2 * Copyright (c) 2000 Marcel Moolenaar
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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.
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.
29 #include <sys/cdefs.h>
30 __FBSDID("$FreeBSD$");
32 #include <sys/param.h>
33 #include <sys/systm.h>
34 #include <sys/capsicum.h>
36 #include <sys/fcntl.h>
37 #include <sys/imgact.h>
39 #include <sys/malloc.h>
41 #include <sys/mutex.h>
45 #include <sys/queue.h>
46 #include <sys/resource.h>
47 #include <sys/resourcevar.h>
48 #include <sys/signalvar.h>
49 #include <sys/syscallsubr.h>
50 #include <sys/sysproto.h>
51 #include <sys/unistd.h>
53 #include <sys/sched.h>
55 #include <machine/frame.h>
56 #include <machine/psl.h>
57 #include <machine/segments.h>
58 #include <machine/sysarch.h>
62 #include <vm/vm_map.h>
64 #include <i386/linux/linux.h>
65 #include <i386/linux/linux_proto.h>
66 #include <compat/linux/linux_ipc.h>
67 #include <compat/linux/linux_misc.h>
68 #include <compat/linux/linux_mmap.h>
69 #include <compat/linux/linux_signal.h>
70 #include <compat/linux/linux_util.h>
71 #include <compat/linux/linux_emul.h>
73 #include <i386/include/pcb.h> /* needed for pcb definition in linux_set_thread_area */
75 #include "opt_posix.h"
77 extern struct sysentvec elf32_freebsd_sysvec; /* defined in i386/i386/elf_machdep.c */
85 l_uint read_exec_only:1;
86 l_uint limit_in_pages:1;
87 l_uint seg_not_present:1;
91 struct l_old_select_argv {
96 struct l_timeval *timeout;
101 linux_execve(struct thread *td, struct linux_execve_args *args)
103 struct image_args eargs;
107 LCONVPATHEXIST(td, args->path, &newpath);
111 printf(ARGS(execve, "%s"), newpath);
114 error = exec_copyin_args(&eargs, newpath, UIO_SYSSPACE,
115 args->argp, args->envp);
116 free(newpath, M_TEMP);
118 error = linux_common_execve(td, &eargs);
122 struct l_ipc_kludge {
123 struct l_msgbuf *msgp;
128 linux_ipc(struct thread *td, struct linux_ipc_args *args)
131 switch (args->what & 0xFFFF) {
133 struct linux_semop_args a;
135 a.semid = args->arg1;
137 a.nsops = args->arg2;
138 return (linux_semop(td, &a));
141 struct linux_semget_args a;
144 a.nsems = args->arg2;
145 a.semflg = args->arg3;
146 return (linux_semget(td, &a));
149 struct linux_semctl_args a;
152 a.semid = args->arg1;
153 a.semnum = args->arg2;
155 error = copyin(args->ptr, &a.arg, sizeof(a.arg));
158 return (linux_semctl(td, &a));
161 struct linux_msgsnd_args a;
163 a.msqid = args->arg1;
165 a.msgsz = args->arg2;
166 a.msgflg = args->arg3;
167 return (linux_msgsnd(td, &a));
170 struct linux_msgrcv_args a;
172 a.msqid = args->arg1;
173 a.msgsz = args->arg2;
174 a.msgflg = args->arg3;
175 if ((args->what >> 16) == 0) {
176 struct l_ipc_kludge tmp;
179 if (args->ptr == NULL)
181 error = copyin(args->ptr, &tmp, sizeof(tmp));
185 a.msgtyp = tmp.msgtyp;
188 a.msgtyp = args->arg5;
190 return (linux_msgrcv(td, &a));
193 struct linux_msgget_args a;
196 a.msgflg = args->arg2;
197 return (linux_msgget(td, &a));
200 struct linux_msgctl_args a;
202 a.msqid = args->arg1;
205 return (linux_msgctl(td, &a));
208 struct linux_shmat_args a;
210 a.shmid = args->arg1;
211 a.shmaddr = args->ptr;
212 a.shmflg = args->arg2;
213 a.raddr = (l_ulong *)args->arg3;
214 return (linux_shmat(td, &a));
217 struct linux_shmdt_args a;
219 a.shmaddr = args->ptr;
220 return (linux_shmdt(td, &a));
223 struct linux_shmget_args a;
227 a.shmflg = args->arg3;
228 return (linux_shmget(td, &a));
231 struct linux_shmctl_args a;
233 a.shmid = args->arg1;
236 return (linux_shmctl(td, &a));
246 linux_old_select(struct thread *td, struct linux_old_select_args *args)
248 struct l_old_select_argv linux_args;
249 struct linux_select_args newsel;
253 if (ldebug(old_select))
254 printf(ARGS(old_select, "%p"), args->ptr);
257 error = copyin(args->ptr, &linux_args, sizeof(linux_args));
261 newsel.nfds = linux_args.nfds;
262 newsel.readfds = linux_args.readfds;
263 newsel.writefds = linux_args.writefds;
264 newsel.exceptfds = linux_args.exceptfds;
265 newsel.timeout = linux_args.timeout;
266 return (linux_select(td, &newsel));
270 linux_set_cloned_tls(struct thread *td, void *desc)
272 struct segment_descriptor sd;
273 struct l_user_desc info;
277 error = copyin(desc, &info, sizeof(struct l_user_desc));
279 printf(LMSG("copyin failed!"));
281 idx = info.entry_number;
284 * looks like we're getting the idx we returned
285 * in the set_thread_area() syscall
287 if (idx != 6 && idx != 3) {
288 printf(LMSG("resetting idx!"));
292 /* this doesnt happen in practice */
294 /* we might copy out the entry_number as 3 */
295 info.entry_number = 3;
296 error = copyout(&info, desc, sizeof(struct l_user_desc));
298 printf(LMSG("copyout failed!"));
301 a[0] = LINUX_LDT_entry_a(&info);
302 a[1] = LINUX_LDT_entry_b(&info);
304 memcpy(&sd, &a, sizeof(a));
307 printf("Segment created in clone with "
308 "CLONE_SETTLS: lobase: %x, hibase: %x, "
309 "lolimit: %x, hilimit: %x, type: %i, "
310 "dpl: %i, p: %i, xx: %i, def32: %i, "
311 "gran: %i\n", sd.sd_lobase, sd.sd_hibase,
312 sd.sd_lolimit, sd.sd_hilimit, sd.sd_type,
313 sd.sd_dpl, sd.sd_p, sd.sd_xx,
314 sd.sd_def32, sd.sd_gran);
318 td->td_pcb->pcb_gsd = sd;
319 td->td_pcb->pcb_gs = GSEL(GUGS_SEL, SEL_UPL);
326 linux_set_upcall_kse(struct thread *td, register_t stack)
330 td->td_frame->tf_esp = stack;
333 * The newly created Linux thread returns
334 * to the user space by the same path that a parent do.
336 td->td_frame->tf_eax = 0;
341 linux_mmap2(struct thread *td, struct linux_mmap2_args *args)
346 printf(ARGS(mmap2, "%p, %d, %d, 0x%08x, %d, %d"),
347 (void *)args->addr, args->len, args->prot,
348 args->flags, args->fd, args->pgoff);
351 return (linux_mmap_common(td, args->addr, args->len, args->prot,
352 args->flags, args->fd, (uint64_t)(uint32_t)args->pgoff *
357 linux_mmap(struct thread *td, struct linux_mmap_args *args)
360 struct l_mmap_argv linux_args;
362 error = copyin(args->ptr, &linux_args, sizeof(linux_args));
368 printf(ARGS(mmap, "%p, %d, %d, 0x%08x, %d, %d"),
369 (void *)linux_args.addr, linux_args.len, linux_args.prot,
370 linux_args.flags, linux_args.fd, linux_args.pgoff);
373 return (linux_mmap_common(td, linux_args.addr, linux_args.len,
374 linux_args.prot, linux_args.flags, linux_args.fd,
375 (uint32_t)linux_args.pgoff));
379 linux_mprotect(struct thread *td, struct linux_mprotect_args *uap)
382 return (linux_mprotect_common(td, PTROUT(uap->addr), uap->len, uap->prot));
386 linux_ioperm(struct thread *td, struct linux_ioperm_args *args)
389 struct i386_ioperm_args iia;
391 iia.start = args->start;
392 iia.length = args->length;
393 iia.enable = args->enable;
394 error = i386_set_ioperm(td, &iia);
399 linux_iopl(struct thread *td, struct linux_iopl_args *args)
403 if (args->level < 0 || args->level > 3)
405 if ((error = priv_check(td, PRIV_IO)) != 0)
407 if ((error = securelevel_gt(td->td_ucred, 0)) != 0)
409 td->td_frame->tf_eflags = (td->td_frame->tf_eflags & ~PSL_IOPL) |
410 (args->level * (PSL_IOPL / 3));
415 linux_modify_ldt(struct thread *td, struct linux_modify_ldt_args *uap)
418 struct i386_ldt_args ldt;
419 struct l_descriptor ld;
420 union descriptor desc;
424 case 0x00: /* read_ldt */
426 ldt.descs = uap->ptr;
427 ldt.num = uap->bytecount / sizeof(union descriptor);
428 error = i386_get_ldt(td, &ldt);
429 td->td_retval[0] *= sizeof(union descriptor);
431 case 0x02: /* read_default_ldt = 0 */
432 size = 5*sizeof(struct l_desc_struct);
433 if (size > uap->bytecount)
434 size = uap->bytecount;
435 for (written = error = 0; written < size && error == 0; written++)
436 error = subyte((char *)uap->ptr + written, 0);
437 td->td_retval[0] = written;
439 case 0x01: /* write_ldt */
440 case 0x11: /* write_ldt */
441 if (uap->bytecount != sizeof(ld))
444 error = copyin(uap->ptr, &ld, sizeof(ld));
448 ldt.start = ld.entry_number;
451 desc.sd.sd_lolimit = (ld.limit & 0x0000ffff);
452 desc.sd.sd_hilimit = (ld.limit & 0x000f0000) >> 16;
453 desc.sd.sd_lobase = (ld.base_addr & 0x00ffffff);
454 desc.sd.sd_hibase = (ld.base_addr & 0xff000000) >> 24;
455 desc.sd.sd_type = SDT_MEMRO | ((ld.read_exec_only ^ 1) << 1) |
458 desc.sd.sd_p = (ld.seg_not_present ^ 1);
460 desc.sd.sd_def32 = ld.seg_32bit;
461 desc.sd.sd_gran = ld.limit_in_pages;
462 error = i386_set_ldt(td, &ldt, &desc);
469 if (error == EOPNOTSUPP) {
470 printf("linux: modify_ldt needs kernel option USER_LDT\n");
478 linux_sigaction(struct thread *td, struct linux_sigaction_args *args)
481 l_sigaction_t act, oact;
485 if (ldebug(sigaction))
486 printf(ARGS(sigaction, "%d, %p, %p"),
487 args->sig, (void *)args->nsa, (void *)args->osa);
490 if (args->nsa != NULL) {
491 error = copyin(args->nsa, &osa, sizeof(l_osigaction_t));
494 act.lsa_handler = osa.lsa_handler;
495 act.lsa_flags = osa.lsa_flags;
496 act.lsa_restorer = osa.lsa_restorer;
497 LINUX_SIGEMPTYSET(act.lsa_mask);
498 act.lsa_mask.__mask = osa.lsa_mask;
501 error = linux_do_sigaction(td, args->sig, args->nsa ? &act : NULL,
502 args->osa ? &oact : NULL);
504 if (args->osa != NULL && !error) {
505 osa.lsa_handler = oact.lsa_handler;
506 osa.lsa_flags = oact.lsa_flags;
507 osa.lsa_restorer = oact.lsa_restorer;
508 osa.lsa_mask = oact.lsa_mask.__mask;
509 error = copyout(&osa, args->osa, sizeof(l_osigaction_t));
516 * Linux has two extra args, restart and oldmask. We dont use these,
517 * but it seems that "restart" is actually a context pointer that
518 * enables the signal to happen with a different register set.
521 linux_sigsuspend(struct thread *td, struct linux_sigsuspend_args *args)
527 if (ldebug(sigsuspend))
528 printf(ARGS(sigsuspend, "%08lx"), (unsigned long)args->mask);
531 LINUX_SIGEMPTYSET(mask);
532 mask.__mask = args->mask;
533 linux_to_bsd_sigset(&mask, &sigmask);
534 return (kern_sigsuspend(td, sigmask));
538 linux_rt_sigsuspend(struct thread *td, struct linux_rt_sigsuspend_args *uap)
545 if (ldebug(rt_sigsuspend))
546 printf(ARGS(rt_sigsuspend, "%p, %d"),
547 (void *)uap->newset, uap->sigsetsize);
550 if (uap->sigsetsize != sizeof(l_sigset_t))
553 error = copyin(uap->newset, &lmask, sizeof(l_sigset_t));
557 linux_to_bsd_sigset(&lmask, &sigmask);
558 return (kern_sigsuspend(td, sigmask));
562 linux_pause(struct thread *td, struct linux_pause_args *args)
564 struct proc *p = td->td_proc;
569 printf(ARGS(pause, ""));
573 sigmask = td->td_sigmask;
575 return (kern_sigsuspend(td, sigmask));
579 linux_sigaltstack(struct thread *td, struct linux_sigaltstack_args *uap)
586 if (ldebug(sigaltstack))
587 printf(ARGS(sigaltstack, "%p, %p"), uap->uss, uap->uoss);
590 if (uap->uss != NULL) {
591 error = copyin(uap->uss, &lss, sizeof(l_stack_t));
595 ss.ss_sp = lss.ss_sp;
596 ss.ss_size = lss.ss_size;
597 ss.ss_flags = linux_to_bsd_sigaltstack(lss.ss_flags);
599 error = kern_sigaltstack(td, (uap->uss != NULL) ? &ss : NULL,
600 (uap->uoss != NULL) ? &oss : NULL);
601 if (!error && uap->uoss != NULL) {
602 lss.ss_sp = oss.ss_sp;
603 lss.ss_size = oss.ss_size;
604 lss.ss_flags = bsd_to_linux_sigaltstack(oss.ss_flags);
605 error = copyout(&lss, uap->uoss, sizeof(l_stack_t));
612 linux_ftruncate64(struct thread *td, struct linux_ftruncate64_args *args)
614 struct ftruncate_args sa;
617 if (ldebug(ftruncate64))
618 printf(ARGS(ftruncate64, "%u, %jd"), args->fd,
619 (intmax_t)args->length);
623 sa.length = args->length;
624 return sys_ftruncate(td, &sa);
628 linux_set_thread_area(struct thread *td, struct linux_set_thread_area_args *args)
630 struct l_user_desc info;
634 struct segment_descriptor sd;
636 error = copyin(args->desc, &info, sizeof(struct l_user_desc));
641 if (ldebug(set_thread_area))
642 printf(ARGS(set_thread_area, "%i, %x, %x, %i, %i, %i, %i, %i, %i\n"),
650 info.seg_not_present,
654 idx = info.entry_number;
656 * Semantics of linux version: every thread in the system has array of
657 * 3 tls descriptors. 1st is GLIBC TLS, 2nd is WINE, 3rd unknown. This
658 * syscall loads one of the selected tls decriptors with a value and
659 * also loads GDT descriptors 6, 7 and 8 with the content of the
660 * per-thread descriptors.
662 * Semantics of fbsd version: I think we can ignore that linux has 3
663 * per-thread descriptors and use just the 1st one. The tls_array[]
664 * is used only in set/get-thread_area() syscalls and for loading the
665 * GDT descriptors. In fbsd we use just one GDT descriptor for TLS so
666 * we will load just one.
668 * XXX: this doesn't work when a user space process tries to use more
669 * than 1 TLS segment. Comment in the linux sources says wine might do
674 * we support just GLIBC TLS now
675 * we should let 3 proceed as well because we use this segment so
676 * if code does two subsequent calls it should succeed
678 if (idx != 6 && idx != -1 && idx != 3)
682 * we have to copy out the GDT entry we use
683 * FreeBSD uses GDT entry #3 for storing %gs so load that
685 * XXX: what if a user space program doesn't check this value and tries
688 idx = info.entry_number = 3;
689 error = copyout(&info, args->desc, sizeof(struct l_user_desc));
693 if (LINUX_LDT_empty(&info)) {
697 a[0] = LINUX_LDT_entry_a(&info);
698 a[1] = LINUX_LDT_entry_b(&info);
701 memcpy(&sd, &a, sizeof(a));
703 if (ldebug(set_thread_area))
704 printf("Segment created in set_thread_area: lobase: %x, hibase: %x, lolimit: %x, hilimit: %x, type: %i, dpl: %i, p: %i, xx: %i, def32: %i, gran: %i\n", sd.sd_lobase,
716 /* this is taken from i386 version of cpu_set_user_tls() */
719 td->td_pcb->pcb_gsd = sd;
720 PCPU_GET(fsgs_gdt)[1] = sd;
721 load_gs(GSEL(GUGS_SEL, SEL_UPL));
728 linux_get_thread_area(struct thread *td, struct linux_get_thread_area_args *args)
731 struct l_user_desc info;
734 struct l_desc_struct desc;
735 struct segment_descriptor sd;
738 if (ldebug(get_thread_area))
739 printf(ARGS(get_thread_area, "%p"), args->desc);
742 error = copyin(args->desc, &info, sizeof(struct l_user_desc));
746 idx = info.entry_number;
747 /* XXX: I am not sure if we want 3 to be allowed too. */
748 if (idx != 6 && idx != 3)
753 memset(&info, 0, sizeof(info));
755 sd = PCPU_GET(fsgs_gdt)[1];
757 memcpy(&desc, &sd, sizeof(desc));
759 info.entry_number = idx;
760 info.base_addr = LINUX_GET_BASE(&desc);
761 info.limit = LINUX_GET_LIMIT(&desc);
762 info.seg_32bit = LINUX_GET_32BIT(&desc);
763 info.contents = LINUX_GET_CONTENTS(&desc);
764 info.read_exec_only = !LINUX_GET_WRITABLE(&desc);
765 info.limit_in_pages = LINUX_GET_LIMIT_PAGES(&desc);
766 info.seg_not_present = !LINUX_GET_PRESENT(&desc);
767 info.useable = LINUX_GET_USEABLE(&desc);
769 error = copyout(&info, args->desc, sizeof(struct l_user_desc));
776 /* XXX: this wont work with module - convert it */
778 linux_mq_open(struct thread *td, struct linux_mq_open_args *args)
780 #ifdef P1003_1B_MQUEUE
781 return sys_kmq_open(td, (struct kmq_open_args *) args);
788 linux_mq_unlink(struct thread *td, struct linux_mq_unlink_args *args)
790 #ifdef P1003_1B_MQUEUE
791 return sys_kmq_unlink(td, (struct kmq_unlink_args *) args);
798 linux_mq_timedsend(struct thread *td, struct linux_mq_timedsend_args *args)
800 #ifdef P1003_1B_MQUEUE
801 return sys_kmq_timedsend(td, (struct kmq_timedsend_args *) args);
808 linux_mq_timedreceive(struct thread *td, struct linux_mq_timedreceive_args *args)
810 #ifdef P1003_1B_MQUEUE
811 return sys_kmq_timedreceive(td, (struct kmq_timedreceive_args *) args);
818 linux_mq_notify(struct thread *td, struct linux_mq_notify_args *args)
820 #ifdef P1003_1B_MQUEUE
821 return sys_kmq_notify(td, (struct kmq_notify_args *) args);
828 linux_mq_getsetattr(struct thread *td, struct linux_mq_getsetattr_args *args)
830 #ifdef P1003_1B_MQUEUE
831 return sys_kmq_setattr(td, (struct kmq_setattr_args *) args);