procfs_linux.c revision 1.39.6.3 1 /* $NetBSD: procfs_linux.c,v 1.39.6.3 2007/11/14 19:04:48 joerg Exp $ */
2
3 /*
4 * Copyright (c) 2001 Wasabi Systems, Inc.
5 * All rights reserved.
6 *
7 * Written by Frank van der Linden for Wasabi Systems, Inc.
8 *
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
11 * are met:
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. All advertising materials mentioning features or use of this software
18 * must display the following acknowledgement:
19 * This product includes software developed for the NetBSD Project by
20 * Wasabi Systems, Inc.
21 * 4. The name of Wasabi Systems, Inc. may not be used to endorse
22 * or promote products derived from this software without specific prior
23 * written permission.
24 *
25 * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
26 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
27 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
28 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC
29 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
32 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
33 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
34 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35 * POSSIBILITY OF SUCH DAMAGE.
36 */
37
38 #include <sys/cdefs.h>
39 __KERNEL_RCSID(0, "$NetBSD: procfs_linux.c,v 1.39.6.3 2007/11/14 19:04:48 joerg Exp $");
40
41 #include <sys/param.h>
42 #include <sys/systm.h>
43 #include <sys/time.h>
44 #include <sys/kernel.h>
45 #include <sys/proc.h>
46 #include <sys/vnode.h>
47 #include <sys/exec.h>
48 #include <sys/resource.h>
49 #include <sys/resourcevar.h>
50 #include <sys/signal.h>
51 #include <sys/signalvar.h>
52 #include <sys/tty.h>
53 #include <sys/malloc.h>
54 #include <sys/mount.h>
55 #include <sys/conf.h>
56
57 #include <miscfs/procfs/procfs.h>
58 #include <compat/linux/common/linux_exec.h>
59
60 #include <uvm/uvm_extern.h>
61 #include <uvm/uvm.h>
62
63 extern struct devsw_conv *devsw_conv;
64 extern int max_devsw_convs;
65
66 #define PGTOB(p) ((unsigned long)(p) << PAGE_SHIFT)
67 #define PGTOKB(p) ((unsigned long)(p) << (PAGE_SHIFT - 10))
68
69 #define LBFSZ (8 * 1024)
70
71 static void
72 get_proc_size_info(struct lwp *l, unsigned long *stext, unsigned long *etext, unsigned long *sstack)
73 {
74 struct proc *p = l->l_proc;
75 struct vmspace *vm;
76 struct vm_map *map;
77 struct vm_map_entry *entry;
78
79 *stext = 0;
80 *etext = 0;
81 *sstack = 0;
82
83 proc_vmspace_getref(p, &vm);
84 map = &vm->vm_map;
85 vm_map_lock_read(map);
86
87 for (entry = map->header.next; entry != &map->header;
88 entry = entry->next) {
89 if (UVM_ET_ISSUBMAP(entry))
90 continue;
91 /* assume text is the first entry */
92 if (*stext == *etext) {
93 *stext = entry->start;
94 *etext = entry->end;
95 break;
96 }
97 }
98 #ifdef LINUX_USRSTACK32
99 if (strcmp(p->p_emul->e_name, "linux32") == 0 &&
100 LINUX_USRSTACK32 < USRSTACK32)
101 *sstack = (unsigned long)LINUX_USRSTACK32;
102 else
103 #endif
104 #ifdef LINUX_USRSTACK
105 if (strcmp(p->p_emul->e_name, "linux") == 0 &&
106 LINUX_USRSTACK < USRSTACK)
107 *sstack = (unsigned long)LINUX_USRSTACK;
108 else
109 #endif
110 #ifdef USRSTACK32
111 if (strstr(p->p_emul->e_name, "32") != NULL)
112 *sstack = (unsigned long)USRSTACK32;
113 else
114 #endif
115 *sstack = (unsigned long)USRSTACK;
116
117 /*
118 * jdk 1.6 compares low <= addr && addr < high
119 * if we put addr == high, then the test fails
120 * so eat one page.
121 */
122 *sstack -= PAGE_SIZE;
123
124 vm_map_unlock_read(map);
125 uvmspace_free(vm);
126 }
127
128 /*
129 * Linux compatible /proc/meminfo. Only active when the -o linux
130 * mountflag is used.
131 */
132 int
133 procfs_domeminfo(struct lwp *curl, struct proc *p,
134 struct pfsnode *pfs, struct uio *uio)
135 {
136 char *bf;
137 int len;
138 int error = 0;
139
140 bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
141
142 len = snprintf(bf, LBFSZ,
143 " total: used: free: shared: buffers: cached:\n"
144 "Mem: %8lu %8lu %8lu %8lu %8lu %8lu\n"
145 "Swap: %8lu %8lu %8lu\n"
146 "MemTotal: %8lu kB\n"
147 "MemFree: %8lu kB\n"
148 "MemShared: %8lu kB\n"
149 "Buffers: %8lu kB\n"
150 "Cached: %8lu kB\n"
151 "SwapTotal: %8lu kB\n"
152 "SwapFree: %8lu kB\n",
153 PGTOB(uvmexp.npages),
154 PGTOB(uvmexp.npages - uvmexp.free),
155 PGTOB(uvmexp.free),
156 0L,
157 PGTOB(uvmexp.filepages),
158 PGTOB(uvmexp.anonpages + uvmexp.filepages + uvmexp.execpages),
159 PGTOB(uvmexp.swpages),
160 PGTOB(uvmexp.swpginuse),
161 PGTOB(uvmexp.swpages - uvmexp.swpginuse),
162 PGTOKB(uvmexp.npages),
163 PGTOKB(uvmexp.free),
164 0L,
165 PGTOKB(uvmexp.filepages),
166 PGTOKB(uvmexp.anonpages + uvmexp.filepages + uvmexp.execpages),
167 PGTOKB(uvmexp.swpages),
168 PGTOKB(uvmexp.swpages - uvmexp.swpginuse));
169
170 if (len == 0)
171 goto out;
172
173 error = uiomove_frombuf(bf, len, uio);
174 out:
175 free(bf, M_TEMP);
176 return error;
177 }
178
179 /*
180 * Linux compatible /proc/devices. Only active when the -o linux
181 * mountflag is used.
182 */
183 int
184 procfs_dodevices(struct lwp *curl, struct proc *p,
185 struct pfsnode *pfs, struct uio *uio)
186 {
187 char *bf;
188 int offset = 0;
189 int i, error = ENAMETOOLONG;
190 extern kmutex_t devsw_lock;
191
192 /* XXX elad - may need filtering. */
193
194 bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
195
196 offset += snprintf(&bf[offset], LBFSZ - offset, "Character devices:\n");
197 if (offset >= LBFSZ)
198 goto out;
199
200 mutex_enter(&devsw_lock);
201 for (i = 0; i < max_devsw_convs; i++) {
202 if ((devsw_conv[i].d_name == NULL) ||
203 (devsw_conv[i].d_cmajor == -1))
204 continue;
205
206 offset += snprintf(&bf[offset], LBFSZ - offset,
207 "%3d %s\n", devsw_conv[i].d_cmajor, devsw_conv[i].d_name);
208 if (offset >= LBFSZ) {
209 mutex_exit(&devsw_lock);
210 goto out;
211 }
212 }
213
214 offset += snprintf(&bf[offset], LBFSZ - offset, "\nBlock devices:\n");
215 if (offset >= LBFSZ) {
216 mutex_exit(&devsw_lock);
217 goto out;
218 }
219
220 for (i = 0; i < max_devsw_convs; i++) {
221 if ((devsw_conv[i].d_name == NULL) ||
222 (devsw_conv[i].d_bmajor == -1))
223 continue;
224
225 offset += snprintf(&bf[offset], LBFSZ - offset,
226 "%3d %s\n", devsw_conv[i].d_bmajor, devsw_conv[i].d_name);
227 if (offset >= LBFSZ) {
228 mutex_exit(&devsw_lock);
229 goto out;
230 }
231 }
232 mutex_exit(&devsw_lock);
233
234 error = uiomove_frombuf(bf, offset, uio);
235 out:
236 free(bf, M_TEMP);
237 return error;
238 }
239
240 /*
241 * Linux compatible /proc/stat. Only active when the -o linux
242 * mountflag is used.
243 */
244 int
245 procfs_docpustat(struct lwp *curl, struct proc *p,
246 struct pfsnode *pfs, struct uio *uio)
247 {
248 struct timeval runtime;
249 char *bf;
250 int error;
251 int len;
252 #if defined(MULTIPROCESSOR)
253 struct cpu_info *ci;
254 CPU_INFO_ITERATOR cii;
255 #endif
256 int i;
257
258 error = ENAMETOOLONG;
259 bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
260
261 len = snprintf(bf, LBFSZ,
262 "cpu %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64 "\n",
263 curcpu()->ci_schedstate.spc_cp_time[CP_USER],
264 curcpu()->ci_schedstate.spc_cp_time[CP_NICE],
265 curcpu()->ci_schedstate.spc_cp_time[CP_SYS] /*+ [CP_INTR]*/,
266 curcpu()->ci_schedstate.spc_cp_time[CP_IDLE]);
267 if (len == 0)
268 goto out;
269
270 #if defined(MULTIPROCESSOR)
271 #define ALLCPUS CPU_INFO_FOREACH(cii, ci)
272 #define CPUNAME ci
273 #else
274 #define ALLCPUS ; i < 1 ;
275 #define CPUNAME curcpu()
276 #endif
277
278 i = 0;
279 for (ALLCPUS) {
280 len += snprintf(&bf[len], LBFSZ - len,
281 "cpu%d %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64
282 "\n", i,
283 CPUNAME->ci_schedstate.spc_cp_time[CP_USER],
284 CPUNAME->ci_schedstate.spc_cp_time[CP_NICE],
285 CPUNAME->ci_schedstate.spc_cp_time[CP_SYS],
286 CPUNAME->ci_schedstate.spc_cp_time[CP_IDLE]);
287 if (len >= LBFSZ)
288 goto out;
289 i += 1;
290 }
291
292 timersub(&curlwp->l_stime, &boottime, &runtime);
293 len += snprintf(&bf[len], LBFSZ - len,
294 "disk 0 0 0 0\n"
295 "page %u %u\n"
296 "swap %u %u\n"
297 "intr %u\n"
298 "ctxt %u\n"
299 "btime %lld\n",
300 uvmexp.pageins, uvmexp.pdpageouts,
301 uvmexp.pgswapin, uvmexp.pgswapout,
302 uvmexp.intrs,
303 uvmexp.swtch,
304 (long long)boottime.tv_sec);
305 if (len >= LBFSZ)
306 goto out;
307
308 error = uiomove_frombuf(bf, len, uio);
309 out:
310 free(bf, M_TEMP);
311 return error;
312 }
313
314 /*
315 * Linux compatible /proc/loadavg. Only active when the -o linux
316 * mountflag is used.
317 */
318 int
319 procfs_doloadavg(struct lwp *curl, struct proc *p,
320 struct pfsnode *pfs, struct uio *uio)
321 {
322 char *bf;
323 int error;
324 int len;
325
326 error = ENAMETOOLONG;
327 bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
328
329 averunnable.fscale = FSCALE;
330 len = snprintf(bf, LBFSZ,
331 "%d.%02d %d.%02d %d.%02d %d/%d %d\n",
332 (int)(averunnable.ldavg[0] / averunnable.fscale),
333 (int)(averunnable.ldavg[0] * 100 / averunnable.fscale % 100),
334 (int)(averunnable.ldavg[1] / averunnable.fscale),
335 (int)(averunnable.ldavg[1] * 100 / averunnable.fscale % 100),
336 (int)(averunnable.ldavg[2] / averunnable.fscale),
337 (int)(averunnable.ldavg[2] * 100 / averunnable.fscale % 100),
338 1, /* number of ONPROC processes */
339 nprocs,
340 30000); /* last pid */
341 if (len == 0)
342 goto out;
343
344 error = uiomove_frombuf(bf, len, uio);
345 out:
346 free(bf, M_TEMP);
347 return error;
348 }
349
350 /*
351 * Linux compatible /proc/<pid>/statm. Only active when the -o linux
352 * mountflag is used.
353 */
354 int
355 procfs_do_pid_statm(struct lwp *curl, struct lwp *l,
356 struct pfsnode *pfs, struct uio *uio)
357 {
358 struct vmspace *vm;
359 struct proc *p = l->l_proc;
360 struct rusage *ru = &p->p_stats->p_ru;
361 char *bf;
362 int error;
363 int len;
364
365 error = ENAMETOOLONG;
366 bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
367
368 /* XXX - we use values from vmspace, since dsl says that ru figures
369 are always 0 except for zombies. See kvm_proc.c::kvm_getproc2() */
370 if ((error = proc_vmspace_getref(p, &vm)) != 0) {
371 goto out;
372 }
373
374 len = snprintf(bf, LBFSZ,
375 "%lu %lu %lu %lu %lu %lu %lu\n",
376 (unsigned long)(vm->vm_tsize + vm->vm_dsize + vm->vm_ssize), /* size */
377 (unsigned long)(vm->vm_rssize), /* resident */
378 (unsigned long)(ru->ru_ixrss), /* shared */
379 (unsigned long)(vm->vm_tsize), /* text size in pages */
380 (unsigned long)(vm->vm_dsize), /* data size in pages */
381 (unsigned long)(vm->vm_ssize), /* stack size in pages */
382 (unsigned long) 0);
383
384 if (len == 0)
385 goto out;
386
387 error = uiomove_frombuf(bf, len, uio);
388 out:
389 free(bf, M_TEMP);
390 return error;
391 }
392
393 #define USEC_2_TICKS(x) ((x) / 10000)
394
395 /*
396 * Linux compatible /proc/<pid>/stat. Only active when the -o linux
397 * mountflag is used.
398 */
399 int
400 procfs_do_pid_stat(struct lwp *curl, struct lwp *l,
401 struct pfsnode *pfs, struct uio *uio)
402 {
403 char *bf;
404 struct proc *p = l->l_proc;
405 int len;
406 struct tty *tty = p->p_session->s_ttyp;
407 struct rusage *ru = &p->p_stats->p_ru;
408 struct rusage *cru = &p->p_stats->p_cru;
409 unsigned long stext = 0, etext = 0, sstack = 0;
410 struct timeval rt;
411 struct vmspace *vm;
412 int error = 0;
413
414 bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
415
416 if ((error = proc_vmspace_getref(p, &vm)) != 0) {
417 goto out;
418 }
419
420 get_proc_size_info(l, &stext, &etext, &sstack);
421
422 mutex_enter(&proclist_lock);
423 mutex_enter(&p->p_mutex);
424 mutex_enter(&p->p_smutex);
425
426 calcru(p, NULL, NULL, NULL, &rt);
427
428 len = snprintf(bf, LBFSZ,
429 "%d (%s) %c %d %d %d %d %d "
430 "%u "
431 "%lu %lu %lu %lu %lu %lu %lu %lu "
432 "%d %d %d "
433 "%lu %lu %lu %lu %" PRIu64 " "
434 "%lu %lu %lu "
435 "%u %u "
436 "%u %u %u %u "
437 "%lu %lu %lu %d %d\n",
438
439 p->p_pid,
440 p->p_comm,
441 "0IR3SZD"[(p->p_stat > 6) ? 0 : (int)p->p_stat],
442 (p->p_pptr != NULL) ? p->p_pptr->p_pid : 0,
443
444 p->p_pgid,
445 p->p_session->s_sid,
446 tty ? tty->t_dev : 0,
447 (tty && tty->t_pgrp) ? tty->t_pgrp->pg_id : 0,
448
449 p->p_flag,
450
451 ru->ru_minflt,
452 cru->ru_minflt,
453 ru->ru_majflt,
454 cru->ru_majflt,
455 USEC_2_TICKS(ru->ru_utime.tv_usec),
456 USEC_2_TICKS(ru->ru_stime.tv_usec),
457 USEC_2_TICKS(cru->ru_utime.tv_usec),
458 USEC_2_TICKS(cru->ru_stime.tv_usec),
459
460 l->l_priority, /* XXX: priority */
461 p->p_nice - 20,
462 0,
463
464 rt.tv_sec,
465 p->p_stats->p_start.tv_sec,
466 (unsigned long)(vm->vm_tsize + vm->vm_dsize + vm->vm_ssize), /* size */
467 (unsigned long)(vm->vm_rssize), /* resident */
468 p->p_rlimit[RLIMIT_RSS].rlim_cur,
469
470 stext, /* start code */
471 etext, /* end code */
472 sstack, /* mm start stack */
473 0, /* XXX: pc */
474 0, /* XXX: sp */
475 p->p_sigpend.sp_set.__bits[0], /* XXX: pending */
476 0, /* XXX: held */
477 p->p_sigctx.ps_sigignore.__bits[0], /* ignored */
478 p->p_sigctx.ps_sigcatch.__bits[0], /* caught */
479
480 (unsigned long)(intptr_t)l->l_wchan,
481 ru->ru_nvcsw,
482 ru->ru_nivcsw,
483 p->p_exitsig,
484 0); /* XXX: processor */
485
486 mutex_exit(&p->p_smutex);
487 mutex_exit(&p->p_mutex);
488 mutex_exit(&proclist_lock);
489
490 if (len == 0)
491 goto out;
492
493 error = uiomove_frombuf(bf, len, uio);
494 out:
495 free(bf, M_TEMP);
496 return error;
497 }
498
499 int
500 procfs_docpuinfo(struct lwp *curl, struct proc *p,
501 struct pfsnode *pfs, struct uio *uio)
502 {
503 int len = LBFSZ;
504 char *bf = malloc(len, M_TEMP, M_WAITOK);
505 int error;
506
507 if (procfs_getcpuinfstr(bf, &len) < 0) {
508 error = ENOSPC;
509 goto done;
510 }
511
512 if (len == 0) {
513 error = 0;
514 goto done;
515 }
516
517 error = uiomove_frombuf(bf, len, uio);
518 done:
519 free(bf, M_TEMP);
520 return error;
521 }
522
523 int
524 procfs_douptime(struct lwp *curl, struct proc *p,
525 struct pfsnode *pfs, struct uio *uio)
526 {
527 char *bf;
528 int len;
529 struct timeval runtime;
530 u_int64_t idle;
531 int error = 0;
532
533 bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
534
535 timersub(&curlwp->l_stime, &boottime, &runtime);
536 idle = curcpu()->ci_schedstate.spc_cp_time[CP_IDLE];
537 len = snprintf(bf, LBFSZ,
538 "%lu.%02lu %" PRIu64 ".%02" PRIu64 "\n",
539 runtime.tv_sec, runtime.tv_usec / 10000,
540 idle / hz, (((idle % hz) * 100) / hz) % 100);
541
542 if (len == 0)
543 goto out;
544
545 error = uiomove_frombuf(bf, len, uio);
546 out:
547 free(bf, M_TEMP);
548 return error;
549 }
550
551 int
552 procfs_domounts(struct lwp *curl, struct proc *p,
553 struct pfsnode *pfs, struct uio *uio)
554 {
555 char *bf, *mtab = NULL;
556 const char *fsname;
557 size_t len, mtabsz = 0;
558 struct mount *mp, *nmp;
559 struct statvfs *sfs;
560 int error = 0;
561
562 bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
563 mutex_enter(&mountlist_lock);
564 for (mp = CIRCLEQ_FIRST(&mountlist); mp != (void *)&mountlist;
565 mp = nmp) {
566 if (vfs_busy(mp, LK_NOWAIT, &mountlist_lock)) {
567 nmp = CIRCLEQ_NEXT(mp, mnt_list);
568 continue;
569 }
570
571 sfs = &mp->mnt_stat;
572
573 /* Linux uses different names for some filesystems */
574 fsname = sfs->f_fstypename;
575 if (strcmp(fsname, "procfs") == 0)
576 fsname = "proc";
577 else if (strcmp(fsname, "ext2fs") == 0)
578 fsname = "ext2";
579
580 len = snprintf(bf, LBFSZ, "%s %s %s %s%s%s%s%s%s 0 0\n",
581 sfs->f_mntfromname,
582 sfs->f_mntonname,
583 fsname,
584 (mp->mnt_flag & MNT_RDONLY) ? "ro" : "rw",
585 (mp->mnt_flag & MNT_NOSUID) ? ",nosuid" : "",
586 (mp->mnt_flag & MNT_NOEXEC) ? ",noexec" : "",
587 (mp->mnt_flag & MNT_NODEV) ? ",nodev" : "",
588 (mp->mnt_flag & MNT_SYNCHRONOUS) ? ",sync" : "",
589 (mp->mnt_flag & MNT_NOATIME) ? ",noatime" : ""
590 );
591
592 mtab = realloc(mtab, mtabsz + len, M_TEMP, M_WAITOK);
593 memcpy(mtab + mtabsz, bf, len);
594 mtabsz += len;
595
596 mutex_enter(&mountlist_lock);
597 nmp = CIRCLEQ_NEXT(mp, mnt_list);
598 vfs_unbusy(mp);
599 }
600 mutex_exit(&mountlist_lock);
601 free(bf, M_TEMP);
602
603 if (mtabsz > 0) {
604 error = uiomove_frombuf(mtab, mtabsz, uio);
605 free(mtab, M_TEMP);
606 }
607
608 return error;
609 }
610