Modify kern/makesyscall.sh to prefix all kernel system call procedures
[dragonfly.git] / sys / kern / sys_process.c
CommitLineData
984263bc
MD
1/*
2 * Copyright (c) 1994, Sean Eric Fagan
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 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by Sean Eric Fagan.
16 * 4. The name of the author may not be used to endorse or promote products
17 * derived from this software without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 *
31 * $FreeBSD: src/sys/kern/sys_process.c,v 1.51.2.6 2003/01/08 03:06:45 kan Exp $
753fd850 32 * $DragonFly: src/sys/kern/sys_process.c,v 1.20 2006/06/05 07:26:10 dillon Exp $
984263bc
MD
33 */
34
35#include <sys/param.h>
36#include <sys/systm.h>
37#include <sys/sysproto.h>
38#include <sys/proc.h>
39#include <sys/vnode.h>
40#include <sys/ptrace.h>
41
42#include <machine/reg.h>
43#include <vm/vm.h>
44#include <sys/lock.h>
45#include <vm/pmap.h>
46#include <vm/vm_map.h>
47#include <vm/vm_page.h>
48
49#include <sys/user.h>
1f2de5d4 50#include <vfs/procfs/procfs.h>
a2a5ad0d 51#include <sys/thread2.h>
984263bc
MD
52
53/* use the equivalent procfs code */
54#if 0
55static int
56pread (struct proc *procp, unsigned int addr, unsigned int *retval) {
57 int rv;
58 vm_map_t map, tmap;
59 vm_object_t object;
60 vm_offset_t kva = 0;
61 int page_offset; /* offset into page */
62 vm_offset_t pageno; /* page number */
63 vm_map_entry_t out_entry;
64 vm_prot_t out_prot;
65 boolean_t wired;
66 vm_pindex_t pindex;
67
68 /* Map page into kernel space */
69
70 map = &procp->p_vmspace->vm_map;
71
72 page_offset = addr - trunc_page(addr);
73 pageno = trunc_page(addr);
74
75 tmap = map;
76 rv = vm_map_lookup (&tmap, pageno, VM_PROT_READ, &out_entry,
77 &object, &pindex, &out_prot, &wired);
78
79 if (rv != KERN_SUCCESS)
80 return EINVAL;
81
a108bf71 82 vm_map_lookup_done (tmap, out_entry, 0);
984263bc
MD
83
84 /* Find space in kernel_map for the page we're interested in */
85 rv = vm_map_find (kernel_map, object, IDX_TO_OFF(pindex),
86 &kva, PAGE_SIZE, 0, VM_PROT_ALL, VM_PROT_ALL, 0);
87
88 if (!rv) {
89 vm_object_reference (object);
90
cde87949 91 rv = vm_map_wire (kernel_map, kva, kva + PAGE_SIZE, 0);
984263bc
MD
92 if (!rv) {
93 *retval = 0;
94 bcopy ((caddr_t)kva + page_offset,
95 retval, sizeof *retval);
96 }
97 vm_map_remove (kernel_map, kva, kva + PAGE_SIZE);
98 }
99
100 return rv;
101}
102
103static int
104pwrite (struct proc *procp, unsigned int addr, unsigned int datum) {
105 int rv;
106 vm_map_t map, tmap;
107 vm_object_t object;
108 vm_offset_t kva = 0;
109 int page_offset; /* offset into page */
110 vm_offset_t pageno; /* page number */
111 vm_map_entry_t out_entry;
112 vm_prot_t out_prot;
113 boolean_t wired;
114 vm_pindex_t pindex;
115 boolean_t fix_prot = 0;
116
117 /* Map page into kernel space */
118
119 map = &procp->p_vmspace->vm_map;
120
121 page_offset = addr - trunc_page(addr);
122 pageno = trunc_page(addr);
123
124 /*
125 * Check the permissions for the area we're interested in.
126 */
127
128 if (vm_map_check_protection (map, pageno, pageno + PAGE_SIZE,
129 VM_PROT_WRITE) == FALSE) {
130 /*
131 * If the page was not writable, we make it so.
132 * XXX It is possible a page may *not* be read/executable,
133 * if a process changes that!
134 */
135 fix_prot = 1;
136 /* The page isn't writable, so let's try making it so... */
137 if ((rv = vm_map_protect (map, pageno, pageno + PAGE_SIZE,
138 VM_PROT_ALL, 0)) != KERN_SUCCESS)
139 return EFAULT; /* I guess... */
140 }
141
142 /*
143 * Now we need to get the page. out_entry, out_prot, wired, and
144 * single_use aren't used. One would think the vm code would be
145 * a *bit* nicer... We use tmap because vm_map_lookup() can
146 * change the map argument.
147 */
148
149 tmap = map;
150 rv = vm_map_lookup (&tmap, pageno, VM_PROT_WRITE, &out_entry,
151 &object, &pindex, &out_prot, &wired);
152 if (rv != KERN_SUCCESS) {
153 return EINVAL;
154 }
155
156 /*
157 * Okay, we've got the page. Let's release tmap.
158 */
159
a108bf71 160 vm_map_lookup_done (tmap, out_entry, 0);
984263bc
MD
161
162 /*
163 * Fault the page in...
164 */
165
166 rv = vm_fault(map, pageno, VM_PROT_WRITE|VM_PROT_READ, FALSE);
167 if (rv != KERN_SUCCESS)
168 return EFAULT;
169
170 /* Find space in kernel_map for the page we're interested in */
171 rv = vm_map_find (kernel_map, object, IDX_TO_OFF(pindex),
172 &kva, PAGE_SIZE, 0,
173 VM_PROT_ALL, VM_PROT_ALL, 0);
174 if (!rv) {
175 vm_object_reference (object);
176
cde87949 177 rv = vm_map_wire (kernel_map, kva, kva + PAGE_SIZE, 0);
984263bc
MD
178 if (!rv) {
179 bcopy (&datum, (caddr_t)kva + page_offset, sizeof datum);
180 }
181 vm_map_remove (kernel_map, kva, kva + PAGE_SIZE);
182 }
183
184 if (fix_prot)
185 vm_map_protect (map, pageno, pageno + PAGE_SIZE,
186 VM_PROT_READ|VM_PROT_EXECUTE, 0);
187 return rv;
188}
189#endif
190
191/*
192 * Process debugging system call.
193 */
984263bc 194int
753fd850 195sys_ptrace(struct ptrace_args *uap)
984263bc 196{
41c20dac
MD
197 struct proc *p = curproc;
198
984263bc
MD
199 /*
200 * XXX this obfuscation is to reduce stack usage, but the register
201 * structs may be too large to put on the stack anyway.
202 */
203 union {
204 struct ptrace_io_desc piod;
205 struct dbreg dbreg;
206 struct fpreg fpreg;
207 struct reg reg;
208 } r;
209 void *addr;
210 int error = 0;
211
212 addr = &r;
213 switch (uap->req) {
214 case PT_GETREGS:
215 case PT_GETFPREGS:
216#ifdef PT_GETDBREGS
217 case PT_GETDBREGS:
218#endif
219 break;
220 case PT_SETREGS:
221 error = copyin(uap->addr, &r.reg, sizeof r.reg);
222 break;
223 case PT_SETFPREGS:
224 error = copyin(uap->addr, &r.fpreg, sizeof r.fpreg);
225 break;
226#ifdef PT_SETDBREGS
227 case PT_SETDBREGS:
228 error = copyin(uap->addr, &r.dbreg, sizeof r.dbreg);
229 break;
230#endif
231 case PT_IO:
232 error = copyin(uap->addr, &r.piod, sizeof r.piod);
233 break;
234 default:
235 addr = uap->addr;
236 }
237 if (error)
238 return (error);
239
90b9818c 240 error = kern_ptrace(p, uap->req, uap->pid, addr, uap->data,
c7114eea 241 &uap->sysmsg_result);
984263bc
MD
242 if (error)
243 return (error);
244
245 switch (uap->req) {
246 case PT_IO:
247 (void)copyout(&r.piod, uap->addr, sizeof r.piod);
248 break;
249 case PT_GETREGS:
250 error = copyout(&r.reg, uap->addr, sizeof r.reg);
251 break;
252 case PT_GETFPREGS:
253 error = copyout(&r.fpreg, uap->addr, sizeof r.fpreg);
254 break;
255#ifdef PT_GETDBREGS
256 case PT_GETDBREGS:
257 error = copyout(&r.dbreg, uap->addr, sizeof r.dbreg);
258 break;
259#endif
260 }
261
262 return (error);
263}
264
265int
90b9818c 266kern_ptrace(struct proc *curp, int req, pid_t pid, void *addr, int data, int *res)
984263bc
MD
267{
268 struct proc *p, *pp;
269 struct iovec iov;
270 struct uio uio;
271 struct ptrace_io_desc *piod;
272 int error = 0;
d767fe64 273 int write, tmp;
984263bc
MD
274
275 write = 0;
1301b4d8 276 if (req == PT_TRACE_ME) {
984263bc 277 p = curp;
1301b4d8 278 } else {
984263bc
MD
279 if ((p = pfind(pid)) == NULL)
280 return ESRCH;
281 }
41c20dac 282 if (!PRISON_CHECK(curp->p_ucred, p->p_ucred))
984263bc
MD
283 return (ESRCH);
284
285 /* Can't trace a process that's currently exec'ing. */
286 if ((p->p_flag & P_INEXEC) != 0)
287 return EAGAIN;
288
289 /*
290 * Permissions check
291 */
292 switch (req) {
293 case PT_TRACE_ME:
294 /* Always legal. */
295 break;
296
297 case PT_ATTACH:
298 /* Self */
299 if (p->p_pid == curp->p_pid)
300 return EINVAL;
301
302 /* Already traced */
303 if (p->p_flag & P_TRACED)
304 return EBUSY;
305
306 if (curp->p_flag & P_TRACED)
307 for (pp = curp->p_pptr; pp != NULL; pp = pp->p_pptr)
308 if (pp == p)
309 return (EINVAL);
310
311 /* not owned by you, has done setuid (unless you're root) */
41c20dac 312 if ((p->p_ucred->cr_ruid != curp->p_ucred->cr_ruid) ||
984263bc 313 (p->p_flag & P_SUGID)) {
dadab5e9 314 if ((error = suser(curp->p_thread)) != 0)
984263bc
MD
315 return error;
316 }
317
318 /* can't trace init when securelevel > 0 */
319 if (securelevel > 0 && p->p_pid == 1)
320 return EPERM;
321
322 /* OK */
323 break;
324
325 case PT_READ_I:
326 case PT_READ_D:
327 case PT_WRITE_I:
328 case PT_WRITE_D:
329 case PT_IO:
330 case PT_CONTINUE:
331 case PT_KILL:
332 case PT_STEP:
333 case PT_DETACH:
334#ifdef PT_GETREGS
335 case PT_GETREGS:
336#endif
337#ifdef PT_SETREGS
338 case PT_SETREGS:
339#endif
340#ifdef PT_GETFPREGS
341 case PT_GETFPREGS:
342#endif
343#ifdef PT_SETFPREGS
344 case PT_SETFPREGS:
345#endif
346#ifdef PT_GETDBREGS
347 case PT_GETDBREGS:
348#endif
349#ifdef PT_SETDBREGS
350 case PT_SETDBREGS:
351#endif
352 /* not being traced... */
353 if ((p->p_flag & P_TRACED) == 0)
354 return EPERM;
355
356 /* not being traced by YOU */
357 if (p->p_pptr != curp)
358 return EBUSY;
359
360 /* not currently stopped */
344ad853
MD
361 if ((p->p_flag & P_STOPPED) == 0 ||
362 (p->p_flag & P_WAITED) == 0) {
984263bc 363 return EBUSY;
344ad853 364 }
984263bc
MD
365
366 /* OK */
367 break;
368
369 default:
370 return EINVAL;
371 }
372
373#ifdef FIX_SSTEP
374 /*
375 * Single step fixup ala procfs
376 */
377 FIX_SSTEP(p);
378#endif
379
380 /*
381 * Actually do the requests
382 */
383
90b9818c 384 *res = 0;
984263bc
MD
385
386 switch (req) {
387 case PT_TRACE_ME:
388 /* set my trace flag and "owner" so it can read/write me */
389 p->p_flag |= P_TRACED;
390 p->p_oppid = p->p_pptr->p_pid;
391 return 0;
392
393 case PT_ATTACH:
394 /* security check done above */
395 p->p_flag |= P_TRACED;
396 p->p_oppid = p->p_pptr->p_pid;
397 if (p->p_pptr != curp)
398 proc_reparent(p, curp);
399 data = SIGSTOP;
400 goto sendsig; /* in PT_CONTINUE below */
401
402 case PT_STEP:
403 case PT_CONTINUE:
404 case PT_DETACH:
43300f51
DR
405 /* Zero means do not send any signal */
406 if (data < 0 || data > _SIG_MAXSIG)
984263bc
MD
407 return EINVAL;
408
409 PHOLD(p);
410
411 if (req == PT_STEP) {
e9182c58 412 if ((error = ptrace_single_step (&p->p_lwp))) {
984263bc
MD
413 PRELE(p);
414 return error;
415 }
416 }
417
418 if (addr != (void *)1) {
419 if ((error = ptrace_set_pc (p,
420 (u_long)(uintfptr_t)addr))) {
421 PRELE(p);
422 return error;
423 }
424 }
425 PRELE(p);
426
427 if (req == PT_DETACH) {
428 /* reset process parent */
429 if (p->p_oppid != p->p_pptr->p_pid) {
430 struct proc *pp;
431
432 pp = pfind(p->p_oppid);
433 proc_reparent(p, pp ? pp : initproc);
434 }
435
436 p->p_flag &= ~(P_TRACED | P_WAITED);
437 p->p_oppid = 0;
438
439 /* should we send SIGCHLD? */
440 }
441
442 sendsig:
344ad853
MD
443 /*
444 * Deliver or queue signal. If the process is stopped
445 * force it to SRUN again.
446 */
e43a034f 447 crit_enter();
344ad853 448 if (p->p_flag & P_STOPPED) {
984263bc 449 p->p_xstat = data;
344ad853 450 p->p_flag |= P_BREAKTSLEEP;
984263bc
MD
451 setrunnable(p);
452 } else if (data) {
453 psignal(p, data);
454 }
e43a034f 455 crit_exit();
984263bc
MD
456 return 0;
457
458 case PT_WRITE_I:
459 case PT_WRITE_D:
460 write = 1;
461 /* fallthrough */
462 case PT_READ_I:
463 case PT_READ_D:
1301b4d8
MD
464 /*
465 * NOTE! uio_offset represents the offset in the target
466 * process. The iov is in the current process (the guy
467 * making the ptrace call) so uio_td must be the current
468 * process (though for a SYSSPACE transfer it doesn't
469 * really matter).
470 */
984263bc
MD
471 tmp = 0;
472 /* write = 0 set above */
473 iov.iov_base = write ? (caddr_t)&data : (caddr_t)&tmp;
474 iov.iov_len = sizeof(int);
475 uio.uio_iov = &iov;
476 uio.uio_iovcnt = 1;
477 uio.uio_offset = (off_t)(uintptr_t)addr;
478 uio.uio_resid = sizeof(int);
479 uio.uio_segflg = UIO_SYSSPACE;
480 uio.uio_rw = write ? UIO_WRITE : UIO_READ;
1301b4d8 481 uio.uio_td = curp->p_thread;
984263bc
MD
482 error = procfs_domem(curp, p, NULL, &uio);
483 if (uio.uio_resid != 0) {
484 /*
485 * XXX procfs_domem() doesn't currently return ENOSPC,
486 * so I think write() can bogusly return 0.
487 * XXX what happens for short writes? We don't want
488 * to write partial data.
489 * XXX procfs_domem() returns EPERM for other invalid
490 * addresses. Convert this to EINVAL. Does this
491 * clobber returns of EPERM for other reasons?
492 */
493 if (error == 0 || error == ENOSPC || error == EPERM)
494 error = EINVAL; /* EOF */
495 }
496 if (!write)
90b9818c 497 *res = tmp;
984263bc
MD
498 return (error);
499
500 case PT_IO:
1301b4d8
MD
501 /*
502 * NOTE! uio_offset represents the offset in the target
503 * process. The iov is in the current process (the guy
504 * making the ptrace call) so uio_td must be the current
505 * process.
506 */
984263bc
MD
507 piod = addr;
508 iov.iov_base = piod->piod_addr;
509 iov.iov_len = piod->piod_len;
510 uio.uio_iov = &iov;
511 uio.uio_iovcnt = 1;
512 uio.uio_offset = (off_t)(uintptr_t)piod->piod_offs;
513 uio.uio_resid = piod->piod_len;
514 uio.uio_segflg = UIO_USERSPACE;
1301b4d8 515 uio.uio_td = curp->p_thread;
984263bc
MD
516 switch (piod->piod_op) {
517 case PIOD_READ_D:
518 case PIOD_READ_I:
519 uio.uio_rw = UIO_READ;
520 break;
521 case PIOD_WRITE_D:
522 case PIOD_WRITE_I:
523 uio.uio_rw = UIO_WRITE;
524 break;
525 default:
526 return (EINVAL);
527 }
528 error = procfs_domem(curp, p, NULL, &uio);
529 piod->piod_len -= uio.uio_resid;
530 return (error);
531
532 case PT_KILL:
533 data = SIGKILL;
534 goto sendsig; /* in PT_CONTINUE above */
535
536#ifdef PT_SETREGS
537 case PT_SETREGS:
538 write = 1;
539 /* fallthrough */
540#endif /* PT_SETREGS */
541#ifdef PT_GETREGS
542 case PT_GETREGS:
543 /* write = 0 above */
544#endif /* PT_SETREGS */
545#if defined(PT_SETREGS) || defined(PT_GETREGS)
546 if (!procfs_validregs(p)) /* no P_SYSTEM procs please */
547 return EINVAL;
548 else {
549 iov.iov_base = addr;
550 iov.iov_len = sizeof(struct reg);
551 uio.uio_iov = &iov;
552 uio.uio_iovcnt = 1;
553 uio.uio_offset = 0;
554 uio.uio_resid = sizeof(struct reg);
555 uio.uio_segflg = UIO_SYSSPACE;
556 uio.uio_rw = write ? UIO_WRITE : UIO_READ;
dadab5e9 557 uio.uio_td = curp->p_thread;
984263bc
MD
558 return (procfs_doregs(curp, p, NULL, &uio));
559 }
560#endif /* defined(PT_SETREGS) || defined(PT_GETREGS) */
561
562#ifdef PT_SETFPREGS
563 case PT_SETFPREGS:
564 write = 1;
565 /* fallthrough */
566#endif /* PT_SETFPREGS */
567#ifdef PT_GETFPREGS
568 case PT_GETFPREGS:
569 /* write = 0 above */
570#endif /* PT_SETFPREGS */
571#if defined(PT_SETFPREGS) || defined(PT_GETFPREGS)
572 if (!procfs_validfpregs(p)) /* no P_SYSTEM procs please */
573 return EINVAL;
574 else {
575 iov.iov_base = addr;
576 iov.iov_len = sizeof(struct fpreg);
577 uio.uio_iov = &iov;
578 uio.uio_iovcnt = 1;
579 uio.uio_offset = 0;
580 uio.uio_resid = sizeof(struct fpreg);
581 uio.uio_segflg = UIO_SYSSPACE;
582 uio.uio_rw = write ? UIO_WRITE : UIO_READ;
dadab5e9 583 uio.uio_td = curp->p_thread;
984263bc
MD
584 return (procfs_dofpregs(curp, p, NULL, &uio));
585 }
586#endif /* defined(PT_SETFPREGS) || defined(PT_GETFPREGS) */
587
588#ifdef PT_SETDBREGS
589 case PT_SETDBREGS:
590 write = 1;
591 /* fallthrough */
592#endif /* PT_SETDBREGS */
593#ifdef PT_GETDBREGS
594 case PT_GETDBREGS:
595 /* write = 0 above */
596#endif /* PT_SETDBREGS */
597#if defined(PT_SETDBREGS) || defined(PT_GETDBREGS)
598 if (!procfs_validdbregs(p)) /* no P_SYSTEM procs please */
599 return EINVAL;
600 else {
601 iov.iov_base = addr;
602 iov.iov_len = sizeof(struct dbreg);
603 uio.uio_iov = &iov;
604 uio.uio_iovcnt = 1;
605 uio.uio_offset = 0;
606 uio.uio_resid = sizeof(struct dbreg);
607 uio.uio_segflg = UIO_SYSSPACE;
608 uio.uio_rw = write ? UIO_WRITE : UIO_READ;
dadab5e9 609 uio.uio_td = curp->p_thread;
984263bc
MD
610 return (procfs_dodbregs(curp, p, NULL, &uio));
611 }
612#endif /* defined(PT_SETDBREGS) || defined(PT_GETDBREGS) */
613
614 default:
615 break;
616 }
617
618 return 0;
619}
620
621int
622trace_req(p)
623 struct proc *p;
624{
625 return 1;
626}
627
628/*
629 * stopevent()
270ac911
MD
630 *
631 * Stop a process because of a procfs event. Stay stopped until p->p_step
632 * is cleared (cleared by PIOCCONT in procfs).
633 *
634 * MPSAFE
984263bc 635 */
984263bc 636void
a2a5ad0d
MD
637stopevent(struct proc *p, unsigned int event, unsigned int val)
638{
984263bc
MD
639 p->p_step = 1;
640
641 do {
a2a5ad0d
MD
642 crit_enter();
643 wakeup(&p->p_stype); /* Wake up any PIOCWAIT'ing procs */
984263bc
MD
644 p->p_xstat = val;
645 p->p_stype = event; /* Which event caused the stop? */
377d4740 646 tsleep(&p->p_step, 0, "stopevent", 0);
a2a5ad0d 647 crit_exit();
984263bc
MD
648 } while (p->p_step);
649}
a2a5ad0d 650