xref: /netbsd-src/sys/miscfs/procfs/procfs_linux.c (revision 4b8a875ae268d9fd57ce79aae2093671f6afc1bc)
1 /*      $NetBSD: procfs_linux.c,v 1.85 2020/06/11 19:20:46 ad 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 1.85 2020/06/11 19:20:46 ad Exp $");
40 
41 #include <sys/param.h>
42 #include <sys/systm.h>
43 #include <sys/atomic.h>
44 #include <sys/time.h>
45 #include <sys/cpu.h>
46 #include <sys/kernel.h>
47 #include <sys/proc.h>
48 #include <sys/vnode.h>
49 #include <sys/exec.h>
50 #include <sys/resource.h>
51 #include <sys/resourcevar.h>
52 #include <sys/signal.h>
53 #include <sys/signalvar.h>
54 #include <sys/tty.h>
55 #include <sys/malloc.h>
56 #include <sys/mount.h>
57 #include <sys/conf.h>
58 #include <sys/sysctl.h>
59 #include <sys/kauth.h>
60 #include <sys/filedesc.h>
61 
62 #include <miscfs/procfs/procfs.h>
63 
64 #include <compat/linux/common/linux_exec.h>
65 #include <compat/linux32/common/linux32_sysctl.h>
66 
67 #include <uvm/uvm_extern.h>
68 #include <uvm/uvm.h>
69 
70 extern struct devsw_conv *devsw_conv;
71 extern int max_devsw_convs;
72 
73 #define PGTOB(p)	((unsigned long)(p) << PAGE_SHIFT)
74 #define PGTOKB(p)	((unsigned long)(p) << (PAGE_SHIFT - 10))
75 
76 #define LBFSZ (8 * 1024)
77 
78 static void
79 get_proc_size_info(struct proc *p, struct vm_map *map, unsigned long *stext,
80     unsigned long *etext, unsigned long *sstack)
81 {
82 	struct vm_map_entry *entry;
83 
84 	*stext = 0;
85 	*etext = 0;
86 	*sstack = 0;
87 
88 	vm_map_lock_read(map);
89 
90 	for (entry = map->header.next; entry != &map->header;
91 	    entry = entry->next) {
92 		if (UVM_ET_ISSUBMAP(entry))
93 			continue;
94 		/* assume text is the first entry */
95 		if (*stext == *etext) {
96 			*stext = entry->start;
97 			*etext = entry->end;
98 			break;
99 		}
100 	}
101 #if defined(LINUX_USRSTACK32) && defined(USRSTACK32)
102 	if (strcmp(p->p_emul->e_name, "linux32") == 0 &&
103 	    LINUX_USRSTACK32 < USRSTACK32)
104 		*sstack = (unsigned long)LINUX_USRSTACK32;
105 	else
106 #endif
107 #ifdef LINUX_USRSTACK
108 	if (strcmp(p->p_emul->e_name, "linux") == 0 &&
109 	    LINUX_USRSTACK < USRSTACK)
110 		*sstack = (unsigned long)LINUX_USRSTACK;
111 	else
112 #endif
113 #ifdef	USRSTACK32
114 	if (strstr(p->p_emul->e_name, "32") != NULL)
115 		*sstack = (unsigned long)USRSTACK32;
116 	else
117 #endif
118 		*sstack = (unsigned long)USRSTACK;
119 
120 	/*
121 	 * jdk 1.6 compares low <= addr && addr < high
122 	 * if we put addr == high, then the test fails
123 	 * so eat one page.
124 	 */
125 	*sstack -= PAGE_SIZE;
126 
127 	vm_map_unlock_read(map);
128 }
129 
130 /*
131  * Linux compatible /proc/meminfo. Only active when the -o linux
132  * mountflag is used.
133  */
134 int
135 procfs_domeminfo(struct lwp *curl, struct proc *p,
136     struct pfsnode *pfs, struct uio *uio)
137 {
138 	char *bf;
139 	int len;
140 	int error = 0;
141 	long filepg, anonpg, execpg, freepg;
142 
143 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
144 
145 	cpu_count_sync_all();
146 	freepg = (long)uvm_availmem(true);
147 	filepg = (long)cpu_count_get(CPU_COUNT_FILEPAGES);
148 	anonpg = (long)cpu_count_get(CPU_COUNT_ANONPAGES);
149 	execpg = (long)cpu_count_get(CPU_COUNT_EXECPAGES);
150 
151 	len = snprintf(bf, LBFSZ,
152 		"        total:    used:    free:  shared: buffers: cached:\n"
153 		"Mem:  %8lu %8lu %8lu %8lu %8lu %8lu\n"
154 		"Swap: %8lu %8lu %8lu\n"
155 		"MemTotal:  %8lu kB\n"
156 		"MemFree:   %8lu kB\n"
157 		"MemShared: %8lu kB\n"
158 		"Buffers:   %8lu kB\n"
159 		"Cached:    %8lu kB\n"
160 		"SwapTotal: %8lu kB\n"
161 		"SwapFree:  %8lu kB\n",
162 		PGTOB(uvmexp.npages),
163 		PGTOB(uvmexp.npages - freepg),
164 		PGTOB(freepg),
165 		0L,
166 		PGTOB(filepg),
167 		PGTOB(anonpg + filepg + execpg),
168 		PGTOB(uvmexp.swpages),
169 		PGTOB(uvmexp.swpginuse),
170 		PGTOB(uvmexp.swpages - uvmexp.swpginuse),
171 		PGTOKB(uvmexp.npages),
172 		PGTOKB(freepg),
173 		0L,
174 		PGTOKB(freepg),
175 		PGTOKB(anonpg + filepg + execpg),
176 		PGTOKB(uvmexp.swpages),
177 		PGTOKB(uvmexp.swpages - uvmexp.swpginuse));
178 
179 	if (len == 0)
180 		goto out;
181 
182 	error = uiomove_frombuf(bf, len, uio);
183 out:
184 	free(bf, M_TEMP);
185 	return error;
186 }
187 
188 /*
189  * Linux compatible /proc/devices. Only active when the -o linux
190  * mountflag is used.
191  */
192 int
193 procfs_dodevices(struct lwp *curl, struct proc *p,
194     struct pfsnode *pfs, struct uio *uio)
195 {
196 	char *bf;
197 	int offset = 0;
198 	int i, error = ENAMETOOLONG;
199 
200 	/* XXX elad - may need filtering. */
201 
202 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
203 
204 	offset += snprintf(&bf[offset], LBFSZ - offset, "Character devices:\n");
205 	if (offset >= LBFSZ)
206 		goto out;
207 
208 	mutex_enter(&device_lock);
209 	for (i = 0; i < max_devsw_convs; i++) {
210 		if ((devsw_conv[i].d_name == NULL) ||
211 		    (devsw_conv[i].d_cmajor == -1))
212 			continue;
213 
214 		offset += snprintf(&bf[offset], LBFSZ - offset,
215 		    "%3d %s\n", devsw_conv[i].d_cmajor, devsw_conv[i].d_name);
216 		if (offset >= LBFSZ) {
217 			mutex_exit(&device_lock);
218 			goto out;
219 		}
220 	}
221 
222 	offset += snprintf(&bf[offset], LBFSZ - offset, "\nBlock devices:\n");
223 	if (offset >= LBFSZ) {
224 		mutex_exit(&device_lock);
225 		goto out;
226 	}
227 
228 	for (i = 0; i < max_devsw_convs; i++) {
229 		if ((devsw_conv[i].d_name == NULL) ||
230 		    (devsw_conv[i].d_bmajor == -1))
231 			continue;
232 
233 		offset += snprintf(&bf[offset], LBFSZ - offset,
234 		    "%3d %s\n", devsw_conv[i].d_bmajor, devsw_conv[i].d_name);
235 		if (offset >= LBFSZ) {
236 			mutex_exit(&device_lock);
237 			goto out;
238 		}
239 	}
240 	mutex_exit(&device_lock);
241 
242 	error = uiomove_frombuf(bf, offset, uio);
243 out:
244 	free(bf, M_TEMP);
245 	return error;
246 }
247 
248 /*
249  * Linux compatible /proc/stat. Only active when the -o linux
250  * mountflag is used.
251  */
252 int
253 procfs_docpustat(struct lwp *curl, struct proc *p,
254     struct pfsnode *pfs, struct uio *uio)
255 {
256 	char		*bf;
257 	int	 	 error;
258 	int	 	 len;
259 #if defined(MULTIPROCESSOR)
260         struct cpu_info *ci;
261         CPU_INFO_ITERATOR cii;
262 #endif
263 	int	 	 i;
264 
265 	error = ENAMETOOLONG;
266 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
267 
268 	len = snprintf(bf, LBFSZ,
269 		"cpu %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64 "\n",
270 		curcpu()->ci_schedstate.spc_cp_time[CP_USER],
271 		curcpu()->ci_schedstate.spc_cp_time[CP_NICE],
272 		curcpu()->ci_schedstate.spc_cp_time[CP_SYS] /*+ [CP_INTR]*/,
273 		curcpu()->ci_schedstate.spc_cp_time[CP_IDLE]);
274 	if (len == 0)
275 		goto out;
276 
277 #if defined(MULTIPROCESSOR)
278 #define ALLCPUS	CPU_INFO_FOREACH(cii, ci)
279 #define CPUNAME	ci
280 #else
281 #define ALLCPUS	; i < 1 ;
282 #define CPUNAME	curcpu()
283 #endif
284 
285 	i = 0;
286 	for (ALLCPUS) {
287 		len += snprintf(&bf[len], LBFSZ - len,
288 			"cpu%d %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64
289 			"\n", i,
290 			CPUNAME->ci_schedstate.spc_cp_time[CP_USER],
291 			CPUNAME->ci_schedstate.spc_cp_time[CP_NICE],
292 			CPUNAME->ci_schedstate.spc_cp_time[CP_SYS],
293 			CPUNAME->ci_schedstate.spc_cp_time[CP_IDLE]);
294 		if (len >= LBFSZ)
295 			goto out;
296 		i += 1;
297 	}
298 
299 	cpu_count_sync_all();
300 
301 	struct timeval btv;
302 	getmicroboottime(&btv);
303 
304 	len += snprintf(&bf[len], LBFSZ - len,
305 			"disk 0 0 0 0\n"
306 			"page %u %u\n"
307 			"swap %u %u\n"
308 			"intr %"PRId64"\n"
309 			"ctxt %"PRId64"\n"
310 			"btime %"PRId64"\n",
311 			uvmexp.pageins, uvmexp.pdpageouts,
312 			uvmexp.pgswapin, uvmexp.pgswapout,
313 			cpu_count_get(CPU_COUNT_NINTR),
314 			cpu_count_get(CPU_COUNT_NSWTCH),
315 			btv.tv_sec);
316 	if (len >= LBFSZ)
317 		goto out;
318 
319 	error = uiomove_frombuf(bf, len, uio);
320 out:
321 	free(bf, M_TEMP);
322 	return error;
323 }
324 
325 /*
326  * Linux compatible /proc/loadavg. Only active when the -o linux
327  * mountflag is used.
328  */
329 int
330 procfs_doloadavg(struct lwp *curl, struct proc *p,
331     struct pfsnode *pfs, struct uio *uio)
332 {
333 	char	*bf;
334 	int 	 error;
335 	int 	 len;
336 
337 	error = ENAMETOOLONG;
338 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
339 
340 	averunnable.fscale = FSCALE;
341 	len = snprintf(bf, LBFSZ,
342 	        "%d.%02d %d.%02d %d.%02d %d/%d %d\n",
343 		(int)(averunnable.ldavg[0] / averunnable.fscale),
344 		(int)(averunnable.ldavg[0] * 100 / averunnable.fscale % 100),
345 		(int)(averunnable.ldavg[1] / averunnable.fscale),
346 		(int)(averunnable.ldavg[1] * 100 / averunnable.fscale % 100),
347 		(int)(averunnable.ldavg[2] / averunnable.fscale),
348 		(int)(averunnable.ldavg[2] * 100 / averunnable.fscale % 100),
349 		1,		/* number of ONPROC processes */
350 		atomic_load_relaxed(&nprocs),
351 		30000);		/* last pid */
352 	if (len == 0)
353 		goto out;
354 
355 	error = uiomove_frombuf(bf, len, uio);
356 out:
357 	free(bf, M_TEMP);
358 	return error;
359 }
360 
361 /*
362  * Linux compatible /proc/<pid>/statm. Only active when the -o linux
363  * mountflag is used.
364  */
365 int
366 procfs_do_pid_statm(struct lwp *curl, struct lwp *l,
367     struct pfsnode *pfs, struct uio *uio)
368 {
369 	struct vmspace	*vm;
370 	struct proc	*p = l->l_proc;
371 	char		*bf;
372 	int	 	 error;
373 	int	 	 len;
374 	struct kinfo_proc2 ki;
375 
376 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
377 
378 	/* XXX - we use values from vmspace, since dsl says that ru figures
379 	   are always 0 except for zombies. See kvm_proc.c::kvm_getproc2() */
380 	if ((error = proc_vmspace_getref(p, &vm)) != 0) {
381 		goto out;
382 	}
383 
384 	mutex_enter(&proc_lock);
385 	mutex_enter(p->p_lock);
386 
387 	/* retrieve RSS size */
388 	memset(&ki, 0, sizeof(ki));
389 	fill_kproc2(p, &ki, false, false);
390 
391 	mutex_exit(p->p_lock);
392 	mutex_exit(&proc_lock);
393 
394 	uvmspace_free(vm);
395 
396 	len = snprintf(bf, LBFSZ,
397 	        "%lu %lu %lu %lu %lu %lu %lu\n",
398 		(unsigned long)(ki.p_vm_msize),	/* size */
399 		(unsigned long)(ki.p_vm_rssize),/* resident */
400 		(unsigned long)(ki.p_uru_ixrss),/* shared */
401 		(unsigned long)(ki.p_vm_tsize),	/* text */
402 		(unsigned long) 0,		/* library (unused) */
403 		(unsigned long)(ki.p_vm_dsize + ki.p_vm_ssize),	/* data+stack */
404 		(unsigned long) 0);		/* dirty */
405 
406 	if (len == 0)
407 		goto out;
408 
409 	error = uiomove_frombuf(bf, len, uio);
410 out:
411 	free(bf, M_TEMP);
412 	return error;
413 }
414 
415 #define UTIME2TICKS(s,u)	(((uint64_t)(s) * 1000000 + (u)) / 10000)
416 
417 /*
418  * Linux compatible /proc/<pid>/stat. Only active when the -o linux
419  * mountflag is used.
420  */
421 int
422 procfs_do_pid_stat(struct lwp *curl, struct lwp *l,
423     struct pfsnode *pfs, struct uio *uio)
424 {
425 	char *bf;
426 	struct proc *p = l->l_proc;
427 	int len;
428 	struct rusage *cru = &p->p_stats->p_cru;
429 	unsigned long stext = 0, etext = 0, sstack = 0;
430 	struct timeval rt;
431 	struct vmspace	*vm;
432 	struct kinfo_proc2 ki;
433 	int error;
434 
435 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
436 
437 	if ((error = proc_vmspace_getref(p, &vm)) != 0) {
438 		goto out;
439 	}
440 
441 	get_proc_size_info(p, &vm->vm_map, &stext, &etext, &sstack);
442 
443 	mutex_enter(&proc_lock);
444 	mutex_enter(p->p_lock);
445 
446 	memset(&ki, 0, sizeof(ki));
447 	fill_kproc2(p, &ki, false, false);
448 	calcru(p, NULL, NULL, NULL, &rt);
449 
450 	len = snprintf(bf, LBFSZ,
451 	    "%d (%s) %c %d %d %d %u %d "
452 	    "%u "
453 	    "%"PRIu64" %lu %"PRIu64" %lu %"PRIu64" %"PRIu64" %"PRIu64" %"PRIu64" "
454 	    "%d %d %"PRIu64" "
455 	    "%lld %"PRIu64" %"PRId64" %lu %"PRIu64" "
456 	    "%lu %lu %lu "
457 	    "%u %u "
458 	    "%u %u %u %u "
459 	    "%"PRIu64" %"PRIu64" %"PRIu64" %d %"PRIu64"\n",
460 
461 	    ki.p_pid,						/* 1 pid */
462 	    ki.p_comm,						/* 2 tcomm */
463 	    "0RRSTZXR8"[(ki.p_stat > 8) ? 0 : (int)ki.p_stat],	/* 3 state */
464 	    ki.p_ppid,						/* 4 ppid */
465 	    ki.p__pgid,						/* 5 pgrp */
466 	    ki.p_sid,						/* 6 sid */
467 	    (ki.p_tdev != (uint32_t)NODEV) ? ki.p_tdev : 0,	/* 7 tty_nr */
468 	    ki.p_tpgid,						/* 8 tty_pgrp */
469 
470 	    ki.p_flag,						/* 9 flags */
471 
472 	    ki.p_uru_minflt,					/* 10 min_flt */
473 	    cru->ru_minflt,
474 	    ki.p_uru_majflt,					/* 12 maj_flt */
475 	    cru->ru_majflt,
476 	    UTIME2TICKS(ki.p_uutime_sec, ki.p_uutime_usec),	/* 14 utime */
477 	    UTIME2TICKS(ki.p_ustime_sec, ki.p_ustime_usec),	/* 15 stime */
478 	    UTIME2TICKS(cru->ru_utime.tv_sec, cru->ru_utime.tv_usec), /* 16 cutime */
479 	    UTIME2TICKS(cru->ru_stime.tv_sec, cru->ru_stime.tv_usec), /* 17 cstime */
480 
481 	    ki.p_priority,				/* XXX: 18 priority */
482 	    ki.p_nice - NZERO,				/* 19 nice */
483 	    ki.p_nlwps,					/* 20 num_threads */
484 
485 	    (long long)rt.tv_sec,
486 	    UTIME2TICKS(ki.p_ustart_sec, ki.p_ustart_usec), /* 22 start_time */
487 	    ki.p_vm_msize,				/* 23 vsize */
488 	    PGTOKB(ki.p_vm_rssize),			/* 24 rss */
489 	    p->p_rlimit[RLIMIT_RSS].rlim_cur,		/* 25 rsslim */
490 
491 	    stext,					/* 26 start_code */
492 	    etext,					/* 27 end_code */
493 	    sstack,					/* 28 start_stack */
494 
495 	    0,						/* XXX: 29 esp */
496 	    0,						/* XXX: 30 eip */
497 
498 	    ki.p_siglist.__bits[0],			/* XXX: 31 pending */
499 	    0,						/* XXX: 32 blocked */
500 	    ki.p_sigignore.__bits[0],		/* 33 sigign */
501 	    ki.p_sigcatch.__bits[0],		/* 34 sigcatch */
502 
503 	    ki.p_wchan,					/* 35 wchan */
504 	    ki.p_uru_nvcsw,
505 	    ki.p_uru_nivcsw,
506 	    ki.p_exitsig,				/* 38 exit_signal */
507 	    ki.p_cpuid);				/* 39 task_cpu */
508 
509 	mutex_exit(p->p_lock);
510 	mutex_exit(&proc_lock);
511 
512 	uvmspace_free(vm);
513 
514 	if (len == 0)
515 		goto out;
516 
517 	error = uiomove_frombuf(bf, len, uio);
518 out:
519 	free(bf, M_TEMP);
520 	return error;
521 }
522 
523 int
524 procfs_docpuinfo(struct lwp *curl, struct proc *p,
525     struct pfsnode *pfs, struct uio *uio)
526 {
527 	size_t len = LBFSZ;
528 	char *bf = NULL;
529 	int error;
530 
531 	do {
532 		if (bf)
533 			free(bf, M_TEMP);
534 		bf = malloc(len, M_TEMP, M_WAITOK);
535 	} while (procfs_getcpuinfstr(bf, &len) < 0);
536 
537 	if (len == 0) {
538 		error = 0;
539 		goto done;
540 	}
541 
542 	error = uiomove_frombuf(bf, len, uio);
543 done:
544 	free(bf, M_TEMP);
545 	return error;
546 }
547 
548 int
549 procfs_douptime(struct lwp *curl, struct proc *p,
550     struct pfsnode *pfs, struct uio *uio)
551 {
552 	char *bf;
553 	int len;
554 	struct timeval runtime;
555 	u_int64_t idle;
556 	int error = 0;
557 
558 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
559 
560 	microuptime(&runtime);
561 	idle = curcpu()->ci_schedstate.spc_cp_time[CP_IDLE];
562 	len = snprintf(bf, LBFSZ,
563 	    "%lld.%02lu %" PRIu64 ".%02" PRIu64 "\n",
564 	    (long long)runtime.tv_sec, (long)runtime.tv_usec / 10000,
565 	    idle / hz, (((idle % hz) * 100) / hz) % 100);
566 
567 	if (len == 0)
568 		goto out;
569 
570 	error = uiomove_frombuf(bf, len, uio);
571 out:
572 	free(bf, M_TEMP);
573 	return error;
574 }
575 
576 static int
577 procfs_format_sfs(char **mtab, size_t *mlen, char *buf, size_t blen,
578     const struct statvfs *sfs, struct lwp *curl, int suser)
579 {
580 	const char *fsname;
581 
582 	/* Linux uses different names for some filesystems */
583 	fsname = sfs->f_fstypename;
584 	if (strcmp(fsname, "procfs") == 0)
585 		fsname = "proc";
586 	else if (strcmp(fsname, "ext2fs") == 0)
587 		fsname = "ext2";
588 
589 	blen = snprintf(buf, blen, "%s %s %s %s%s%s%s%s%s 0 0\n",
590 	    sfs->f_mntfromname, sfs->f_mntonname, fsname,
591 	    (sfs->f_flag & ST_RDONLY) ? "ro" : "rw",
592 	    (sfs->f_flag & ST_NOSUID) ? ",nosuid" : "",
593 	    (sfs->f_flag & ST_NOEXEC) ? ",noexec" : "",
594 	    (sfs->f_flag & ST_NODEV) ? ",nodev" : "",
595 	    (sfs->f_flag & ST_SYNCHRONOUS) ? ",sync" : "",
596 	    (sfs->f_flag & ST_NOATIME) ? ",noatime" : "");
597 
598 	*mtab = realloc(*mtab, *mlen + blen, M_TEMP, M_WAITOK);
599 	memcpy(*mtab + *mlen, buf, blen);
600 	*mlen += blen;
601 	return sfs->f_mntonname[0] == '/' && sfs->f_mntonname[1] == '\0';
602 }
603 
604 int
605 procfs_domounts(struct lwp *curl, struct proc *p,
606     struct pfsnode *pfs, struct uio *uio)
607 {
608 	char *bf, *mtab = NULL;
609 	size_t mtabsz = 0;
610 	mount_iterator_t *iter;
611 	struct mount *mp;
612 	int error = 0, root = 0;
613 	struct cwdinfo *cwdi = curl->l_proc->p_cwdi;
614 	struct statvfs *sfs;
615 
616 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
617 
618 	sfs = malloc(sizeof(*sfs), M_TEMP, M_WAITOK);
619 	mountlist_iterator_init(&iter);
620 	while ((mp = mountlist_iterator_next(iter)) != NULL) {
621 		if ((error = dostatvfs(mp, sfs, curl, MNT_WAIT, 0)) == 0)
622 			root |= procfs_format_sfs(&mtab, &mtabsz, bf, LBFSZ,
623 			    sfs, curl, 0);
624 	}
625 	mountlist_iterator_destroy(iter);
626 	free(sfs, M_TEMP);
627 
628 	/*
629 	 * If we are inside a chroot that is not itself a mount point,
630 	 * fake a root entry.
631 	 */
632 	if (!root && cwdi->cwdi_rdir)
633 		(void)procfs_format_sfs(&mtab, &mtabsz, bf, LBFSZ,
634 		    &cwdi->cwdi_rdir->v_mount->mnt_stat, curl, 1);
635 
636 	free(bf, M_TEMP);
637 
638 	if (mtabsz > 0) {
639 		error = uiomove_frombuf(mtab, mtabsz, uio);
640 		free(mtab, M_TEMP);
641 	}
642 
643 	return error;
644 }
645 
646 /*
647  * Linux compatible /proc/version. Only active when the -o linux
648  * mountflag is used.
649  */
650 int
651 procfs_doversion(struct lwp *curl, struct proc *p,
652     struct pfsnode *pfs, struct uio *uio)
653 {
654 	char *bf;
655 	char lostype[20], losrelease[20], lversion[80];
656 	const char *postype, *posrelease, *pversion;
657 	const char *emulname = curlwp->l_proc->p_emul->e_name;
658 	int len;
659 	int error = 0;
660 	int nm[4];
661 	size_t buflen;
662 
663 	CTASSERT(EMUL_LINUX_KERN_OSTYPE == EMUL_LINUX32_KERN_OSTYPE);
664 	CTASSERT(EMUL_LINUX_KERN_OSRELEASE == EMUL_LINUX32_KERN_OSRELEASE);
665 	CTASSERT(EMUL_LINUX_KERN_VERSION == EMUL_LINUX32_KERN_VERSION);
666 
667 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
668 
669 	sysctl_lock(false);
670 
671 	if (strncmp(emulname, "linux", 5) == 0) {
672 		/*
673 		 * Lookup the emulation ostype, osrelease, and version.
674 		 * Since compat_linux and compat_linux32 can be built as
675 		 * modules, we use sysctl to obtain the values instead of
676 		 * using the symbols directly.
677 		 */
678 
679 		if (strcmp(emulname, "linux32") == 0) {
680 			nm[0] = CTL_EMUL;
681 			nm[1] = EMUL_LINUX32;
682 			nm[2] = EMUL_LINUX32_KERN;
683 		} else {
684 			nm[0] = CTL_EMUL;
685 			nm[1] = EMUL_LINUX;
686 			nm[2] = EMUL_LINUX_KERN;
687 		}
688 
689 		nm[3] = EMUL_LINUX_KERN_OSTYPE;
690 		buflen = sizeof(lostype);
691 		error = sysctl_dispatch(nm, __arraycount(nm),
692 		    lostype, &buflen,
693 		    NULL, 0, NULL, NULL, NULL);
694 		if (error)
695 			goto out;
696 
697 		nm[3] = EMUL_LINUX_KERN_OSRELEASE;
698 		buflen = sizeof(losrelease);
699 		error = sysctl_dispatch(nm, __arraycount(nm),
700 		    losrelease, &buflen,
701 		    NULL, 0, NULL, NULL, NULL);
702 		if (error)
703 			goto out;
704 
705 		nm[3] = EMUL_LINUX_KERN_VERSION;
706 		buflen = sizeof(lversion);
707 		error = sysctl_dispatch(nm, __arraycount(nm),
708 		    lversion, &buflen,
709 		    NULL, 0, NULL, NULL, NULL);
710 		if (error)
711 			goto out;
712 
713 		postype = lostype;
714 		posrelease = losrelease;
715 		pversion = lversion;
716 	} else {
717 		postype = ostype;
718 		posrelease = osrelease;
719 		strlcpy(lversion, version, sizeof(lversion));
720 		if (strchr(lversion, '\n'))
721 			*strchr(lversion, '\n') = '\0';
722 		pversion = lversion;
723 	}
724 
725 	len = snprintf(bf, LBFSZ,
726 		"%s version %s (%s@localhost) (gcc version %s) %s\n",
727 		postype, posrelease, emulname,
728 #ifdef __VERSION__
729 		__VERSION__,
730 #else
731 		"unknown",
732 #endif
733 		pversion);
734 
735 	if (len == 0)
736 		goto out;
737 
738 	error = uiomove_frombuf(bf, len, uio);
739 out:
740 	free(bf, M_TEMP);
741 	sysctl_unlock();
742 	return error;
743 }
744