2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
4 * Copyright (c) 2000 Marcel Moolenaar
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 #include <sys/cdefs.h>
30 __FBSDID("$FreeBSD$");
32 #include <sys/param.h>
33 #include <sys/capsicum.h>
34 #include <sys/fcntl.h>
36 #include <sys/imgact.h>
38 #include <sys/malloc.h>
40 #include <sys/mutex.h>
43 #include <sys/queue.h>
44 #include <sys/resource.h>
45 #include <sys/resourcevar.h>
46 #include <sys/sched.h>
47 #include <sys/signalvar.h>
48 #include <sys/syscallsubr.h>
49 #include <sys/sysproto.h>
50 #include <sys/systm.h>
52 #include <sys/unistd.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 <security/audit/audit.h>
66 #include <i386/linux/linux.h>
67 #include <i386/linux/linux_proto.h>
68 #include <compat/linux/linux_emul.h>
69 #include <compat/linux/linux_ipc.h>
70 #include <compat/linux/linux_misc.h>
71 #include <compat/linux/linux_mmap.h>
72 #include <compat/linux/linux_signal.h>
73 #include <compat/linux/linux_util.h>
75 #include <i386/include/pcb.h> /* needed for pcb definition in linux_set_thread_area */
77 #include "opt_posix.h"
79 extern struct sysentvec elf32_freebsd_sysvec; /* defined in i386/i386/elf_machdep.c */
87 l_uint read_exec_only:1;
88 l_uint limit_in_pages:1;
89 l_uint seg_not_present:1;
93 struct l_old_select_argv {
98 struct l_timeval *timeout;
103 linux_execve(struct thread *td, struct linux_execve_args *args)
105 struct image_args eargs;
109 LCONVPATHEXIST(td, args->path, &newpath);
113 printf(ARGS(execve, "%s"), newpath);
116 error = exec_copyin_args(&eargs, newpath, UIO_SYSSPACE,
117 args->argp, args->envp);
118 free(newpath, M_TEMP);
120 error = linux_common_execve(td, &eargs);
121 AUDIT_SYSCALL_EXIT(error == EJUSTRETURN ? 0 : error, td);
125 struct l_ipc_kludge {
126 struct l_msgbuf *msgp;
131 linux_ipc(struct thread *td, struct linux_ipc_args *args)
134 switch (args->what & 0xFFFF) {
136 struct linux_semop_args a;
138 a.semid = args->arg1;
139 a.tsops = PTRIN(args->ptr);
140 a.nsops = args->arg2;
141 return (linux_semop(td, &a));
144 struct linux_semget_args a;
147 a.nsems = args->arg2;
148 a.semflg = args->arg3;
149 return (linux_semget(td, &a));
152 struct linux_semctl_args a;
155 a.semid = args->arg1;
156 a.semnum = args->arg2;
158 error = copyin(PTRIN(args->ptr), &a.arg, sizeof(a.arg));
161 return (linux_semctl(td, &a));
164 struct linux_msgsnd_args a;
166 a.msqid = args->arg1;
167 a.msgp = PTRIN(args->ptr);
168 a.msgsz = args->arg2;
169 a.msgflg = args->arg3;
170 return (linux_msgsnd(td, &a));
173 struct linux_msgrcv_args a;
175 a.msqid = args->arg1;
176 a.msgsz = args->arg2;
177 a.msgflg = args->arg3;
178 if ((args->what >> 16) == 0) {
179 struct l_ipc_kludge tmp;
184 error = copyin(PTRIN(args->ptr), &tmp, sizeof(tmp));
187 a.msgp = PTRIN(tmp.msgp);
188 a.msgtyp = tmp.msgtyp;
190 a.msgp = PTRIN(args->ptr);
191 a.msgtyp = args->arg5;
193 return (linux_msgrcv(td, &a));
196 struct linux_msgget_args a;
199 a.msgflg = args->arg2;
200 return (linux_msgget(td, &a));
203 struct linux_msgctl_args a;
205 a.msqid = args->arg1;
207 a.buf = PTRIN(args->ptr);
208 return (linux_msgctl(td, &a));
211 struct linux_shmat_args a;
215 a.shmid = args->arg1;
216 a.shmaddr = PTRIN(args->ptr);
217 a.shmflg = args->arg2;
218 error = linux_shmat(td, &a);
221 addr = td->td_retval[0];
222 error = copyout(&addr, PTRIN(args->arg3), sizeof(addr));
223 td->td_retval[0] = 0;
227 struct linux_shmdt_args a;
229 a.shmaddr = PTRIN(args->ptr);
230 return (linux_shmdt(td, &a));
233 struct linux_shmget_args a;
237 a.shmflg = args->arg3;
238 return (linux_shmget(td, &a));
241 struct linux_shmctl_args a;
243 a.shmid = args->arg1;
245 a.buf = PTRIN(args->ptr);
246 return (linux_shmctl(td, &a));
256 linux_old_select(struct thread *td, struct linux_old_select_args *args)
258 struct l_old_select_argv linux_args;
259 struct linux_select_args newsel;
263 if (ldebug(old_select))
264 printf(ARGS(old_select, "%p"), args->ptr);
267 error = copyin(args->ptr, &linux_args, sizeof(linux_args));
271 newsel.nfds = linux_args.nfds;
272 newsel.readfds = linux_args.readfds;
273 newsel.writefds = linux_args.writefds;
274 newsel.exceptfds = linux_args.exceptfds;
275 newsel.timeout = linux_args.timeout;
276 return (linux_select(td, &newsel));
280 linux_set_cloned_tls(struct thread *td, void *desc)
282 struct segment_descriptor sd;
283 struct l_user_desc info;
287 error = copyin(desc, &info, sizeof(struct l_user_desc));
289 printf(LMSG("copyin failed!"));
291 idx = info.entry_number;
294 * looks like we're getting the idx we returned
295 * in the set_thread_area() syscall
297 if (idx != 6 && idx != 3) {
298 printf(LMSG("resetting idx!"));
302 /* this doesnt happen in practice */
304 /* we might copy out the entry_number as 3 */
305 info.entry_number = 3;
306 error = copyout(&info, desc, sizeof(struct l_user_desc));
308 printf(LMSG("copyout failed!"));
311 a[0] = LINUX_LDT_entry_a(&info);
312 a[1] = LINUX_LDT_entry_b(&info);
314 memcpy(&sd, &a, sizeof(a));
317 printf("Segment created in clone with "
318 "CLONE_SETTLS: lobase: %x, hibase: %x, "
319 "lolimit: %x, hilimit: %x, type: %i, "
320 "dpl: %i, p: %i, xx: %i, def32: %i, "
321 "gran: %i\n", sd.sd_lobase, sd.sd_hibase,
322 sd.sd_lolimit, sd.sd_hilimit, sd.sd_type,
323 sd.sd_dpl, sd.sd_p, sd.sd_xx,
324 sd.sd_def32, sd.sd_gran);
328 td->td_pcb->pcb_gsd = sd;
329 td->td_pcb->pcb_gs = GSEL(GUGS_SEL, SEL_UPL);
336 linux_set_upcall_kse(struct thread *td, register_t stack)
340 td->td_frame->tf_esp = stack;
343 * The newly created Linux thread returns
344 * to the user space by the same path that a parent do.
346 td->td_frame->tf_eax = 0;
351 linux_mmap2(struct thread *td, struct linux_mmap2_args *args)
356 printf(ARGS(mmap2, "%p, %d, %d, 0x%08x, %d, %d"),
357 (void *)args->addr, args->len, args->prot,
358 args->flags, args->fd, args->pgoff);
361 return (linux_mmap_common(td, args->addr, args->len, args->prot,
362 args->flags, args->fd, (uint64_t)(uint32_t)args->pgoff *
367 linux_mmap(struct thread *td, struct linux_mmap_args *args)
370 struct l_mmap_argv linux_args;
372 error = copyin(args->ptr, &linux_args, sizeof(linux_args));
378 printf(ARGS(mmap, "%p, %d, %d, 0x%08x, %d, %d"),
379 (void *)linux_args.addr, linux_args.len, linux_args.prot,
380 linux_args.flags, linux_args.fd, linux_args.pgoff);
383 return (linux_mmap_common(td, linux_args.addr, linux_args.len,
384 linux_args.prot, linux_args.flags, linux_args.fd,
385 (uint32_t)linux_args.pgoff));
389 linux_mprotect(struct thread *td, struct linux_mprotect_args *uap)
392 return (linux_mprotect_common(td, PTROUT(uap->addr), uap->len, uap->prot));
396 linux_ioperm(struct thread *td, struct linux_ioperm_args *args)
399 struct i386_ioperm_args iia;
401 iia.start = args->start;
402 iia.length = args->length;
403 iia.enable = args->enable;
404 error = i386_set_ioperm(td, &iia);
409 linux_iopl(struct thread *td, struct linux_iopl_args *args)
413 if (args->level < 0 || args->level > 3)
415 if ((error = priv_check(td, PRIV_IO)) != 0)
417 if ((error = securelevel_gt(td->td_ucred, 0)) != 0)
419 td->td_frame->tf_eflags = (td->td_frame->tf_eflags & ~PSL_IOPL) |
420 (args->level * (PSL_IOPL / 3));
425 linux_modify_ldt(struct thread *td, struct linux_modify_ldt_args *uap)
428 struct i386_ldt_args ldt;
429 struct l_descriptor ld;
430 union descriptor desc;
434 case 0x00: /* read_ldt */
436 ldt.descs = uap->ptr;
437 ldt.num = uap->bytecount / sizeof(union descriptor);
438 error = i386_get_ldt(td, &ldt);
439 td->td_retval[0] *= sizeof(union descriptor);
441 case 0x02: /* read_default_ldt = 0 */
442 size = 5*sizeof(struct l_desc_struct);
443 if (size > uap->bytecount)
444 size = uap->bytecount;
445 for (written = error = 0; written < size && error == 0; written++)
446 error = subyte((char *)uap->ptr + written, 0);
447 td->td_retval[0] = written;
449 case 0x01: /* write_ldt */
450 case 0x11: /* write_ldt */
451 if (uap->bytecount != sizeof(ld))
454 error = copyin(uap->ptr, &ld, sizeof(ld));
458 ldt.start = ld.entry_number;
461 desc.sd.sd_lolimit = (ld.limit & 0x0000ffff);
462 desc.sd.sd_hilimit = (ld.limit & 0x000f0000) >> 16;
463 desc.sd.sd_lobase = (ld.base_addr & 0x00ffffff);
464 desc.sd.sd_hibase = (ld.base_addr & 0xff000000) >> 24;
465 desc.sd.sd_type = SDT_MEMRO | ((ld.read_exec_only ^ 1) << 1) |
468 desc.sd.sd_p = (ld.seg_not_present ^ 1);
470 desc.sd.sd_def32 = ld.seg_32bit;
471 desc.sd.sd_gran = ld.limit_in_pages;
472 error = i386_set_ldt(td, &ldt, &desc);
479 if (error == EOPNOTSUPP) {
480 printf("linux: modify_ldt needs kernel option USER_LDT\n");
488 linux_sigaction(struct thread *td, struct linux_sigaction_args *args)
491 l_sigaction_t act, oact;
495 if (ldebug(sigaction))
496 printf(ARGS(sigaction, "%d, %p, %p"),
497 args->sig, (void *)args->nsa, (void *)args->osa);
500 if (args->nsa != NULL) {
501 error = copyin(args->nsa, &osa, sizeof(l_osigaction_t));
504 act.lsa_handler = osa.lsa_handler;
505 act.lsa_flags = osa.lsa_flags;
506 act.lsa_restorer = osa.lsa_restorer;
507 LINUX_SIGEMPTYSET(act.lsa_mask);
508 act.lsa_mask.__mask = osa.lsa_mask;
511 error = linux_do_sigaction(td, args->sig, args->nsa ? &act : NULL,
512 args->osa ? &oact : NULL);
514 if (args->osa != NULL && !error) {
515 osa.lsa_handler = oact.lsa_handler;
516 osa.lsa_flags = oact.lsa_flags;
517 osa.lsa_restorer = oact.lsa_restorer;
518 osa.lsa_mask = oact.lsa_mask.__mask;
519 error = copyout(&osa, args->osa, sizeof(l_osigaction_t));
526 * Linux has two extra args, restart and oldmask. We dont use these,
527 * but it seems that "restart" is actually a context pointer that
528 * enables the signal to happen with a different register set.
531 linux_sigsuspend(struct thread *td, struct linux_sigsuspend_args *args)
537 if (ldebug(sigsuspend))
538 printf(ARGS(sigsuspend, "%08lx"), (unsigned long)args->mask);
541 LINUX_SIGEMPTYSET(mask);
542 mask.__mask = args->mask;
543 linux_to_bsd_sigset(&mask, &sigmask);
544 return (kern_sigsuspend(td, sigmask));
548 linux_rt_sigsuspend(struct thread *td, struct linux_rt_sigsuspend_args *uap)
555 if (ldebug(rt_sigsuspend))
556 printf(ARGS(rt_sigsuspend, "%p, %d"),
557 (void *)uap->newset, uap->sigsetsize);
560 if (uap->sigsetsize != sizeof(l_sigset_t))
563 error = copyin(uap->newset, &lmask, sizeof(l_sigset_t));
567 linux_to_bsd_sigset(&lmask, &sigmask);
568 return (kern_sigsuspend(td, sigmask));
572 linux_pause(struct thread *td, struct linux_pause_args *args)
574 struct proc *p = td->td_proc;
579 printf(ARGS(pause, ""));
583 sigmask = td->td_sigmask;
585 return (kern_sigsuspend(td, sigmask));
589 linux_sigaltstack(struct thread *td, struct linux_sigaltstack_args *uap)
596 if (ldebug(sigaltstack))
597 printf(ARGS(sigaltstack, "%p, %p"), uap->uss, uap->uoss);
600 if (uap->uss != NULL) {
601 error = copyin(uap->uss, &lss, sizeof(l_stack_t));
605 ss.ss_sp = lss.ss_sp;
606 ss.ss_size = lss.ss_size;
607 ss.ss_flags = linux_to_bsd_sigaltstack(lss.ss_flags);
609 error = kern_sigaltstack(td, (uap->uss != NULL) ? &ss : NULL,
610 (uap->uoss != NULL) ? &oss : NULL);
611 if (!error && uap->uoss != NULL) {
612 lss.ss_sp = oss.ss_sp;
613 lss.ss_size = oss.ss_size;
614 lss.ss_flags = bsd_to_linux_sigaltstack(oss.ss_flags);
615 error = copyout(&lss, uap->uoss, sizeof(l_stack_t));
622 linux_ftruncate64(struct thread *td, struct linux_ftruncate64_args *args)
626 if (ldebug(ftruncate64))
627 printf(ARGS(ftruncate64, "%u, %jd"), args->fd,
628 (intmax_t)args->length);
631 return (kern_ftruncate(td, args->fd, args->length));
635 linux_set_thread_area(struct thread *td, struct linux_set_thread_area_args *args)
637 struct l_user_desc info;
641 struct segment_descriptor sd;
643 error = copyin(args->desc, &info, sizeof(struct l_user_desc));
648 if (ldebug(set_thread_area))
649 printf(ARGS(set_thread_area, "%i, %x, %x, %i, %i, %i, %i, %i, %i\n"),
657 info.seg_not_present,
661 idx = info.entry_number;
663 * Semantics of Linux version: every thread in the system has array of
664 * 3 tls descriptors. 1st is GLIBC TLS, 2nd is WINE, 3rd unknown. This
665 * syscall loads one of the selected tls decriptors with a value and
666 * also loads GDT descriptors 6, 7 and 8 with the content of the
667 * per-thread descriptors.
669 * Semantics of FreeBSD version: I think we can ignore that Linux has 3
670 * per-thread descriptors and use just the 1st one. The tls_array[]
671 * is used only in set/get-thread_area() syscalls and for loading the
672 * GDT descriptors. In FreeBSD we use just one GDT descriptor for TLS
673 * so we will load just one.
675 * XXX: this doesn't work when a user space process tries to use more
676 * than 1 TLS segment. Comment in the Linux sources says wine might do
681 * we support just GLIBC TLS now
682 * we should let 3 proceed as well because we use this segment so
683 * if code does two subsequent calls it should succeed
685 if (idx != 6 && idx != -1 && idx != 3)
689 * we have to copy out the GDT entry we use
690 * FreeBSD uses GDT entry #3 for storing %gs so load that
692 * XXX: what if a user space program doesn't check this value and tries
695 idx = info.entry_number = 3;
696 error = copyout(&info, args->desc, sizeof(struct l_user_desc));
700 if (LINUX_LDT_empty(&info)) {
704 a[0] = LINUX_LDT_entry_a(&info);
705 a[1] = LINUX_LDT_entry_b(&info);
708 memcpy(&sd, &a, sizeof(a));
710 if (ldebug(set_thread_area))
711 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,
723 /* this is taken from i386 version of cpu_set_user_tls() */
726 td->td_pcb->pcb_gsd = sd;
727 PCPU_GET(fsgs_gdt)[1] = sd;
728 load_gs(GSEL(GUGS_SEL, SEL_UPL));
735 linux_get_thread_area(struct thread *td, struct linux_get_thread_area_args *args)
738 struct l_user_desc info;
741 struct l_desc_struct desc;
742 struct segment_descriptor sd;
745 if (ldebug(get_thread_area))
746 printf(ARGS(get_thread_area, "%p"), args->desc);
749 error = copyin(args->desc, &info, sizeof(struct l_user_desc));
753 idx = info.entry_number;
754 /* XXX: I am not sure if we want 3 to be allowed too. */
755 if (idx != 6 && idx != 3)
760 memset(&info, 0, sizeof(info));
762 sd = PCPU_GET(fsgs_gdt)[1];
764 memcpy(&desc, &sd, sizeof(desc));
766 info.entry_number = idx;
767 info.base_addr = LINUX_GET_BASE(&desc);
768 info.limit = LINUX_GET_LIMIT(&desc);
769 info.seg_32bit = LINUX_GET_32BIT(&desc);
770 info.contents = LINUX_GET_CONTENTS(&desc);
771 info.read_exec_only = !LINUX_GET_WRITABLE(&desc);
772 info.limit_in_pages = LINUX_GET_LIMIT_PAGES(&desc);
773 info.seg_not_present = !LINUX_GET_PRESENT(&desc);
774 info.useable = LINUX_GET_USEABLE(&desc);
776 error = copyout(&info, args->desc, sizeof(struct l_user_desc));
783 /* XXX: this wont work with module - convert it */
785 linux_mq_open(struct thread *td, struct linux_mq_open_args *args)
787 #ifdef P1003_1B_MQUEUE
788 return (sys_kmq_open(td, (struct kmq_open_args *)args));
795 linux_mq_unlink(struct thread *td, struct linux_mq_unlink_args *args)
797 #ifdef P1003_1B_MQUEUE
798 return (sys_kmq_unlink(td, (struct kmq_unlink_args *)args));
805 linux_mq_timedsend(struct thread *td, struct linux_mq_timedsend_args *args)
807 #ifdef P1003_1B_MQUEUE
808 return (sys_kmq_timedsend(td, (struct kmq_timedsend_args *)args));
815 linux_mq_timedreceive(struct thread *td, struct linux_mq_timedreceive_args *args)
817 #ifdef P1003_1B_MQUEUE
818 return (sys_kmq_timedreceive(td, (struct kmq_timedreceive_args *)args));
825 linux_mq_notify(struct thread *td, struct linux_mq_notify_args *args)
827 #ifdef P1003_1B_MQUEUE
828 return (sys_kmq_notify(td, (struct kmq_notify_args *)args));
835 linux_mq_getsetattr(struct thread *td, struct linux_mq_getsetattr_args *args)
837 #ifdef P1003_1B_MQUEUE
838 return (sys_kmq_setattr(td, (struct kmq_setattr_args *)args));