes <<= 30;
9c059ae3
MD
250 break;
251 case 'm':
252 case 'M':
71152ac6 253 Maxmem_bytes <<= 20;
9c059ae3
MD
254 break;
255 case 'k':
256 case 'K':
71152ac6 257 Maxmem_bytes <<= 10;
9c059ae3
MD
258 break;
259 default:
71152ac6 260 Maxmem_bytes = 0;
278c4e67 261 usage_err("Bad maxmem option");
9c059ae3
MD
262 /* NOT REACHED */
263 break;
264 }
265 }
266 break;
702acf06
JT
267 case 'l':
268 next_cpu = -1;
269 if (strncmp("map", optarg, 3) == 0) {
270 lwp_cpu_lock = LCL_PER_CPU;
92d2f80a 271 if (optarg[3] == ',') {
702acf06 272 next_cpu = strtol(optarg+4, &endp, 0);
92d2f80a 273 if (*endp != '\0')
278c4e67 274 usage_err("Bad target CPU number at '%s'", endp);
92d2f80a
MD
275 } else {
276 next_cpu = 0;
277 }
702acf06 278 if (next_cpu < 0 || next_cpu > real_ncpus - 1)
278c4e67 279 usage_err("Bad target CPU, valid range is 0-%d", real_ncpus - 1);
702acf06
JT
280 } else if (strncmp("any", optarg, 3) == 0) {
281 lwp_cpu_lock = LCL_NONE;
282 } else {
283 lwp_cpu_lock = LCL_SINGLE_CPU;
284 next_cpu = strtol(optarg, &endp, 0);
285 if (*endp != '\0')
278c4e67 286 usage_err("Bad target CPU number at '%s'", endp);
702acf06 287 if (next_cpu < 0 || next_cpu > real_ncpus - 1)
278c4e67 288 usage_err("Bad target CPU, valid range is 0-%d", real_ncpus - 1);
702acf06
JT
289 }
290 break;
c5b0b0ba
MD
291 case 'n':
292 /*
293 * This value is set up by mp_start(), don't just
294 * set ncpus here.
295 */
296#ifdef SMP
297 optcpus = strtol(optarg, NULL, 0);
057f0718 298 if (optcpus < 1 || optcpus > MAXCPU)
278c4e67 299 usage_err("Bad ncpus, valid range is 1-%d", MAXCPU);
c5b0b0ba
MD
300#else
301 if (strtol(optarg, NULL, 0) != 1) {
278c4e67 302 usage_err("You built a UP vkernel, only 1 cpu!");
c5b0b0ba
MD
303 }
304#endif
305
151f45bf 306 break;
b02b4170
MD
307 case 'p':
308 pid_file = optarg;
309 break;
7e42c007
MD
310 case 'U':
311 kernel_mem_readonly = 0;
312 break;
278c4e67
SG
313 case 'h':
314 usage_help(true);
315 break;
316 case '?':
317 default:
318 usage_help(false);
9c059ae3
MD
319 }
320 }
321
b02b4170 322 writepid();
b1b21fa6 323 cpu_disable_intr();
9c059ae3
MD
324 init_sys_memory(memImageFile);
325 init_kern_memory();
8a9d8995 326 init_globaldata();
e4a473f1 327 init_vkernel();
702acf06 328 setrealcpu();
f991237f 329 init_kqueue();
41556a7a 330
870b0161
MD
331 /*
332 * Check TSC
333 */
334 vsize = sizeof(tsc_present);
335 sysctlbyname("hw.tsc_present", &tsc_present, &vsize, NULL, 0);
336 vsize = sizeof(tsc_frequency);
337 sysctlbyname("hw.tsc_frequency", &tsc_frequency, &vsize, NULL, 0);
338 if (tsc_present)
339 cpu_feature |= CPUID_TSC;
340
341 /*
342 * Check SSE
343 */
344 vsize = sizeof(supports_sse);
27c3aec1 345 supports_sse = 0;
870b0161 346 sysctlbyname("hw.instruction_sse", &supports_sse, &vsize, NULL, 0);
aecc8383 347 init_fpu(supports_sse);
870b0161
MD
348 if (supports_sse)
349 cpu_feature |= CPUID_SSE | CPUID_FXSR;
aecc8383 350
41556a7a
MD
351 /*
352 * We boot from the first installed disk.
353 */
354 if (bootOnDisk == 1) {
355 init_disk(diskFile, diskFileNum, VKD_DISK);
356 init_disk(cdFile, cdFileNum, VKD_CD);
357 } else {
358 init_disk(cdFile, cdFileNum, VKD_CD);
359 init_disk(diskFile, diskFileNum, VKD_DISK);
360 }
d53b52ae 361 init_netif(netifFile, netifFileNum);
6092278a 362 init_exceptions();
9c059ae3
MD
363 mi_startup();
364 /* NOT REACHED */
278c4e67 365 exit(EX_SOFTWARE);
9c059ae3
MD
366}
367
368/*
369 * Initialize system memory. This is the virtual kernel's 'RAM'.
370 */
371static
372void
44a09dd6 373init_sys_memory(char *imageFile)
9c059ae3 374{
44a09dd6 375 struct stat st;
3748859a 376 int i;
9c059ae3
MD
377 int fd;
378
379 /*
380 * Figure out the system memory image size. If an image file was
381 * specified and -m was not specified, use the image file's size.
382 */
383
71152ac6
MD
384 if (imageFile && stat(imageFile, &st) == 0 && Maxmem_bytes == 0)
385 Maxmem_bytes = (vm_paddr_t)st.st_size;
386 if ((imageFile == NULL || stat(imageFile, &st) < 0) &&
387 Maxmem_bytes == 0) {
44a09dd6 388 err(1, "Cannot create new memory file %s unless "
9c059ae3
MD
389 "system memory size is specified with -m",
390 imageFile);
391 /* NOT REACHED */
392 }
393
394 /*
395 * Maxmem must be known at this time
396 */
71152ac6 397 if (Maxmem_bytes < 32 * 1024 * 1024 || (Maxmem_bytes & SEG_MASK)) {
9c059ae3
MD
398 err(1, "Bad maxmem specification: 32MB minimum, "
399 "multiples of %dMB only",
400 SEG_SIZE / 1024 / 1024);
401 /* NOT REACHED */
402 }
403
404 /*
405 * Generate an image file name if necessary, then open/create the
406 * file exclusively locked. Do not allow multiple virtual kernels
407 * to use the same image file.
408 */
3748859a
MD
409 if (imageFile == NULL) {
410 for (i = 0; i < 1000000; ++i) {
0171c06b 411 asprintf(&imageFile, "/var/vkernel/memimg.%06d", i);
3748859a
MD
412 fd = open(imageFile,
413 O_RDWR|O_CREAT|O_EXLOCK|O_NONBLOCK, 0644);
0171c06b 414 if (fd < 0 && errno == EWOULDBLOCK) {
3748859a
MD
415 free(imageFile);
416 continue;
417 }
418 break;
419 }
420 } else {
421 fd = open(imageFile, O_RDWR|O_CREAT|O_EXLOCK|O_NONBLOCK, 0644);
422 }
0171c06b 423 printf("Using memory file: %s\n", imageFile);
9c059ae3 424 if (fd < 0 || fstat(fd, &st) < 0) {
5de37c8b 425 err(1, "Unable to open/create %s", imageFile);
9c059ae3
MD
426 /* NOT REACHED */
427 }
428
429 /*
b72e51d1
MD
430 * Truncate or extend the file as necessary. Clean out the contents
431 * of the file, we want it to be full of holes so we don't waste
432 * time reading in data from an old file that we no longer care
433 * about.
9c059ae3 434 */
b72e51d1
MD
435 ftruncate(fd, 0);
436 ftruncate(fd, Maxmem_bytes);
9c059ae3 437
9c059ae3 438 MemImageFd = fd;
71152ac6 439 Maxmem = Maxmem_bytes >> PAGE_SHIFT;
9f90716f 440 physmem = Maxmem;
9c059ae3
MD
441}
442
443/*
444 * Initialize kernel memory. This reserves kernel virtual memory by using
445 * MAP_VPAGETABLE
446 */
0c4c980f 447
9c059ae3
MD
448static
449void
450init_kern_memory(void)
451{
452 void *base;
0c4c980f 453 void *try;
3748859a 454 char *zero;
0c4c980f
MD
455 char dummy;
456 char *topofstack = &dummy;
e4a473f1
MD
457 vpte_t pte;
458 int i;
9c059ae3
MD
459
460 /*
461 * Memory map our kernel virtual memory space. Note that the
462 * kernel image itself is not made part of this memory for the
463 * moment.
71152ac6
MD
464 *
465 * The memory map must be segment-aligned so we can properly
466 * offset KernelPTD.
0c4c980f
MD
467 *
468 * If the system kernel has a different MAXDSIZ, it might not
469 * be possible to map kernel memory in its prefered location.
470 * Try a number of different locations.
9c059ae3 471 */
0c4c980f
MD
472 try = (void *)0x40000000;
473 base = NULL;
474 while ((char *)try + KERNEL_KVA_SIZE < topofstack) {
475 base = mmap(try, KERNEL_KVA_SIZE, PROT_READ|PROT_WRITE,
476 MAP_FILE|MAP_SHARED|MAP_VPAGETABLE,
477 MemImageFd, 0);
478 if (base == try)
479 break;
480 if (base != MAP_FAILED)
481 munmap(base, KERNEL_KVA_SIZE);
482 try = (char *)try + 0x10000000;
483 }
484 if (base != try) {
9c059ae3
MD
485 err(1, "Unable to mmap() kernel virtual memory!");
486 /* NOT REACHED */
487 }
a3c35df6 488 madvise(base, KERNEL_KVA_SIZE, MADV_NOSYNC);
e4a473f1 489 KvaStart = (vm_offset_t)base;
9c059ae3 490 KvaSize = KERNEL_KVA_SIZE;
e4a473f1 491 KvaEnd = KvaStart + KvaSize;
d4f5f27c
MD
492
493 /* cannot use kprintf yet */
0c4c980f 494 printf("KVM mapped at %p-%p\n", (void *)KvaStart, (void *)KvaEnd);
e4a473f1
MD
495
496 /*
497 * Create a top-level page table self-mapping itself.
498 *
499 * Initialize the page directory at physical page index 0 to point
500 * to an array of page table pages starting at physical page index 1
501 */
502 lseek(MemImageFd, 0L, 0);
503 for (i = 0; i < KERNEL_KVA_SIZE / SEG_SIZE; ++i) {
504 pte = ((i + 1) * PAGE_SIZE) | VPTE_V | VPTE_R | VPTE_W;
505 write(MemImageFd, &pte, sizeof(pte));
506 }
507
508 /*
509 * Initialize the PTEs in the page table pages required to map the
510 * page table itself. This includes mapping the page directory page
511 * at the base so we go one more loop then normal.
512 */
513 lseek(MemImageFd, PAGE_SIZE, 0);
71152ac6 514 for (i = 0; i <= KERNEL_KVA_SIZE / SEG_SIZE * sizeof(vpte_t); ++i) {
e4a473f1
MD
515 pte = (i * PAGE_SIZE) | VPTE_V | VPTE_R | VPTE_W;
516 write(MemImageFd, &pte, sizeof(pte));
517 }
518
519 /*
3748859a
MD
520 * Initialize remaining PTEs to 0. We may be reusing a memory image
521 * file. This is approximately a megabyte.
522 */
523 i = (KERNEL_KVA_SIZE / PAGE_SIZE - i) * sizeof(pte);
524 zero = malloc(PAGE_SIZE);
b87d6bd2 525 bzero(zero, PAGE_SIZE);
3748859a
MD
526 while (i) {
527 write(MemImageFd, zero, (i > PAGE_SIZE) ? PAGE_SIZE : i);
528 i = i - ((i > PAGE_SIZE) ? PAGE_SIZE : i);
529 }
530 free(zero);
531
532 /*
e4a473f1
MD
533 * Enable the page table and calculate pointers to our self-map
534 * for easy kernel page table manipulation.
71152ac6
MD
535 *
536 * KernelPTA must be offset so we can do direct VA translations
e4a473f1 537 */
71152ac6
MD
538 mcontrol(base, KERNEL_KVA_SIZE, MADV_SETMAP,
539 0 | VPTE_R | VPTE_W | VPTE_V);
e4a473f1
MD
540 KernelPTD = (vpte_t *)base; /* pg directory */
541 KernelPTA = (vpte_t *)((char *)base + PAGE_SIZE); /* pg table pages */
71152ac6 542 KernelPTA -= KvaStart >> PAGE_SHIFT;
e4a473f1
MD
543
544 /*
545 * phys_avail[] represents unallocated physical memory. MI code
546 * will use phys_avail[] to create the vm_page array.
547 */
548 phys_avail[0] = PAGE_SIZE +
549 KERNEL_KVA_SIZE / PAGE_SIZE * sizeof(vpte_t);
550 phys_avail[0] = (phys_avail[0] + PAGE_MASK) & ~(vm_paddr_t)PAGE_MASK;
71152ac6 551 phys_avail[1] = Maxmem_bytes;
e4a473f1
MD
552
553 /*
554 * (virtual_start, virtual_end) represent unallocated kernel virtual
555 * memory. MI code will create kernel_map using these parameters.
556 */
557 virtual_start = KvaStart + PAGE_SIZE +
558 KERNEL_KVA_SIZE / PAGE_SIZE * sizeof(vpte_t);
559 virtual_start = (virtual_start + PAGE_MASK) & ~(vm_offset_t)PAGE_MASK;
560 virtual_end = KvaStart + KERNEL_KVA_SIZE;
561
562 /*
00835518
MD
563 * kernel_vm_end could be set to virtual_end but we want some
564 * indication of how much of the kernel_map we've used, so
565 * set it low and let pmap_growkernel increase it even though we
566 * don't need to create any new page table pages.
e4a473f1 567 */
00835518 568 kernel_vm_end = virtual_start;
e4a473f1
MD
569
570 /*
571 * Allocate space for process 0's UAREA.
572 */
573 proc0paddr = (void *)virtual_start;
574 for (i = 0; i < UPAGES; ++i) {
575 pmap_kenter_quick(virtual_start, phys_avail[0]);
576 virtual_start += PAGE_SIZE;
577 phys_avail[0] += PAGE_SIZE;
578 }
9c059ae3
MD
579
580 /*
e4a473f1 581 * crashdumpmap
9c059ae3 582 */
e4a473f1
MD
583 crashdumpmap = virtual_start;
584 virtual_start += MAXDUMPPGS * PAGE_SIZE;
585
586 /*
6f7b98e0
MD
587 * msgbufp maps the system message buffer
588 */
589 assert((MSGBUF_SIZE & PAGE_MASK) == 0);
590 msgbufp = (void *)virtual_start;
591 for (i = 0; i < (MSGBUF_SIZE >> PAGE_SHIFT); ++i) {
592 pmap_kenter_quick(virtual_start, phys_avail[0]);
593 virtual_start += PAGE_SIZE;
594 phys_avail[0] += PAGE_SIZE;
595 }
596 msgbufinit(msgbufp, MSGBUF_SIZE);
597
598 /*
599 * used by kern_memio for /dev/mem access
600 */
601 ptvmmap = (caddr_t)virtual_start;
602 virtual_start += PAGE_SIZE;
603
604 /*
e4a473f1
MD
605 * Bootstrap the kernel_pmap
606 */
607 pmap_bootstrap();
9c059ae3
MD
608}
609
610/*
e4a473f1
MD
611 * Map the per-cpu globaldata for cpu #0. Allocate the space using
612 * virtual_start and phys_avail[0]
8a9d8995
MD
613 */
614static
615void
616init_globaldata(void)
617{
e4a473f1
MD
618 int i;
619 vm_paddr_t pa;
620 vm_offset_t va;
621
622 /*
623 * Reserve enough KVA to cover possible cpus. This is a considerable
624 * amount of KVA since the privatespace structure includes two
625 * whole page table mappings.
626 */
627 virtual_start = (virtual_start + SEG_MASK) & ~(vm_offset_t)SEG_MASK;
628 CPU_prvspace = (void *)virtual_start;
629 virtual_start += sizeof(struct privatespace) * SMP_MAXCPU;
630
631 /*
632 * Allocate enough physical memory to cover the mdglobaldata
633 * portion of the space and the idle stack and map the pages
634 * into KVA. For cpu #0 only.
635 */
636 for (i = 0; i < sizeof(struct mdglobaldata); i += PAGE_SIZE) {
637 pa = phys_avail[0];
638 va = (vm_offset_t)&CPU_prvspace[0].mdglobaldata + i;
639 pmap_kenter_quick(va, pa);
640 phys_avail[0] += PAGE_SIZE;
641 }
642 for (i = 0; i < sizeof(CPU_prvspace[0].idlestack); i += PAGE_SIZE) {
643 pa = phys_avail[0];
644 va = (vm_offset_t)&CPU_prvspace[0].idlestack + i;
645 pmap_kenter_quick(va, pa);
646 phys_avail[0] += PAGE_SIZE;
647 }
648
649 /*
27c3aec1
MD
650 * Setup the %fs for cpu #0. The mycpu macro works after this
651 * point. Note that %gs is used by pthreads.
e4a473f1 652 */
9ad680a3 653 tls_set_fs(&CPU_prvspace[0], sizeof(struct privatespace));
e4a473f1
MD
654}
655
656/*
657 * Initialize very low level systems including thread0, proc0, etc.
658 */
659static
660void
661init_vkernel(void)
662{
663 struct mdglobaldata *gd;
664
665 gd = &CPU_prvspace[0].mdglobaldata;
666 bzero(gd, sizeof(*gd));
667
668 gd->mi.gd_curthread = &thread0;
71152ac6 669 thread0.td_gd = &gd->mi;
e4a473f1 670 ncpus = 1;
b45759e1
MD
671 ncpus2 = 1; /* rounded down power of 2 */
672 ncpus_fit = 1; /* rounded up power of 2 */
673 /* ncpus2_mask and ncpus_fit_mask are 0 */
e4a473f1
MD
674 init_param1();
675 gd->mi.gd_prvspace = &CPU_prvspace[0];
676 mi_gdinit(&gd->mi, 0);
677 cpu_gdinit(gd, 0);
678 mi_proc0init(&gd->mi, proc0paddr);
08f2f1bb 679 lwp0.lwp_md.md_regs = &proc0_tf;
252b2227 680
e4a473f1
MD
681 /*init_locks();*/
682 cninit();
683 rand_initialize();
684#if 0 /* #ifdef DDB */
685 kdb_init();
686 if (boothowto & RB_KDB)
687 Debugger("Boot flags requested debugger");
688#endif
514621ec 689 identcpu();
e4a473f1
MD
690#if 0
691 initializecpu(); /* Initialize CPU registers */
692#endif
693 init_param2((phys_avail[1] - phys_avail[0]) / PAGE_SIZE);
694
695#if 0
696 /*
697 * Map the message buffer
698 */
699 for (off = 0; off < round_page(MSGBUF_SIZE); off += PAGE_SIZE)
700 pmap_kenter((vm_offset_t)msgbufp + off, avail_end + off);
701 msgbufinit(msgbufp, MSGBUF_SIZE);
702#endif
703#if 0
704 thread0.td_pcb_cr3 ... MMU
08f2f1bb 705 lwp0.lwp_md.md_regs = &proc0_tf;
e4a473f1 706#endif
8a9d8995
MD
707}
708
709/*
ee63ee00
SW
710 * Filesystem image paths for the virtual kernel are optional.
711 * If specified they each should point to a disk image,
712 * the first of which will become the root disk.
5b85cb7d
MD
713 *
714 * The virtual kernel caches data from our 'disk' just like a normal kernel,
715 * so we do not really want the real kernel to cache the data too. Use
716 * O_DIRECT to remove the duplication.
9c059ae3
MD
717 */
718static
719void
a72d8a9f 720init_disk(char *diskExp[], int diskFileNum, enum vkdisk_type type)
44a09dd6 721{
ee63ee00
SW
722 int i;
723
724 if (diskFileNum == 0)
725 return;
726
727 for(i=0; i < diskFileNum; i++){
728 char *fname;
729 fname = diskExp[i];
730
731 if (fname == NULL) {
732 warnx("Invalid argument to '-r'");
733 continue;
734 }
735
736 if (DiskNum < VKDISK_MAX) {
737 struct stat st;
738 struct vkdisk_info* info = NULL;
739 int fd;
740 size_t l = 0;
741
a72d8a9f 742 if (type == VKD_DISK)
6882b7a5 743 fd = open(fname, O_RDWR|O_DIRECT, 0644);
a72d8a9f
MD
744 else
745 fd = open(fname, O_RDONLY|O_DIRECT, 0644);
ee63ee00
SW
746 if (fd < 0 || fstat(fd, &st) < 0) {
747 err(1, "Unable to open/create %s", fname);
748 /* NOT REACHED */
749 }
6882b7a5
MD
750 if (S_ISREG(st.st_mode)) {
751 if (flock(fd, LOCK_EX|LOCK_NB) < 0) {
752 errx(1, "Disk image %s is already "
753 "in use\n", fname);
754 /* NOT REACHED */
755 }
756 }
ee63ee00
SW
757
758 info = &DiskInfo[DiskNum];
759 l = strlen(fname);
0171c06b 760
ee63ee00
SW
761 info->unit = i;
762 info->fd = fd;
a72d8a9f 763 info->type = type;
ee63ee00
SW
764 memcpy(info->fname, fname, l);
765
41556a7a 766 if (DiskNum == 0) {
1f1bc20e 767 if (type == VKD_CD) {
a72d8a9f 768 rootdevnames[0] = "cd9660:vcd0a";
1f1bc20e 769 } else if (type == VKD_DISK) {
28b44c8f 770 rootdevnames[0] = "ufs:vkd0s0a";
1f1bc20e
MD
771 rootdevnames[1] = "ufs:vkd0s1a";
772 }
a72d8a9f 773 }
ee63ee00
SW
774
775 DiskNum++;
776 } else {
23e2971a 777 warnx("vkd%d (%s) > VKDISK_MAX", DiskNum, fname);
ee63ee00 778 continue;
0171c06b
MD
779 }
780 }
44a09dd6
MD
781}
782
783static
d53b52ae
SZ
784int
785netif_set_tapflags(int tap_unit, int f, int s)
786{
787 struct ifreq ifr;
788 int flags;
789
790 bzero(&ifr, sizeof(ifr));
791
792 snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "tap%d", tap_unit);
793 if (ioctl(s, SIOCGIFFLAGS, &ifr) < 0) {
5de37c8b 794 warn("tap%d: ioctl(SIOCGIFFLAGS) failed", tap_unit);
d53b52ae
SZ
795 return -1;
796 }
797
798 /*
799 * Adjust if_flags
800 *
801 * If the flags are already set/cleared, then we return
802 * immediately to avoid extra syscalls
803 */
804 flags = (ifr.ifr_flags & 0xffff) | (ifr.ifr_flagshigh << 16);
805 if (f < 0) {
806 /* Turn off flags */
807 f = -f;
808 if ((flags & f) == 0)
809 return 0;
810 flags &= ~f;
811 } else {
812 /* Turn on flags */
813 if (flags & f)
814 return 0;
815 flags |= f;
816 }
817
818 /*
819 * Fix up ifreq.ifr_name, since it may be trashed
820 * in previous ioctl(SIOCGIFFLAGS)
821 */
822 snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "tap%d", tap_unit);
823
824 ifr.ifr_flags = flags & 0xffff;
825 ifr.ifr_flagshigh = flags >> 16;
826 if (ioctl(s, SIOCSIFFLAGS, &ifr) < 0) {
5de37c8b 827 warn("tap%d: ioctl(SIOCSIFFLAGS) failed", tap_unit);
d53b52ae
SZ
828 return -1;
829 }
830 return 0;
831}
832
833static
834int
835netif_set_tapaddr(int tap_unit, in_addr_t addr, in_addr_t mask, int s)
836{
837 struct ifaliasreq ifra;
838 struct sockaddr_in *in;
839
840 bzero(&ifra, sizeof(ifra));
841 snprintf(ifra.ifra_name, sizeof(ifra.ifra_name), "tap%d", tap_unit);
842
843 /* Setup address */
844 in = (struct sockaddr_in *)&ifra.ifra_addr;
845 in->sin_family = AF_INET;
846 in->sin_len = sizeof(*in);
847 in->sin_addr.s_addr = addr;
848
849 if (mask != 0) {
850 /* Setup netmask */
851 in = (struct sockaddr_in *)&ifra.ifra_mask;
852 in->sin_len = sizeof(*in);
853 in->sin_addr.s_addr = mask;
854 }
855
856 if (ioctl(s, SIOCAIFADDR, &ifra) < 0) {
5de37c8b 857 warn("tap%d: ioctl(SIOCAIFADDR) failed", tap_unit);
d53b52ae
SZ
858 return -1;
859 }
860 return 0;
861}
862
863static
864int
865netif_add_tap2brg(int tap_unit, const char *ifbridge, int s)
866{
867 struct ifbreq ifbr;
868 struct ifdrv ifd;
869
870 bzero(&ifbr, sizeof(ifbr));
871 snprintf(ifbr.ifbr_ifsname, sizeof(ifbr.ifbr_ifsname),
872 "tap%d", tap_unit);
873
874 bzero(&ifd, sizeof(ifd));
875 strlcpy(ifd.ifd_name, ifbridge, sizeof(ifd.ifd_name));
876 ifd.ifd_cmd = BRDGADD;
877 ifd.ifd_len = sizeof(ifbr);
878 ifd.ifd_data = &ifbr;
879
880 if (ioctl(s, SIOCSDRVSPEC, &ifd) < 0) {
881 /*
882 * 'errno == EEXIST' means that the tap(4) is already
883 * a member of the bridge(4)
884 */
885 if (errno != EEXIST) {
5de37c8b 886 warn("ioctl(%s, SIOCSDRVSPEC) failed", ifbridge);
d53b52ae
SZ
887 return -1;
888 }
889 }
890 return 0;
891}
892
893#define TAPDEV_OFLAGS (O_RDWR | O_NONBLOCK)
894
d53b52ae
SZ
895/*
896 * Locate the first unused tap(4) device file if auto mode is requested,
897 * or open the user supplied device file, and bring up the corresponding
898 * tap(4) interface.
899 *
900 * NOTE: Only tap(4) device file is supported currently
901 */
902static
903int
904netif_open_tap(const char *netif, int *tap_unit, int s)
d869938c 905{
d53b52ae
SZ
906 char tap_dev[MAXPATHLEN];
907 int tap_fd, failed;
908 struct stat st;
235a1432 909 char *dname;
d53b52ae
SZ
910
911 *tap_unit = -1;
912
913 if (strcmp(netif, "auto") == 0) {
d53b52ae
SZ
914 /*
915 * Find first unused tap(4) device file
916 */
235a1432 917 tap_fd = open("/dev/tap", TAPDEV_OFLAGS);
d53b52ae 918 if (tap_fd < 0) {
235a1432 919 warnc(errno, "Unable to find a free tap(4)");
d53b52ae
SZ
920 return -1;
921 }
922 } else {
923 /*
91be174d 924 * User supplied tap(4) device file or unix socket.
d53b52ae
SZ
925 */
926 if (netif[0] == '/') /* Absolute path */
927 strlcpy(tap_dev, netif, sizeof(tap_dev));
928 else
929 snprintf(tap_dev, sizeof(tap_dev), "/dev/%s", netif);
930
931 tap_fd = open(tap_dev, TAPDEV_OFLAGS);
91be174d
MD
932
933 /*
934 * If we cannot open normally try to connect to it.
935 */
936 if (tap_fd < 0)
937 tap_fd = unix_connect(tap_dev);
938
d53b52ae 939 if (tap_fd < 0) {
5de37c8b 940 warn("Unable to open %s", tap_dev);
d53b52ae 941 return -1;
d869938c
SW
942 }
943 }
d53b52ae
SZ
944
945 /*
946 * Check whether the device file is a tap(4)
947 */
235a1432
MD
948 if (fstat(tap_fd, &st) < 0) {
949 failed = 1;
950 } else if (S_ISCHR(st.st_mode)) {
951 dname = fdevname(tap_fd);
952 if (dname)
953 dname = strstr(dname, "tap");
954 if (dname) {
955 /*
956 * Bring up the corresponding tap(4) interface
957 */
958 *tap_unit = strtol(dname + 3, NULL, 10);
959 printf("TAP UNIT %d\n", *tap_unit);
960 if (netif_set_tapflags(*tap_unit, IFF_UP, s) == 0)
961 failed = 0;
3b6e5cb0
SW
962 else
963 failed = 1;
235a1432
MD
964 } else {
965 failed = 1;
966 }
91be174d
MD
967 } else if (S_ISSOCK(st.st_mode)) {
968 /*
969 * Special socket connection (typically to vknet). We
970 * do not have to do anything.
971 */
972 failed = 0;
d53b52ae 973 } else {
235a1432 974 failed = 1;
d53b52ae
SZ
975 }
976
977 if (failed) {
235a1432 978 warnx("%s is not a tap(4) device or socket", tap_dev);
d53b52ae
SZ
979 close(tap_fd);
980 tap_fd = -1;
981 *tap_unit = -1;
982 }
983 return tap_fd;
984}
985
91be174d
MD
986static int
987unix_connect(const char *path)
988{
989 struct sockaddr_un sunx;
990 int len;
991 int net_fd;
3a6117bb
MD
992 int sndbuf = 262144;
993 struct stat st;
91be174d
MD
994
995 snprintf(sunx.sun_path, sizeof(sunx.sun_path), "%s", path);
996 len = offsetof(struct sockaddr_un, sun_path[strlen(sunx.sun_path)]);
997 ++len; /* include nul */
998 sunx.sun_family = AF_UNIX;
999 sunx.sun_len = len;
1000
1001 net_fd = socket(AF_UNIX, SOCK_SEQPACKET, 0);
1002 if (net_fd < 0)
1003 return(-1);
1004 if (connect(net_fd, (void *)&sunx, len) < 0) {
1005 close(net_fd);
1006 return(-1);
1007 }
3a6117bb
MD
1008 setsockopt(net_fd, SOL_SOCKET, SO_SNDBUF, &sndbuf, sizeof(sndbuf));
1009 if (fstat(net_fd, &st) == 0)
1010 printf("Network socket buffer: %d bytes\n", st.st_blksize);
91be174d
MD
1011 fcntl(net_fd, F_SETFL, O_NONBLOCK);
1012 return(net_fd);
1013}
1014
d53b52ae
SZ
1015#undef TAPDEV_MAJOR
1016#undef TAPDEV_MINOR
1017#undef TAPDEV_OFLAGS
1018
1019/*
1020 * Following syntax is supported,
1021 * 1) x.x.x.x tap(4)'s address is x.x.x.x
1022 *
1023 * 2) x.x.x.x/z tap(4)'s address is x.x.x.x
1024 * tap(4)'s netmask len is z
1025 *
1026 * 3) x.x.x.x:y.y.y.y tap(4)'s address is x.x.x.x
1027 * pseudo netif's address is y.y.y.y
1028 *
1029 * 4) x.x.x.x:y.y.y.y/z tap(4)'s address is x.x.x.x
1030 * pseudo netif's address is y.y.y.y
1031 * tap(4) and pseudo netif's netmask len are z
1032 *
1033 * 5) bridgeX tap(4) will be added to bridgeX
1034 *
1035 * 6) bridgeX:y.y.y.y tap(4) will be added to bridgeX
1036 * pseudo netif's address is y.y.y.y
1037 *
1038 * 7) bridgeX:y.y.y.y/z tap(4) will be added to bridgeX
1039 * pseudo netif's address is y.y.y.y
1040 * pseudo netif's netmask len is z
1041 */
1042static
1043int
1044netif_init_tap(int tap_unit, in_addr_t *addr, in_addr_t *mask, int s)
1045{
1046 in_addr_t tap_addr, netmask, netif_addr;
1047 int next_netif_addr;
1048 char *tok, *masklen_str, *ifbridge;
1049
1050 *addr = 0;
1051 *mask = 0;
1052
1053 tok = strtok(NULL, ":/");
1054 if (tok == NULL) {
1055 /*
1056 * Nothing special, simply use tap(4) as backend
1057 */
1058 return 0;
1059 }
1060
1061 if (inet_pton(AF_INET, tok, &tap_addr) > 0) {
1062 /*
1063 * tap(4)'s address is supplied
1064 */
1065 ifbridge = NULL;
1066
1067 /*
1068 * If there is next token, then it may be pseudo
1069 * netif's address or netmask len for tap(4)
1070 */
1071 next_netif_addr = 0;
1072 } else {
1073 /*
1074 * Not tap(4)'s address, assume it as a bridge(4)
1075 * iface name
1076 */
1077 tap_addr = 0;
1078 ifbridge = tok;
1079
1080 /*
1081 * If there is next token, then it must be pseudo
1082 * netif's address
1083 */
1084 next_netif_addr = 1;
1085 }
1086
1087 netmask = netif_addr = 0;
1088
1089 tok = strtok(NULL, ":/");
1090 if (tok == NULL)
1091 goto back;
1092
1093 if (inet_pton(AF_INET, tok, &netif_addr) <= 0) {
1094 if (next_netif_addr) {
1095 warnx("Invalid pseudo netif address: %s", tok);
1096 return -1;
1097 }
1098 netif_addr = 0;
1099
1100 /*
1101 * Current token is not address, then it must be netmask len
1102 */
1103 masklen_str = tok;
1104 } else {
1105 /*
1106 * Current token is pseudo netif address, if there is next token
1107 * it must be netmask len
1108 */
1109 masklen_str = strtok(NULL, "/");
1110 }
1111
1112 /* Calculate netmask */
1113 if (masklen_str != NULL) {
1114 u_long masklen;
1115
1116 masklen = strtoul(masklen_str, NULL, 10);
1117 if (masklen < 32 && masklen > 0) {
1118 netmask = htonl(~((1LL << (32 - masklen)) - 1)
1119 & 0xffffffff);
1120 } else {
1121 warnx("Invalid netmask len: %lu", masklen);
1122 return -1;
1123 }
1124 }
1125
1126 /* Make sure there is no more token left */
1127 if (strtok(NULL, ":/") != NULL) {
1128 warnx("Invalid argument to '-I'");
1129 return -1;
1130 }
1131
1132back:
2c4a81d7
MD
1133 if (tap_unit < 0) {
1134 /* Do nothing */
1135 } else if (ifbridge == NULL) {
d53b52ae
SZ
1136 /* Set tap(4) address/netmask */
1137 if (netif_set_tapaddr(tap_unit, tap_addr, netmask, s) < 0)
1138 return -1;
1139 } else {
1140 /* Tie tap(4) to bridge(4) */
1141 if (netif_add_tap2brg(tap_unit, ifbridge, s) < 0)
1142 return -1;
1143 }
1144
1145 *addr = netif_addr;
1146 *mask = netmask;
1147 return 0;
1148}
1149
1150/*
1151 * NetifInfo[] will be filled for pseudo netif initialization.
1152 * NetifNum will be bumped to reflect the number of valid entries
1153 * in NetifInfo[].
1154 */
1155static
1156void
1157init_netif(char *netifExp[], int netifExpNum)
1158{
1159 int i, s;
1160
1161 if (netifExpNum == 0)
1162 return;
1163
1164 s = socket(AF_INET, SOCK_DGRAM, 0); /* for ioctl(SIOC) */
1165 if (s < 0)
1166 return;
1167
1168 for (i = 0; i < netifExpNum; ++i) {
1169 struct vknetif_info *info;
1170 in_addr_t netif_addr, netif_mask;
1171 int tap_fd, tap_unit;
1172 char *netif;
1173
1174 netif = strtok(netifExp[i], ":");
1175 if (netif == NULL) {
ee63ee00 1176 warnx("Invalid argument to '-I'");
d53b52ae
SZ
1177 continue;
1178 }
1179
1180 /*
1181 * Open tap(4) device file and bring up the
1182 * corresponding interface
1183 */
1184 tap_fd = netif_open_tap(netif, &tap_unit, s);
1185 if (tap_fd < 0)
1186 continue;
1187
1188 /*
1189 * Initialize tap(4) and get address/netmask
1190 * for pseudo netif
1191 *
1192 * NB: Rest part of netifExp[i] is passed
1193 * to netif_init_tap() implicitly.
1194 */
1195 if (netif_init_tap(tap_unit, &netif_addr, &netif_mask, s) < 0) {
1196 /*
1197 * NB: Closing tap(4) device file will bring
1198 * down the corresponding interface
1199 */
1200 close(tap_fd);
1201 continue;
1202 }
1203
1204 info = &NetifInfo[NetifNum];
1205 info->tap_fd = tap_fd;
1206 info->tap_unit = tap_unit;
1207 info->netif_addr = netif_addr;
1208 info->netif_mask = netif_mask;
1209
1210 NetifNum++;
1211 if (NetifNum >= VKNETIF_MAX) /* XXX will this happen? */
1212 break;
1213 }
1214 close(s);
d869938c
SW
1215}
1216
1217static
1218void
b02b4170
MD
1219writepid( void )
1220{
1221 pid_t self;
1222 FILE *fp;
1223
1224 if (pid_file != NULL) {
1225 self = getpid();
1226 fp = fopen(pid_file, "w");
1227
1228 if (fp != NULL) {
1229 fprintf(fp, "%ld\n", (long)self);
1230 fclose(fp);
1231 }
1232 else {
1233 perror("Warning: couldn't open pidfile");
1234 }
1235 }
1236}
1237
1238static
1239void
1240cleanpid( void )
1241{
1242 if (pid_file != NULL) {
1243 if ( unlink(pid_file) != 0 )
1244 perror("Warning: couldn't remove pidfile");
1245 }
1246}
1247
1248static
1249void
278c4e67 1250usage_err(const char *ctl, ...)
9c059ae3 1251{
057f0718
MD
1252 va_list va;
1253
1254 va_start(va, ctl);
1255 vfprintf(stderr, ctl, va);
1256 va_end(va);
1257 fprintf(stderr, "\n");
278c4e67
SG
1258 exit(EX_USAGE);
1259}
1260
1261static
1262void
1263usage_help(_Bool help)
1264{
1265 fprintf(stderr, "Usage: %s [-hsUv] [-c file] [-e name=value:name=value:...]\n"
1266 "\t[-i file] [-I interface[:address1[:address2][/netmask]]] [-l cpulock]\n"
1267 "\t[-m size] [-n numcpus] [-p file] [-r file]\n", save_av[0]);
1268
1269 if (help)
1270 fprintf(stderr, "\nArguments:\n"
1271 "\t-c\tSpecify a readonly CD-ROM image file to be used by the kernel.\n"
1272 "\t-e\tSpecify an environment to be used by the kernel.\n"
1273 "\t-h\tThis list of options.\n"
1274 "\t-i\tSpecify a memory image file to be used by the virtual kernel.\n"
1275 "\t-I\tCreate a virtual network device.\n"
1276 "\t-l\tSpecify which, if any, real CPUs to lock virtual CPUs to.\n"
1277 "\t-m\tSpecify the amount of memory to be used by the kernel in bytes.\n"
1278 "\t-n\tSpecify the number of CPUs you wish to emulate.\n"
1279 "\t-p\tSpecify a file in which to store the process ID.\n"
1280 "\t-r\tSpecify a R/W disk image file to be used by the kernel.\n"
1281 "\t-s\tBoot into single-user mode.\n"
1282 "\t-U\tEnable writing to kernel memory and module loading.\n"
1283 "\t-v\tTurn on verbose booting.\n");
1284
1285 exit(EX_USAGE);
9c059ae3
MD
1286}
1287
8a9d8995
MD
1288void
1289cpu_reset(void)
1290{
40d0276c
MD
1291 kprintf("cpu reset, rebooting vkernel\n");
1292 closefrom(3);
b02b4170 1293 cleanpid();
40d0276c 1294 execv(save_av[0], save_av);
8a9d8995 1295}
6f7b98e0
MD
1296
1297void
1298cpu_halt(void)
1299{
40d0276c 1300 kprintf("cpu halt, exiting vkernel\n");
b02b4170 1301 cleanpid();
278c4e67 1302 exit(EX_OK);
6f7b98e0 1303}
702acf06
JT
1304
1305void
1306setrealcpu(void)
1307{
1308 switch(lwp_cpu_lock) {
1309 case LCL_PER_CPU:
1310 if (bootverbose)
1311 kprintf("Locking CPU%d to real cpu %d\n",
1312 mycpuid, next_cpu);
1313 usched_set(getpid(), USCHED_SET_CPU, &next_cpu, sizeof(next_cpu));
1314 next_cpu++;
1315 if (next_cpu >= real_ncpus)
1316 next_cpu = 0;
1317 break;
1318 case LCL_SINGLE_CPU:
1319 if (bootverbose)
1320 kprintf("Locking CPU%d to real cpu %d\n",
1321 mycpuid, next_cpu);
1322 usched_set(getpid(), USCHED_SET_CPU, &next_cpu, sizeof(next_cpu));
1323 break;
1324 default:
1325 /* do not map virtual cpus to real cpus */
1326 break;
1327 }
1328}
1329