?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
sbin/hammer: Directly access volume in volume list
[dragonfly.git] / sys / emulation / linux / linux_misc.c
blob3d1575029ebf09f0e8b5b8ca6704ea05cc0a31e8
1 /*-
2 * Copyright (c) 1994-1995 Søren Schmidt
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer
10 * in this position and unchanged.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 * 3. The name of the author may not be used to endorse or promote products
15 * derived from this software without specific prior written permission
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
18 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
19 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
20 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
21 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
22 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
23 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
24 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
25 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
26 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27 *
28 * $FreeBSD: src/sys/compat/linux/linux_misc.c,v 1.85.2.9 2002/09/24 08:11:41 mdodd Exp $
29 */
30
31 #include "opt_compat.h"
32
33 #include <sys/param.h>
34 #include <sys/systm.h>
35 #include <sys/fcntl.h>
36 #include <sys/imgact_aout.h>
37 #include <sys/kernel.h>
38 #include <sys/kern_syscall.h>
39 #include <sys/lock.h>
40 #include <sys/mman.h>
41 #include <sys/mount.h>
42 #include <sys/poll.h>
43 #include <sys/proc.h>
44 #include <sys/priv.h>
45 #include <sys/nlookup.h>
46 #include <sys/blist.h>
47 #include <sys/reboot.h>
48 #include <sys/resourcevar.h>
49 #include <sys/signalvar.h>
50 #include <sys/stat.h>
51 #include <sys/sysctl.h>
52 #include <sys/sysproto.h>
53 #include <sys/time.h>
54 #include <sys/unistd.h>
55 #include <sys/vmmeter.h>
56 #include <sys/vnode.h>
57 #include <sys/wait.h>
58
59 #include <sys/signal2.h>
60 #include <sys/thread2.h>
61 #include <sys/mplock2.h>
62 #include <sys/spinlock2.h>
63
64 #include <vm/vm.h>
65 #include <vm/pmap.h>
66 #include <vm/vm_kern.h>
67 #include <vm/vm_map.h>
68 #include <vm/vm_extern.h>
69 #include <vm/vm_object.h>
70 #include <vm/vm_zone.h>
71 #include <vm/swap_pager.h>
72
73 #include <machine/frame.h>
74 #include <machine/limits.h>
75 #include <machine/psl.h>
76 #include <machine/sysarch.h>
77 #ifdef __i386__
78 #include <machine/segments.h>
79 #endif
80
81 #include <sys/sched.h>
82
83 #include <emulation/linux/linux_sysproto.h>
84 #include <arch_linux/linux.h>
85 #include <arch_linux/linux_proto.h>
86 #include "linux_mib.h"
87 #include "linux_util.h"
88 #include "linux_emuldata.h"
89 #include "i386/linux.h"
90
91 #define BSD_TO_LINUX_SIGNAL(sig) \
92 (((sig) <= LINUX_SIGTBLSZ) ? bsd_to_linux_signal[_SIG_IDX(sig)] : sig)
93
94 static unsigned int linux_to_bsd_resource[LINUX_RLIM_NLIMITS] = {
95 RLIMIT_CPU, RLIMIT_FSIZE, RLIMIT_DATA, RLIMIT_STACK,
96 RLIMIT_CORE, RLIMIT_RSS, RLIMIT_NPROC, RLIMIT_NOFILE,
97 RLIMIT_MEMLOCK, -1
98 };
99
100 struct l_sysinfo {
101 l_long uptime; /* Seconds since boot */
102 l_ulong loads[3]; /* 1, 5, and 15 minute load averages */
103 l_ulong totalram; /* Total usable main memory size */
104 l_ulong freeram; /* Available memory size */
105 l_ulong sharedram; /* Amount of shared memory */
106 l_ulong bufferram; /* Memory used by buffers */
107 l_ulong totalswap; /* Total swap space size */
108 l_ulong freeswap; /* swap space still available */
109 l_ushort procs; /* Number of current processes */
110 l_ushort pad; /* explicit padding */
111 l_ulong totalhigh; /* Total high memory size */
112 l_ulong freehigh; /* Available high memory size */
113 l_uint mem_unit; /* Memory unit size in bytes */
114 char _f[20-2*sizeof(l_long)-sizeof(l_int)]; /* Padding for libc5 */
115 };
116
117 int
118 sys_linux_madvise(struct linux_madvise_args *args)
119 {
120 return 0;
121 }
122
123 /*
124 * MPALMOSTSAFE
125 */
126 int
127 sys_linux_sysinfo(struct linux_sysinfo_args *args)
128 {
129 struct l_sysinfo sysinfo;
130 struct timespec ts;
131 int error;
132 int i;
133 int n;
134
135 /* Uptime is copied out of print_uptime() in kern_shutdown.c */
136 getnanouptime(&ts);
137 i = 0;
138 if (ts.tv_sec >= 86400) {
139 ts.tv_sec %= 86400;
140 i = 1;
141 }
142 if (i || ts.tv_sec >= 3600) {
143 ts.tv_sec %= 3600;
144 i = 1;
145 }
146 if (i || ts.tv_sec >= 60) {
147 ts.tv_sec %= 60;
148 i = 1;
149 }
150
151 bzero(&sysinfo, sizeof(sysinfo));
152 sysinfo.uptime=ts.tv_sec;
153
154 /* Use the information from the mib to get our load averages */
155 for (i = 0; i < 3; i++)
156 sysinfo.loads[i] = averunnable.ldavg[i];
157
158 sysinfo.totalram = Maxmem * PAGE_SIZE;
159 sysinfo.freeram = sysinfo.totalram - vmstats.v_wire_count * PAGE_SIZE;
160 sysinfo.sharedram = 0;
161
162 for (n = 0; n < ncpus; ++n) {
163 globaldata_t gd = globaldata_find(n);
164
165 sysinfo.sharedram += gd->gd_vmtotal.t_avmshr;
166 }
167 sysinfo.sharedram *= PAGE_SIZE;
168 sysinfo.bufferram = 0;
169
170 if (swapblist == NULL) {
171 sysinfo.totalswap= 0;
172 sysinfo.freeswap = 0;
173 } else {
174 sysinfo.totalswap = swapblist->bl_blocks * 1024;
175 sysinfo.freeswap = swapblist->bl_root->u.bmu_avail * PAGE_SIZE;
176 }
177
178 sysinfo.procs = nprocs;
179 sysinfo.totalhigh = 0;
180 sysinfo.freehigh = 0;
181 sysinfo.mem_unit = 1; /* Set the basic mem unit to 1 */
182
183 error = copyout(&sysinfo, (caddr_t)args->info, sizeof(sysinfo));
184 return (error);
185 }
186
187 /*
188 * MPALMOSTSAFE
189 */
190 int
191 sys_linux_alarm(struct linux_alarm_args *args)
192 {
193 struct thread *td = curthread;
194 struct proc *p = td->td_proc;
195 struct itimerval it, old_it;
196 struct timeval tv;
197
198 #ifdef DEBUG
199 if (ldebug(alarm))
200 kprintf(ARGS(alarm, "%u"), args->secs);
201 #endif
202
203 if (args->secs > 100000000)
204 return EINVAL;
205
206 it.it_value.tv_sec = (long)args->secs;
207 it.it_value.tv_usec = 0;
208 it.it_interval.tv_sec = 0;
209 it.it_interval.tv_usec = 0;
210 get_mplock();
211 crit_enter();
212 old_it = p->p_realtimer;
213 getmicrouptime(&tv);
214 if (timevalisset(&old_it.it_value))
215 callout_stop(&p->p_ithandle);
216 if (it.it_value.tv_sec != 0) {
217 callout_reset(&p->p_ithandle, tvtohz_high(&it.it_value),
218 realitexpire, p);
219 timevaladd(&it.it_value, &tv);
220 }
221 p->p_realtimer = it;
222 crit_exit();
223 rel_mplock();
224 if (timevalcmp(&old_it.it_value, &tv, >)) {
225 timevalsub(&old_it.it_value, &tv);
226 if (old_it.it_value.tv_usec != 0)
227 old_it.it_value.tv_sec++;
228 args->sysmsg_result = old_it.it_value.tv_sec;
229 }
230 return 0;
231 }
232
233 /*
234 * MPALMOSTSAFE
235 */
236 int
237 sys_linux_brk(struct linux_brk_args *args)
238 {
239 struct thread *td = curthread;
240 struct proc *p = td->td_proc;
241 struct vmspace *vm;
242 vm_offset_t new, old;
243 struct obreak_args bsd_args;
244
245 get_mplock();
246 vm = p->p_vmspace;
247 #ifdef DEBUG
248 if (ldebug(brk))
249 kprintf(ARGS(brk, "%p"), (void *)args->dsend);
250 #endif
251 old = (vm_offset_t)vm->vm_daddr + ctob(vm->vm_dsize);
252 new = (vm_offset_t)args->dsend;
253 bsd_args.sysmsg_result = 0;
254 bsd_args.nsize = (char *) new;
255 bsd_args.sysmsg_result = 0;
256 if (((caddr_t)new > vm->vm_daddr) && !sys_obreak(&bsd_args))
257 args->sysmsg_result = (long)new;
258 else
259 args->sysmsg_result = (long)old;
260 rel_mplock();
261
262 return 0;
263 }
264
265 /*
266 * MPALMOSTSAFE
267 */
268 int
269 sys_linux_uselib(struct linux_uselib_args *args)
270 {
271 struct thread *td = curthread;
272 struct proc *p;
273 struct nlookupdata nd;
274 struct vnode *vp;
275 struct exec *a_out;
276 struct vattr attr;
277 vm_offset_t vmaddr;
278 unsigned long file_offset;
279 vm_offset_t buffer;
280 unsigned long bss_size;
281 int error;
282 int locked;
283 char *path;
284
285 p = td->td_proc;
286
287 error = linux_copyin_path(args->library, &path, LINUX_PATH_EXISTS);
288 if (error)
289 return (error);
290 #ifdef DEBUG
291 if (ldebug(uselib))
292 kprintf(ARGS(uselib, "%s"), path);
293 #endif
294
295 a_out = NULL;
296 locked = 0;
297 vp = NULL;
298
299 get_mplock();
300 error = nlookup_init(&nd, path, UIO_SYSSPACE, NLC_FOLLOW);
301 nd.nl_flags |= NLC_EXEC;
302 if (error == 0)
303 error = nlookup(&nd);
304 if (error == 0)
305 error = cache_vget(&nd.nl_nch, nd.nl_cred, LK_EXCLUSIVE, &vp);
306 if (error)
307 goto cleanup;
308 /*
309 * From here on down, we have a locked vnode that must be unlocked.
310 */
311 locked = 1;
312
313 /* Writable? */
314 if (vp->v_writecount) {
315 error = ETXTBSY;
316 goto cleanup;
317 }
318
319 /* Executable? */
320 error = VOP_GETATTR(vp, &attr);
321 if (error)
322 goto cleanup;
323
324 if ((vp->v_mount->mnt_flag & MNT_NOEXEC) ||
325 ((attr.va_mode & 0111) == 0) || (attr.va_type != VREG)) {
326 error = ENOEXEC;
327 goto cleanup;
328 }
329
330 /* Sensible size? */
331 if (attr.va_size == 0) {
332 error = ENOEXEC;
333 goto cleanup;
334 }
335
336 error = VOP_OPEN(vp, FREAD, td->td_ucred, NULL);
337 if (error)
338 goto cleanup;
339
340 /*
341 * Lock no longer needed
342 */
343 vn_unlock(vp);
344 locked = 0;
345
346 /* Pull in executable header into kernel_map */
347 error = vm_mmap(&kernel_map, (vm_offset_t *)&a_out, PAGE_SIZE,
348 VM_PROT_READ, VM_PROT_READ, 0, (caddr_t)vp, 0);
349 if (error)
350 goto cleanup;
351
352 /* Is it a Linux binary ? */
353 if (((a_out->a_magic >> 16) & 0xff) != 0x64) {
354 error = ENOEXEC;
355 goto cleanup;
356 }
357
358 /*
359 * While we are here, we should REALLY do some more checks
360 */
361
362 /* Set file/virtual offset based on a.out variant. */
363 switch ((int)(a_out->a_magic & 0xffff)) {
364 case 0413: /* ZMAGIC */
365 file_offset = 1024;
366 break;
367 case 0314: /* QMAGIC */
368 file_offset = 0;
369 break;
370 default:
371 error = ENOEXEC;
372 goto cleanup;
373 }
374
375 bss_size = round_page(a_out->a_bss);
376
377 /* Check various fields in header for validity/bounds. */
378 if (a_out->a_text & PAGE_MASK || a_out->a_data & PAGE_MASK) {
379 error = ENOEXEC;
380 goto cleanup;
381 }
382
383 /* text + data can't exceed file size */
384 if (a_out->a_data + a_out->a_text > attr.va_size) {
385 error = EFAULT;
386 goto cleanup;
387 }
388
389 /*
390 * text/data/bss must not exceed limits
391 * XXX - this is not complete. it should check current usage PLUS
392 * the resources needed by this library.
393 */
394 if (a_out->a_text > maxtsiz ||
395 a_out->a_data + bss_size > p->p_rlimit[RLIMIT_DATA].rlim_cur) {
396 error = ENOMEM;
397 goto cleanup;
398 }
399
400 /* prevent more writers */
401 vsetflags(vp, VTEXT);
402
403 /*
404 * Check if file_offset page aligned. Currently we cannot handle
405 * misalinged file offsets, and so we read in the entire image
406 * (what a waste).
407 */
408 if (file_offset & PAGE_MASK) {
409 #ifdef DEBUG
410 kprintf("uselib: Non page aligned binary %lu\n", file_offset);
411 #endif
412 /* Map text+data read/write/execute */
413
414 /* a_entry is the load address and is page aligned */
415 vmaddr = trunc_page(a_out->a_entry);
416
417 /* get anon user mapping, read+write+execute */
418 error = vm_map_find(&p->p_vmspace->vm_map, NULL, NULL,
419 0, &vmaddr, a_out->a_text + a_out->a_data,
420 PAGE_SIZE,
421 FALSE, VM_MAPTYPE_NORMAL,
422 VM_PROT_ALL, VM_PROT_ALL, 0);
423 if (error)
424 goto cleanup;
425
426 /* map file into kernel_map */
427 error = vm_mmap(&kernel_map, &buffer,
428 round_page(a_out->a_text + a_out->a_data + file_offset),
429 VM_PROT_READ, VM_PROT_READ, 0, (caddr_t)vp,
430 trunc_page(file_offset));
431 if (error)
432 goto cleanup;
433
434 /* copy from kernel VM space to user space */
435 error = copyout((caddr_t)(uintptr_t)(buffer + file_offset),
436 (caddr_t)vmaddr, a_out->a_text + a_out->a_data);
437
438 /* release temporary kernel space */
439 vm_map_remove(&kernel_map, buffer, buffer +
440 round_page(a_out->a_text + a_out->a_data + file_offset));
441
442 if (error)
443 goto cleanup;
444 } else {
445 #ifdef DEBUG
446 kprintf("uselib: Page aligned binary %lu\n", file_offset);
447 #endif
448 /*
449 * for QMAGIC, a_entry is 20 bytes beyond the load address
450 * to skip the executable header
451 */
452 vmaddr = trunc_page(a_out->a_entry);
453
454 /*
455 * Map it all into the process's space as a single
456 * copy-on-write "data" segment.
457 */
458 error = vm_mmap(&p->p_vmspace->vm_map, &vmaddr,
459 a_out->a_text + a_out->a_data, VM_PROT_ALL, VM_PROT_ALL,
460 MAP_PRIVATE | MAP_FIXED, (caddr_t)vp, file_offset);
461 if (error)
462 goto cleanup;
463 }
464 #ifdef DEBUG
465 kprintf("mem=%08lx = %08lx %08lx\n", (long)vmaddr, ((long*)vmaddr)[0],
466 ((long*)vmaddr)[1]);
467 #endif
468 if (bss_size != 0) {
469 /* Calculate BSS start address */
470 vmaddr = trunc_page(a_out->a_entry) + a_out->a_text +
471 a_out->a_data;
472
473 /* allocate some 'anon' space */
474 error = vm_map_find(&p->p_vmspace->vm_map, NULL, NULL,
475 0, &vmaddr, bss_size,
476 PAGE_SIZE,
477 FALSE, VM_MAPTYPE_NORMAL,
478 VM_PROT_ALL, VM_PROT_ALL, 0);
479 if (error)
480 goto cleanup;
481 }
482
483 cleanup:
484 /* Unlock/release vnode */
485 if (vp) {
486 if (locked)
487 vn_unlock(vp);
488 vrele(vp);
489 }
490 /* Release the kernel mapping. */
491 if (a_out) {
492 vm_map_remove(&kernel_map, (vm_offset_t)a_out,
493 (vm_offset_t)a_out + PAGE_SIZE);
494 }
495 nlookup_done(&nd);
496 rel_mplock();
497 linux_free_path(&path);
498 return (error);
499 }
500
501 /*
502 * MPSAFE
503 */
504 int
505 sys_linux_select(struct linux_select_args *args)
506 {
507 struct select_args bsa;
508 struct timeval tv0, tv1, utv, *tvp;
509 caddr_t sg;
510 int error;
511
512 #ifdef DEBUG
513 if (ldebug(select))
514 kprintf(ARGS(select, "%d, %p, %p, %p, %p"), args->nfds,
515 (void *)args->readfds, (void *)args->writefds,
516 (void *)args->exceptfds, (void *)args->timeout);
517 #endif
518
519 error = 0;
520 bsa.sysmsg_result = 0;
521 bsa.nd = args->nfds;
522 bsa.in = args->readfds;
523 bsa.ou = args->writefds;
524 bsa.ex = args->exceptfds;
525 bsa.tv = (struct timeval *)args->timeout;
526
527 /*
528 * Store current time for computation of the amount of
529 * time left.
530 */
531 if (args->timeout) {
532 if ((error = copyin((caddr_t)args->timeout, &utv,
533 sizeof(utv))))
534 goto select_out;
535 #ifdef DEBUG
536 if (ldebug(select))
537 kprintf(LMSG("incoming timeout (%ld/%ld)"),
538 utv.tv_sec, utv.tv_usec);
539 #endif
540
541 if (itimerfix(&utv)) {
542 /*
543 * The timeval was invalid. Convert it to something
544 * valid that will act as it does under Linux.
545 */
546 sg = stackgap_init();
547 tvp = stackgap_alloc(&sg, sizeof(utv));
548 utv.tv_sec += utv.tv_usec / 1000000;
549 utv.tv_usec %= 1000000;
550 if (utv.tv_usec < 0) {
551 utv.tv_sec -= 1;
552 utv.tv_usec += 1000000;
553 }
554 if (utv.tv_sec < 0)
555 timevalclear(&utv);
556 if ((error = copyout(&utv, tvp, sizeof(utv))))
557 goto select_out;
558 bsa.tv = tvp;
559 }
560 microtime(&tv0);
561 }
562
563 error = sys_select(&bsa);
564 args->sysmsg_result = bsa.sysmsg_result;
565 #ifdef DEBUG
566 if (ldebug(select))
567 kprintf(LMSG("real select returns %d"), error);
568 #endif
569 if (error) {
570 /*
571 * See fs/select.c in the Linux kernel. Without this,
572 * Maelstrom doesn't work.
573 */
574 if (error == ERESTART)
575 error = EINTR;
576 goto select_out;
577 }
578
579 if (args->timeout) {
580 if (args->sysmsg_result) {
581 /*
582 * Compute how much time was left of the timeout,
583 * by subtracting the current time and the time
584 * before we started the call, and subtracting
585 * that result from the user-supplied value.
586 */
587 microtime(&tv1);
588 timevalsub(&tv1, &tv0);
589 timevalsub(&utv, &tv1);
590 if (utv.tv_sec < 0)
591 timevalclear(&utv);
592 } else
593 timevalclear(&utv);
594 #ifdef DEBUG
595 if (ldebug(select))
596 kprintf(LMSG("outgoing timeout (%ld/%ld)"),
597 utv.tv_sec, utv.tv_usec);
598 #endif
599 if ((error = copyout(&utv, (caddr_t)args->timeout,
600 sizeof(utv))))
601 goto select_out;
602 }
603
604 select_out:
605 #ifdef DEBUG
606 if (ldebug(select))
607 kprintf(LMSG("select_out -> %d"), error);
608 #endif
609 return error;
610 }
611
612 /*
613 * MPSAFE
614 */
615 int
616 sys_linux_mremap(struct linux_mremap_args *args)
617 {
618 struct munmap_args bsd_args;
619 int error = 0;
620
621 #ifdef DEBUG
622 if (ldebug(mremap))
623 kprintf(ARGS(mremap, "%p, %08lx, %08lx, %08lx"),
624 (void *)args->addr,
625 (unsigned long)args->old_len,
626 (unsigned long)args->new_len,
627 (unsigned long)args->flags);
628 #endif
629 if (args->flags & ~(LINUX_MREMAP_FIXED | LINUX_MREMAP_MAYMOVE)) {
630 args->sysmsg_resultp = NULL;
631 return (EINVAL);
632 }
633
634 /*
635 * Check for the page alignment.
636 * Linux defines PAGE_MASK to be FreeBSD ~PAGE_MASK.
637 */
638 if (args->addr & PAGE_MASK) {
639 args->sysmsg_resultp = NULL;
640 return (EINVAL);
641 }
642
643 args->new_len = round_page(args->new_len);
644 args->old_len = round_page(args->old_len);
645
646 if (args->new_len > args->old_len) {
647 args->sysmsg_result = 0;
648 return ENOMEM;
649 }
650
651 if (args->new_len < args->old_len) {
652 bsd_args.sysmsg_result = 0;
653 bsd_args.addr = (caddr_t)(args->addr + args->new_len);
654 bsd_args.len = args->old_len - args->new_len;
655 error = sys_munmap(&bsd_args);
656 }
657
658 args->sysmsg_resultp = error ? NULL : (void *)args->addr;
659 return error;
660 }
661
662 #define LINUX_MS_ASYNC 0x0001
663 #define LINUX_MS_INVALIDATE 0x0002
664 #define LINUX_MS_SYNC 0x0004
665
666 /*
667 * MPSAFE
668 */
669 int
670 sys_linux_msync(struct linux_msync_args *args)
671 {
672 struct msync_args bsd_args;
673 int error;
674
675 bsd_args.addr = (caddr_t)args->addr;
676 bsd_args.len = args->len;
677 bsd_args.flags = args->fl & ~LINUX_MS_SYNC;
678 bsd_args.sysmsg_result = 0;
679
680 error = sys_msync(&bsd_args);
681 args->sysmsg_result = bsd_args.sysmsg_result;
682 return(error);
683 }
684
685 /*
686 * MPSAFE
687 */
688 int
689 sys_linux_time(struct linux_time_args *args)
690 {
691 struct timeval tv;
692 l_time_t tm;
693 int error;
694
695 #ifdef DEBUG
696 if (ldebug(time))
697 kprintf(ARGS(time, "*"));
698 #endif
699
700 microtime(&tv);
701 tm = tv.tv_sec;
702 if (args->tm && (error = copyout(&tm, (caddr_t)args->tm, sizeof(tm))))
703 return error;
704 args->sysmsg_lresult = tm;
705 return 0;
706 }
707
708 struct l_times_argv {
709 l_long tms_utime;
710 l_long tms_stime;
711 l_long tms_cutime;
712 l_long tms_cstime;
713 };
714
715 #define CLK_TCK 100 /* Linux uses 100 */
716
717 #define CONVTCK(r) (r.tv_sec * CLK_TCK + r.tv_usec / (1000000 / CLK_TCK))
718
719 /*
720 * MPALMOSTSAFE
721 */
722 int
723 sys_linux_times(struct linux_times_args *args)
724 {
725 struct thread *td = curthread;
726 struct proc *p = td->td_proc;
727 struct timeval tv;
728 struct l_times_argv tms;
729 struct rusage ru;
730 int error;
731
732 #ifdef DEBUG
733 if (ldebug(times))
734 kprintf(ARGS(times, "*"));
735 #endif
736
737 get_mplock();
738 calcru_proc(p, &ru);
739 rel_mplock();
740
741 tms.tms_utime = CONVTCK(ru.ru_utime);
742 tms.tms_stime = CONVTCK(ru.ru_stime);
743
744 tms.tms_cutime = CONVTCK(p->p_cru.ru_utime);
745 tms.tms_cstime = CONVTCK(p->p_cru.ru_stime);
746
747 if ((error = copyout(&tms, (caddr_t)args->buf, sizeof(tms))))
748 return error;
749
750 microuptime(&tv);
751 args->sysmsg_result = (int)CONVTCK(tv);
752 return 0;
753 }
754
755 /*
756 * MPALMOSTSAFE
757 */
758 int
759 sys_linux_newuname(struct linux_newuname_args *args)
760 {
761 struct thread *td = curthread;
762 struct l_new_utsname utsname;
763 char *osrelease, *osname;
764
765 #ifdef DEBUG
766 if (ldebug(newuname))
767 kprintf(ARGS(newuname, "*"));
768 #endif
769
770 get_mplock();
771 osname = linux_get_osname(td);
772 osrelease = linux_get_osrelease(td);
773
774 bzero(&utsname, sizeof(utsname));
775 strncpy(utsname.sysname, osname, LINUX_MAX_UTSNAME-1);
776 strncpy(utsname.nodename, hostname, LINUX_MAX_UTSNAME-1);
777 strncpy(utsname.release, osrelease, LINUX_MAX_UTSNAME-1);
778 strncpy(utsname.version, version, LINUX_MAX_UTSNAME-1);
779 strncpy(utsname.machine, machine, LINUX_MAX_UTSNAME-1);
780 strncpy(utsname.domainname, domainname, LINUX_MAX_UTSNAME-1);
781 rel_mplock();
782
783 return (copyout(&utsname, (caddr_t)args->buf, sizeof(utsname)));
784 }
785
786 /* XXX: why would this be i386-only? most of these are wrong! */
787 #if defined(__i386__)
788 struct l_utimbuf {
789 l_time_t l_actime;
790 l_time_t l_modtime;
791 };
792
793 /*
794 * MPALMOSTSAFE
795 */
796 int
797 sys_linux_utime(struct linux_utime_args *args)
798 {
799 struct timeval tv[2];
800 struct l_utimbuf lut;
801 struct nlookupdata nd;
802 char *path;
803 int error;
804
805 error = linux_copyin_path(args->fname, &path, LINUX_PATH_EXISTS);
806 if (error)
807 return (error);
808 #ifdef DEBUG
809 if (ldebug(utime))
810 kprintf(ARGS(utime, "%s, *"), path);
811 #endif
812
813 if (args->times) {
814 error = copyin(args->times, &lut, sizeof(lut));
815 if (error)
816 goto cleanup;
817 tv[0].tv_sec = lut.l_actime;
818 tv[0].tv_usec = 0;
819 tv[1].tv_sec = lut.l_modtime;
820 tv[1].tv_usec = 0;
821 }
822 get_mplock();
823 error = nlookup_init(&nd, path, UIO_SYSSPACE, NLC_FOLLOW);
824 if (error == 0)
825 error = kern_utimes(&nd, args->times ? tv : NULL);
826 nlookup_done(&nd);
827 rel_mplock();
828 cleanup:
829 linux_free_path(&path);
830 return (error);
831 }
832
833 int
834 sys_linux_utimes(struct linux_utimes_args *args)
835 {
836 l_timeval ltv[2];
837 struct timeval tv[2], *tvp = NULL;
838 struct nlookupdata nd;
839 char *path;
840 int error;
841
842 error = linux_copyin_path(args->fname, &path, LINUX_PATH_EXISTS);
843 if (error)
844 return (error);
845 #ifdef DEBUG
846 if (ldebug(utimes))
847 kprintf(ARGS(utimes, "%s, *"), path);
848 #endif
849
850 if (args->tptr) {
851 error = copyin(args->tptr, ltv, sizeof(ltv));
852 if (error)
853 goto cleanup;
854 tv[0].tv_sec = ltv[0].tv_sec;
855 tv[0].tv_usec = ltv[0].tv_usec;
856 tv[1].tv_sec = ltv[1].tv_sec;
857 tv[1].tv_usec = ltv[1].tv_usec;
858 tvp = tv;
859 }
860 get_mplock();
861 error = nlookup_init(&nd, path, UIO_SYSSPACE, NLC_FOLLOW);
862 if (error == 0)
863 error = kern_utimes(&nd, tvp);
864 nlookup_done(&nd);
865 rel_mplock();
866 cleanup:
867 linux_free_path(&path);
868 return (error);
869 }
870
871 int
872 sys_linux_futimesat(struct linux_futimesat_args *args)
873 {
874 l_timeval ltv[2];
875 struct timeval tv[2], *tvp = NULL;
876 struct file *fp;
877 struct nlookupdata nd;
878 char *path;
879 int dfd,error;
880
881 error = linux_copyin_path(args->fname, &path, LINUX_PATH_EXISTS);
882 if (error)
883 return (error);
884 #ifdef DEBUG
885 if (ldebug(futimesat))
886 kprintf(ARGS(futimesat, "%s, *"), path);
887 #endif
888 if (args->tptr) {
889 error = copyin(args->tptr, ltv, sizeof(ltv));
890 if (error)
891 goto cleanup;
892 tv[0].tv_sec = ltv[0].tv_sec;
893 tv[0].tv_usec = ltv[0].tv_usec;
894 tv[1].tv_sec = ltv[1].tv_sec;
895 tv[1].tv_usec = ltv[1].tv_usec;
896 tvp = tv;
897 }
898 dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd;
899 get_mplock();
900 error = nlookup_init_at(&nd, &fp, dfd, path, UIO_SYSSPACE, NLC_FOLLOW);
901 if (error == 0)
902 error = kern_utimes(&nd, tvp);
903 nlookup_done_at(&nd, fp);
904 rel_mplock();
905 cleanup:
906 linux_free_path(&path);
907 return (error);
908 }
909
910
911 int
912 sys_linux_utimensat(struct linux_utimensat_args *args)
913 {
914 struct l_timespec ltv[2];
915 struct timeval tv[2], *tvp = NULL;
916 struct file *fp;
917 struct nlookupdata nd;
918 char *path;
919 int dfd, flags, error = 0;
920
921 if (args->flag & ~LINUX_AT_SYMLINK_NOFOLLOW)
922 return (EINVAL);
923
924 if (args->dfd == LINUX_AT_FDCWD && args->fname == NULL)
925 return (EINVAL);
926
927 if (args->fname) {
928 error = linux_copyin_path(args->fname, &path, LINUX_PATH_EXISTS);
929 if (error)
930 return (error);
931 }
932 #ifdef DEBUG
933 if (ldebug(utimensat))
934 kprintf(ARGS(utimensat, "%s, *"), path);
935 #endif
936 if (args->tptr) {
937 error = copyin(args->tptr, ltv, sizeof(ltv));
938 if (error)
939 goto cleanup;
940
941 if (ltv[0].tv_sec == LINUX_UTIME_NOW) {
942 microtime(&tv[0]);
943 } else if (ltv[0].tv_sec == LINUX_UTIME_OMIT) {
944 /* XXX: this is not right, but will do for now */
945 microtime(&tv[0]);
946 } else {
947 tv[0].tv_sec = ltv[0].tv_sec;
948 /* XXX: we lose precision here, as we don't have ns */
949 tv[0].tv_usec = ltv[0].tv_nsec/1000;
950 }
951 if (ltv[1].tv_sec == LINUX_UTIME_NOW) {
952 microtime(&tv[1]);
953 } else if (ltv[1].tv_sec == LINUX_UTIME_OMIT) {
954 /* XXX: this is not right, but will do for now */
955 microtime(&tv[1]);
956 } else {
957 tv[1].tv_sec = ltv[1].tv_sec;
958 /* XXX: we lose precision here, as we don't have ns */
959 tv[1].tv_usec = ltv[1].tv_nsec/1000;
960 }
961 tvp = tv;
962 }
963
964 dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd;
965 flags = (args->flag & LINUX_AT_SYMLINK_NOFOLLOW) ? 0 : NLC_FOLLOW;
966
967 get_mplock();
968 if (args->fname) {
969 error = nlookup_init_at(&nd, &fp, dfd, path, UIO_SYSSPACE, flags);
970 if (error == 0)
971 error = kern_utimes(&nd, tvp);
972 nlookup_done_at(&nd, fp);
973 } else {
974 /* Thank you, Linux, for another non-standard "feature" */
975 KKASSERT(dfd != AT_FDCWD);
976 error = kern_futimes(dfd, tvp);
977 }
978 rel_mplock();
979 cleanup:
980 if (args->fname)
981 linux_free_path(&path);
982
983 return (error);
984 }
985 #endif /* __i386__ */
986
987 #define __WCLONE 0x80000000
988
989 /*
990 * MPALMOSTSAFE
991 */
992 int
993 sys_linux_waitpid(struct linux_waitpid_args *args)
994 {
995 int error, options, status;
996
997 #ifdef DEBUG
998 if (ldebug(waitpid))
999 kprintf(ARGS(waitpid, "%d, %p, %d"),
1000 args->pid, (void *)args->status, args->options);
1001 #endif
1002 options = args->options & (WNOHANG | WUNTRACED);
1003 /* WLINUXCLONE should be equal to __WCLONE, but we make sure */
1004 if (args->options & __WCLONE)
1005 options |= WLINUXCLONE;
1006
1007 error = kern_wait(args->pid, args->status ? &status : NULL, options,
1008 NULL, &args->sysmsg_result);
1009
1010 if (error == 0 && args->status) {
1011 status &= 0xffff;
1012 if (WIFSIGNALED(status))
1013 status = (status & 0xffffff80) |
1014 BSD_TO_LINUX_SIGNAL(WTERMSIG(status));
1015 else if (WIFSTOPPED(status))
1016 status = (status & 0xffff00ff) |
1017 (BSD_TO_LINUX_SIGNAL(WSTOPSIG(status)) << 8);
1018 error = copyout(&status, args->status, sizeof(status));
1019 }
1020
1021 return (error);
1022 }
1023
1024 /*
1025 * MPALMOSTSAFE
1026 */
1027 int
1028 sys_linux_wait4(struct linux_wait4_args *args)
1029 {
1030 struct thread *td = curthread;
1031 struct lwp *lp = td->td_lwp;
1032 struct rusage rusage;
1033 int error, options, status;
1034
1035 #ifdef DEBUG
1036 if (ldebug(wait4))
1037 kprintf(ARGS(wait4, "%d, %p, %d, %p"),
1038 args->pid, (void *)args->status, args->options,
1039 (void *)args->rusage);
1040 #endif
1041 options = args->options & (WNOHANG | WUNTRACED);
1042 /* WLINUXCLONE should be equal to __WCLONE, but we make sure */
1043 if (args->options & __WCLONE)
1044 options |= WLINUXCLONE;
1045
1046 error = kern_wait(args->pid, args->status ? &status : NULL, options,
1047 args->rusage ? &rusage : NULL, &args->sysmsg_result);
1048
1049 if (error == 0) {
1050 spin_lock(&lp->lwp_spin);
1051 lwp_delsig(lp, SIGCHLD);
1052 spin_unlock(&lp->lwp_spin);
1053 }
1054
1055 if (error == 0 && args->status) {
1056 status &= 0xffff;
1057 if (WIFSIGNALED(status))
1058 status = (status & 0xffffff80) |
1059 BSD_TO_LINUX_SIGNAL(WTERMSIG(status));
1060 else if (WIFSTOPPED(status))
1061 status = (status & 0xffff00ff) |
1062 (BSD_TO_LINUX_SIGNAL(WSTOPSIG(status)) << 8);
1063 error = copyout(&status, args->status, sizeof(status));
1064 }
1065 if (error == 0 && args->rusage)
1066 error = copyout(&rusage, args->rusage, sizeof(rusage));
1067
1068 return (error);
1069 }
1070
1071 /*
1072 * MPALMOSTSAFE
1073 */
1074 int
1075 sys_linux_mknod(struct linux_mknod_args *args)
1076 {
1077 struct nlookupdata nd;
1078 char *path;
1079 int error;
1080
1081 error = linux_copyin_path(args->path, &path, LINUX_PATH_CREATE);
1082 if (error)
1083 return (error);
1084 #ifdef DEBUG
1085 if (ldebug(mknod))
1086 kprintf(ARGS(mknod, "%s, %d, %d"),
1087 path, args->mode, args->dev);
1088 #endif
1089 get_mplock();
1090 error = nlookup_init(&nd, path, UIO_SYSSPACE, 0);
1091 if (error == 0) {
1092 if (args->mode & S_IFIFO) {
1093 error = kern_mkfifo(&nd, args->mode);
1094 } else {
1095 error = kern_mknod(&nd, args->mode,
1096 umajor(args->dev),
1097 uminor(args->dev));
1098 }
1099 }
1100 nlookup_done(&nd);
1101 rel_mplock();
1102
1103 linux_free_path(&path);
1104 return(error);
1105 }
1106
1107 int
1108 sys_linux_mknodat(struct linux_mknodat_args *args)
1109 {
1110 struct nlookupdata nd;
1111 struct file *fp;
1112 char *path;
1113 int dfd, error;
1114
1115 error = linux_copyin_path(args->path, &path, LINUX_PATH_CREATE);
1116 if (error)
1117 return (error);
1118 #ifdef DEBUG
1119 if (ldebug(mknod))
1120 kprintf(ARGS(mknod, "%s, %d, %d"),
1121 path, args->mode, args->dev);
1122 #endif
1123 get_mplock();
1124 dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd;
1125 error = nlookup_init_at(&nd, &fp, dfd, path, UIO_SYSSPACE, 0);
1126 if (error == 0) {
1127 if (args->mode & S_IFIFO) {
1128 error = kern_mkfifo(&nd, args->mode);
1129 } else {
1130 error = kern_mknod(&nd, args->mode,
1131 umajor(args->dev),
1132 uminor(args->dev));
1133 }
1134 }
1135 nlookup_done_at(&nd, fp);
1136 rel_mplock();
1137
1138 linux_free_path(&path);
1139 return(error);
1140 }
1141
1142 /*
1143 * UGH! This is just about the dumbest idea I've ever heard!!
1144 *
1145 * MPSAFE
1146 */
1147 int
1148 sys_linux_personality(struct linux_personality_args *args)
1149 {
1150 #ifdef DEBUG
1151 if (ldebug(personality))
1152 kprintf(ARGS(personality, "%d"), args->per);
1153 #endif
1154 if (args->per != 0)
1155 return EINVAL;
1156
1157 /* Yes Jim, it's still a Linux... */
1158 args->sysmsg_result = 0;
1159 return 0;
1160 }
1161
1162 /*
1163 * Wrappers for get/setitimer for debugging..
1164 *
1165 * MPSAFE
1166 */
1167 int
1168 sys_linux_setitimer(struct linux_setitimer_args *args)
1169 {
1170 struct setitimer_args bsa;
1171 struct itimerval foo;
1172 int error;
1173
1174 #ifdef DEBUG
1175 if (ldebug(setitimer))
1176 kprintf(ARGS(setitimer, "%p, %p"),
1177 (void *)args->itv, (void *)args->oitv);
1178 #endif
1179 bsa.which = args->which;
1180 bsa.itv = (struct itimerval *)args->itv;
1181 bsa.oitv = (struct itimerval *)args->oitv;
1182 bsa.sysmsg_result = 0;
1183 if (args->itv) {
1184 if ((error = copyin((caddr_t)args->itv, &foo, sizeof(foo))))
1185 return error;
1186 #ifdef DEBUG
1187 if (ldebug(setitimer)) {
1188 kprintf("setitimer: value: sec: %ld, usec: %ld\n",
1189 foo.it_value.tv_sec, foo.it_value.tv_usec);
1190 kprintf("setitimer: interval: sec: %ld, usec: %ld\n",
1191 foo.it_interval.tv_sec, foo.it_interval.tv_usec);
1192 }
1193 #endif
1194 }
1195 error = sys_setitimer(&bsa);
1196 args->sysmsg_result = bsa.sysmsg_result;
1197 return(error);
1198 }
1199
1200 /*
1201 * MPSAFE
1202 */
1203 int
1204 sys_linux_getitimer(struct linux_getitimer_args *args)
1205 {
1206 struct getitimer_args bsa;
1207 int error;
1208 #ifdef DEBUG
1209 if (ldebug(getitimer))
1210 kprintf(ARGS(getitimer, "%p"), (void *)args->itv);
1211 #endif
1212 bsa.which = args->which;
1213 bsa.itv = (struct itimerval *)args->itv;
1214 bsa.sysmsg_result = 0;
1215 error = sys_getitimer(&bsa);
1216 args->sysmsg_result = bsa.sysmsg_result;
1217 return(error);
1218 }
1219
1220 /*
1221 * MPSAFE
1222 */
1223 int
1224 sys_linux_nice(struct linux_nice_args *args)
1225 {
1226 struct setpriority_args bsd_args;
1227 int error;
1228
1229 bsd_args.which = PRIO_PROCESS;
1230 bsd_args.who = 0; /* current process */
1231 bsd_args.prio = args->inc;
1232 bsd_args.sysmsg_result = 0;
1233 error = sys_setpriority(&bsd_args);
1234 args->sysmsg_result = bsd_args.sysmsg_result;
1235 return(error);
1236 }
1237
1238 /*
1239 * MPALMOSTSAFE
1240 */
1241 int
1242 sys_linux_setgroups(struct linux_setgroups_args *args)
1243 {
1244 struct thread *td = curthread;
1245 struct proc *p = td->td_proc;
1246 struct ucred *newcred, *oldcred;
1247 l_gid_t linux_gidset[NGROUPS];
1248 gid_t *bsd_gidset;
1249 int ngrp, error;
1250
1251 ngrp = args->gidsetsize;
1252 oldcred = td->td_ucred;
1253
1254 /*
1255 * cr_groups[0] holds egid. Setting the whole set from
1256 * the supplied set will cause egid to be changed too.
1257 * Keep cr_groups[0] unchanged to prevent that.
1258 */
1259
1260 if ((error = priv_check_cred(oldcred, PRIV_CRED_SETGROUPS, 0)) != 0)
1261 return (error);
1262
1263 if ((u_int)ngrp >= NGROUPS)
1264 return (EINVAL);
1265
1266 get_mplock();
1267 newcred = crdup(oldcred);
1268 if (ngrp > 0) {
1269 error = copyin((caddr_t)args->grouplist, linux_gidset,
1270 ngrp * sizeof(l_gid_t));
1271 if (error) {
1272 crfree(newcred);
1273 rel_mplock();
1274 return (error);
1275 }
1276
1277 newcred->cr_ngroups = ngrp + 1;
1278
1279 bsd_gidset = newcred->cr_groups;
1280 ngrp--;
1281 while (ngrp >= 0) {
1282 bsd_gidset[ngrp + 1] = linux_gidset[ngrp];
1283 ngrp--;
1284 }
1285 } else {
1286 newcred->cr_ngroups = 1;
1287 }
1288
1289 setsugid();
1290 oldcred = p->p_ucred; /* reload, deal with threads race */
1291 p->p_ucred = newcred;
1292 crfree(oldcred);
1293 rel_mplock();
1294 return (0);
1295 }
1296
1297 /*
1298 * MPSAFE
1299 */
1300 int
1301 sys_linux_getgroups(struct linux_getgroups_args *args)
1302 {
1303 struct thread *td = curthread;
1304 struct ucred *cred;
1305 l_gid_t linux_gidset[NGROUPS];
1306 gid_t *bsd_gidset;
1307 int bsd_gidsetsz, ngrp, error;
1308
1309 cred = td->td_ucred;
1310 bsd_gidset = cred->cr_groups;
1311 bsd_gidsetsz = cred->cr_ngroups - 1;
1312
1313 /*
1314 * cr_groups[0] holds egid. Returning the whole set
1315 * here will cause a duplicate. Exclude cr_groups[0]
1316 * to prevent that.
1317 */
1318
1319 if ((ngrp = args->gidsetsize) == 0) {
1320 args->sysmsg_result = bsd_gidsetsz;
1321 return (0);
1322 }
1323
1324 if ((u_int)ngrp < bsd_gidsetsz)
1325 return (EINVAL);
1326
1327 ngrp = 0;
1328 while (ngrp < bsd_gidsetsz) {
1329 linux_gidset[ngrp] = bsd_gidset[ngrp + 1];
1330 ngrp++;
1331 }
1332
1333 if ((error = copyout(linux_gidset, args->grouplist,
1334 ngrp * sizeof(l_gid_t)))) {
1335 return (error);
1336 }
1337
1338 args->sysmsg_result = ngrp;
1339 return (0);
1340 }
1341
1342 /*
1343 * MPSAFE
1344 */
1345 int
1346 sys_linux_setrlimit(struct linux_setrlimit_args *args)
1347 {
1348 struct l_rlimit linux_rlim;
1349 struct rlimit rlim;
1350 u_int which;
1351 int error;
1352
1353 #ifdef DEBUG
1354 if (ldebug(setrlimit))
1355 kprintf(ARGS(setrlimit, "%d, %p"),
1356 args->resource, (void *)args->rlim);
1357 #endif
1358 if (args->resource >= LINUX_RLIM_NLIMITS)
1359 return (EINVAL);
1360 which = linux_to_bsd_resource[args->resource];
1361 if (which == -1)
1362 return (EINVAL);
1363
1364 error = copyin(args->rlim, &linux_rlim, sizeof(linux_rlim));
1365 if (error)
1366 return (error);
1367 rlim.rlim_cur = (rlim_t)linux_rlim.rlim_cur;
1368 rlim.rlim_max = (rlim_t)linux_rlim.rlim_max;
1369
1370 error = kern_setrlimit(which, &rlim);
1371
1372 return(error);
1373 }
1374
1375 /*
1376 * MPSAFE
1377 */
1378 int
1379 sys_linux_old_getrlimit(struct linux_old_getrlimit_args *args)
1380 {
1381 struct l_rlimit linux_rlim;
1382 struct rlimit rlim;
1383 u_int which;
1384 int error;
1385
1386 #ifdef DEBUG
1387 if (ldebug(old_getrlimit))
1388 kprintf(ARGS(old_getrlimit, "%d, %p"),
1389 args->resource, (void *)args->rlim);
1390 #endif
1391 if (args->resource >= LINUX_RLIM_NLIMITS)
1392 return (EINVAL);
1393 which = linux_to_bsd_resource[args->resource];
1394 if (which == -1)
1395 return (EINVAL);
1396
1397 error = kern_getrlimit(which, &rlim);
1398
1399 if (error == 0) {
1400 linux_rlim.rlim_cur = (l_ulong)rlim.rlim_cur;
1401 if (linux_rlim.rlim_cur == ULONG_MAX)
1402 linux_rlim.rlim_cur = LONG_MAX;
1403 linux_rlim.rlim_max = (l_ulong)rlim.rlim_max;
1404 if (linux_rlim.rlim_max == ULONG_MAX)
1405 linux_rlim.rlim_max = LONG_MAX;
1406 error = copyout(&linux_rlim, args->rlim, sizeof(linux_rlim));
1407 }
1408 return (error);
1409 }
1410
1411 /*
1412 * MPSAFE
1413 */
1414 int
1415 sys_linux_getrlimit(struct linux_getrlimit_args *args)
1416 {
1417 struct l_rlimit linux_rlim;
1418 struct rlimit rlim;
1419 u_int which;
1420 int error;
1421
1422 #ifdef DEBUG
1423 if (ldebug(getrlimit))
1424 kprintf(ARGS(getrlimit, "%d, %p"),
1425 args->resource, (void *)args->rlim);
1426 #endif
1427 if (args->resource >= LINUX_RLIM_NLIMITS)
1428 return (EINVAL);
1429 which = linux_to_bsd_resource[args->resource];
1430 if (which == -1)
1431 return (EINVAL);
1432
1433 error = kern_getrlimit(which, &rlim);
1434
1435 if (error == 0) {
1436 linux_rlim.rlim_cur = (l_ulong)rlim.rlim_cur;
1437 linux_rlim.rlim_max = (l_ulong)rlim.rlim_max;
1438 error = copyout(&linux_rlim, args->rlim, sizeof(linux_rlim));
1439 }
1440 return (error);
1441 }
1442
1443 /*
1444 * MPSAFE
1445 */
1446 int
1447 sys_linux_sched_setscheduler(struct linux_sched_setscheduler_args *args)
1448 {
1449 struct sched_setscheduler_args bsd;
1450 int error;
1451
1452 #ifdef DEBUG
1453 if (ldebug(sched_setscheduler))
1454 kprintf(ARGS(sched_setscheduler, "%d, %d, %p"),
1455 args->pid, args->policy, (const void *)args->param);
1456 #endif
1457
1458 switch (args->policy) {
1459 case LINUX_SCHED_OTHER:
1460 bsd.policy = SCHED_OTHER;
1461 break;
1462 case LINUX_SCHED_FIFO:
1463 bsd.policy = SCHED_FIFO;
1464 break;
1465 case LINUX_SCHED_RR:
1466 bsd.policy = SCHED_RR;
1467 break;
1468 default:
1469 return EINVAL;
1470 }
1471
1472 bsd.pid = args->pid;
1473 bsd.param = (struct sched_param *)args->param;
1474 bsd.sysmsg_result = 0;
1475
1476 error = sys_sched_setscheduler(&bsd);
1477 args->sysmsg_result = bsd.sysmsg_result;
1478 return(error);
1479 }
1480
1481 /*
1482 * MPSAFE
1483 */
1484 int
1485 sys_linux_sched_getscheduler(struct linux_sched_getscheduler_args *args)
1486 {
1487 struct sched_getscheduler_args bsd;
1488 int error;
1489
1490 #ifdef DEBUG
1491 if (ldebug(sched_getscheduler))
1492 kprintf(ARGS(sched_getscheduler, "%d"), args->pid);
1493 #endif
1494
1495 bsd.sysmsg_result = 0;
1496 bsd.pid = args->pid;
1497 error = sys_sched_getscheduler(&bsd);
1498 args->sysmsg_result = bsd.sysmsg_result;
1499
1500 switch (args->sysmsg_result) {
1501 case SCHED_OTHER:
1502 args->sysmsg_result = LINUX_SCHED_OTHER;
1503 break;
1504 case SCHED_FIFO:
1505 args->sysmsg_result = LINUX_SCHED_FIFO;
1506 break;
1507 case SCHED_RR:
1508 args->sysmsg_result = LINUX_SCHED_RR;
1509 break;
1510 }
1511 return error;
1512 }
1513
1514 /*
1515 * MPSAFE
1516 */
1517 int
1518 sys_linux_sched_get_priority_max(struct linux_sched_get_priority_max_args *args)
1519 {
1520 struct sched_get_priority_max_args bsd;
1521 int error;
1522
1523 #ifdef DEBUG
1524 if (ldebug(sched_get_priority_max))
1525 kprintf(ARGS(sched_get_priority_max, "%d"), args->policy);
1526 #endif
1527
1528 switch (args->policy) {
1529 case LINUX_SCHED_OTHER:
1530 bsd.policy = SCHED_OTHER;
1531 break;
1532 case LINUX_SCHED_FIFO:
1533 bsd.policy = SCHED_FIFO;
1534 break;
1535 case LINUX_SCHED_RR:
1536 bsd.policy = SCHED_RR;
1537 break;
1538 default:
1539 return EINVAL;
1540 }
1541 bsd.sysmsg_result = 0;
1542
1543 error = sys_sched_get_priority_max(&bsd);
1544 args->sysmsg_result = bsd.sysmsg_result;
1545 return(error);
1546 }
1547
1548 /*
1549 * MPSAFE
1550 */
1551 int
1552 sys_linux_sched_get_priority_min(struct linux_sched_get_priority_min_args *args)
1553 {
1554 struct sched_get_priority_min_args bsd;
1555 int error;
1556
1557 #ifdef DEBUG
1558 if (ldebug(sched_get_priority_min))
1559 kprintf(ARGS(sched_get_priority_min, "%d"), args->policy);
1560 #endif
1561
1562 switch (args->policy) {
1563 case LINUX_SCHED_OTHER:
1564 bsd.policy = SCHED_OTHER;
1565 break;
1566 case LINUX_SCHED_FIFO:
1567 bsd.policy = SCHED_FIFO;
1568 break;
1569 case LINUX_SCHED_RR:
1570 bsd.policy = SCHED_RR;
1571 break;
1572 default:
1573 return EINVAL;
1574 }
1575 bsd.sysmsg_result = 0;
1576
1577 error = sys_sched_get_priority_min(&bsd);
1578 args->sysmsg_result = bsd.sysmsg_result;
1579 return(error);
1580 }
1581
1582 #define REBOOT_CAD_ON 0x89abcdef
1583 #define REBOOT_CAD_OFF 0
1584 #define REBOOT_HALT 0xcdef0123
1585 #define REBOOT_RESTART 0x01234567
1586 #define REBOOT_RESTART2 0xA1B2C3D4
1587 #define REBOOT_POWEROFF 0x4321FEDC
1588 #define REBOOT_MAGIC1 0xfee1dead
1589 #define REBOOT_MAGIC2 0x28121969
1590 #define REBOOT_MAGIC2A 0x05121996
1591 #define REBOOT_MAGIC2B 0x16041998
1592
1593 /*
1594 * MPSAFE
1595 */
1596 int
1597 sys_linux_reboot(struct linux_reboot_args *args)
1598 {
1599 struct reboot_args bsd_args;
1600 int error;
1601
1602 #ifdef DEBUG
1603 if (ldebug(reboot))
1604 kprintf(ARGS(reboot, "0x%x"), args->cmd);
1605 #endif
1606
1607 if ((args->magic1 != REBOOT_MAGIC1) ||
1608 ((args->magic2 != REBOOT_MAGIC2) &&
1609 (args->magic2 != REBOOT_MAGIC2A) &&
1610 (args->magic2 != REBOOT_MAGIC2B)))
1611 return EINVAL;
1612
1613 switch (args->cmd) {
1614 case REBOOT_CAD_ON:
1615 case REBOOT_CAD_OFF:
1616 return (priv_check(curthread, PRIV_REBOOT));
1617 /* NOTREACHED */
1618 case REBOOT_HALT:
1619 bsd_args.opt = RB_HALT;
1620 break;
1621 case REBOOT_RESTART:
1622 case REBOOT_RESTART2:
1623 bsd_args.opt = 0;
1624 break;
1625 case REBOOT_POWEROFF:
1626 bsd_args.opt = RB_POWEROFF;
1627 break;
1628 default:
1629 return EINVAL;
1630 /* NOTREACHED */
1631 }
1632
1633 bsd_args.sysmsg_result = 0;
1634
1635 error = sys_reboot(&bsd_args);
1636 args->sysmsg_result = bsd_args.sysmsg_result;
1637 return(error);
1638 }
1639
1640 /*
1641 * The FreeBSD native getpid(2), getgid(2) and getuid(2) also modify
1642 * p->p_retval[1] when COMPAT_43 is defined. This
1643 * globbers registers that are assumed to be preserved. The following
1644 * lightweight syscalls fixes this. See also linux_getgid16() and
1645 * linux_getuid16() in linux_uid16.c.
1646 *
1647 * linux_getpid() - MP SAFE
1648 * linux_getgid() - MP SAFE
1649 * linux_getuid() - MP SAFE
1650 */
1651
1652 /*
1653 * MPALMOSTSAFE
1654 */
1655 int
1656 sys_linux_getpid(struct linux_getpid_args *args)
1657 {
1658 struct linux_emuldata *em;
1659 struct proc *p = curproc;
1660
1661
1662 EMUL_LOCK();
1663 em = emuldata_get(p);
1664 if (em == NULL) /* this should never happen */
1665 args->sysmsg_result = p->p_pid;
1666 else
1667 args->sysmsg_result = em->s->group_pid;
1668 EMUL_UNLOCK();
1669
1670 return (0);
1671 }
1672
1673 /*
1674 * MPALMOSTSAFE
1675 */
1676 int
1677 sys_linux_getppid(struct linux_getppid_args *args)
1678 {
1679 struct linux_emuldata *em;
1680 struct proc *parent;
1681 struct proc *p;
1682 pid_t group_pid;
1683
1684 EMUL_LOCK();
1685 em = emuldata_get(curproc);
1686 KKASSERT(em != NULL);
1687 group_pid = em->s->group_pid;
1688 EMUL_UNLOCK();
1689
1690 p = pfind(group_pid);
1691 /* We are not allowed to fail */
1692 if (p == NULL)
1693 goto out;
1694
1695 parent = p->p_pptr;
1696 if (parent->p_sysent == &elf_linux_sysvec) {
1697 EMUL_LOCK();
1698 em = emuldata_get(parent);
1699 args->sysmsg_result = em->s->group_pid;
1700 EMUL_UNLOCK();
1701 } else {
1702 args->sysmsg_result = parent->p_pid;
1703 }
1704 PRELE(p);
1705
1706 out:
1707 return (0);
1708 }
1709
1710 /*
1711 * MPSAFE
1712 */
1713 int
1714 sys_linux_getgid(struct linux_getgid_args *args)
1715 {
1716 struct thread *td = curthread;
1717
1718 args->sysmsg_result = td->td_ucred->cr_rgid;
1719 return (0);
1720 }
1721
1722 /*
1723 * MPSAFE
1724 */
1725 int
1726 sys_linux_getuid(struct linux_getuid_args *args)
1727 {
1728 struct thread *td = curthread;
1729
1730 args->sysmsg_result = td->td_ucred->cr_ruid;
1731 return (0);
1732 }
1733
1734 /*
1735 * MPSAFE
1736 */
1737 int
1738 sys_linux_getsid(struct linux_getsid_args *args)
1739 {
1740 struct getsid_args bsd;
1741 int error;
1742
1743 bsd.sysmsg_result = 0;
1744 bsd.pid = args->pid;
1745 error = sys_getsid(&bsd);
1746 args->sysmsg_result = bsd.sysmsg_result;
1747 return(error);
1748 }
1749
1750 /*
1751 * MPSAFE
1752 */
1753 int
1754 linux_nosys(struct nosys_args *args)
1755 {
1756 /* XXX */
1757 return (ENOSYS);
1758 }
1759
1760 int
1761 sys_linux_mq_open(struct linux_mq_open_args *args)
1762 {
1763 struct mq_open_args moa;
1764 int error, oflag;
1765
1766 oflag = 0;
1767 if (args->oflag & LINUX_O_RDONLY)
1768 oflag |= O_RDONLY;
1769 if (args->oflag & LINUX_O_WRONLY)
1770 oflag |= O_WRONLY;
1771 if (args->oflag & LINUX_O_RDWR)
1772 oflag |= O_RDWR;
1773
1774 if (args->oflag & LINUX_O_NONBLOCK)
1775 oflag |= O_NONBLOCK;
1776 if (args->oflag & LINUX_O_CREAT)
1777 oflag |= O_CREAT;
1778 if (args->oflag & LINUX_O_EXCL)
1779 oflag |= O_EXCL;
1780
1781 moa.name = args->name;
1782 moa.oflag = oflag;
1783 moa.mode = args->mode;
1784 moa.attr = args->attr;
1785
1786 error = sys_mq_open(&moa);
1787
1788 return (error);
1789 }
1790
1791 int
1792 sys_linux_mq_getsetattr(struct linux_mq_getsetattr_args *args)
1793 {
1794 struct mq_getattr_args gaa;
1795 struct mq_setattr_args saa;
1796 int error;
1797
1798 gaa.mqdes = args->mqd;
1799 gaa.mqstat = args->oattr;
1800
1801 saa.mqdes = args->mqd;
1802 saa.mqstat = args->attr;
1803 saa.mqstat = args->oattr;
1804
1805 if (args->attr != NULL) {
1806 error = sys_mq_setattr(&saa);
1807 } else {
1808 error = sys_mq_getattr(&gaa);
1809 }
1810
1811 return error;
1812 }
1813
1814 /*
1815 * Get affinity of a process.
1816 */
1817 int
1818 sys_linux_sched_getaffinity(struct linux_sched_getaffinity_args *args)
1819 {
1820 cpumask_t mask;
1821 struct proc *p;
1822 struct lwp *lp;
1823 int error = 0;
1824
1825 #ifdef DEBUG
1826 if (ldebug(sched_getaffinity))
1827 kprintf(ARGS(sched_getaffinity, "%d, %d, *"), args->pid,
1828 args->len);
1829 #endif
1830 if (args->len < sizeof(cpumask_t))
1831 return (EINVAL);
1832 #if 0
1833 if ((error = priv_check(curthread, PRIV_SCHED_CPUSET)) != 0)
1834 return (EPERM);
1835 #endif
1836 /* Get the mplock to ensure that the proc is not running */
1837 get_mplock();
1838 if (args->pid == 0) {
1839 p = curproc;
1840 PHOLD(p);
1841 } else {
1842 p = pfind(args->pid);
1843 if (p == NULL) {
1844 error = ESRCH;
1845 goto done;
1846 }
1847 }
1848
1849 lp = FIRST_LWP_IN_PROC(p);
1850 /*
1851 * XXX: if lwp_cpumask is ever changed to support more than
1852 * 32 processors, this needs to be changed to a bcopy.
1853 */
1854 mask = lp->lwp_cpumask;
1855 if ((error = copyout(&mask, args->user_mask_ptr, sizeof(cpumask_t))))
1856 error = EFAULT;
1857 done:
1858 rel_mplock();
1859 #if 0
1860 if (error == 0)
1861 args->sysmsg_iresult = sizeof(cpumask_t);
1862 #endif
1863 if (p)
1864 PRELE(p);
1865 return (error);
1866 }
1867
1868 /*
1869 * Set affinity of a process.
1870 */
1871 int
1872 sys_linux_sched_setaffinity(struct linux_sched_setaffinity_args *args)
1873 {
1874 #ifdef DEBUG
1875 if (ldebug(sched_setaffinity))
1876 kprintf(ARGS(sched_setaffinity, "%d, %d, *"), args->pid,
1877 args->len);
1878 #endif
1879 /*
1880 * From Linux man page:
1881 * sched_setaffinity() sets the CPU affinity mask of the process
1882 * whose ID is pid to the value specified by mask. If pid is zero,
1883 * then the calling process is used. The argument cpusetsize is
1884 * the length (in bytes) of the data pointed to by mask. Normally
1885 * this argument would be specified as sizeof(cpu_set_t).
1886 *
1887 * If the process specified by pid is not currently running on one
1888 * of the CPUs specified in mask, then that process is migrated to
1889 * one of the CPUs specified in mask.
1890 */
1891 /*
1892 * About our implementation: I don't think that it is too important
1893 * to have a working implementation, but if it was ever needed,
1894 * the best approach would be to implement the whole mechanism
1895 * properly in kern_usched.
1896 * The idea has to be to change the affinity mask AND migrate the
1897 * lwp to one of the new valid CPUs for the lwp, in case the current
1898 * CPU isn't anymore in the affinity mask passed in.
1899 * For now, we'll just signal success even if we didn't do anything.
1900 */
1901 return 0;
1902 }
1903
1904 int
1905 sys_linux_gettid(struct linux_gettid_args *args)
1906 {
1907 args->sysmsg_iresult = curproc->p_pid;
1908 return 0;
1909 }
1910
1911 int
1912 sys_linux_getcpu(struct linux_getcpu_args *args)
1913 {
1914 struct globaldata *gd;
1915 l_uint node = 0;
1916 int error;
1917
1918 gd = mycpu;
1919 error = copyout(&gd->gd_cpuid, args->pcpu, sizeof(gd->gd_cpuid));
1920 if (error)
1921 return (error);
1922 /*
1923 * XXX: this should be the NUMA node, but since we don't implement it,
1924 * just return 0 for it.
1925 */
1926 error = copyout(&node, args->pnode, sizeof(node));
1927 return (error);
1928 }
1929
1930 int
1931 sys_linux_sethostname(struct linux_sethostname_args *uap)
1932 {
1933 struct thread *td = curthread;
1934 size_t len;
1935 char *hostname;
1936 int name[2];
1937 int error;
1938
1939 name[0] = CTL_KERN;
1940 name[1] = KERN_HOSTNAME;
1941 error = priv_check_cred(td->td_ucred, PRIV_SETHOSTNAME, 0);
1942 if (error)
1943 return (error);
1944 len = MIN(uap->len, MAXHOSTNAMELEN);
1945 hostname = kmalloc(MAXHOSTNAMELEN, M_TEMP, M_WAITOK);
1946
1947 error = copyin(uap->hostname, hostname, len);
1948 if (error) {
1949 kfree(hostname, M_TEMP);
1950 return (error);
1951 }
1952
1953 get_mplock();
1954 error = kernel_sysctl(name, 2, NULL, 0, hostname, len, NULL);
1955 rel_mplock();
1956
1957 kfree(hostname, M_TEMP);
1958 return (error);
1959 }
DragonFly BSD