Initial import from FreeBSD RELENG_4:
[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 $
32 */
33
34#include <sys/param.h>
35#include <sys/systm.h>
36#include <sys/sysproto.h>
37#include <sys/proc.h>
38#include <sys/vnode.h>
39#include <sys/ptrace.h>
40
41#include <machine/reg.h>
42#include <vm/vm.h>
43#include <sys/lock.h>
44#include <vm/pmap.h>
45#include <vm/vm_map.h>
46#include <vm/vm_page.h>
47
48#include <sys/user.h>
49#include <miscfs/procfs/procfs.h>
50
51/* use the equivalent procfs code */
52#if 0
53static int
54pread (struct proc *procp, unsigned int addr, unsigned int *retval) {
55 int rv;
56 vm_map_t map, tmap;
57 vm_object_t object;
58 vm_offset_t kva = 0;
59 int page_offset; /* offset into page */
60 vm_offset_t pageno; /* page number */
61 vm_map_entry_t out_entry;
62 vm_prot_t out_prot;
63 boolean_t wired;
64 vm_pindex_t pindex;
65
66 /* Map page into kernel space */
67
68 map = &procp->p_vmspace->vm_map;
69
70 page_offset = addr - trunc_page(addr);
71 pageno = trunc_page(addr);
72
73 tmap = map;
74 rv = vm_map_lookup (&tmap, pageno, VM_PROT_READ, &out_entry,
75 &object, &pindex, &out_prot, &wired);
76
77 if (rv != KERN_SUCCESS)
78 return EINVAL;
79
80 vm_map_lookup_done (tmap, out_entry);
81
82 /* Find space in kernel_map for the page we're interested in */
83 rv = vm_map_find (kernel_map, object, IDX_TO_OFF(pindex),
84 &kva, PAGE_SIZE, 0, VM_PROT_ALL, VM_PROT_ALL, 0);
85
86 if (!rv) {
87 vm_object_reference (object);
88
89 rv = vm_map_pageable (kernel_map, kva, kva + PAGE_SIZE, 0);
90 if (!rv) {
91 *retval = 0;
92 bcopy ((caddr_t)kva + page_offset,
93 retval, sizeof *retval);
94 }
95 vm_map_remove (kernel_map, kva, kva + PAGE_SIZE);
96 }
97
98 return rv;
99}
100
101static int
102pwrite (struct proc *procp, unsigned int addr, unsigned int datum) {
103 int rv;
104 vm_map_t map, tmap;
105 vm_object_t object;
106 vm_offset_t kva = 0;
107 int page_offset; /* offset into page */
108 vm_offset_t pageno; /* page number */
109 vm_map_entry_t out_entry;
110 vm_prot_t out_prot;
111 boolean_t wired;
112 vm_pindex_t pindex;
113 boolean_t fix_prot = 0;
114
115 /* Map page into kernel space */
116
117 map = &procp->p_vmspace->vm_map;
118
119 page_offset = addr - trunc_page(addr);
120 pageno = trunc_page(addr);
121
122 /*
123 * Check the permissions for the area we're interested in.
124 */
125
126 if (vm_map_check_protection (map, pageno, pageno + PAGE_SIZE,
127 VM_PROT_WRITE) == FALSE) {
128 /*
129 * If the page was not writable, we make it so.
130 * XXX It is possible a page may *not* be read/executable,
131 * if a process changes that!
132 */
133 fix_prot = 1;
134 /* The page isn't writable, so let's try making it so... */
135 if ((rv = vm_map_protect (map, pageno, pageno + PAGE_SIZE,
136 VM_PROT_ALL, 0)) != KERN_SUCCESS)
137 return EFAULT; /* I guess... */
138 }
139
140 /*
141 * Now we need to get the page. out_entry, out_prot, wired, and
142 * single_use aren't used. One would think the vm code would be
143 * a *bit* nicer... We use tmap because vm_map_lookup() can
144 * change the map argument.
145 */
146
147 tmap = map;
148 rv = vm_map_lookup (&tmap, pageno, VM_PROT_WRITE, &out_entry,
149 &object, &pindex, &out_prot, &wired);
150 if (rv != KERN_SUCCESS) {
151 return EINVAL;
152 }
153
154 /*
155 * Okay, we've got the page. Let's release tmap.
156 */
157
158 vm_map_lookup_done (tmap, out_entry);
159
160 /*
161 * Fault the page in...
162 */
163
164 rv = vm_fault(map, pageno, VM_PROT_WRITE|VM_PROT_READ, FALSE);
165 if (rv != KERN_SUCCESS)
166 return EFAULT;
167
168 /* Find space in kernel_map for the page we're interested in */
169 rv = vm_map_find (kernel_map, object, IDX_TO_OFF(pindex),
170 &kva, PAGE_SIZE, 0,
171 VM_PROT_ALL, VM_PROT_ALL, 0);
172 if (!rv) {
173 vm_object_reference (object);
174
175 rv = vm_map_pageable (kernel_map, kva, kva + PAGE_SIZE, 0);
176 if (!rv) {
177 bcopy (&datum, (caddr_t)kva + page_offset, sizeof datum);
178 }
179 vm_map_remove (kernel_map, kva, kva + PAGE_SIZE);
180 }
181
182 if (fix_prot)
183 vm_map_protect (map, pageno, pageno + PAGE_SIZE,
184 VM_PROT_READ|VM_PROT_EXECUTE, 0);
185 return rv;
186}
187#endif
188
189/*
190 * Process debugging system call.
191 */
192#ifndef _SYS_SYSPROTO_H_
193struct ptrace_args {
194 int req;
195 pid_t pid;
196 caddr_t addr;
197 int data;
198};
199#endif
200
201int
202ptrace(struct proc *p, struct ptrace_args *uap)
203{
204 /*
205 * XXX this obfuscation is to reduce stack usage, but the register
206 * structs may be too large to put on the stack anyway.
207 */
208 union {
209 struct ptrace_io_desc piod;
210 struct dbreg dbreg;
211 struct fpreg fpreg;
212 struct reg reg;
213 } r;
214 void *addr;
215 int error = 0;
216
217 addr = &r;
218 switch (uap->req) {
219 case PT_GETREGS:
220 case PT_GETFPREGS:
221#ifdef PT_GETDBREGS
222 case PT_GETDBREGS:
223#endif
224 break;
225 case PT_SETREGS:
226 error = copyin(uap->addr, &r.reg, sizeof r.reg);
227 break;
228 case PT_SETFPREGS:
229 error = copyin(uap->addr, &r.fpreg, sizeof r.fpreg);
230 break;
231#ifdef PT_SETDBREGS
232 case PT_SETDBREGS:
233 error = copyin(uap->addr, &r.dbreg, sizeof r.dbreg);
234 break;
235#endif
236 case PT_IO:
237 error = copyin(uap->addr, &r.piod, sizeof r.piod);
238 break;
239 default:
240 addr = uap->addr;
241 }
242 if (error)
243 return (error);
244
245 error = kern_ptrace(p, uap->req, uap->pid, addr, uap->data);
246 if (error)
247 return (error);
248
249 switch (uap->req) {
250 case PT_IO:
251 (void)copyout(&r.piod, uap->addr, sizeof r.piod);
252 break;
253 case PT_GETREGS:
254 error = copyout(&r.reg, uap->addr, sizeof r.reg);
255 break;
256 case PT_GETFPREGS:
257 error = copyout(&r.fpreg, uap->addr, sizeof r.fpreg);
258 break;
259#ifdef PT_GETDBREGS
260 case PT_GETDBREGS:
261 error = copyout(&r.dbreg, uap->addr, sizeof r.dbreg);
262 break;
263#endif
264 }
265
266 return (error);
267}
268
269int
270kern_ptrace(struct proc *curp, int req, pid_t pid, void *addr, int data)
271{
272 struct proc *p, *pp;
273 struct iovec iov;
274 struct uio uio;
275 struct ptrace_io_desc *piod;
276 int error = 0;
277 int write, tmp, s;
278
279 write = 0;
280 if (req == PT_TRACE_ME)
281 p = curp;
282 else {
283 if ((p = pfind(pid)) == NULL)
284 return ESRCH;
285 }
286 if (!PRISON_CHECK(curp, p))
287 return (ESRCH);
288
289 /* Can't trace a process that's currently exec'ing. */
290 if ((p->p_flag & P_INEXEC) != 0)
291 return EAGAIN;
292
293 /*
294 * Permissions check
295 */
296 switch (req) {
297 case PT_TRACE_ME:
298 /* Always legal. */
299 break;
300
301 case PT_ATTACH:
302 /* Self */
303 if (p->p_pid == curp->p_pid)
304 return EINVAL;
305
306 /* Already traced */
307 if (p->p_flag & P_TRACED)
308 return EBUSY;
309
310 if (curp->p_flag & P_TRACED)
311 for (pp = curp->p_pptr; pp != NULL; pp = pp->p_pptr)
312 if (pp == p)
313 return (EINVAL);
314
315 /* not owned by you, has done setuid (unless you're root) */
316 if ((p->p_cred->p_ruid != curp->p_cred->p_ruid) ||
317 (p->p_flag & P_SUGID)) {
318 if ((error = suser(curp)) != 0)
319 return error;
320 }
321
322 /* can't trace init when securelevel > 0 */
323 if (securelevel > 0 && p->p_pid == 1)
324 return EPERM;
325
326 /* OK */
327 break;
328
329 case PT_READ_I:
330 case PT_READ_D:
331 case PT_WRITE_I:
332 case PT_WRITE_D:
333 case PT_IO:
334 case PT_CONTINUE:
335 case PT_KILL:
336 case PT_STEP:
337 case PT_DETACH:
338#ifdef PT_GETREGS
339 case PT_GETREGS:
340#endif
341#ifdef PT_SETREGS
342 case PT_SETREGS:
343#endif
344#ifdef PT_GETFPREGS
345 case PT_GETFPREGS:
346#endif
347#ifdef PT_SETFPREGS
348 case PT_SETFPREGS:
349#endif
350#ifdef PT_GETDBREGS
351 case PT_GETDBREGS:
352#endif
353#ifdef PT_SETDBREGS
354 case PT_SETDBREGS:
355#endif
356 /* not being traced... */
357 if ((p->p_flag & P_TRACED) == 0)
358 return EPERM;
359
360 /* not being traced by YOU */
361 if (p->p_pptr != curp)
362 return EBUSY;
363
364 /* not currently stopped */
365 if (p->p_stat != SSTOP || (p->p_flag & P_WAITED) == 0)
366 return EBUSY;
367
368 /* OK */
369 break;
370
371 default:
372 return EINVAL;
373 }
374
375#ifdef FIX_SSTEP
376 /*
377 * Single step fixup ala procfs
378 */
379 FIX_SSTEP(p);
380#endif
381
382 /*
383 * Actually do the requests
384 */
385
386 curp->p_retval[0] = 0;
387
388 switch (req) {
389 case PT_TRACE_ME:
390 /* set my trace flag and "owner" so it can read/write me */
391 p->p_flag |= P_TRACED;
392 p->p_oppid = p->p_pptr->p_pid;
393 return 0;
394
395 case PT_ATTACH:
396 /* security check done above */
397 p->p_flag |= P_TRACED;
398 p->p_oppid = p->p_pptr->p_pid;
399 if (p->p_pptr != curp)
400 proc_reparent(p, curp);
401 data = SIGSTOP;
402 goto sendsig; /* in PT_CONTINUE below */
403
404 case PT_STEP:
405 case PT_CONTINUE:
406 case PT_DETACH:
407 if ((req != PT_STEP) && ((unsigned)data > _SIG_MAXSIG))
408 return EINVAL;
409
410 PHOLD(p);
411
412 if (req == PT_STEP) {
413 if ((error = ptrace_single_step (p))) {
414 PRELE(p);
415 return error;
416 }
417 }
418
419 if (addr != (void *)1) {
420 if ((error = ptrace_set_pc (p,
421 (u_long)(uintfptr_t)addr))) {
422 PRELE(p);
423 return error;
424 }
425 }
426 PRELE(p);
427
428 if (req == PT_DETACH) {
429 /* reset process parent */
430 if (p->p_oppid != p->p_pptr->p_pid) {
431 struct proc *pp;
432
433 pp = pfind(p->p_oppid);
434 proc_reparent(p, pp ? pp : initproc);
435 }
436
437 p->p_flag &= ~(P_TRACED | P_WAITED);
438 p->p_oppid = 0;
439
440 /* should we send SIGCHLD? */
441 }
442
443 sendsig:
444 /* deliver or queue signal */
445 s = splhigh();
446 if (p->p_stat == SSTOP) {
447 p->p_xstat = data;
448 setrunnable(p);
449 } else if (data) {
450 psignal(p, data);
451 }
452 splx(s);
453 return 0;
454
455 case PT_WRITE_I:
456 case PT_WRITE_D:
457 write = 1;
458 /* fallthrough */
459 case PT_READ_I:
460 case PT_READ_D:
461 tmp = 0;
462 /* write = 0 set above */
463 iov.iov_base = write ? (caddr_t)&data : (caddr_t)&tmp;
464 iov.iov_len = sizeof(int);
465 uio.uio_iov = &iov;
466 uio.uio_iovcnt = 1;
467 uio.uio_offset = (off_t)(uintptr_t)addr;
468 uio.uio_resid = sizeof(int);
469 uio.uio_segflg = UIO_SYSSPACE;
470 uio.uio_rw = write ? UIO_WRITE : UIO_READ;
471 uio.uio_procp = p;
472 error = procfs_domem(curp, p, NULL, &uio);
473 if (uio.uio_resid != 0) {
474 /*
475 * XXX procfs_domem() doesn't currently return ENOSPC,
476 * so I think write() can bogusly return 0.
477 * XXX what happens for short writes? We don't want
478 * to write partial data.
479 * XXX procfs_domem() returns EPERM for other invalid
480 * addresses. Convert this to EINVAL. Does this
481 * clobber returns of EPERM for other reasons?
482 */
483 if (error == 0 || error == ENOSPC || error == EPERM)
484 error = EINVAL; /* EOF */
485 }
486 if (!write)
487 curp->p_retval[0] = tmp;
488 return (error);
489
490 case PT_IO:
491 piod = addr;
492 iov.iov_base = piod->piod_addr;
493 iov.iov_len = piod->piod_len;
494 uio.uio_iov = &iov;
495 uio.uio_iovcnt = 1;
496 uio.uio_offset = (off_t)(uintptr_t)piod->piod_offs;
497 uio.uio_resid = piod->piod_len;
498 uio.uio_segflg = UIO_USERSPACE;
499 uio.uio_procp = p;
500 switch (piod->piod_op) {
501 case PIOD_READ_D:
502 case PIOD_READ_I:
503 uio.uio_rw = UIO_READ;
504 break;
505 case PIOD_WRITE_D:
506 case PIOD_WRITE_I:
507 uio.uio_rw = UIO_WRITE;
508 break;
509 default:
510 return (EINVAL);
511 }
512 error = procfs_domem(curp, p, NULL, &uio);
513 piod->piod_len -= uio.uio_resid;
514 return (error);
515
516 case PT_KILL:
517 data = SIGKILL;
518 goto sendsig; /* in PT_CONTINUE above */
519
520#ifdef PT_SETREGS
521 case PT_SETREGS:
522 write = 1;
523 /* fallthrough */
524#endif /* PT_SETREGS */
525#ifdef PT_GETREGS
526 case PT_GETREGS:
527 /* write = 0 above */
528#endif /* PT_SETREGS */
529#if defined(PT_SETREGS) || defined(PT_GETREGS)
530 if (!procfs_validregs(p)) /* no P_SYSTEM procs please */
531 return EINVAL;
532 else {
533 iov.iov_base = addr;
534 iov.iov_len = sizeof(struct reg);
535 uio.uio_iov = &iov;
536 uio.uio_iovcnt = 1;
537 uio.uio_offset = 0;
538 uio.uio_resid = sizeof(struct reg);
539 uio.uio_segflg = UIO_SYSSPACE;
540 uio.uio_rw = write ? UIO_WRITE : UIO_READ;
541 uio.uio_procp = curp;
542 return (procfs_doregs(curp, p, NULL, &uio));
543 }
544#endif /* defined(PT_SETREGS) || defined(PT_GETREGS) */
545
546#ifdef PT_SETFPREGS
547 case PT_SETFPREGS:
548 write = 1;
549 /* fallthrough */
550#endif /* PT_SETFPREGS */
551#ifdef PT_GETFPREGS
552 case PT_GETFPREGS:
553 /* write = 0 above */
554#endif /* PT_SETFPREGS */
555#if defined(PT_SETFPREGS) || defined(PT_GETFPREGS)
556 if (!procfs_validfpregs(p)) /* no P_SYSTEM procs please */
557 return EINVAL;
558 else {
559 iov.iov_base = addr;
560 iov.iov_len = sizeof(struct fpreg);
561 uio.uio_iov = &iov;
562 uio.uio_iovcnt = 1;
563 uio.uio_offset = 0;
564 uio.uio_resid = sizeof(struct fpreg);
565 uio.uio_segflg = UIO_SYSSPACE;
566 uio.uio_rw = write ? UIO_WRITE : UIO_READ;
567 uio.uio_procp = curp;
568 return (procfs_dofpregs(curp, p, NULL, &uio));
569 }
570#endif /* defined(PT_SETFPREGS) || defined(PT_GETFPREGS) */
571
572#ifdef PT_SETDBREGS
573 case PT_SETDBREGS:
574 write = 1;
575 /* fallthrough */
576#endif /* PT_SETDBREGS */
577#ifdef PT_GETDBREGS
578 case PT_GETDBREGS:
579 /* write = 0 above */
580#endif /* PT_SETDBREGS */
581#if defined(PT_SETDBREGS) || defined(PT_GETDBREGS)
582 if (!procfs_validdbregs(p)) /* no P_SYSTEM procs please */
583 return EINVAL;
584 else {
585 iov.iov_base = addr;
586 iov.iov_len = sizeof(struct dbreg);
587 uio.uio_iov = &iov;
588 uio.uio_iovcnt = 1;
589 uio.uio_offset = 0;
590 uio.uio_resid = sizeof(struct dbreg);
591 uio.uio_segflg = UIO_SYSSPACE;
592 uio.uio_rw = write ? UIO_WRITE : UIO_READ;
593 uio.uio_procp = curp;
594 return (procfs_dodbregs(curp, p, NULL, &uio));
595 }
596#endif /* defined(PT_SETDBREGS) || defined(PT_GETDBREGS) */
597
598 default:
599 break;
600 }
601
602 return 0;
603}
604
605int
606trace_req(p)
607 struct proc *p;
608{
609 return 1;
610}
611
612/*
613 * stopevent()
614 * Stop a process because of a procfs event;
615 * stay stopped until p->p_step is cleared
616 * (cleared by PIOCCONT in procfs).
617 */
618
619void
620stopevent(struct proc *p, unsigned int event, unsigned int val) {
621 p->p_step = 1;
622
623 do {
624 p->p_xstat = val;
625 p->p_stype = event; /* Which event caused the stop? */
626 wakeup(&p->p_stype); /* Wake up any PIOCWAIT'ing procs */
627 tsleep(&p->p_step, PWAIT, "stopevent", 0);
628 } while (p->p_step);
629}