xref: /freebsd/sys/compat/linprocfs/linprocfs.c (revision eac7052fdebb90caf2f653e06187bdbca837b9c7)
1 /*-
2  * SPDX-License-Identifier: BSD-4-Clause
3  *
4  * Copyright (c) 2000 Dag-Erling Coïdan Smørgrav
5  * Copyright (c) 1999 Pierre Beyssac
6  * Copyright (c) 1993 Jan-Simon Pendry
7  * Copyright (c) 1993
8  *	The Regents of the University of California.  All rights reserved.
9  *
10  * This code is derived from software contributed to Berkeley by
11  * Jan-Simon Pendry.
12  *
13  * Redistribution and use in source and binary forms, with or without
14  * modification, are permitted provided that the following conditions
15  * are met:
16  * 1. Redistributions of source code must retain the above copyright
17  *    notice, this list of conditions and the following disclaimer.
18  * 2. Redistributions in binary form must reproduce the above copyright
19  *    notice, this list of conditions and the following disclaimer in the
20  *    documentation and/or other materials provided with the distribution.
21  * 3. All advertising materials mentioning features or use of this software
22  *    must display the following acknowledgement:
23  *	This product includes software developed by the University of
24  *	California, Berkeley and its contributors.
25  * 4. Neither the name of the University nor the names of its contributors
26  *    may be used to endorse or promote products derived from this software
27  *    without specific prior written permission.
28  *
29  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
30  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
31  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
32  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
33  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
34  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
35  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
36  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
37  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
38  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
39  * SUCH DAMAGE.
40  *
41  *	@(#)procfs_status.c	8.4 (Berkeley) 6/15/94
42  */
43 
44 #include <sys/cdefs.h>
45 __FBSDID("$FreeBSD$");
46 
47 #include <sys/param.h>
48 #include <sys/queue.h>
49 #include <sys/blist.h>
50 #include <sys/conf.h>
51 #include <sys/exec.h>
52 #include <sys/fcntl.h>
53 #include <sys/filedesc.h>
54 #include <sys/jail.h>
55 #include <sys/kernel.h>
56 #include <sys/limits.h>
57 #include <sys/linker.h>
58 #include <sys/lock.h>
59 #include <sys/malloc.h>
60 #include <sys/msg.h>
61 #include <sys/mutex.h>
62 #include <sys/namei.h>
63 #include <sys/proc.h>
64 #include <sys/ptrace.h>
65 #include <sys/resourcevar.h>
66 #include <sys/resource.h>
67 #include <sys/sbuf.h>
68 #include <sys/sem.h>
69 #include <sys/shm.h>
70 #include <sys/smp.h>
71 #include <sys/socket.h>
72 #include <sys/syscallsubr.h>
73 #include <sys/sysctl.h>
74 #include <sys/sysent.h>
75 #include <sys/systm.h>
76 #include <sys/time.h>
77 #include <sys/tty.h>
78 #include <sys/user.h>
79 #include <sys/uuid.h>
80 #include <sys/vmmeter.h>
81 #include <sys/vnode.h>
82 #include <sys/bus.h>
83 
84 #include <net/if.h>
85 #include <net/if_var.h>
86 #include <net/if_types.h>
87 
88 #include <vm/vm.h>
89 #include <vm/vm_extern.h>
90 #include <vm/pmap.h>
91 #include <vm/vm_map.h>
92 #include <vm/vm_param.h>
93 #include <vm/vm_object.h>
94 #include <vm/swap_pager.h>
95 
96 #include <machine/clock.h>
97 
98 #include <geom/geom.h>
99 #include <geom/geom_int.h>
100 
101 #if defined(__i386__) || defined(__amd64__)
102 #include <machine/cputypes.h>
103 #include <machine/md_var.h>
104 #endif /* __i386__ || __amd64__ */
105 
106 #include <compat/linux/linux.h>
107 #include <compat/linux/linux_mib.h>
108 #include <compat/linux/linux_misc.h>
109 #include <compat/linux/linux_util.h>
110 #include <fs/pseudofs/pseudofs.h>
111 #include <fs/procfs/procfs.h>
112 
113 /*
114  * Various conversion macros
115  */
116 #define T2J(x) ((long)(((x) * 100ULL) / (stathz ? stathz : hz)))	/* ticks to jiffies */
117 #define T2CS(x) ((unsigned long)(((x) * 100ULL) / (stathz ? stathz : hz)))	/* ticks to centiseconds */
118 #define T2S(x) ((x) / (stathz ? stathz : hz))		/* ticks to seconds */
119 #define B2K(x) ((x) >> 10)				/* bytes to kbytes */
120 #define B2P(x) ((x) >> PAGE_SHIFT)			/* bytes to pages */
121 #define P2B(x) ((x) << PAGE_SHIFT)			/* pages to bytes */
122 #define P2K(x) ((x) << (PAGE_SHIFT - 10))		/* pages to kbytes */
123 #define TV2J(x)	((x)->tv_sec * 100UL + (x)->tv_usec / 10000)
124 
125 /**
126  * @brief Mapping of ki_stat in struct kinfo_proc to the linux state
127  *
128  * The linux procfs state field displays one of the characters RSDZTW to
129  * denote running, sleeping in an interruptible wait, waiting in an
130  * uninterruptible disk sleep, a zombie process, process is being traced
131  * or stopped, or process is paging respectively.
132  *
133  * Our struct kinfo_proc contains the variable ki_stat which contains a
134  * value out of SIDL, SRUN, SSLEEP, SSTOP, SZOMB, SWAIT and SLOCK.
135  *
136  * This character array is used with ki_stati-1 as an index and tries to
137  * map our states to suitable linux states.
138  */
139 static char linux_state[] = "RRSTZDD";
140 
141 /*
142  * Filler function for proc/meminfo
143  */
144 static int
145 linprocfs_domeminfo(PFS_FILL_ARGS)
146 {
147 	unsigned long memtotal;		/* total memory in bytes */
148 	unsigned long memfree;		/* free memory in bytes */
149 	unsigned long cached;		/* page cache */
150 	unsigned long buffers;		/* buffer cache */
151 	unsigned long long swaptotal;	/* total swap space in bytes */
152 	unsigned long long swapused;	/* used swap space in bytes */
153 	unsigned long long swapfree;	/* free swap space in bytes */
154 	size_t sz;
155 	int error, i, j;
156 
157 	memtotal = physmem * PAGE_SIZE;
158 	memfree = (unsigned long)vm_free_count() * PAGE_SIZE;
159 	swap_pager_status(&i, &j);
160 	swaptotal = (unsigned long long)i * PAGE_SIZE;
161 	swapused = (unsigned long long)j * PAGE_SIZE;
162 	swapfree = swaptotal - swapused;
163 
164 	/*
165 	 * This value may exclude wired pages, but we have no good way of
166 	 * accounting for that.
167 	 */
168 	cached =
169 	    (vm_active_count() + vm_inactive_count() + vm_laundry_count()) *
170 	    PAGE_SIZE;
171 
172 	sz = sizeof(buffers);
173 	error = kernel_sysctlbyname(curthread, "vfs.bufspace", &buffers, &sz,
174 	    NULL, 0, 0, 0);
175 	if (error != 0)
176 		buffers = 0;
177 
178 	sbuf_printf(sb,
179 	    "MemTotal: %9lu kB\n"
180 	    "MemFree:  %9lu kB\n"
181 	    "Buffers:  %9lu kB\n"
182 	    "Cached:   %9lu kB\n"
183 	    "SwapTotal:%9llu kB\n"
184 	    "SwapFree: %9llu kB\n",
185 	    B2K(memtotal), B2K(memfree), B2K(buffers),
186 	    B2K(cached), B2K(swaptotal), B2K(swapfree));
187 
188 	return (0);
189 }
190 
191 #if defined(__i386__) || defined(__amd64__)
192 /*
193  * Filler function for proc/cpuinfo (i386 & amd64 version)
194  */
195 static int
196 linprocfs_docpuinfo(PFS_FILL_ARGS)
197 {
198 	int hw_model[2];
199 	char model[128];
200 	uint64_t freq;
201 	size_t size;
202 	u_int cache_size[4];
203 	int fqmhz, fqkhz;
204 	int i, j;
205 
206 	/*
207 	 * We default the flags to include all non-conflicting flags,
208 	 * and the Intel versions of conflicting flags.
209 	 */
210 	static char *cpu_feature_names[] = {
211 		/*  0 */ "fpu", "vme", "de", "pse",
212 		/*  4 */ "tsc", "msr", "pae", "mce",
213 		/*  8 */ "cx8", "apic", "", "sep",
214 		/* 12 */ "mtrr", "pge", "mca", "cmov",
215 		/* 16 */ "pat", "pse36", "pn", "clflush",
216 		/* 20 */ "", "dts", "acpi", "mmx",
217 		/* 24 */ "fxsr", "sse", "sse2", "ss",
218 		/* 28 */ "ht", "tm", "ia64", "pbe"
219 	};
220 
221 	static char *amd_feature_names[] = {
222 		/*  0 */ "", "", "", "",
223 		/*  4 */ "", "", "", "",
224 		/*  8 */ "", "", "", "syscall",
225 		/* 12 */ "", "", "", "",
226 		/* 16 */ "", "", "", "mp",
227 		/* 20 */ "nx", "", "mmxext", "",
228 		/* 24 */ "", "fxsr_opt", "pdpe1gb", "rdtscp",
229 		/* 28 */ "", "lm", "3dnowext", "3dnow"
230 	};
231 
232 	static char *cpu_feature2_names[] = {
233 		/*  0 */ "pni", "pclmulqdq", "dtes64", "monitor",
234 		/*  4 */ "ds_cpl", "vmx", "smx", "est",
235 		/*  8 */ "tm2", "ssse3", "cid", "sdbg",
236 		/* 12 */ "fma", "cx16", "xtpr", "pdcm",
237 		/* 16 */ "", "pcid", "dca", "sse4_1",
238 		/* 20 */ "sse4_2", "x2apic", "movbe", "popcnt",
239 		/* 24 */ "tsc_deadline_timer", "aes", "xsave", "",
240 		/* 28 */ "avx", "f16c", "rdrand", "hypervisor"
241 	};
242 
243 	static char *amd_feature2_names[] = {
244 		/*  0 */ "lahf_lm", "cmp_legacy", "svm", "extapic",
245 		/*  4 */ "cr8_legacy", "abm", "sse4a", "misalignsse",
246 		/*  8 */ "3dnowprefetch", "osvw", "ibs", "xop",
247 		/* 12 */ "skinit", "wdt", "", "lwp",
248 		/* 16 */ "fma4", "tce", "", "nodeid_msr",
249 		/* 20 */ "", "tbm", "topoext", "perfctr_core",
250 		/* 24 */ "perfctr_nb", "", "bpext", "ptsc",
251 		/* 28 */ "perfctr_llc", "mwaitx", "", ""
252 	};
253 
254 	static char *cpu_stdext_feature_names[] = {
255 		/*  0 */ "fsgsbase", "tsc_adjust", "", "bmi1",
256 		/*  4 */ "hle", "avx2", "", "smep",
257 		/*  8 */ "bmi2", "erms", "invpcid", "rtm",
258 		/* 12 */ "cqm", "", "mpx", "rdt_a",
259 		/* 16 */ "avx512f", "avx512dq", "rdseed", "adx",
260 		/* 20 */ "smap", "avx512ifma", "", "clflushopt",
261 		/* 24 */ "clwb", "intel_pt", "avx512pf", "avx512er",
262 		/* 28 */ "avx512cd", "sha_ni", "avx512bw", "avx512vl"
263 	};
264 
265 	static char *power_flags[] = {
266 		"ts",           "fid",          "vid",
267 		"ttp",          "tm",           "stc",
268 		"100mhzsteps",  "hwpstate",     "",
269 		"cpb",          "eff_freq_ro",  "proc_feedback",
270 		"acc_power",
271 	};
272 
273 	hw_model[0] = CTL_HW;
274 	hw_model[1] = HW_MODEL;
275 	model[0] = '\0';
276 	size = sizeof(model);
277 	if (kernel_sysctl(td, hw_model, 2, &model, &size, 0, 0, 0, 0) != 0)
278 		strcpy(model, "unknown");
279 #ifdef __i386__
280 	switch (cpu_vendor_id) {
281 	case CPU_VENDOR_AMD:
282 		if (cpu_class < CPUCLASS_686)
283 			cpu_feature_names[16] = "fcmov";
284 		break;
285 	case CPU_VENDOR_CYRIX:
286 		cpu_feature_names[24] = "cxmmx";
287 		break;
288 	}
289 #endif
290 	if (cpu_exthigh >= 0x80000006)
291 		do_cpuid(0x80000006, cache_size);
292 	else
293 		memset(cache_size, 0, sizeof(cache_size));
294 	for (i = 0; i < mp_ncpus; ++i) {
295 		fqmhz = 0;
296 		fqkhz = 0;
297 		freq = atomic_load_acq_64(&tsc_freq);
298 		if (freq != 0) {
299 			fqmhz = (freq + 4999) / 1000000;
300 			fqkhz = ((freq + 4999) / 10000) % 100;
301 		}
302 		sbuf_printf(sb,
303 		    "processor\t: %d\n"
304 		    "vendor_id\t: %.20s\n"
305 		    "cpu family\t: %u\n"
306 		    "model\t\t: %u\n"
307 		    "model name\t: %s\n"
308 		    "stepping\t: %u\n"
309 		    "cpu MHz\t\t: %d.%02d\n"
310 		    "cache size\t: %d KB\n"
311 		    "physical id\t: %d\n"
312 		    "siblings\t: %d\n"
313 		    "core id\t\t: %d\n"
314 		    "cpu cores\t: %d\n"
315 		    "apicid\t\t: %d\n"
316 		    "initial apicid\t: %d\n"
317 		    "fpu\t\t: %s\n"
318 		    "fpu_exception\t: %s\n"
319 		    "cpuid level\t: %d\n"
320 		    "wp\t\t: %s\n",
321 		    i, cpu_vendor, CPUID_TO_FAMILY(cpu_id),
322 		    CPUID_TO_MODEL(cpu_id), model, cpu_id & CPUID_STEPPING,
323 		    fqmhz, fqkhz,
324 		    (cache_size[2] >> 16), 0, mp_ncpus, i, mp_ncpus,
325 		    i, i, /*cpu_id & CPUID_LOCAL_APIC_ID ??*/
326 		    (cpu_feature & CPUID_FPU) ? "yes" : "no", "yes",
327 		    CPUID_TO_FAMILY(cpu_id), "yes");
328 		sbuf_cat(sb, "flags\t\t:");
329 		for (j = 0; j < nitems(cpu_feature_names); j++)
330 			if (cpu_feature & (1 << j) &&
331 			    cpu_feature_names[j][0] != '\0')
332 				sbuf_printf(sb, " %s", cpu_feature_names[j]);
333 		for (j = 0; j < nitems(amd_feature_names); j++)
334 			if (amd_feature & (1 << j) &&
335 			    amd_feature_names[j][0] != '\0')
336 				sbuf_printf(sb, " %s", amd_feature_names[j]);
337 		for (j = 0; j < nitems(cpu_feature2_names); j++)
338 			if (cpu_feature2 & (1 << j) &&
339 			    cpu_feature2_names[j][0] != '\0')
340 				sbuf_printf(sb, " %s", cpu_feature2_names[j]);
341 		for (j = 0; j < nitems(amd_feature2_names); j++)
342 			if (amd_feature2 & (1 << j) &&
343 			    amd_feature2_names[j][0] != '\0')
344 				sbuf_printf(sb, " %s", amd_feature2_names[j]);
345 		for (j = 0; j < nitems(cpu_stdext_feature_names); j++)
346 			if (cpu_stdext_feature & (1 << j) &&
347 			    cpu_stdext_feature_names[j][0] != '\0')
348 				sbuf_printf(sb, " %s",
349 				    cpu_stdext_feature_names[j]);
350 		sbuf_cat(sb, "\n");
351 		sbuf_printf(sb,
352 		    "bugs\t\t: %s\n"
353 		    "bogomips\t: %d.%02d\n"
354 		    "clflush size\t: %d\n"
355 		    "cache_alignment\t: %d\n"
356 		    "address sizes\t: %d bits physical, %d bits virtual\n",
357 #if defined(I586_CPU) && !defined(NO_F00F_HACK)
358 		    (has_f00f_bug) ? "Intel F00F" : "",
359 #else
360 		    "",
361 #endif
362 		    fqmhz * 2, fqkhz,
363 		    cpu_clflush_line_size, cpu_clflush_line_size,
364 		    cpu_maxphyaddr,
365 		    (cpu_maxphyaddr > 32) ? 48 : 0);
366 		sbuf_cat(sb, "power management: ");
367 		for (j = 0; j < nitems(power_flags); j++)
368 			if (amd_pminfo & (1 << j))
369 				sbuf_printf(sb, " %s", power_flags[j]);
370 		sbuf_cat(sb, "\n\n");
371 
372 		/* XXX per-cpu vendor / class / model / id? */
373 	}
374 	sbuf_cat(sb, "\n");
375 
376 	return (0);
377 }
378 #else
379 /* ARM64TODO: implement non-stubbed linprocfs_docpuinfo */
380 static int
381 linprocfs_docpuinfo(PFS_FILL_ARGS)
382 {
383 	int i;
384 
385 	for (i = 0; i < mp_ncpus; ++i) {
386 		sbuf_printf(sb,
387 		    "processor\t: %d\n"
388 		    "BogoMIPS\t: %d.%02d\n",
389 		    i, 0, 0);
390 		sbuf_cat(sb, "Features\t: ");
391 		sbuf_cat(sb, "\n");
392 		sbuf_printf(sb,
393 		    "CPU implementer\t: \n"
394 		    "CPU architecture: \n"
395 		    "CPU variant\t: 0x%x\n"
396 		    "CPU part\t: 0x%x\n"
397 		    "CPU revision\t: %d\n",
398 		    0, 0, 0);
399 		sbuf_cat(sb, "\n");
400 	}
401 
402 	return (0);
403 }
404 #endif /* __i386__ || __amd64__ */
405 
406 /*
407  * Filler function for proc/mtab
408  *
409  * This file doesn't exist in Linux' procfs, but is included here so
410  * users can symlink /compat/linux/etc/mtab to /proc/mtab
411  */
412 static int
413 linprocfs_domtab(PFS_FILL_ARGS)
414 {
415 	struct nameidata nd;
416 	const char *lep;
417 	char *dlep, *flep, *mntto, *mntfrom, *fstype;
418 	size_t lep_len;
419 	int error;
420 	struct statfs *buf, *sp;
421 	size_t count;
422 
423 	/* resolve symlinks etc. in the emulation tree prefix */
424 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, linux_emul_path, td);
425 	flep = NULL;
426 	error = namei(&nd);
427 	lep = linux_emul_path;
428 	if (error == 0) {
429 		if (vn_fullpath(nd.ni_vp, &dlep, &flep) == 0)
430 			lep = dlep;
431 		vrele(nd.ni_vp);
432 	}
433 	lep_len = strlen(lep);
434 
435 	buf = NULL;
436 	error = kern_getfsstat(td, &buf, SIZE_T_MAX, &count,
437 	    UIO_SYSSPACE, MNT_WAIT);
438 	if (error != 0) {
439 		free(buf, M_TEMP);
440 		free(flep, M_TEMP);
441 		return (error);
442 	}
443 
444 	for (sp = buf; count > 0; sp++, count--) {
445 		/* determine device name */
446 		mntfrom = sp->f_mntfromname;
447 
448 		/* determine mount point */
449 		mntto = sp->f_mntonname;
450 		if (strncmp(mntto, lep, lep_len) == 0 && mntto[lep_len] == '/')
451 			mntto += lep_len;
452 
453 		/* determine fs type */
454 		fstype = sp->f_fstypename;
455 		if (strcmp(fstype, pn->pn_info->pi_name) == 0)
456 			mntfrom = fstype = "proc";
457 		else if (strcmp(fstype, "procfs") == 0)
458 			continue;
459 
460 		if (strcmp(fstype, "autofs") == 0) {
461 			/*
462 			 * FreeBSD uses eg "map -hosts", whereas Linux
463 			 * expects just "-hosts".
464 			 */
465 			if (strncmp(mntfrom, "map ", 4) == 0)
466 				mntfrom += 4;
467 		}
468 
469 		if (strcmp(fstype, "linsysfs") == 0) {
470 			sbuf_printf(sb, "/sys %s sysfs %s", mntto,
471 			    sp->f_flags & MNT_RDONLY ? "ro" : "rw");
472 		} else {
473 			/* For Linux msdosfs is called vfat */
474 			if (strcmp(fstype, "msdosfs") == 0)
475 				fstype = "vfat";
476 			sbuf_printf(sb, "%s %s %s %s", mntfrom, mntto, fstype,
477 			    sp->f_flags & MNT_RDONLY ? "ro" : "rw");
478 		}
479 #define ADD_OPTION(opt, name) \
480 	if (sp->f_flags & (opt)) sbuf_printf(sb, "," name);
481 		ADD_OPTION(MNT_SYNCHRONOUS,	"sync");
482 		ADD_OPTION(MNT_NOEXEC,		"noexec");
483 		ADD_OPTION(MNT_NOSUID,		"nosuid");
484 		ADD_OPTION(MNT_UNION,		"union");
485 		ADD_OPTION(MNT_ASYNC,		"async");
486 		ADD_OPTION(MNT_SUIDDIR,		"suiddir");
487 		ADD_OPTION(MNT_NOSYMFOLLOW,	"nosymfollow");
488 		ADD_OPTION(MNT_NOATIME,		"noatime");
489 #undef ADD_OPTION
490 		/* a real Linux mtab will also show NFS options */
491 		sbuf_printf(sb, " 0 0\n");
492 	}
493 
494 	free(buf, M_TEMP);
495 	free(flep, M_TEMP);
496 	return (error);
497 }
498 
499 /*
500  * Filler function for proc/partitions
501  */
502 static int
503 linprocfs_dopartitions(PFS_FILL_ARGS)
504 {
505 	struct g_class *cp;
506 	struct g_geom *gp;
507 	struct g_provider *pp;
508 	int major, minor;
509 
510 	g_topology_lock();
511 	sbuf_printf(sb, "major minor  #blocks  name rio rmerge rsect "
512 	    "ruse wio wmerge wsect wuse running use aveq\n");
513 
514 	LIST_FOREACH(cp, &g_classes, class) {
515 		if (strcmp(cp->name, "DISK") == 0 ||
516 		    strcmp(cp->name, "PART") == 0)
517 			LIST_FOREACH(gp, &cp->geom, geom) {
518 				LIST_FOREACH(pp, &gp->provider, provider) {
519 					if (linux_driver_get_major_minor(
520 					    pp->name, &major, &minor) != 0) {
521 						major = 0;
522 						minor = 0;
523 					}
524 					sbuf_printf(sb, "%d %d %lld %s "
525 					    "%d %d %d %d %d "
526 					     "%d %d %d %d %d %d\n",
527 					     major, minor,
528 					     (long long)pp->mediasize, pp->name,
529 					     0, 0, 0, 0, 0,
530 					     0, 0, 0, 0, 0, 0);
531 				}
532 			}
533 	}
534 	g_topology_unlock();
535 
536 	return (0);
537 }
538 
539 /*
540  * Filler function for proc/stat
541  *
542  * Output depends on kernel version:
543  *
544  * v2.5.40 <=
545  *   user nice system idle
546  * v2.5.41
547  *   user nice system idle iowait
548  * v2.6.11
549  *   user nice system idle iowait irq softirq steal
550  * v2.6.24
551  *   user nice system idle iowait irq softirq steal guest
552  * v2.6.33 >=
553  *   user nice system idle iowait irq softirq steal guest guest_nice
554  */
555 static int
556 linprocfs_dostat(PFS_FILL_ARGS)
557 {
558 	struct pcpu *pcpu;
559 	long cp_time[CPUSTATES];
560 	long *cp;
561 	struct timeval boottime;
562 	int i;
563 	char *zero_pad;
564 	bool has_intr = true;
565 
566 	if (linux_kernver(td) >= LINUX_KERNVER(2,6,33)) {
567 		zero_pad = " 0 0 0 0\n";
568 	} else if (linux_kernver(td) >= LINUX_KERNVER(2,6,24)) {
569 		zero_pad = " 0 0 0\n";
570 	} else if (linux_kernver(td) >= LINUX_KERNVER(2,6,11)) {
571 		zero_pad = " 0 0\n";
572 	} else if (linux_kernver(td) >= LINUX_KERNVER(2,5,41)) {
573 		has_intr = false;
574 		zero_pad = " 0\n";
575 	} else {
576 		has_intr = false;
577 		zero_pad = "\n";
578 	}
579 
580 	read_cpu_time(cp_time);
581 	getboottime(&boottime);
582 	/* Parameters common to all versions */
583 	sbuf_printf(sb, "cpu %lu %lu %lu %lu",
584 	    T2J(cp_time[CP_USER]),
585 	    T2J(cp_time[CP_NICE]),
586 	    T2J(cp_time[CP_SYS]),
587 	    T2J(cp_time[CP_IDLE]));
588 
589 	/* Print interrupt stats if available */
590 	if (has_intr) {
591 		sbuf_printf(sb, " 0 %lu", T2J(cp_time[CP_INTR]));
592 	}
593 
594 	/* Pad out remaining fields depending on version */
595 	sbuf_printf(sb, "%s", zero_pad);
596 
597 	CPU_FOREACH(i) {
598 		pcpu = pcpu_find(i);
599 		cp = pcpu->pc_cp_time;
600 		sbuf_printf(sb, "cpu%d %lu %lu %lu %lu", i,
601 		    T2J(cp[CP_USER]),
602 		    T2J(cp[CP_NICE]),
603 		    T2J(cp[CP_SYS]),
604 		    T2J(cp[CP_IDLE]));
605 
606 		if (has_intr) {
607 			sbuf_printf(sb, " 0 %lu", T2J(cp[CP_INTR]));
608 		}
609 
610 		sbuf_printf(sb, "%s", zero_pad);
611 	}
612 	sbuf_printf(sb,
613 	    "disk 0 0 0 0\n"
614 	    "page %ju %ju\n"
615 	    "swap %ju %ju\n"
616 	    "intr %ju\n"
617 	    "ctxt %ju\n"
618 	    "btime %lld\n",
619 	    (uintmax_t)VM_CNT_FETCH(v_vnodepgsin),
620 	    (uintmax_t)VM_CNT_FETCH(v_vnodepgsout),
621 	    (uintmax_t)VM_CNT_FETCH(v_swappgsin),
622 	    (uintmax_t)VM_CNT_FETCH(v_swappgsout),
623 	    (uintmax_t)VM_CNT_FETCH(v_intr),
624 	    (uintmax_t)VM_CNT_FETCH(v_swtch),
625 	    (long long)boottime.tv_sec);
626 	return (0);
627 }
628 
629 static int
630 linprocfs_doswaps(PFS_FILL_ARGS)
631 {
632 	struct xswdev xsw;
633 	uintmax_t total, used;
634 	int n;
635 	char devname[SPECNAMELEN + 1];
636 
637 	sbuf_printf(sb, "Filename\t\t\t\tType\t\tSize\tUsed\tPriority\n");
638 	for (n = 0; ; n++) {
639 		if (swap_dev_info(n, &xsw, devname, sizeof(devname)) != 0)
640 			break;
641 		total = (uintmax_t)xsw.xsw_nblks * PAGE_SIZE / 1024;
642 		used  = (uintmax_t)xsw.xsw_used * PAGE_SIZE / 1024;
643 
644 		/*
645 		 * The space and not tab after the device name is on
646 		 * purpose.  Linux does so.
647 		 */
648 		sbuf_printf(sb, "/dev/%-34s unknown\t\t%jd\t%jd\t-1\n",
649 		    devname, total, used);
650 	}
651 	return (0);
652 }
653 
654 /*
655  * Filler function for proc/uptime
656  */
657 static int
658 linprocfs_douptime(PFS_FILL_ARGS)
659 {
660 	long cp_time[CPUSTATES];
661 	struct timeval tv;
662 
663 	getmicrouptime(&tv);
664 	read_cpu_time(cp_time);
665 	sbuf_printf(sb, "%lld.%02ld %ld.%02lu\n",
666 	    (long long)tv.tv_sec, tv.tv_usec / 10000,
667 	    T2S(cp_time[CP_IDLE] / mp_ncpus),
668 	    T2CS(cp_time[CP_IDLE] / mp_ncpus) % 100);
669 	return (0);
670 }
671 
672 /*
673  * Get OS build date
674  */
675 static void
676 linprocfs_osbuild(struct thread *td, struct sbuf *sb)
677 {
678 #if 0
679 	char osbuild[256];
680 	char *cp1, *cp2;
681 
682 	strncpy(osbuild, version, 256);
683 	osbuild[255] = '\0';
684 	cp1 = strstr(osbuild, "\n");
685 	cp2 = strstr(osbuild, ":");
686 	if (cp1 && cp2) {
687 		*cp1 = *cp2 = '\0';
688 		cp1 = strstr(osbuild, "#");
689 	} else
690 		cp1 = NULL;
691 	if (cp1)
692 		sbuf_printf(sb, "%s%s", cp1, cp2 + 1);
693 	else
694 #endif
695 		sbuf_cat(sb, "#4 Sun Dec 18 04:30:00 CET 1977");
696 }
697 
698 /*
699  * Get OS builder
700  */
701 static void
702 linprocfs_osbuilder(struct thread *td, struct sbuf *sb)
703 {
704 #if 0
705 	char builder[256];
706 	char *cp;
707 
708 	cp = strstr(version, "\n    ");
709 	if (cp) {
710 		strncpy(builder, cp + 5, 256);
711 		builder[255] = '\0';
712 		cp = strstr(builder, ":");
713 		if (cp)
714 			*cp = '\0';
715 	}
716 	if (cp)
717 		sbuf_cat(sb, builder);
718 	else
719 #endif
720 		sbuf_cat(sb, "des@freebsd.org");
721 }
722 
723 /*
724  * Filler function for proc/version
725  */
726 static int
727 linprocfs_doversion(PFS_FILL_ARGS)
728 {
729 	char osname[LINUX_MAX_UTSNAME];
730 	char osrelease[LINUX_MAX_UTSNAME];
731 
732 	linux_get_osname(td, osname);
733 	linux_get_osrelease(td, osrelease);
734 	sbuf_printf(sb, "%s version %s (", osname, osrelease);
735 	linprocfs_osbuilder(td, sb);
736 	sbuf_cat(sb, ") (gcc version " __VERSION__ ") ");
737 	linprocfs_osbuild(td, sb);
738 	sbuf_cat(sb, "\n");
739 
740 	return (0);
741 }
742 
743 /*
744  * Filler function for proc/loadavg
745  */
746 static int
747 linprocfs_doloadavg(PFS_FILL_ARGS)
748 {
749 
750 	sbuf_printf(sb,
751 	    "%d.%02d %d.%02d %d.%02d %d/%d %d\n",
752 	    (int)(averunnable.ldavg[0] / averunnable.fscale),
753 	    (int)(averunnable.ldavg[0] * 100 / averunnable.fscale % 100),
754 	    (int)(averunnable.ldavg[1] / averunnable.fscale),
755 	    (int)(averunnable.ldavg[1] * 100 / averunnable.fscale % 100),
756 	    (int)(averunnable.ldavg[2] / averunnable.fscale),
757 	    (int)(averunnable.ldavg[2] * 100 / averunnable.fscale % 100),
758 	    1,				/* number of running tasks */
759 	    nprocs,			/* number of tasks */
760 	    lastpid			/* the last pid */
761 	);
762 	return (0);
763 }
764 
765 static int
766 linprocfs_get_tty_nr(struct proc *p)
767 {
768 	struct session *sp;
769 	const char *ttyname;
770 	int error, major, minor, nr;
771 
772 	PROC_LOCK_ASSERT(p, MA_OWNED);
773 	sx_assert(&proctree_lock, SX_LOCKED);
774 
775 	if ((p->p_flag & P_CONTROLT) == 0)
776 		return (-1);
777 
778 	sp = p->p_pgrp->pg_session;
779 	if (sp == NULL)
780 		return (-1);
781 
782 	ttyname = devtoname(sp->s_ttyp->t_dev);
783 	error = linux_driver_get_major_minor(ttyname, &major, &minor);
784 	if (error != 0)
785 		return (-1);
786 
787 	nr = makedev(major, minor);
788 	return (nr);
789 }
790 
791 /*
792  * Filler function for proc/pid/stat
793  */
794 static int
795 linprocfs_doprocstat(PFS_FILL_ARGS)
796 {
797 	struct kinfo_proc kp;
798 	struct timeval boottime;
799 	char state;
800 	static int ratelimit = 0;
801 	int tty_nr;
802 	vm_offset_t startcode, startdata;
803 
804 	getboottime(&boottime);
805 	sx_slock(&proctree_lock);
806 	PROC_LOCK(p);
807 	fill_kinfo_proc(p, &kp);
808 	tty_nr = linprocfs_get_tty_nr(p);
809 	sx_sunlock(&proctree_lock);
810 	if (p->p_vmspace) {
811 	   startcode = (vm_offset_t)p->p_vmspace->vm_taddr;
812 	   startdata = (vm_offset_t)p->p_vmspace->vm_daddr;
813 	} else {
814 	   startcode = 0;
815 	   startdata = 0;
816 	}
817 	sbuf_printf(sb, "%d", p->p_pid);
818 #define PS_ADD(name, fmt, arg) sbuf_printf(sb, " " fmt, arg)
819 	PS_ADD("comm",		"(%s)",	p->p_comm);
820 	if (kp.ki_stat > sizeof(linux_state)) {
821 		state = 'R';
822 
823 		if (ratelimit == 0) {
824 			printf("linprocfs: don't know how to handle unknown FreeBSD state %d/%zd, mapping to R\n",
825 			    kp.ki_stat, sizeof(linux_state));
826 			++ratelimit;
827 		}
828 	} else
829 		state = linux_state[kp.ki_stat - 1];
830 	PS_ADD("state",		"%c",	state);
831 	PS_ADD("ppid",		"%d",	p->p_pptr ? p->p_pptr->p_pid : 0);
832 	PS_ADD("pgrp",		"%d",	p->p_pgid);
833 	PS_ADD("session",	"%d",	p->p_session->s_sid);
834 	PROC_UNLOCK(p);
835 	PS_ADD("tty",		"%d",	tty_nr);
836 	PS_ADD("tpgid",		"%d",	kp.ki_tpgid);
837 	PS_ADD("flags",		"%u",	0); /* XXX */
838 	PS_ADD("minflt",	"%lu",	kp.ki_rusage.ru_minflt);
839 	PS_ADD("cminflt",	"%lu",	kp.ki_rusage_ch.ru_minflt);
840 	PS_ADD("majflt",	"%lu",	kp.ki_rusage.ru_majflt);
841 	PS_ADD("cmajflt",	"%lu",	kp.ki_rusage_ch.ru_majflt);
842 	PS_ADD("utime",		"%ld",	TV2J(&kp.ki_rusage.ru_utime));
843 	PS_ADD("stime",		"%ld",	TV2J(&kp.ki_rusage.ru_stime));
844 	PS_ADD("cutime",	"%ld",	TV2J(&kp.ki_rusage_ch.ru_utime));
845 	PS_ADD("cstime",	"%ld",	TV2J(&kp.ki_rusage_ch.ru_stime));
846 	PS_ADD("priority",	"%d",	kp.ki_pri.pri_user);
847 	PS_ADD("nice",		"%d",	kp.ki_nice); /* 19 (nicest) to -19 */
848 	PS_ADD("0",		"%d",	0); /* removed field */
849 	PS_ADD("itrealvalue",	"%d",	0); /* XXX */
850 	PS_ADD("starttime",	"%lu",	TV2J(&kp.ki_start) - TV2J(&boottime));
851 	PS_ADD("vsize",		"%ju",	P2K((uintmax_t)kp.ki_size));
852 	PS_ADD("rss",		"%ju",	(uintmax_t)kp.ki_rssize);
853 	PS_ADD("rlim",		"%lu",	kp.ki_rusage.ru_maxrss);
854 	PS_ADD("startcode",	"%ju",	(uintmax_t)startcode);
855 	PS_ADD("endcode",	"%ju",	(uintmax_t)startdata);
856 	PS_ADD("startstack",	"%u",	0); /* XXX */
857 	PS_ADD("kstkesp",	"%u",	0); /* XXX */
858 	PS_ADD("kstkeip",	"%u",	0); /* XXX */
859 	PS_ADD("signal",	"%u",	0); /* XXX */
860 	PS_ADD("blocked",	"%u",	0); /* XXX */
861 	PS_ADD("sigignore",	"%u",	0); /* XXX */
862 	PS_ADD("sigcatch",	"%u",	0); /* XXX */
863 	PS_ADD("wchan",		"%u",	0); /* XXX */
864 	PS_ADD("nswap",		"%lu",	kp.ki_rusage.ru_nswap);
865 	PS_ADD("cnswap",	"%lu",	kp.ki_rusage_ch.ru_nswap);
866 	PS_ADD("exitsignal",	"%d",	0); /* XXX */
867 	PS_ADD("processor",	"%u",	kp.ki_lastcpu);
868 	PS_ADD("rt_priority",	"%u",	0); /* XXX */ /* >= 2.5.19 */
869 	PS_ADD("policy",	"%u",	kp.ki_pri.pri_class); /* >= 2.5.19 */
870 #undef PS_ADD
871 	sbuf_putc(sb, '\n');
872 
873 	return (0);
874 }
875 
876 /*
877  * Filler function for proc/pid/statm
878  */
879 static int
880 linprocfs_doprocstatm(PFS_FILL_ARGS)
881 {
882 	struct kinfo_proc kp;
883 	segsz_t lsize;
884 
885 	sx_slock(&proctree_lock);
886 	PROC_LOCK(p);
887 	fill_kinfo_proc(p, &kp);
888 	PROC_UNLOCK(p);
889 	sx_sunlock(&proctree_lock);
890 
891 	/*
892 	 * See comments in linprocfs_doprocstatus() regarding the
893 	 * computation of lsize.
894 	 */
895 	/* size resident share trs drs lrs dt */
896 	sbuf_printf(sb, "%ju ", B2P((uintmax_t)kp.ki_size));
897 	sbuf_printf(sb, "%ju ", (uintmax_t)kp.ki_rssize);
898 	sbuf_printf(sb, "%ju ", (uintmax_t)0); /* XXX */
899 	sbuf_printf(sb, "%ju ",	(uintmax_t)kp.ki_tsize);
900 	sbuf_printf(sb, "%ju ", (uintmax_t)(kp.ki_dsize + kp.ki_ssize));
901 	lsize = B2P(kp.ki_size) - kp.ki_dsize -
902 	    kp.ki_ssize - kp.ki_tsize - 1;
903 	sbuf_printf(sb, "%ju ", (uintmax_t)lsize);
904 	sbuf_printf(sb, "%ju\n", (uintmax_t)0); /* XXX */
905 
906 	return (0);
907 }
908 
909 /*
910  * Filler function for proc/pid/status
911  */
912 static int
913 linprocfs_doprocstatus(PFS_FILL_ARGS)
914 {
915 	struct kinfo_proc kp;
916 	char *state;
917 	segsz_t lsize;
918 	struct thread *td2;
919 	struct sigacts *ps;
920 	l_sigset_t siglist, sigignore, sigcatch;
921 	int i;
922 
923 	sx_slock(&proctree_lock);
924 	PROC_LOCK(p);
925 	td2 = FIRST_THREAD_IN_PROC(p); /* XXXKSE pretend only one thread */
926 
927 	if (P_SHOULDSTOP(p)) {
928 		state = "T (stopped)";
929 	} else {
930 		switch(p->p_state) {
931 		case PRS_NEW:
932 			state = "I (idle)";
933 			break;
934 		case PRS_NORMAL:
935 			if (p->p_flag & P_WEXIT) {
936 				state = "X (exiting)";
937 				break;
938 			}
939 			switch(td2->td_state) {
940 			case TDS_INHIBITED:
941 				state = "S (sleeping)";
942 				break;
943 			case TDS_RUNQ:
944 			case TDS_RUNNING:
945 				state = "R (running)";
946 				break;
947 			default:
948 				state = "? (unknown)";
949 				break;
950 			}
951 			break;
952 		case PRS_ZOMBIE:
953 			state = "Z (zombie)";
954 			break;
955 		default:
956 			state = "? (unknown)";
957 			break;
958 		}
959 	}
960 
961 	fill_kinfo_proc(p, &kp);
962 	sx_sunlock(&proctree_lock);
963 
964 	sbuf_printf(sb, "Name:\t%s\n",		p->p_comm); /* XXX escape */
965 	sbuf_printf(sb, "State:\t%s\n",		state);
966 
967 	/*
968 	 * Credentials
969 	 */
970 	sbuf_printf(sb, "Tgid:\t%d\n",		p->p_pid);
971 	sbuf_printf(sb, "Pid:\t%d\n",		p->p_pid);
972 	sbuf_printf(sb, "PPid:\t%d\n",		kp.ki_ppid );
973 	sbuf_printf(sb, "TracerPid:\t%d\n",	kp.ki_tracer );
974 	sbuf_printf(sb, "Uid:\t%d %d %d %d\n",	p->p_ucred->cr_ruid,
975 						p->p_ucred->cr_uid,
976 						p->p_ucred->cr_svuid,
977 						/* FreeBSD doesn't have fsuid */
978 						p->p_ucred->cr_uid);
979 	sbuf_printf(sb, "Gid:\t%d %d %d %d\n",	p->p_ucred->cr_rgid,
980 						p->p_ucred->cr_gid,
981 						p->p_ucred->cr_svgid,
982 						/* FreeBSD doesn't have fsgid */
983 						p->p_ucred->cr_gid);
984 	sbuf_cat(sb, "Groups:\t");
985 	for (i = 0; i < p->p_ucred->cr_ngroups; i++)
986 		sbuf_printf(sb, "%d ",		p->p_ucred->cr_groups[i]);
987 	PROC_UNLOCK(p);
988 	sbuf_putc(sb, '\n');
989 
990 	/*
991 	 * Memory
992 	 *
993 	 * While our approximation of VmLib may not be accurate (I
994 	 * don't know of a simple way to verify it, and I'm not sure
995 	 * it has much meaning anyway), I believe it's good enough.
996 	 *
997 	 * The same code that could (I think) accurately compute VmLib
998 	 * could also compute VmLck, but I don't really care enough to
999 	 * implement it. Submissions are welcome.
1000 	 */
1001 	sbuf_printf(sb, "VmSize:\t%8ju kB\n",	B2K((uintmax_t)kp.ki_size));
1002 	sbuf_printf(sb, "VmLck:\t%8u kB\n",	P2K(0)); /* XXX */
1003 	sbuf_printf(sb, "VmRSS:\t%8ju kB\n",	P2K((uintmax_t)kp.ki_rssize));
1004 	sbuf_printf(sb, "VmData:\t%8ju kB\n",	P2K((uintmax_t)kp.ki_dsize));
1005 	sbuf_printf(sb, "VmStk:\t%8ju kB\n",	P2K((uintmax_t)kp.ki_ssize));
1006 	sbuf_printf(sb, "VmExe:\t%8ju kB\n",	P2K((uintmax_t)kp.ki_tsize));
1007 	lsize = B2P(kp.ki_size) - kp.ki_dsize -
1008 	    kp.ki_ssize - kp.ki_tsize - 1;
1009 	sbuf_printf(sb, "VmLib:\t%8ju kB\n",	P2K((uintmax_t)lsize));
1010 
1011 	/*
1012 	 * Signal masks
1013 	 */
1014 	PROC_LOCK(p);
1015 	bsd_to_linux_sigset(&p->p_siglist, &siglist);
1016 	ps = p->p_sigacts;
1017 	mtx_lock(&ps->ps_mtx);
1018 	bsd_to_linux_sigset(&ps->ps_sigignore, &sigignore);
1019 	bsd_to_linux_sigset(&ps->ps_sigcatch, &sigcatch);
1020 	mtx_unlock(&ps->ps_mtx);
1021 	PROC_UNLOCK(p);
1022 
1023 	sbuf_printf(sb, "SigPnd:\t%016jx\n",	siglist.__mask);
1024 	/*
1025 	 * XXX. SigBlk - target thread's signal mask, td_sigmask.
1026 	 * To implement SigBlk pseudofs should support proc/tid dir entries.
1027 	 */
1028 	sbuf_printf(sb, "SigBlk:\t%016x\n",	0);
1029 	sbuf_printf(sb, "SigIgn:\t%016jx\n",	sigignore.__mask);
1030 	sbuf_printf(sb, "SigCgt:\t%016jx\n",	sigcatch.__mask);
1031 
1032 	/*
1033 	 * Linux also prints the capability masks, but we don't have
1034 	 * capabilities yet, and when we do get them they're likely to
1035 	 * be meaningless to Linux programs, so we lie. XXX
1036 	 */
1037 	sbuf_printf(sb, "CapInh:\t%016x\n",	0);
1038 	sbuf_printf(sb, "CapPrm:\t%016x\n",	0);
1039 	sbuf_printf(sb, "CapEff:\t%016x\n",	0);
1040 
1041 	return (0);
1042 }
1043 
1044 /*
1045  * Filler function for proc/pid/cwd
1046  */
1047 static int
1048 linprocfs_doproccwd(PFS_FILL_ARGS)
1049 {
1050 	struct pwd *pwd;
1051 	char *fullpath = "unknown";
1052 	char *freepath = NULL;
1053 
1054 	pwd = pwd_hold(td);
1055 	vn_fullpath(pwd->pwd_cdir, &fullpath, &freepath);
1056 	sbuf_printf(sb, "%s", fullpath);
1057 	if (freepath)
1058 		free(freepath, M_TEMP);
1059 	pwd_drop(pwd);
1060 	return (0);
1061 }
1062 
1063 /*
1064  * Filler function for proc/pid/root
1065  */
1066 static int
1067 linprocfs_doprocroot(PFS_FILL_ARGS)
1068 {
1069 	struct pwd *pwd;
1070 	struct vnode *vp;
1071 	char *fullpath = "unknown";
1072 	char *freepath = NULL;
1073 
1074 	pwd = pwd_hold(td);
1075 	vp = jailed(p->p_ucred) ? pwd->pwd_jdir : pwd->pwd_rdir;
1076 	vn_fullpath(vp, &fullpath, &freepath);
1077 	sbuf_printf(sb, "%s", fullpath);
1078 	if (freepath)
1079 		free(freepath, M_TEMP);
1080 	pwd_drop(pwd);
1081 	return (0);
1082 }
1083 
1084 /*
1085  * Filler function for proc/pid/cmdline
1086  */
1087 static int
1088 linprocfs_doproccmdline(PFS_FILL_ARGS)
1089 {
1090 	int ret;
1091 
1092 	PROC_LOCK(p);
1093 	if ((ret = p_cansee(td, p)) != 0) {
1094 		PROC_UNLOCK(p);
1095 		return (ret);
1096 	}
1097 
1098 	/*
1099 	 * Mimic linux behavior and pass only processes with usermode
1100 	 * address space as valid.  Return zero silently otherwize.
1101 	 */
1102 	if (p->p_vmspace == &vmspace0) {
1103 		PROC_UNLOCK(p);
1104 		return (0);
1105 	}
1106 	if (p->p_args != NULL) {
1107 		sbuf_bcpy(sb, p->p_args->ar_args, p->p_args->ar_length);
1108 		PROC_UNLOCK(p);
1109 		return (0);
1110 	}
1111 
1112 	if ((p->p_flag & P_SYSTEM) != 0) {
1113 		PROC_UNLOCK(p);
1114 		return (0);
1115 	}
1116 
1117 	PROC_UNLOCK(p);
1118 
1119 	ret = proc_getargv(td, p, sb);
1120 	return (ret);
1121 }
1122 
1123 /*
1124  * Filler function for proc/pid/environ
1125  */
1126 static int
1127 linprocfs_doprocenviron(PFS_FILL_ARGS)
1128 {
1129 
1130 	/*
1131 	 * Mimic linux behavior and pass only processes with usermode
1132 	 * address space as valid.  Return zero silently otherwize.
1133 	 */
1134 	if (p->p_vmspace == &vmspace0)
1135 		return (0);
1136 
1137 	return (proc_getenvv(td, p, sb));
1138 }
1139 
1140 static char l32_map_str[] = "%08lx-%08lx %s%s%s%s %08lx %02x:%02x %lu%s%s\n";
1141 static char l64_map_str[] = "%016lx-%016lx %s%s%s%s %08lx %02x:%02x %lu%s%s\n";
1142 static char vdso_str[] = "      [vdso]";
1143 static char stack_str[] = "      [stack]";
1144 
1145 /*
1146  * Filler function for proc/pid/maps
1147  */
1148 static int
1149 linprocfs_doprocmaps(PFS_FILL_ARGS)
1150 {
1151 	struct vmspace *vm;
1152 	vm_map_t map;
1153 	vm_map_entry_t entry, tmp_entry;
1154 	vm_object_t obj, tobj, lobj;
1155 	vm_offset_t e_start, e_end;
1156 	vm_ooffset_t off;
1157 	vm_prot_t e_prot;
1158 	unsigned int last_timestamp;
1159 	char *name = "", *freename = NULL;
1160 	const char *l_map_str;
1161 	ino_t ino;
1162 	int ref_count, shadow_count, flags;
1163 	int error;
1164 	struct vnode *vp;
1165 	struct vattr vat;
1166 	bool private;
1167 
1168 	PROC_LOCK(p);
1169 	error = p_candebug(td, p);
1170 	PROC_UNLOCK(p);
1171 	if (error)
1172 		return (error);
1173 
1174 	if (uio->uio_rw != UIO_READ)
1175 		return (EOPNOTSUPP);
1176 
1177 	error = 0;
1178 	vm = vmspace_acquire_ref(p);
1179 	if (vm == NULL)
1180 		return (ESRCH);
1181 
1182 	if (SV_CURPROC_FLAG(SV_LP64))
1183 		l_map_str = l64_map_str;
1184 	else
1185 		l_map_str = l32_map_str;
1186 	map = &vm->vm_map;
1187 	vm_map_lock_read(map);
1188 	VM_MAP_ENTRY_FOREACH(entry, map) {
1189 		name = "";
1190 		freename = NULL;
1191 		if (entry->eflags & MAP_ENTRY_IS_SUB_MAP)
1192 			continue;
1193 		e_prot = entry->protection;
1194 		e_start = entry->start;
1195 		e_end = entry->end;
1196 		obj = entry->object.vm_object;
1197 		off = entry->offset;
1198 		for (lobj = tobj = obj; tobj != NULL;
1199 		    lobj = tobj, tobj = tobj->backing_object) {
1200 			VM_OBJECT_RLOCK(tobj);
1201 			off += lobj->backing_object_offset;
1202 			if (lobj != obj)
1203 				VM_OBJECT_RUNLOCK(lobj);
1204 		}
1205 		private = (entry->eflags & MAP_ENTRY_COW) != 0 || obj == NULL ||
1206 		    (obj->flags & OBJ_ANON) != 0;
1207 		last_timestamp = map->timestamp;
1208 		vm_map_unlock_read(map);
1209 		ino = 0;
1210 		if (lobj) {
1211 			vp = vm_object_vnode(lobj);
1212 			if (vp != NULL)
1213 				vref(vp);
1214 			if (lobj != obj)
1215 				VM_OBJECT_RUNLOCK(lobj);
1216 			flags = obj->flags;
1217 			ref_count = obj->ref_count;
1218 			shadow_count = obj->shadow_count;
1219 			VM_OBJECT_RUNLOCK(obj);
1220 			if (vp != NULL) {
1221 				vn_fullpath(vp, &name, &freename);
1222 				vn_lock(vp, LK_SHARED | LK_RETRY);
1223 				VOP_GETATTR(vp, &vat, td->td_ucred);
1224 				ino = vat.va_fileid;
1225 				vput(vp);
1226 			} else if (SV_PROC_ABI(p) == SV_ABI_LINUX) {
1227 				if (e_start == p->p_sysent->sv_shared_page_base)
1228 					name = vdso_str;
1229 				if (e_end == p->p_sysent->sv_usrstack)
1230 					name = stack_str;
1231 			}
1232 		} else {
1233 			flags = 0;
1234 			ref_count = 0;
1235 			shadow_count = 0;
1236 		}
1237 
1238 		/*
1239 		 * format:
1240 		 *  start, end, access, offset, major, minor, inode, name.
1241 		 */
1242 		error = sbuf_printf(sb, l_map_str,
1243 		    (u_long)e_start, (u_long)e_end,
1244 		    (e_prot & VM_PROT_READ)?"r":"-",
1245 		    (e_prot & VM_PROT_WRITE)?"w":"-",
1246 		    (e_prot & VM_PROT_EXECUTE)?"x":"-",
1247 		    private ? "p" : "s",
1248 		    (u_long)off,
1249 		    0,
1250 		    0,
1251 		    (u_long)ino,
1252 		    *name ? "     " : "",
1253 		    name
1254 		    );
1255 		if (freename)
1256 			free(freename, M_TEMP);
1257 		vm_map_lock_read(map);
1258 		if (error == -1) {
1259 			error = 0;
1260 			break;
1261 		}
1262 		if (last_timestamp != map->timestamp) {
1263 			/*
1264 			 * Look again for the entry because the map was
1265 			 * modified while it was unlocked.  Specifically,
1266 			 * the entry may have been clipped, merged, or deleted.
1267 			 */
1268 			vm_map_lookup_entry(map, e_end - 1, &tmp_entry);
1269 			entry = tmp_entry;
1270 		}
1271 	}
1272 	vm_map_unlock_read(map);
1273 	vmspace_free(vm);
1274 
1275 	return (error);
1276 }
1277 
1278 /*
1279  * Criteria for interface name translation
1280  */
1281 #define IFP_IS_ETH(ifp) (ifp->if_type == IFT_ETHER)
1282 
1283 static int
1284 linux_ifname(struct ifnet *ifp, char *buffer, size_t buflen)
1285 {
1286 	struct ifnet *ifscan;
1287 	int ethno;
1288 
1289 	IFNET_RLOCK_ASSERT();
1290 
1291 	/* Short-circuit non ethernet interfaces */
1292 	if (!IFP_IS_ETH(ifp))
1293 		return (strlcpy(buffer, ifp->if_xname, buflen));
1294 
1295 	/* Determine the (relative) unit number for ethernet interfaces */
1296 	ethno = 0;
1297 	CK_STAILQ_FOREACH(ifscan, &V_ifnet, if_link) {
1298 		if (ifscan == ifp)
1299 			return (snprintf(buffer, buflen, "eth%d", ethno));
1300 		if (IFP_IS_ETH(ifscan))
1301 			ethno++;
1302 	}
1303 
1304 	return (0);
1305 }
1306 
1307 /*
1308  * Filler function for proc/net/dev
1309  */
1310 static int
1311 linprocfs_donetdev(PFS_FILL_ARGS)
1312 {
1313 	char ifname[16]; /* XXX LINUX_IFNAMSIZ */
1314 	struct ifnet *ifp;
1315 
1316 	sbuf_printf(sb, "%6s|%58s|%s\n"
1317 	    "%6s|%58s|%58s\n",
1318 	    "Inter-", "   Receive", "  Transmit",
1319 	    " face",
1320 	    "bytes    packets errs drop fifo frame compressed multicast",
1321 	    "bytes    packets errs drop fifo colls carrier compressed");
1322 
1323 	CURVNET_SET(TD_TO_VNET(curthread));
1324 	IFNET_RLOCK();
1325 	CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
1326 		linux_ifname(ifp, ifname, sizeof ifname);
1327 		sbuf_printf(sb, "%6.6s: ", ifname);
1328 		sbuf_printf(sb, "%7ju %7ju %4ju %4ju %4lu %5lu %10lu %9ju ",
1329 		    (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IBYTES),
1330 		    (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IPACKETS),
1331 		    (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IERRORS),
1332 		    (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IQDROPS),
1333 							/* rx_missed_errors */
1334 		    0UL,				/* rx_fifo_errors */
1335 		    0UL,				/* rx_length_errors +
1336 							 * rx_over_errors +
1337 							 * rx_crc_errors +
1338 							 * rx_frame_errors */
1339 		    0UL,				/* rx_compressed */
1340 		    (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IMCASTS));
1341 							/* XXX-BZ rx only? */
1342 		sbuf_printf(sb, "%8ju %7ju %4ju %4ju %4lu %5ju %7lu %10lu\n",
1343 		    (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_OBYTES),
1344 		    (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_OPACKETS),
1345 		    (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_OERRORS),
1346 		    (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_OQDROPS),
1347 		    0UL,				/* tx_fifo_errors */
1348 		    (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_COLLISIONS),
1349 		    0UL,				/* tx_carrier_errors +
1350 							 * tx_aborted_errors +
1351 							 * tx_window_errors +
1352 							 * tx_heartbeat_errors*/
1353 		    0UL);				/* tx_compressed */
1354 	}
1355 	IFNET_RUNLOCK();
1356 	CURVNET_RESTORE();
1357 
1358 	return (0);
1359 }
1360 
1361 /*
1362  * Filler function for proc/sys/kernel/osrelease
1363  */
1364 static int
1365 linprocfs_doosrelease(PFS_FILL_ARGS)
1366 {
1367 	char osrelease[LINUX_MAX_UTSNAME];
1368 
1369 	linux_get_osrelease(td, osrelease);
1370 	sbuf_printf(sb, "%s\n", osrelease);
1371 
1372 	return (0);
1373 }
1374 
1375 /*
1376  * Filler function for proc/sys/kernel/ostype
1377  */
1378 static int
1379 linprocfs_doostype(PFS_FILL_ARGS)
1380 {
1381 	char osname[LINUX_MAX_UTSNAME];
1382 
1383 	linux_get_osname(td, osname);
1384 	sbuf_printf(sb, "%s\n", osname);
1385 
1386 	return (0);
1387 }
1388 
1389 /*
1390  * Filler function for proc/sys/kernel/version
1391  */
1392 static int
1393 linprocfs_doosbuild(PFS_FILL_ARGS)
1394 {
1395 
1396 	linprocfs_osbuild(td, sb);
1397 	sbuf_cat(sb, "\n");
1398 	return (0);
1399 }
1400 
1401 /*
1402  * Filler function for proc/sys/kernel/msgmax
1403  */
1404 static int
1405 linprocfs_domsgmax(PFS_FILL_ARGS)
1406 {
1407 
1408 	sbuf_printf(sb, "%d\n", msginfo.msgmax);
1409 	return (0);
1410 }
1411 
1412 /*
1413  * Filler function for proc/sys/kernel/msgmni
1414  */
1415 static int
1416 linprocfs_domsgmni(PFS_FILL_ARGS)
1417 {
1418 
1419 	sbuf_printf(sb, "%d\n", msginfo.msgmni);
1420 	return (0);
1421 }
1422 
1423 /*
1424  * Filler function for proc/sys/kernel/msgmnb
1425  */
1426 static int
1427 linprocfs_domsgmnb(PFS_FILL_ARGS)
1428 {
1429 
1430 	sbuf_printf(sb, "%d\n", msginfo.msgmnb);
1431 	return (0);
1432 }
1433 
1434 /*
1435  * Filler function for proc/sys/kernel/pid_max
1436  */
1437 static int
1438 linprocfs_dopid_max(PFS_FILL_ARGS)
1439 {
1440 
1441 	sbuf_printf(sb, "%i\n", PID_MAX);
1442 	return (0);
1443 }
1444 
1445 /*
1446  * Filler function for proc/sys/kernel/sem
1447  */
1448 static int
1449 linprocfs_dosem(PFS_FILL_ARGS)
1450 {
1451 
1452 	sbuf_printf(sb, "%d %d %d %d\n", seminfo.semmsl, seminfo.semmns,
1453 	    seminfo.semopm, seminfo.semmni);
1454 	return (0);
1455 }
1456 
1457 /*
1458  * Filler function for proc/sys/kernel/shmall
1459  */
1460 static int
1461 linprocfs_doshmall(PFS_FILL_ARGS)
1462 {
1463 
1464 	sbuf_printf(sb, "%lu\n", shminfo.shmall);
1465 	return (0);
1466 }
1467 
1468 /*
1469  * Filler function for proc/sys/kernel/shmmax
1470  */
1471 static int
1472 linprocfs_doshmmax(PFS_FILL_ARGS)
1473 {
1474 
1475 	sbuf_printf(sb, "%lu\n", shminfo.shmmax);
1476 	return (0);
1477 }
1478 
1479 /*
1480  * Filler function for proc/sys/kernel/shmmni
1481  */
1482 static int
1483 linprocfs_doshmmni(PFS_FILL_ARGS)
1484 {
1485 
1486 	sbuf_printf(sb, "%lu\n", shminfo.shmmni);
1487 	return (0);
1488 }
1489 
1490 /*
1491  * Filler function for proc/sys/kernel/tainted
1492  */
1493 static int
1494 linprocfs_dotainted(PFS_FILL_ARGS)
1495 {
1496 
1497 	sbuf_printf(sb, "0\n");
1498 	return (0);
1499 }
1500 
1501 /*
1502  * Filler function for proc/sys/vm/min_free_kbytes
1503  *
1504  * This mirrors the approach in illumos to return zero for reads. Effectively,
1505  * it says, no memory is kept in reserve for "atomic allocations". This class
1506  * of allocation can be used at times when a thread cannot be suspended.
1507  */
1508 static int
1509 linprocfs_dominfree(PFS_FILL_ARGS)
1510 {
1511 
1512 	sbuf_printf(sb, "%d\n", 0);
1513 	return (0);
1514 }
1515 
1516 /*
1517  * Filler function for proc/scsi/device_info
1518  */
1519 static int
1520 linprocfs_doscsidevinfo(PFS_FILL_ARGS)
1521 {
1522 
1523 	return (0);
1524 }
1525 
1526 /*
1527  * Filler function for proc/scsi/scsi
1528  */
1529 static int
1530 linprocfs_doscsiscsi(PFS_FILL_ARGS)
1531 {
1532 
1533 	return (0);
1534 }
1535 
1536 /*
1537  * Filler function for proc/devices
1538  */
1539 static int
1540 linprocfs_dodevices(PFS_FILL_ARGS)
1541 {
1542 	char *char_devices;
1543 	sbuf_printf(sb, "Character devices:\n");
1544 
1545 	char_devices = linux_get_char_devices();
1546 	sbuf_printf(sb, "%s", char_devices);
1547 	linux_free_get_char_devices(char_devices);
1548 
1549 	sbuf_printf(sb, "\nBlock devices:\n");
1550 
1551 	return (0);
1552 }
1553 
1554 /*
1555  * Filler function for proc/cmdline
1556  */
1557 static int
1558 linprocfs_docmdline(PFS_FILL_ARGS)
1559 {
1560 
1561 	sbuf_printf(sb, "BOOT_IMAGE=%s", kernelname);
1562 	sbuf_printf(sb, " ro root=302\n");
1563 	return (0);
1564 }
1565 
1566 /*
1567  * Filler function for proc/filesystems
1568  */
1569 static int
1570 linprocfs_dofilesystems(PFS_FILL_ARGS)
1571 {
1572 	struct vfsconf *vfsp;
1573 
1574 	vfsconf_slock();
1575 	TAILQ_FOREACH(vfsp, &vfsconf, vfc_list) {
1576 		if (vfsp->vfc_flags & VFCF_SYNTHETIC)
1577 			sbuf_printf(sb, "nodev");
1578 		sbuf_printf(sb, "\t%s\n", vfsp->vfc_name);
1579 	}
1580 	vfsconf_sunlock();
1581 	return(0);
1582 }
1583 
1584 /*
1585  * Filler function for proc/modules
1586  */
1587 static int
1588 linprocfs_domodules(PFS_FILL_ARGS)
1589 {
1590 #if 0
1591 	struct linker_file *lf;
1592 
1593 	TAILQ_FOREACH(lf, &linker_files, link) {
1594 		sbuf_printf(sb, "%-20s%8lu%4d\n", lf->filename,
1595 		    (unsigned long)lf->size, lf->refs);
1596 	}
1597 #endif
1598 	return (0);
1599 }
1600 
1601 /*
1602  * Filler function for proc/pid/fd
1603  */
1604 static int
1605 linprocfs_dofdescfs(PFS_FILL_ARGS)
1606 {
1607 
1608 	if (p == curproc)
1609 		sbuf_printf(sb, "/dev/fd");
1610 	else
1611 		sbuf_printf(sb, "unknown");
1612 	return (0);
1613 }
1614 
1615 /*
1616  * Filler function for proc/pid/limits
1617  */
1618 static const struct linux_rlimit_ident {
1619 	const char	*desc;
1620 	const char	*unit;
1621 	unsigned int	rlim_id;
1622 } linux_rlimits_ident[] = {
1623 	{ "Max cpu time",	"seconds",	RLIMIT_CPU },
1624 	{ "Max file size", 	"bytes",	RLIMIT_FSIZE },
1625 	{ "Max data size",	"bytes", 	RLIMIT_DATA },
1626 	{ "Max stack size",	"bytes", 	RLIMIT_STACK },
1627 	{ "Max core file size",  "bytes",	RLIMIT_CORE },
1628 	{ "Max resident set",	"bytes",	RLIMIT_RSS },
1629 	{ "Max processes",	"processes",	RLIMIT_NPROC },
1630 	{ "Max open files",	"files",	RLIMIT_NOFILE },
1631 	{ "Max locked memory",	"bytes",	RLIMIT_MEMLOCK },
1632 	{ "Max address space",	"bytes",	RLIMIT_AS },
1633 	{ "Max file locks",	"locks",	LINUX_RLIMIT_LOCKS },
1634 	{ "Max pending signals", "signals",	LINUX_RLIMIT_SIGPENDING },
1635 	{ "Max msgqueue size",	"bytes",	LINUX_RLIMIT_MSGQUEUE },
1636 	{ "Max nice priority", 		"",	LINUX_RLIMIT_NICE },
1637 	{ "Max realtime priority",	"",	LINUX_RLIMIT_RTPRIO },
1638 	{ "Max realtime timeout",	"us",	LINUX_RLIMIT_RTTIME },
1639 	{ 0, 0, 0 }
1640 };
1641 
1642 static int
1643 linprocfs_doproclimits(PFS_FILL_ARGS)
1644 {
1645 	const struct linux_rlimit_ident *li;
1646 	struct plimit *limp;
1647 	struct rlimit rl;
1648 	ssize_t size;
1649 	int res, error;
1650 
1651 	error = 0;
1652 
1653 	PROC_LOCK(p);
1654 	limp = lim_hold(p->p_limit);
1655 	PROC_UNLOCK(p);
1656 	size = sizeof(res);
1657 	sbuf_printf(sb, "%-26s%-21s%-21s%-21s\n", "Limit", "Soft Limit",
1658 			"Hard Limit", "Units");
1659 	for (li = linux_rlimits_ident; li->desc != NULL; ++li) {
1660 		switch (li->rlim_id)
1661 		{
1662 		case LINUX_RLIMIT_LOCKS:
1663 			/* FALLTHROUGH */
1664 		case LINUX_RLIMIT_RTTIME:
1665 			rl.rlim_cur = RLIM_INFINITY;
1666 			break;
1667 		case LINUX_RLIMIT_SIGPENDING:
1668 			error = kernel_sysctlbyname(td,
1669 			    "kern.sigqueue.max_pending_per_proc",
1670 			    &res, &size, 0, 0, 0, 0);
1671 			if (error != 0)
1672 				goto out;
1673 			rl.rlim_cur = res;
1674 			rl.rlim_max = res;
1675 			break;
1676 		case LINUX_RLIMIT_MSGQUEUE:
1677 			error = kernel_sysctlbyname(td,
1678 			    "kern.ipc.msgmnb", &res, &size, 0, 0, 0, 0);
1679 			if (error != 0)
1680 				goto out;
1681 			rl.rlim_cur = res;
1682 			rl.rlim_max = res;
1683 			break;
1684 		case LINUX_RLIMIT_NICE:
1685 			/* FALLTHROUGH */
1686 		case LINUX_RLIMIT_RTPRIO:
1687 			rl.rlim_cur = 0;
1688 			rl.rlim_max = 0;
1689 			break;
1690 		default:
1691 			rl = limp->pl_rlimit[li->rlim_id];
1692 			break;
1693 		}
1694 		if (rl.rlim_cur == RLIM_INFINITY)
1695 			sbuf_printf(sb, "%-26s%-21s%-21s%-10s\n",
1696 			    li->desc, "unlimited", "unlimited", li->unit);
1697 		else
1698 			sbuf_printf(sb, "%-26s%-21llu%-21llu%-10s\n",
1699 			    li->desc, (unsigned long long)rl.rlim_cur,
1700 			    (unsigned long long)rl.rlim_max, li->unit);
1701 	}
1702 out:
1703 	lim_free(limp);
1704 	return (error);
1705 }
1706 
1707 /*
1708  * The point of the following two functions is to work around
1709  * an assertion in Chromium; see kern/240991 for details.
1710  */
1711 static int
1712 linprocfs_dotaskattr(PFS_ATTR_ARGS)
1713 {
1714 
1715 	vap->va_nlink = 3;
1716 	return (0);
1717 }
1718 
1719 /*
1720  * Filler function for proc/<pid>/task/.dummy
1721  */
1722 static int
1723 linprocfs_dotaskdummy(PFS_FILL_ARGS)
1724 {
1725 
1726 	return (0);
1727 }
1728 
1729 /*
1730  * Filler function for proc/sys/kernel/random/uuid
1731  */
1732 static int
1733 linprocfs_douuid(PFS_FILL_ARGS)
1734 {
1735 	struct uuid uuid;
1736 
1737 	kern_uuidgen(&uuid, 1);
1738 	sbuf_printf_uuid(sb, &uuid);
1739 	sbuf_printf(sb, "\n");
1740 	return(0);
1741 }
1742 
1743 /*
1744  * Filler function for proc/pid/auxv
1745  */
1746 static int
1747 linprocfs_doauxv(PFS_FILL_ARGS)
1748 {
1749 	struct sbuf *asb;
1750 	off_t buflen, resid;
1751 	int error;
1752 
1753 	/*
1754 	 * Mimic linux behavior and pass only processes with usermode
1755 	 * address space as valid. Return zero silently otherwise.
1756 	 */
1757 	if (p->p_vmspace == &vmspace0)
1758 		return (0);
1759 
1760 	if (uio->uio_resid == 0)
1761 		return (0);
1762 	if (uio->uio_offset < 0 || uio->uio_resid < 0)
1763 		return (EINVAL);
1764 
1765 	asb = sbuf_new_auto();
1766 	if (asb == NULL)
1767 		return (ENOMEM);
1768 	error = proc_getauxv(td, p, asb);
1769 	if (error == 0)
1770 		error = sbuf_finish(asb);
1771 
1772 	resid = sbuf_len(asb) - uio->uio_offset;
1773 	if (resid > uio->uio_resid)
1774 		buflen = uio->uio_resid;
1775 	else
1776 		buflen = resid;
1777 	if (buflen > IOSIZE_MAX)
1778 		return (EINVAL);
1779 	if (buflen > MAXPHYS)
1780 		buflen = MAXPHYS;
1781 	if (resid <= 0)
1782 		return (0);
1783 
1784 	if (error == 0)
1785 		error = uiomove(sbuf_data(asb) + uio->uio_offset, buflen, uio);
1786 	sbuf_delete(asb);
1787 	return (error);
1788 }
1789 
1790 /*
1791  * Constructor
1792  */
1793 static int
1794 linprocfs_init(PFS_INIT_ARGS)
1795 {
1796 	struct pfs_node *root;
1797 	struct pfs_node *dir;
1798 	struct pfs_node *sys;
1799 
1800 	root = pi->pi_root;
1801 
1802 	/* /proc/... */
1803 	pfs_create_file(root, "cmdline", &linprocfs_docmdline,
1804 	    NULL, NULL, NULL, PFS_RD);
1805 	pfs_create_file(root, "cpuinfo", &linprocfs_docpuinfo,
1806 	    NULL, NULL, NULL, PFS_RD);
1807 	pfs_create_file(root, "devices", &linprocfs_dodevices,
1808 	    NULL, NULL, NULL, PFS_RD);
1809 	pfs_create_file(root, "filesystems", &linprocfs_dofilesystems,
1810 	    NULL, NULL, NULL, PFS_RD);
1811 	pfs_create_file(root, "loadavg", &linprocfs_doloadavg,
1812 	    NULL, NULL, NULL, PFS_RD);
1813 	pfs_create_file(root, "meminfo", &linprocfs_domeminfo,
1814 	    NULL, NULL, NULL, PFS_RD);
1815 	pfs_create_file(root, "modules", &linprocfs_domodules,
1816 	    NULL, NULL, NULL, PFS_RD);
1817 	pfs_create_file(root, "mounts", &linprocfs_domtab,
1818 	    NULL, NULL, NULL, PFS_RD);
1819 	pfs_create_file(root, "mtab", &linprocfs_domtab,
1820 	    NULL, NULL, NULL, PFS_RD);
1821 	pfs_create_file(root, "partitions", &linprocfs_dopartitions,
1822 	    NULL, NULL, NULL, PFS_RD);
1823 	pfs_create_link(root, "self", &procfs_docurproc,
1824 	    NULL, NULL, NULL, 0);
1825 	pfs_create_file(root, "stat", &linprocfs_dostat,
1826 	    NULL, NULL, NULL, PFS_RD);
1827 	pfs_create_file(root, "swaps", &linprocfs_doswaps,
1828 	    NULL, NULL, NULL, PFS_RD);
1829 	pfs_create_file(root, "uptime", &linprocfs_douptime,
1830 	    NULL, NULL, NULL, PFS_RD);
1831 	pfs_create_file(root, "version", &linprocfs_doversion,
1832 	    NULL, NULL, NULL, PFS_RD);
1833 
1834 	/* /proc/bus/... */
1835 	dir = pfs_create_dir(root, "bus", NULL, NULL, NULL, 0);
1836 	dir = pfs_create_dir(dir, "pci", NULL, NULL, NULL, 0);
1837 	dir = pfs_create_dir(dir, "devices", NULL, NULL, NULL, 0);
1838 
1839 	/* /proc/net/... */
1840 	dir = pfs_create_dir(root, "net", NULL, NULL, NULL, 0);
1841 	pfs_create_file(dir, "dev", &linprocfs_donetdev,
1842 	    NULL, NULL, NULL, PFS_RD);
1843 
1844 	/* /proc/<pid>/... */
1845 	dir = pfs_create_dir(root, "pid", NULL, NULL, NULL, PFS_PROCDEP);
1846 	pfs_create_file(dir, "cmdline", &linprocfs_doproccmdline,
1847 	    NULL, NULL, NULL, PFS_RD);
1848 	pfs_create_link(dir, "cwd", &linprocfs_doproccwd,
1849 	    NULL, NULL, NULL, 0);
1850 	pfs_create_file(dir, "environ", &linprocfs_doprocenviron,
1851 	    NULL, &procfs_candebug, NULL, PFS_RD);
1852 	pfs_create_link(dir, "exe", &procfs_doprocfile,
1853 	    NULL, &procfs_notsystem, NULL, 0);
1854 	pfs_create_file(dir, "maps", &linprocfs_doprocmaps,
1855 	    NULL, NULL, NULL, PFS_RD);
1856 	pfs_create_file(dir, "mem", &procfs_doprocmem,
1857 	    procfs_attr_rw, &procfs_candebug, NULL, PFS_RDWR | PFS_RAW);
1858 	pfs_create_file(dir, "mounts", &linprocfs_domtab,
1859 	    NULL, NULL, NULL, PFS_RD);
1860 	pfs_create_link(dir, "root", &linprocfs_doprocroot,
1861 	    NULL, NULL, NULL, 0);
1862 	pfs_create_file(dir, "stat", &linprocfs_doprocstat,
1863 	    NULL, NULL, NULL, PFS_RD);
1864 	pfs_create_file(dir, "statm", &linprocfs_doprocstatm,
1865 	    NULL, NULL, NULL, PFS_RD);
1866 	pfs_create_file(dir, "status", &linprocfs_doprocstatus,
1867 	    NULL, NULL, NULL, PFS_RD);
1868 	pfs_create_link(dir, "fd", &linprocfs_dofdescfs,
1869 	    NULL, NULL, NULL, 0);
1870 	pfs_create_file(dir, "auxv", &linprocfs_doauxv,
1871 	    NULL, &procfs_candebug, NULL, PFS_RD|PFS_RAWRD);
1872 	pfs_create_file(dir, "limits", &linprocfs_doproclimits,
1873 	    NULL, NULL, NULL, PFS_RD);
1874 
1875 	/* /proc/<pid>/task/... */
1876 	dir = pfs_create_dir(dir, "task", linprocfs_dotaskattr, NULL, NULL, 0);
1877 	pfs_create_file(dir, ".dummy", &linprocfs_dotaskdummy,
1878 	    NULL, NULL, NULL, PFS_RD);
1879 
1880 	/* /proc/scsi/... */
1881 	dir = pfs_create_dir(root, "scsi", NULL, NULL, NULL, 0);
1882 	pfs_create_file(dir, "device_info", &linprocfs_doscsidevinfo,
1883 	    NULL, NULL, NULL, PFS_RD);
1884 	pfs_create_file(dir, "scsi", &linprocfs_doscsiscsi,
1885 	    NULL, NULL, NULL, PFS_RD);
1886 
1887 	/* /proc/sys/... */
1888 	sys = pfs_create_dir(root, "sys", NULL, NULL, NULL, 0);
1889 
1890 	/* /proc/sys/kernel/... */
1891 	dir = pfs_create_dir(sys, "kernel", NULL, NULL, NULL, 0);
1892 	pfs_create_file(dir, "osrelease", &linprocfs_doosrelease,
1893 	    NULL, NULL, NULL, PFS_RD);
1894 	pfs_create_file(dir, "ostype", &linprocfs_doostype,
1895 	    NULL, NULL, NULL, PFS_RD);
1896 	pfs_create_file(dir, "version", &linprocfs_doosbuild,
1897 	    NULL, NULL, NULL, PFS_RD);
1898 	pfs_create_file(dir, "msgmax", &linprocfs_domsgmax,
1899 	    NULL, NULL, NULL, PFS_RD);
1900 	pfs_create_file(dir, "msgmni", &linprocfs_domsgmni,
1901 	    NULL, NULL, NULL, PFS_RD);
1902 	pfs_create_file(dir, "msgmnb", &linprocfs_domsgmnb,
1903 	    NULL, NULL, NULL, PFS_RD);
1904 	pfs_create_file(dir, "pid_max", &linprocfs_dopid_max,
1905 	    NULL, NULL, NULL, PFS_RD);
1906 	pfs_create_file(dir, "sem", &linprocfs_dosem,
1907 	    NULL, NULL, NULL, PFS_RD);
1908 	pfs_create_file(dir, "shmall", &linprocfs_doshmall,
1909 	    NULL, NULL, NULL, PFS_RD);
1910 	pfs_create_file(dir, "shmmax", &linprocfs_doshmmax,
1911 	    NULL, NULL, NULL, PFS_RD);
1912 	pfs_create_file(dir, "shmmni", &linprocfs_doshmmni,
1913 	    NULL, NULL, NULL, PFS_RD);
1914 	pfs_create_file(dir, "tainted", &linprocfs_dotainted,
1915 	    NULL, NULL, NULL, PFS_RD);
1916 
1917 	/* /proc/sys/kernel/random/... */
1918 	dir = pfs_create_dir(dir, "random", NULL, NULL, NULL, 0);
1919 	pfs_create_file(dir, "uuid", &linprocfs_douuid,
1920 	    NULL, NULL, NULL, PFS_RD);
1921 
1922 	/* /proc/sys/vm/.... */
1923 	dir = pfs_create_dir(sys, "vm", NULL, NULL, NULL, 0);
1924 	pfs_create_file(dir, "min_free_kbytes", &linprocfs_dominfree,
1925 	    NULL, NULL, NULL, PFS_RD);
1926 
1927 	return (0);
1928 }
1929 
1930 /*
1931  * Destructor
1932  */
1933 static int
1934 linprocfs_uninit(PFS_INIT_ARGS)
1935 {
1936 
1937 	/* nothing to do, pseudofs will GC */
1938 	return (0);
1939 }
1940 
1941 PSEUDOFS(linprocfs, 1, VFCF_JAIL);
1942 #if defined(__aarch64__) || defined(__amd64__)
1943 MODULE_DEPEND(linprocfs, linux_common, 1, 1, 1);
1944 #else
1945 MODULE_DEPEND(linprocfs, linux, 1, 1, 1);
1946 #endif
1947 MODULE_DEPEND(linprocfs, procfs, 1, 1, 1);
1948 MODULE_DEPEND(linprocfs, sysvmsg, 1, 1, 1);
1949 MODULE_DEPEND(linprocfs, sysvsem, 1, 1, 1);
1950 MODULE_DEPEND(linprocfs, sysvshm, 1, 1, 1);
1951