2 * Copyright (c) 1990 The Regents of the University of California.
5 * This code is derived from software contributed to Berkeley by
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
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 the
15 * documentation and/or other materials provided with the distribution.
16 * 3. All advertising materials mentioning features or use of this software
17 * must display the following acknowledgement:
18 * This product includes software developed by the University of
19 * California, Berkeley and its contributors.
20 * 4. Neither the name of the University nor the names of its contributors
21 * may be used to endorse or promote products derived from this software
22 * without specific prior written permission.
24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36 * from: @(#)autoconf.c 7.1 (Berkeley) 5/9/91
37 * $FreeBSD: src/sys/i386/i386/autoconf.c,v 1.146.2.2 2001/06/07 06:05:58 dd Exp $
41 * Setup the system to run on the current machine.
43 * Configure() is called at boot time and initializes the vba
44 * device tables and the memory controller monitoring. Available
45 * devices are determined (from possibilities mentioned in ioconf.c),
46 * and the drivers are initialized.
48 #include "opt_bootp.h"
50 #include "opt_cd9660.h"
52 #include "opt_nfsroot.h"
53 #include "opt_rootdevname.h"
57 #include <sys/param.h>
58 #include <sys/systm.h>
59 #include <sys/bootmaj.h>
63 #include <sys/diskslice.h>
64 #include <sys/reboot.h>
65 #include <sys/kernel.h>
66 #include <sys/malloc.h>
67 #include <sys/mount.h>
69 #include <sys/thread.h>
70 #include <sys/device.h>
71 #include <sys/machintr.h>
74 #include <vm/vm_kern.h>
75 #include <vm/vm_extern.h>
76 #include <vm/vm_pager.h>
79 #include <machine/pcb.h>
80 #include <machine/pcb_ext.h>
81 #include <machine/vm86.h>
83 #include <machine/smp.h>
84 #include <machine/globaldata.h>
85 #include <machine/md_var.h>
88 #include <bus/isa/isavar.h>
90 device_t isa_bus_device = NULL;
93 static void cpu_startup (void *);
94 static void configure_first (void *);
95 static void configure (void *);
96 static void configure_final (void *);
98 #if defined(FFS) && defined(FFS_ROOT)
99 static void setroot (void);
102 #if defined(NFS) && defined(NFS_ROOT)
103 #if !defined(BOOTP_NFSROOT)
104 static void pxe_setup_nfsdiskless(void);
108 SYSINIT(cpu, SI_BOOT2_START_CPU, SI_ORDER_FIRST, cpu_startup, NULL);
109 SYSINIT(configure1, SI_SUB_CONFIGURE, SI_ORDER_FIRST, configure_first, NULL);
110 /* SI_ORDER_SECOND is hookable */
111 SYSINIT(configure2, SI_SUB_CONFIGURE, SI_ORDER_THIRD, configure, NULL);
112 /* SI_ORDER_MIDDLE is hookable */
113 SYSINIT(configure3, SI_SUB_CONFIGURE, SI_ORDER_ANY, configure_final, NULL);
115 cdev_t rootdev = NULL;
116 cdev_t dumpdev = NULL;
122 cpu_startup(void *dummy)
124 vm_offset_t buffer_sva;
125 vm_offset_t buffer_eva;
126 vm_offset_t pager_sva;
127 vm_offset_t pager_eva;
129 kprintf("%s", version);
130 kprintf("real memory = %ju (%ju MB)\n",
131 (uintmax_t)ptoa(Maxmem),
132 (uintmax_t)ptoa(Maxmem) / 1024 / 1024);
135 int factor = 4 * BKVASIZE / 1024;
136 int kbytes = Maxmem * (PAGE_SIZE / 1024);
140 nbuf += min((kbytes - 4096) / factor, 65536 / factor);
142 nbuf += (kbytes - 65536) * 2 / (factor * 5);
143 if (maxbcache && nbuf > maxbcache / BKVASIZE)
144 nbuf = maxbcache / BKVASIZE;
146 if (nbuf > (virtual_end - virtual_start) / (BKVASIZE * 2)) {
147 nbuf = (virtual_end - virtual_start) / (BKVASIZE * 2);
148 kprintf("Warning: nbufs capped at %d\n", nbuf);
151 nswbuf = lmax(lmin(nbuf / 4, 256), 16);
153 if (nswbuf < NSWBUF_MIN)
158 * Allocate memory for the buffer cache
160 buf = (void *)kmem_alloc(&kernel_map, nbuf * sizeof(struct buf));
161 swbuf = (void *)kmem_alloc(&kernel_map, nswbuf * sizeof(struct buf));
167 kmem_suballoc(&kernel_map, &clean_map, &clean_sva, &clean_eva,
168 (nbuf*BKVASIZE) + (nswbuf*MAXPHYS) + pager_map_size);
169 kmem_suballoc(&clean_map, &buffer_map, &buffer_sva, &buffer_eva,
171 buffer_map.system_map = 1;
172 kmem_suballoc(&clean_map, &pager_map, &pager_sva, &pager_eva,
173 (nswbuf*MAXPHYS) + pager_map_size);
174 pager_map.system_map = 1;
175 #if defined(USERCONFIG)
177 cninit(); /* the preferred console may have changed */
179 kprintf("avail memory = %ju (%ju MB)\n",
180 (uintmax_t)ptoa(vmstats.v_free_count),
181 (uintmax_t)ptoa(vmstats.v_free_count) / 1024 / 1024);
183 vm_pager_bufferinit();
192 * Determine i/o configuration for a machine.
195 configure_first(void *dummy)
200 configure(void *dummy)
203 * Final interrupt support acviation, then enable hardware interrupts.
205 MachIntrABI.finalize();
209 * This will configure all devices, generally starting with the
210 * nexus (i386/i386/nexus.c). The nexus ISA code explicitly
211 * dummies up the attach in order to delay legacy initialization
212 * until after all other busses/subsystems have had a chance
213 * at those resources.
215 root_bus_configure();
219 * Explicitly probe and attach ISA last. The isa bus saves
220 * it's device node at attach time for us here.
223 isa_probe_children(isa_bus_device);
227 * Allow lowering of the ipl to the lowest kernel level if we
228 * panic (or call tsleep() before clearing `cold'). No level is
229 * completely safe (since a panic may occur in a critical region
230 * at splhigh()), but we want at least bio interrupts to work.
232 safepri = TDPRI_KERN_USER;
236 configure_final(void *dummy)
241 kprintf("Device configuration finished.\n");
245 void bootpc_init(void);
248 * Do legacy root filesystem discovery.
256 #if defined(NFS) && defined(NFS_ROOT)
257 #if !defined(BOOTP_NFSROOT)
258 pxe_setup_nfsdiskless();
259 if (nfs_diskless_valid)
261 rootdevnames[0] = "nfs:";
263 #if defined(FFS) && defined(FFS_ROOT)
264 if (!rootdevnames[0])
268 SYSINIT(cpu_rootconf, SI_SUB_ROOT_CONF, SI_ORDER_FIRST, cpu_rootconf, NULL)
270 u_long bootdev = 0; /* not a cdev_t - encoding is different */
272 #if defined(FFS) && defined(FFS_ROOT)
275 * The boot code uses old block device major numbers to pass bootdev to
276 * us. We have to translate these to character device majors because
277 * we don't have block devices any more.
280 boot_translate_majdev(int bmajor)
282 static int conv[] = { BOOTMAJOR_CONVARY };
284 if (bmajor >= 0 && bmajor < NELEM(conv))
285 return(conv[bmajor]);
290 * Attempt to find the device from which we were booted.
291 * If we can do so, and not instructed not to do so,
292 * set rootdevs[] and rootdevnames[] to correspond to the
295 * This code survives in order to allow the system to be
296 * booted from legacy environments that do not correctly
297 * populate the kernel environment. There are significant
298 * restrictions on the bootability of the system in this
299 * situation; it can only be mounting root from a 'da'
300 * 'wd' or 'fd' device, and the root filesystem must be ufs.
305 int majdev, mindev, unit, slice, part;
306 cdev_t newrootdev, dev;
310 if ((bootdev & B_MAGICMASK) != B_DEVMAGIC) {
311 kprintf("no B_DEVMAGIC (bootdev=%#lx)\n", bootdev);
314 majdev = boot_translate_majdev(B_TYPE(bootdev));
316 kprintf("bootdev: %08lx type=%ld unit=%ld "
317 "slice=%ld part=%ld major=%d\n",
318 bootdev, B_TYPE(bootdev), B_UNIT(bootdev),
319 B_SLICE(bootdev), B_PARTITION(bootdev), majdev);
321 dev = udev2dev(makeudev(majdev, 0), 0);
322 if (!dev_is_good(dev))
324 unit = B_UNIT(bootdev);
325 slice = B_SLICE(bootdev);
326 if (slice == WHOLE_DISK_SLICE)
327 slice = COMPATIBILITY_SLICE;
328 if (slice < 0 || slice >= MAX_SLICES) {
329 kprintf("bad slice\n");
333 part = B_PARTITION(bootdev);
334 mindev = dkmakeminor(unit, slice, part);
335 newrootdev = udev2dev(makeudev(majdev, mindev), 0);
336 if (!dev_is_good(newrootdev))
338 sname = dsname(newrootdev, unit, slice, part, partname);
339 rootdevnames[0] = kmalloc(strlen(sname) + 6, M_DEVBUF, M_WAITOK);
340 ksprintf(rootdevnames[0], "ufs:%s%s", sname, partname);
343 * For properly dangerously dedicated disks (ones with a historical
344 * bogus partition table), the boot blocks will give slice = 4, but
345 * the kernel will only provide the compatibility slice since it
346 * knows that slice 4 is not a real slice. Arrange to try mounting
347 * the compatibility slice as root if mounting the slice passed by
348 * the boot blocks fails. This handles the dangerously dedicated
349 * case and perhaps others.
351 if (slice == COMPATIBILITY_SLICE)
353 slice = COMPATIBILITY_SLICE;
354 sname = dsname(newrootdev, unit, slice, part, partname);
355 rootdevnames[1] = kmalloc(strlen(sname) + 6, M_DEVBUF, M_WAITOK);
356 ksprintf(rootdevnames[1], "ufs:%s%s", sname, partname);
360 #if defined(NFS) && defined(NFS_ROOT)
361 #if !defined(BOOTP_NFSROOT)
363 #include <sys/socket.h>
365 #include <net/if_dl.h>
366 #include <net/if_types.h>
367 #include <net/if_var.h>
368 #include <net/ethernet.h>
369 #include <netinet/in.h>
370 #include <vfs/nfs/rpcv2.h>
371 #include <vfs/nfs/nfsproto.h>
372 #include <vfs/nfs/nfs.h>
373 #include <vfs/nfs/nfsdiskless.h>
375 extern struct nfs_diskless nfs_diskless;
378 * Convert a kenv variable to a sockaddr. If the kenv variable does not
379 * exist the sockaddr will remain zerod out (callers typically just check
380 * sin_len). A network address of 0.0.0.0 is equivalent to failure.
383 inaddr_to_sockaddr(char *ev, struct sockaddr_in *sa)
388 bzero(sa, sizeof(*sa));
390 if ((cp = kgetenv(ev)) == NULL)
392 if (ksscanf(cp, "%d.%d.%d.%d", &a[0], &a[1], &a[2], &a[3]) != 4)
394 if (a[0] == 0 && a[1] == 0 && a[2] == 0 && a[3] == 0)
396 /* XXX is this ordering correct? */
397 sa->sin_addr.s_addr = (a[3] << 24) + (a[2] << 16) + (a[1] << 8) + a[0];
398 sa->sin_len = sizeof(*sa);
399 sa->sin_family = AF_INET;
404 hwaddr_to_sockaddr(char *ev, struct sockaddr_dl *sa)
409 bzero(sa, sizeof(*sa));
410 sa->sdl_len = sizeof(*sa);
411 sa->sdl_family = AF_LINK;
412 sa->sdl_type = IFT_ETHER;
413 sa->sdl_alen = ETHER_ADDR_LEN;
414 if ((cp = kgetenv(ev)) == NULL)
416 if (ksscanf(cp, "%x:%x:%x:%x:%x:%x", &a[0], &a[1], &a[2], &a[3], &a[4], &a[5]) != 6)
418 sa->sdl_data[0] = a[0];
419 sa->sdl_data[1] = a[1];
420 sa->sdl_data[2] = a[2];
421 sa->sdl_data[3] = a[3];
422 sa->sdl_data[4] = a[4];
423 sa->sdl_data[5] = a[5];
428 decode_nfshandle(char *ev, u_char *fh)
433 if (((cp = kgetenv(ev)) == NULL) || (strlen(cp) < 2) || (*cp != 'X'))
440 if ((ksscanf(cp, "%2x", &val) != 1) || (val > 0xff))
451 * Populate the essential fields in the nfsv3_diskless structure.
453 * The loader is expected to export the following environment variables:
455 * boot.netif.ip IP address on boot interface
456 * boot.netif.netmask netmask on boot interface
457 * boot.netif.gateway default gateway (optional)
458 * boot.netif.hwaddr hardware address of boot interface
459 * boot.netif.name name of boot interface (instead of hw addr)
460 * boot.nfsroot.server IP address of root filesystem server
461 * boot.nfsroot.path path of the root filesystem on server
462 * boot.nfsroot.nfshandle NFS handle for root filesystem on server
465 pxe_setup_nfsdiskless(void)
467 struct nfs_diskless *nd = &nfs_diskless;
470 struct sockaddr_dl *sdl, ourdl;
471 struct sockaddr_in myaddr, netmask;
474 /* set up interface */
475 if (inaddr_to_sockaddr("boot.netif.ip", &myaddr))
477 if (inaddr_to_sockaddr("boot.netif.netmask", &netmask)) {
478 kprintf("PXE: no netmask\n");
481 bcopy(&myaddr, &nd->myif.ifra_addr, sizeof(myaddr));
482 bcopy(&myaddr, &nd->myif.ifra_broadaddr, sizeof(myaddr));
483 ((struct sockaddr_in *) &nd->myif.ifra_broadaddr)->sin_addr.s_addr =
484 myaddr.sin_addr.s_addr | ~ netmask.sin_addr.s_addr;
485 bcopy(&netmask, &nd->myif.ifra_mask, sizeof(netmask));
487 if ((cp = kgetenv("boot.netif.name")) != NULL) {
488 TAILQ_FOREACH(ifp, &ifnet, if_link) {
489 if (strcmp(cp, ifp->if_xname) == 0)
494 kprintf("PXE: cannot find interface %s\n", cp);
498 if (hwaddr_to_sockaddr("boot.netif.hwaddr", &ourdl)) {
499 kprintf("PXE: no hardware address\n");
503 TAILQ_FOREACH(ifp, &ifnet, if_link) {
504 struct ifaddr_container *ifac;
506 TAILQ_FOREACH(ifac, &ifp->if_addrheads[mycpuid], ifa_link) {
509 if ((ifa->ifa_addr->sa_family == AF_LINK) &&
510 (sdl = ((struct sockaddr_dl *)ifa->ifa_addr))) {
511 if ((sdl->sdl_type == ourdl.sdl_type) &&
512 (sdl->sdl_alen == ourdl.sdl_alen) &&
513 !bcmp(sdl->sdl_data + sdl->sdl_nlen,
514 ourdl.sdl_data + ourdl.sdl_nlen,
520 kprintf("PXE: no interface\n");
521 return; /* no matching interface */
523 strlcpy(nd->myif.ifra_name, ifp->if_xname, sizeof(nd->myif.ifra_name));
526 inaddr_to_sockaddr("boot.netif.gateway", &nd->mygateway);
528 /* XXX set up swap? */
530 /* set up root mount */
531 nd->root_args.rsize = 8192; /* XXX tunable? */
532 nd->root_args.wsize = 8192;
533 nd->root_args.sotype = SOCK_STREAM;
534 nd->root_args.flags = NFSMNT_WSIZE | NFSMNT_RSIZE | NFSMNT_RESVPORT;
535 if (inaddr_to_sockaddr("boot.nfsroot.server", &nd->root_saddr)) {
536 kprintf("PXE: no server\n");
539 nd->root_saddr.sin_port = htons(NFS_PORT);
542 * A tftp-only loader may pass NFS path information without a
543 * root handle. Generate a warning but continue configuring.
545 if (decode_nfshandle("boot.nfsroot.nfshandle", &nd->root_fh[0]) == 0) {
546 kprintf("PXE: Warning, no NFS handle passed from loader\n");
548 if ((cp = kgetenv("boot.nfsroot.path")) != NULL)
549 strncpy(nd->root_hostnam, cp, MNAMELEN - 1);
551 nfs_diskless_valid = 1;