sys/compat/linprocfs/linprocfs.c

   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 "opt_inet.h"
  45
  46 #include <sys/cdefs.h>
  47 __FBSDID("$FreeBSD$");
  48
  49 #include <sys/param.h>
  50 #include <sys/queue.h>
  51 #include <sys/blist.h>
  52 #include <sys/conf.h>
  53 #include <sys/exec.h>
  54 #include <sys/fcntl.h>
  55 #include <sys/filedesc.h>
  56 #include <sys/jail.h>
  57 #include <sys/kernel.h>
  58 #include <sys/limits.h>
  59 #include <sys/linker.h>
  60 #include <sys/lock.h>
  61 #include <sys/malloc.h>
  62 #include <sys/msg.h>
  63 #include <sys/mutex.h>
  64 #include <sys/namei.h>
  65 #include <sys/proc.h>
  66 #include <sys/ptrace.h>
  67 #include <sys/resourcevar.h>
  68 #include <sys/resource.h>
  69 #include <sys/sbuf.h>
  70 #include <sys/sem.h>
  71 #include <sys/shm.h>
  72 #include <sys/smp.h>
  73 #include <sys/socket.h>
  74 #include <sys/syscallsubr.h>
  75 #include <sys/sysctl.h>
  76 #include <sys/sysent.h>
  77 #include <sys/systm.h>
  78 #include <sys/time.h>
  79 #include <sys/tty.h>
  80 #include <sys/user.h>
  81 #include <sys/uuid.h>
  82 #include <sys/vmmeter.h>
  83 #include <sys/vnode.h>
  84 #include <sys/bus.h>
  85 #include <sys/uio.h>
  86
  87 #include <net/if.h>
  88 #include <net/if_var.h>
  89 #include <net/if_types.h>
  90
  91 #include <net/route.h>
  92 #include <net/route/nhop.h>
  93 #include <net/route/route_ctl.h>
  94
  95 #include <vm/vm.h>
  96 #include <vm/vm_extern.h>
  97 #include <vm/pmap.h>
  98 #include <vm/vm_map.h>
  99 #include <vm/vm_param.h>
 100 #include <vm/vm_object.h>
 101 #include <vm/swap_pager.h>
 102
 103 #include <machine/clock.h>
 104
 105 #include <geom/geom.h>
 106 #include <geom/geom_int.h>
 107
 108 #if defined(__i386__) || defined(__amd64__)
 109 #include <machine/cputypes.h>
 110 #include <machine/md_var.h>
 111 #endif /* __i386__ || __amd64__ */
 112
 113 #include <compat/linux/linux.h>
 114 #include <compat/linux/linux_emul.h>
 115 #include <compat/linux/linux_mib.h>
 116 #include <compat/linux/linux_misc.h>
 117 #include <compat/linux/linux_util.h>
 118 #include <fs/pseudofs/pseudofs.h>
 119 #include <fs/procfs/procfs.h>
 120
 121 /*
 122  * Various conversion macros
 123  */
 124 #define T2J(x) ((long)(((x) * 100ULL) / (stathz ? stathz : hz)))        /* ticks to jiffies */
 125 #define T2CS(x) ((unsigned long)(((x) * 100ULL) / (stathz ? stathz : hz)))      /* ticks to centiseconds */
 126 #define T2S(x) ((x) / (stathz ? stathz : hz))           /* ticks to seconds */
 127 #define B2K(x) ((x) >> 10)                              /* bytes to kbytes */
 128 #define B2P(x) ((x) >> PAGE_SHIFT)                      /* bytes to pages */
 129 #define P2B(x) ((x) << PAGE_SHIFT)                      /* pages to bytes */
 130 #define P2K(x) ((x) << (PAGE_SHIFT - 10))               /* pages to kbytes */
 131 #define TV2J(x) ((x)->tv_sec * 100UL + (x)->tv_usec / 10000)
 132
 133 /**
 134  * @brief Mapping of ki_stat in struct kinfo_proc to the linux state
 135  *
 136  * The linux procfs state field displays one of the characters RSDZTW to
 137  * denote running, sleeping in an interruptible wait, waiting in an
 138  * uninterruptible disk sleep, a zombie process, process is being traced
 139  * or stopped, or process is paging respectively.
 140  *
 141  * Our struct kinfo_proc contains the variable ki_stat which contains a
 142  * value out of SIDL, SRUN, SSLEEP, SSTOP, SZOMB, SWAIT and SLOCK.
 143  *
 144  * This character array is used with ki_stati-1 as an index and tries to
 145  * map our states to suitable linux states.
 146  */
 147 static char linux_state[] = "RRSTZDD";
 148
 149 /*
 150  * Filler function for proc/meminfo
 151  */
 152 static int
 153 linprocfs_domeminfo(PFS_FILL_ARGS)
 154 {
 155         unsigned long memtotal;         /* total memory in bytes */
 156         unsigned long memfree;          /* free memory in bytes */
 157         unsigned long cached;           /* page cache */
 158         unsigned long buffers;          /* buffer cache */
 159         unsigned long long swaptotal;   /* total swap space in bytes */
 160         unsigned long long swapused;    /* used swap space in bytes */
 161         unsigned long long swapfree;    /* free swap space in bytes */
 162         size_t sz;
 163         int error, i, j;
 164
 165         memtotal = physmem * PAGE_SIZE;
 166         memfree = (unsigned long)vm_free_count() * PAGE_SIZE;
 167         swap_pager_status(&i, &j);
 168         swaptotal = (unsigned long long)i * PAGE_SIZE;
 169         swapused = (unsigned long long)j * PAGE_SIZE;
 170         swapfree = swaptotal - swapused;
 171
 172         /*
 173          * This value may exclude wired pages, but we have no good way of
 174          * accounting for that.
 175          */
 176         cached =
 177             (vm_active_count() + vm_inactive_count() + vm_laundry_count()) *
 178             PAGE_SIZE;
 179
 180         sz = sizeof(buffers);
 181         error = kernel_sysctlbyname(curthread, "vfs.bufspace", &buffers, &sz,
 182             NULL, 0, 0, 0);
 183         if (error != 0)
 184                 buffers = 0;
 185
 186         sbuf_printf(sb,
 187             "MemTotal: %9lu kB\n"
 188             "MemFree:  %9lu kB\n"
 189             "Buffers:  %9lu kB\n"
 190             "Cached:   %9lu kB\n"
 191             "SwapTotal:%9llu kB\n"
 192             "SwapFree: %9llu kB\n",
 193             B2K(memtotal), B2K(memfree), B2K(buffers),
 194             B2K(cached), B2K(swaptotal), B2K(swapfree));
 195
 196         return (0);
 197 }
 198
 199 #if defined(__i386__) || defined(__amd64__)
 200 /*
 201  * Filler function for proc/cpuinfo (i386 & amd64 version)
 202  */
 203 static int
 204 linprocfs_docpuinfo(PFS_FILL_ARGS)
 205 {
 206         int hw_model[2];
 207         char model[128];
 208         uint64_t freq;
 209         size_t size;
 210         u_int cache_size[4];
 211         u_int regs[4] = { 0 };
 212         int fqmhz, fqkhz;
 213         int i, j;
 214
 215         /*
 216          * We default the flags to include all non-conflicting flags,
 217          * and the Intel versions of conflicting flags.
 218          */
 219         static char *cpu_feature_names[] = {
 220                 /*  0 */ "fpu", "vme", "de", "pse",
 221                 /*  4 */ "tsc", "msr", "pae", "mce",
 222                 /*  8 */ "cx8", "apic", "", "sep",
 223                 /* 12 */ "mtrr", "pge", "mca", "cmov",
 224                 /* 16 */ "pat", "pse36", "pn", "clflush",
 225                 /* 20 */ "", "dts", "acpi", "mmx",
 226                 /* 24 */ "fxsr", "sse", "sse2", "ss",
 227                 /* 28 */ "ht", "tm", "ia64", "pbe"
 228         };
 229
 230         static char *amd_feature_names[] = {
 231                 /*  0 */ "", "", "", "",
 232                 /*  4 */ "", "", "", "",
 233                 /*  8 */ "", "", "", "syscall",
 234                 /* 12 */ "", "", "", "",
 235                 /* 16 */ "", "", "", "mp",
 236                 /* 20 */ "nx", "", "mmxext", "",
 237                 /* 24 */ "", "fxsr_opt", "pdpe1gb", "rdtscp",
 238                 /* 28 */ "", "lm", "3dnowext", "3dnow"
 239         };
 240
 241         static char *cpu_feature2_names[] = {
 242                 /*  0 */ "pni", "pclmulqdq", "dtes64", "monitor",
 243                 /*  4 */ "ds_cpl", "vmx", "smx", "est",
 244                 /*  8 */ "tm2", "ssse3", "cid", "sdbg",
 245                 /* 12 */ "fma", "cx16", "xtpr", "pdcm",
 246                 /* 16 */ "", "pcid", "dca", "sse4_1",
 247                 /* 20 */ "sse4_2", "x2apic", "movbe", "popcnt",
 248                 /* 24 */ "tsc_deadline_timer", "aes", "xsave", "",
 249                 /* 28 */ "avx", "f16c", "rdrand", "hypervisor"
 250         };
 251
 252         static char *amd_feature2_names[] = {
 253                 /*  0 */ "lahf_lm", "cmp_legacy", "svm", "extapic",
 254                 /*  4 */ "cr8_legacy", "abm", "sse4a", "misalignsse",
 255                 /*  8 */ "3dnowprefetch", "osvw", "ibs", "xop",
 256                 /* 12 */ "skinit", "wdt", "", "lwp",
 257                 /* 16 */ "fma4", "tce", "", "nodeid_msr",
 258                 /* 20 */ "", "tbm", "topoext", "perfctr_core",
 259                 /* 24 */ "perfctr_nb", "", "bpext", "ptsc",
 260                 /* 28 */ "perfctr_llc", "mwaitx", "", ""
 261         };
 262
 263         static char *cpu_stdext_feature_names[] = {
 264                 /*  0 */ "fsgsbase", "tsc_adjust", "sgx", "bmi1",
 265                 /*  4 */ "hle", "avx2", "", "smep",
 266                 /*  8 */ "bmi2", "erms", "invpcid", "rtm",
 267                 /* 12 */ "cqm", "", "mpx", "rdt_a",
 268                 /* 16 */ "avx512f", "avx512dq", "rdseed", "adx",
 269                 /* 20 */ "smap", "avx512ifma", "", "clflushopt",
 270                 /* 24 */ "clwb", "intel_pt", "avx512pf", "avx512er",
 271                 /* 28 */ "avx512cd", "sha_ni", "avx512bw", "avx512vl"
 272         };
 273
 274         static char *cpu_stdext_feature2_names[] = {
 275                 /*  0 */ "prefetchwt1", "avx512vbmi", "umip", "pku",
 276                 /*  4 */ "ospke", "waitpkg", "avx512_vbmi2", "",
 277                 /*  8 */ "gfni", "vaes", "vpclmulqdq", "avx512_vnni",
 278                 /* 12 */ "avx512_bitalg", "", "avx512_vpopcntdq", "",
 279                 /* 16 */ "", "", "", "",
 280                 /* 20 */ "", "", "rdpid", "",
 281                 /* 24 */ "", "cldemote", "", "movdiri",
 282                 /* 28 */ "movdir64b", "enqcmd", "sgx_lc", ""
 283         };
 284
 285         static char *cpu_stdext_feature3_names[] = {
 286                 /*  0 */ "", "", "avx512_4vnniw", "avx512_4fmaps",
 287                 /*  4 */ "fsrm", "", "", "",
 288                 /*  8 */ "avx512_vp2intersect", "", "md_clear", "",
 289                 /* 12 */ "", "", "", "",
 290                 /* 16 */ "", "", "pconfig", "",
 291                 /* 20 */ "", "", "", "",
 292                 /* 24 */ "", "", "ibrs", "stibp",
 293                 /* 28 */ "flush_l1d", "arch_capabilities", "core_capabilities", "ssbd"
 294         };
 295
 296         static char *cpu_stdext_feature_l1_names[] = {
 297                 /*  0 */ "xsaveopt", "xsavec", "xgetbv1", "xsaves",
 298                 /*  4 */ "xfd"
 299         };
 300
 301         static char *power_flags[] = {
 302                 "ts",           "fid",          "vid",
 303                 "ttp",          "tm",           "stc",
 304                 "100mhzsteps",  "hwpstate",     "",
 305                 "cpb",          "eff_freq_ro",  "proc_feedback",
 306                 "acc_power",
 307         };
 308
 309         hw_model[0] = CTL_HW;
 310         hw_model[1] = HW_MODEL;
 311         model[0] = '\0';
 312         size = sizeof(model);
 313         if (kernel_sysctl(td, hw_model, 2, &model, &size, 0, 0, 0, 0) != 0)
 314                 strcpy(model, "unknown");
 315 #ifdef __i386__
 316         switch (cpu_vendor_id) {
 317         case CPU_VENDOR_AMD:
 318                 if (cpu_class < CPUCLASS_686)
 319                         cpu_feature_names[16] = "fcmov";
 320                 break;
 321         case CPU_VENDOR_CYRIX:
 322                 cpu_feature_names[24] = "cxmmx";
 323                 break;
 324         }
 325 #endif
 326         if (cpu_exthigh >= 0x80000006)
 327                 do_cpuid(0x80000006, cache_size);
 328         else
 329                 memset(cache_size, 0, sizeof(cache_size));
 330         for (i = 0; i < mp_ncpus; ++i) {
 331                 fqmhz = 0;
 332                 fqkhz = 0;
 333                 freq = atomic_load_acq_64(&tsc_freq);
 334                 if (freq != 0) {
 335                         fqmhz = (freq + 4999) / 1000000;
 336                         fqkhz = ((freq + 4999) / 10000) % 100;
 337                 }
 338                 sbuf_printf(sb,
 339                     "processor\t: %d\n"
 340                     "vendor_id\t: %.20s\n"
 341                     "cpu family\t: %u\n"
 342                     "model\t\t: %u\n"
 343                     "model name\t: %s\n"
 344                     "stepping\t: %u\n"
 345                     "cpu MHz\t\t: %d.%02d\n"
 346                     "cache size\t: %d KB\n"
 347                     "physical id\t: %d\n"
 348                     "siblings\t: %d\n"
 349                     "core id\t\t: %d\n"
 350                     "cpu cores\t: %d\n"
 351                     "apicid\t\t: %d\n"
 352                     "initial apicid\t: %d\n"
 353                     "fpu\t\t: %s\n"
 354                     "fpu_exception\t: %s\n"
 355                     "cpuid level\t: %d\n"
 356                     "wp\t\t: %s\n",
 357                     i, cpu_vendor, CPUID_TO_FAMILY(cpu_id),
 358                     CPUID_TO_MODEL(cpu_id), model, cpu_id & CPUID_STEPPING,
 359                     fqmhz, fqkhz,
 360                     (cache_size[2] >> 16), 0, mp_ncpus, i, mp_ncpus,
 361                     i, i, /*cpu_id & CPUID_LOCAL_APIC_ID ??*/
 362                     (cpu_feature & CPUID_FPU) ? "yes" : "no", "yes",
 363                     CPUID_TO_FAMILY(cpu_id), "yes");
 364                 sbuf_cat(sb, "flags\t\t:");
 365                 for (j = 0; j < nitems(cpu_feature_names); j++)
 366                         if (cpu_feature & (1 << j) &&
 367                             cpu_feature_names[j][0] != '\0')
 368                                 sbuf_printf(sb, " %s", cpu_feature_names[j]);
 369                 for (j = 0; j < nitems(amd_feature_names); j++)
 370                         if (amd_feature & (1 << j) &&
 371                             amd_feature_names[j][0] != '\0')
 372                                 sbuf_printf(sb, " %s", amd_feature_names[j]);
 373                 for (j = 0; j < nitems(cpu_feature2_names); j++)
 374                         if (cpu_feature2 & (1 << j) &&
 375                             cpu_feature2_names[j][0] != '\0')
 376                                 sbuf_printf(sb, " %s", cpu_feature2_names[j]);
 377                 for (j = 0; j < nitems(amd_feature2_names); j++)
 378                         if (amd_feature2 & (1 << j) &&
 379                             amd_feature2_names[j][0] != '\0')
 380                                 sbuf_printf(sb, " %s", amd_feature2_names[j]);
 381                 for (j = 0; j < nitems(cpu_stdext_feature_names); j++)
 382                         if (cpu_stdext_feature & (1 << j) &&
 383                             cpu_stdext_feature_names[j][0] != '\0')
 384                                 sbuf_printf(sb, " %s",
 385                                     cpu_stdext_feature_names[j]);
 386                 if (tsc_is_invariant)
 387                         sbuf_cat(sb, " constant_tsc");
 388                 for (j = 0; j < nitems(cpu_stdext_feature2_names); j++)
 389                         if (cpu_stdext_feature2 & (1 << j) &&
 390                             cpu_stdext_feature2_names[j][0] != '\0')
 391                                 sbuf_printf(sb, " %s",
 392                                     cpu_stdext_feature2_names[j]);
 393                 for (j = 0; j < nitems(cpu_stdext_feature3_names); j++)
 394                         if (cpu_stdext_feature3 & (1 << j) &&
 395                             cpu_stdext_feature3_names[j][0] != '\0')
 396                                 sbuf_printf(sb, " %s",
 397                                     cpu_stdext_feature3_names[j]);
 398                 if ((cpu_feature2 & CPUID2_XSAVE) != 0) {
 399                         cpuid_count(0xd, 0x1, regs);
 400                         for (j = 0; j < nitems(cpu_stdext_feature_l1_names); j++)
 401                                 if (regs[0] & (1 << j) &&
 402                                     cpu_stdext_feature_l1_names[j][0] != '\0')
 403                                         sbuf_printf(sb, " %s",
 404                                             cpu_stdext_feature_l1_names[j]);
 405                 }
 406                 sbuf_cat(sb, "\n");
 407                 sbuf_printf(sb,
 408                     "bugs\t\t: %s\n"
 409                     "bogomips\t: %d.%02d\n"
 410                     "clflush size\t: %d\n"
 411                     "cache_alignment\t: %d\n"
 412                     "address sizes\t: %d bits physical, %d bits virtual\n",
 413 #if defined(I586_CPU) && !defined(NO_F00F_HACK)
 414                     (has_f00f_bug) ? "Intel F00F" : "",
 415 #else
 416                     "",
 417 #endif
 418                     fqmhz * 2, fqkhz,
 419                     cpu_clflush_line_size, cpu_clflush_line_size,
 420                     cpu_maxphyaddr,
 421                     (cpu_maxphyaddr > 32) ? 48 : 0);
 422                 sbuf_cat(sb, "power management: ");
 423                 for (j = 0; j < nitems(power_flags); j++)
 424                         if (amd_pminfo & (1 << j))
 425                                 sbuf_printf(sb, " %s", power_flags[j]);
 426                 sbuf_cat(sb, "\n\n");
 427
 428                 /* XXX per-cpu vendor / class / model / id? */
 429         }
 430         sbuf_cat(sb, "\n");
 431
 432         return (0);
 433 }
 434 #else
 435 /* ARM64TODO: implement non-stubbed linprocfs_docpuinfo */
 436 static int
 437 linprocfs_docpuinfo(PFS_FILL_ARGS)
 438 {
 439         int i;
 440
 441         for (i = 0; i < mp_ncpus; ++i) {
 442                 sbuf_printf(sb,
 443                     "processor\t: %d\n"
 444                     "BogoMIPS\t: %d.%02d\n",
 445                     i, 0, 0);
 446                 sbuf_cat(sb, "Features\t: ");
 447                 sbuf_cat(sb, "\n");
 448                 sbuf_printf(sb,
 449                     "CPU implementer\t: \n"
 450                     "CPU architecture: \n"
 451                     "CPU variant\t: 0x%x\n"
 452                     "CPU part\t: 0x%x\n"
 453                     "CPU revision\t: %d\n",
 454                     0, 0, 0);
 455                 sbuf_cat(sb, "\n");
 456         }
 457
 458         return (0);
 459 }
 460 #endif /* __i386__ || __amd64__ */
 461
 462 static const char *path_slash_sys = "/sys";
 463 static const char *fstype_sysfs = "sysfs";
 464
 465 static int
 466 _mtab_helper(const struct pfs_node *pn, const struct statfs *sp,
 467     const char **mntfrom, const char **mntto, const char **fstype)
 468 {
 469         /* determine device name */
 470         *mntfrom = sp->f_mntfromname;
 471
 472         /* determine mount point */
 473         *mntto = sp->f_mntonname;
 474
 475         /* determine fs type */
 476         *fstype = sp->f_fstypename;
 477         if (strcmp(*fstype, pn->pn_info->pi_name) == 0)
 478                 *mntfrom = *fstype = "proc";
 479         else if (strcmp(*fstype, "procfs") == 0)
 480                 return (ECANCELED);
 481
 482         if (strcmp(*fstype, "autofs") == 0) {
 483                 /*
 484                  * FreeBSD uses eg "map -hosts", whereas Linux
 485                  * expects just "-hosts".
 486                  */
 487                 if (strncmp(*mntfrom, "map ", 4) == 0)
 488                         *mntfrom += 4;
 489         }
 490
 491         if (strcmp(*fstype, "linsysfs") == 0) {
 492                 *mntfrom = path_slash_sys;
 493                 *fstype = fstype_sysfs;
 494         } else {
 495                 /* For Linux msdosfs is called vfat */
 496                 if (strcmp(*fstype, "msdosfs") == 0)
 497                         *fstype = "vfat";
 498         }
 499         return (0);
 500 }
 501
 502 static void
 503 _sbuf_mntoptions_helper(struct sbuf *sb, uint64_t f_flags)
 504 {
 505         sbuf_cat(sb, (f_flags & MNT_RDONLY) ? "ro" : "rw");
 506 #define ADD_OPTION(opt, name) \
 507         if (f_flags & (opt)) sbuf_cat(sb, "," name);
 508         ADD_OPTION(MNT_SYNCHRONOUS,     "sync");
 509         ADD_OPTION(MNT_NOEXEC,          "noexec");
 510         ADD_OPTION(MNT_NOSUID,          "nosuid");
 511         ADD_OPTION(MNT_UNION,           "union");
 512         ADD_OPTION(MNT_ASYNC,           "async");
 513         ADD_OPTION(MNT_SUIDDIR,         "suiddir");
 514         ADD_OPTION(MNT_NOSYMFOLLOW,     "nosymfollow");
 515         ADD_OPTION(MNT_NOATIME,         "noatime");
 516 #undef ADD_OPTION
 517 }
 518
 519 /*
 520  * Filler function for proc/mtab and proc/<pid>/mounts.
 521  *
 522  * /proc/mtab doesn't exist in Linux' procfs, but is included here so
 523  * users can symlink /compat/linux/etc/mtab to /proc/mtab
 524  */
 525 static int
 526 linprocfs_domtab(PFS_FILL_ARGS)
 527 {
 528         struct nameidata nd;
 529         const char *lep, *mntto, *mntfrom, *fstype;
 530         char *dlep, *flep;
 531         size_t lep_len;
 532         int error;
 533         struct statfs *buf, *sp;
 534         size_t count;
 535
 536         /* resolve symlinks etc. in the emulation tree prefix */
 537         /*
 538          * Ideally, this would use the current chroot rather than some
 539          * hardcoded path.
 540          */
 541         NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, linux_emul_path);
 542         flep = NULL;
 543         error = namei(&nd);
 544         lep = linux_emul_path;
 545         if (error == 0) {
 546                 if (vn_fullpath(nd.ni_vp, &dlep, &flep) == 0)
 547                         lep = dlep;
 548                 vrele(nd.ni_vp);
 549         }
 550         lep_len = strlen(lep);
 551
 552         buf = NULL;
 553         error = kern_getfsstat(td, &buf, SIZE_T_MAX, &count,
 554             UIO_SYSSPACE, MNT_WAIT);
 555         if (error != 0) {
 556                 free(buf, M_TEMP);
 557                 free(flep, M_TEMP);
 558                 return (error);
 559         }
 560
 561         for (sp = buf; count > 0; sp++, count--) {
 562                 error = _mtab_helper(pn, sp, &mntfrom, &mntto, &fstype);
 563                 if (error != 0) {
 564                         MPASS(error == ECANCELED);
 565                         continue;
 566                 }
 567
 568                 /* determine mount point */
 569                 if (strncmp(mntto, lep, lep_len) == 0 && mntto[lep_len] == '/')
 570                         mntto += lep_len;
 571
 572                 sbuf_printf(sb, "%s %s %s ", mntfrom, mntto, fstype);
 573                 _sbuf_mntoptions_helper(sb, sp->f_flags);
 574                 /* a real Linux mtab will also show NFS options */
 575                 sbuf_printf(sb, " 0 0\n");
 576         }
 577
 578         free(buf, M_TEMP);
 579         free(flep, M_TEMP);
 580         return (error);
 581 }
 582
 583 static int
 584 linprocfs_doprocmountinfo(PFS_FILL_ARGS)
 585 {
 586         struct nameidata nd;
 587         const char *mntfrom, *mntto, *fstype;
 588         const char *lep;
 589         char *dlep, *flep;
 590         struct statfs *buf, *sp;
 591         size_t count, lep_len;
 592         int error;
 593
 594         /*
 595          * Ideally, this would use the current chroot rather than some
 596          * hardcoded path.
 597          */
 598         NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, linux_emul_path);
 599         flep = NULL;
 600         error = namei(&nd);
 601         lep = linux_emul_path;
 602         if (error == 0) {
 603                 if (vn_fullpath(nd.ni_vp, &dlep, &flep) == 0)
 604                         lep = dlep;
 605                 vrele(nd.ni_vp);
 606         }
 607         lep_len = strlen(lep);
 608
 609         buf = NULL;
 610         error = kern_getfsstat(td, &buf, SIZE_T_MAX, &count,
 611             UIO_SYSSPACE, MNT_WAIT);
 612         if (error != 0)
 613                 goto out;
 614
 615         for (sp = buf; count > 0; sp++, count--) {
 616                 error = _mtab_helper(pn, sp, &mntfrom, &mntto, &fstype);
 617                 if (error != 0) {
 618                         MPASS(error == ECANCELED);
 619                         continue;
 620                 }
 621
 622                 if (strncmp(mntto, lep, lep_len) == 0 && mntto[lep_len] == '/')
 623                         mntto += lep_len;
 624 #if 0
 625                 /*
 626                  * If the prefix is a chroot, and this mountpoint is not under
 627                  * the prefix, we should skip it.  Leave it for now for
 628                  * consistency with procmtab above.
 629                  */
 630                 else
 631                         continue;
 632 #endif
 633
 634                 /*
 635                  * (1) mount id
 636                  *
 637                  * (2) parent mount id -- we don't have this cheaply, so
 638                  * provide a dummy value
 639                  *
 640                  * (3) major:minor -- ditto
 641                  *
 642                  * (4) root filesystem mount -- probably a namespaces thing
 643                  *
 644                  * (5) mountto path
 645                  */
 646                 sbuf_printf(sb, "%u 0 0:0 / %s ",
 647                     sp->f_fsid.val[0] ^ sp->f_fsid.val[1], mntto);
 648                 /* (6) mount options */
 649                 _sbuf_mntoptions_helper(sb, sp->f_flags);
 650                 /*
 651                  * (7) zero or more optional fields -- again, namespace related
 652                  *
 653                  * (8) End of variable length fields separator ("-")
 654                  *
 655                  * (9) fstype
 656                  *
 657                  * (10) mount from
 658                  *
 659                  * (11) "superblock" options -- like (6), but different
 660                  * semantics in Linux
 661                  */
 662                 sbuf_printf(sb, " - %s %s %s\n", fstype, mntfrom,
 663                     (sp->f_flags & MNT_RDONLY) ? "ro" : "rw");
 664         }
 665
 666         error = 0;
 667 out:
 668         free(buf, M_TEMP);
 669         free(flep, M_TEMP);
 670         return (error);
 671 }
 672
 673 /*
 674  * Filler function for proc/partitions
 675  */
 676 static int
 677 linprocfs_dopartitions(PFS_FILL_ARGS)
 678 {
 679         struct g_class *cp;
 680         struct g_geom *gp;
 681         struct g_provider *pp;
 682         int major, minor;
 683
 684         g_topology_lock();
 685         sbuf_printf(sb, "major minor  #blocks  name rio rmerge rsect "
 686             "ruse wio wmerge wsect wuse running use aveq\n");
 687
 688         LIST_FOREACH(cp, &g_classes, class) {
 689                 if (strcmp(cp->name, "DISK") == 0 ||
 690                     strcmp(cp->name, "PART") == 0)
 691                         LIST_FOREACH(gp, &cp->geom, geom) {
 692                                 LIST_FOREACH(pp, &gp->provider, provider) {
 693                                         if (linux_driver_get_major_minor(
 694                                             pp->name, &major, &minor) != 0) {
 695                                                 major = 0;
 696                                                 minor = 0;
 697                                         }
 698                                         sbuf_printf(sb, "%d %d %lld %s "
 699                                             "%d %d %d %d %d "
 700                                              "%d %d %d %d %d %d\n",
 701                                              major, minor,
 702                                              (long long)pp->mediasize, pp->name,
 703                                              0, 0, 0, 0, 0,
 704                                              0, 0, 0, 0, 0, 0);
 705                                 }
 706                         }
 707         }
 708         g_topology_unlock();
 709
 710         return (0);
 711 }
 712
 713 /*
 714  * Filler function for proc/stat
 715  *
 716  * Output depends on kernel version:
 717  *
 718  * v2.5.40 <=
 719  *   user nice system idle
 720  * v2.5.41
 721  *   user nice system idle iowait
 722  * v2.6.11
 723  *   user nice system idle iowait irq softirq steal
 724  * v2.6.24
 725  *   user nice system idle iowait irq softirq steal guest
 726  * v2.6.33 >=
 727  *   user nice system idle iowait irq softirq steal guest guest_nice
 728  */
 729 static int
 730 linprocfs_dostat(PFS_FILL_ARGS)
 731 {
 732         struct pcpu *pcpu;
 733         long cp_time[CPUSTATES];
 734         long *cp;
 735         struct timeval boottime;
 736         int i;
 737         char *zero_pad;
 738         bool has_intr = true;
 739
 740         if (linux_kernver(td) >= LINUX_KERNVER(2,6,33)) {
 741                 zero_pad = " 0 0 0 0\n";
 742         } else if (linux_kernver(td) >= LINUX_KERNVER(2,6,24)) {
 743                 zero_pad = " 0 0 0\n";
 744         } else if (linux_kernver(td) >= LINUX_KERNVER(2,6,11)) {
 745                 zero_pad = " 0 0\n";
 746         } else if (linux_kernver(td) >= LINUX_KERNVER(2,5,41)) {
 747                 has_intr = false;
 748                 zero_pad = " 0\n";
 749         } else {
 750                 has_intr = false;
 751                 zero_pad = "\n";
 752         }
 753
 754         read_cpu_time(cp_time);
 755         getboottime(&boottime);
 756         /* Parameters common to all versions */
 757         sbuf_printf(sb, "cpu %lu %lu %lu %lu",
 758             T2J(cp_time[CP_USER]),
 759             T2J(cp_time[CP_NICE]),
 760             T2J(cp_time[CP_SYS]),
 761             T2J(cp_time[CP_IDLE]));
 762
 763         /* Print interrupt stats if available */
 764         if (has_intr) {
 765                 sbuf_printf(sb, " 0 %lu", T2J(cp_time[CP_INTR]));
 766         }
 767
 768         /* Pad out remaining fields depending on version */
 769         sbuf_printf(sb, "%s", zero_pad);
 770
 771         CPU_FOREACH(i) {
 772                 pcpu = pcpu_find(i);
 773                 cp = pcpu->pc_cp_time;
 774                 sbuf_printf(sb, "cpu%d %lu %lu %lu %lu", i,
 775                     T2J(cp[CP_USER]),
 776                     T2J(cp[CP_NICE]),
 777                     T2J(cp[CP_SYS]),
 778                     T2J(cp[CP_IDLE]));
 779
 780                 if (has_intr) {
 781                         sbuf_printf(sb, " 0 %lu", T2J(cp[CP_INTR]));
 782                 }
 783
 784                 sbuf_printf(sb, "%s", zero_pad);
 785         }
 786         sbuf_printf(sb,
 787             "disk 0 0 0 0\n"
 788             "page %ju %ju\n"
 789             "swap %ju %ju\n"
 790             "intr %ju\n"
 791             "ctxt %ju\n"
 792             "btime %lld\n",
 793             (uintmax_t)VM_CNT_FETCH(v_vnodepgsin),
 794             (uintmax_t)VM_CNT_FETCH(v_vnodepgsout),
 795             (uintmax_t)VM_CNT_FETCH(v_swappgsin),
 796             (uintmax_t)VM_CNT_FETCH(v_swappgsout),
 797             (uintmax_t)VM_CNT_FETCH(v_intr),
 798             (uintmax_t)VM_CNT_FETCH(v_swtch),
 799             (long long)boottime.tv_sec);
 800         return (0);
 801 }
 802
 803 static int
 804 linprocfs_doswaps(PFS_FILL_ARGS)
 805 {
 806         struct xswdev xsw;
 807         uintmax_t total, used;
 808         int n;
 809         char devname[SPECNAMELEN + 1];
 810
 811         sbuf_printf(sb, "Filename\t\t\t\tType\t\tSize\tUsed\tPriority\n");
 812         for (n = 0; ; n++) {
 813                 if (swap_dev_info(n, &xsw, devname, sizeof(devname)) != 0)
 814                         break;
 815                 total = (uintmax_t)xsw.xsw_nblks * PAGE_SIZE / 1024;
 816                 used  = (uintmax_t)xsw.xsw_used * PAGE_SIZE / 1024;
 817
 818                 /*
 819                  * The space and not tab after the device name is on
 820                  * purpose.  Linux does so.
 821                  */
 822                 sbuf_printf(sb, "/dev/%-34s unknown\t\t%jd\t%jd\t-1\n",
 823                     devname, total, used);
 824         }
 825         return (0);
 826 }
 827
 828 /*
 829  * Filler function for proc/uptime
 830  */
 831 static int
 832 linprocfs_douptime(PFS_FILL_ARGS)
 833 {
 834         long cp_time[CPUSTATES];
 835         struct timeval tv;
 836
 837         getmicrouptime(&tv);
 838         read_cpu_time(cp_time);
 839         sbuf_printf(sb, "%lld.%02ld %ld.%02lu\n",
 840             (long long)tv.tv_sec, tv.tv_usec / 10000,
 841             T2S(cp_time[CP_IDLE] / mp_ncpus),
 842             T2CS(cp_time[CP_IDLE] / mp_ncpus) % 100);
 843         return (0);
 844 }
 845
 846 /*
 847  * Get OS build date
 848  */
 849 static void
 850 linprocfs_osbuild(struct thread *td, struct sbuf *sb)
 851 {
 852 #if 0
 853         char osbuild[256];
 854         char *cp1, *cp2;
 855
 856         strncpy(osbuild, version, 256);
 857         osbuild[255] = '\0';
 858         cp1 = strstr(osbuild, "\n");
 859         cp2 = strstr(osbuild, ":");
 860         if (cp1 && cp2) {
 861                 *cp1 = *cp2 = '\0';
 862                 cp1 = strstr(osbuild, "#");
 863         } else
 864                 cp1 = NULL;
 865         if (cp1)
 866                 sbuf_printf(sb, "%s%s", cp1, cp2 + 1);
 867         else
 868 #endif
 869                 sbuf_cat(sb, "#4 Sun Dec 18 04:30:00 CET 1977");
 870 }
 871
 872 /*
 873  * Get OS builder
 874  */
 875 static void
 876 linprocfs_osbuilder(struct thread *td, struct sbuf *sb)
 877 {
 878 #if 0
 879         char builder[256];
 880         char *cp;
 881
 882         cp = strstr(version, "\n    ");
 883         if (cp) {
 884                 strncpy(builder, cp + 5, 256);
 885                 builder[255] = '\0';
 886                 cp = strstr(builder, ":");
 887                 if (cp)
 888                         *cp = '\0';
 889         }
 890         if (cp)
 891                 sbuf_cat(sb, builder);
 892         else
 893 #endif
 894                 sbuf_cat(sb, "des@freebsd.org");
 895 }
 896
 897 /*
 898  * Filler function for proc/version
 899  */
 900 static int
 901 linprocfs_doversion(PFS_FILL_ARGS)
 902 {
 903         char osname[LINUX_MAX_UTSNAME];
 904         char osrelease[LINUX_MAX_UTSNAME];
 905
 906         linux_get_osname(td, osname);
 907         linux_get_osrelease(td, osrelease);
 908         sbuf_printf(sb, "%s version %s (", osname, osrelease);
 909         linprocfs_osbuilder(td, sb);
 910         sbuf_cat(sb, ") (gcc version " __VERSION__ ") ");
 911         linprocfs_osbuild(td, sb);
 912         sbuf_cat(sb, "\n");
 913
 914         return (0);
 915 }
 916
 917 /*
 918  * Filler function for proc/loadavg
 919  */
 920 static int
 921 linprocfs_doloadavg(PFS_FILL_ARGS)
 922 {
 923
 924         sbuf_printf(sb,
 925             "%d.%02d %d.%02d %d.%02d %d/%d %d\n",
 926             (int)(averunnable.ldavg[0] / averunnable.fscale),
 927             (int)(averunnable.ldavg[0] * 100 / averunnable.fscale % 100),
 928             (int)(averunnable.ldavg[1] / averunnable.fscale),
 929             (int)(averunnable.ldavg[1] * 100 / averunnable.fscale % 100),
 930             (int)(averunnable.ldavg[2] / averunnable.fscale),
 931             (int)(averunnable.ldavg[2] * 100 / averunnable.fscale % 100),
 932             1,                          /* number of running tasks */
 933             nprocs,                     /* number of tasks */
 934             lastpid                     /* the last pid */
 935         );
 936         return (0);
 937 }
 938
 939 static int
 940 linprocfs_get_tty_nr(struct proc *p)
 941 {
 942         struct session *sp;
 943         const char *ttyname;
 944         int error, major, minor, nr;
 945
 946         PROC_LOCK_ASSERT(p, MA_OWNED);
 947         sx_assert(&proctree_lock, SX_LOCKED);
 948
 949         if ((p->p_flag & P_CONTROLT) == 0)
 950                 return (-1);
 951
 952         sp = p->p_pgrp->pg_session;
 953         if (sp == NULL)
 954                 return (-1);
 955
 956         ttyname = devtoname(sp->s_ttyp->t_dev);
 957         error = linux_driver_get_major_minor(ttyname, &major, &minor);
 958         if (error != 0)
 959                 return (-1);
 960
 961         nr = makedev(major, minor);
 962         return (nr);
 963 }
 964
 965 /*
 966  * Filler function for proc/pid/stat
 967  */
 968 static int
 969 linprocfs_doprocstat(PFS_FILL_ARGS)
 970 {
 971         struct kinfo_proc kp;
 972         struct timeval boottime;
 973         char state;
 974         static int ratelimit = 0;
 975         int tty_nr;
 976         vm_offset_t startcode, startdata;
 977
 978         getboottime(&boottime);
 979         sx_slock(&proctree_lock);
 980         PROC_LOCK(p);
 981         fill_kinfo_proc(p, &kp);
 982         tty_nr = linprocfs_get_tty_nr(p);
 983         sx_sunlock(&proctree_lock);
 984         if (p->p_vmspace) {
 985            startcode = (vm_offset_t)p->p_vmspace->vm_taddr;
 986            startdata = (vm_offset_t)p->p_vmspace->vm_daddr;
 987         } else {
 988            startcode = 0;
 989            startdata = 0;
 990         }
 991         sbuf_printf(sb, "%d", p->p_pid);
 992 #define PS_ADD(name, fmt, arg) sbuf_printf(sb, " " fmt, arg)
 993         PS_ADD("comm",          "(%s)", p->p_comm);
 994         if (kp.ki_stat > sizeof(linux_state)) {
 995                 state = 'R';
 996
 997                 if (ratelimit == 0) {
 998                         printf("linprocfs: don't know how to handle unknown FreeBSD state %d/%zd, mapping to R\n",
 999                             kp.ki_stat, sizeof(linux_state));
1000                         ++ratelimit;
1001                 }
1002         } else
1003                 state = linux_state[kp.ki_stat - 1];
1004         PS_ADD("state",         "%c",   state);
1005         PS_ADD("ppid",          "%d",   p->p_pptr ? p->p_pptr->p_pid : 0);
1006         PS_ADD("pgrp",          "%d",   p->p_pgid);
1007         PS_ADD("session",       "%d",   p->p_session->s_sid);
1008         PROC_UNLOCK(p);
1009         PS_ADD("tty",           "%d",   tty_nr);
1010         PS_ADD("tpgid",         "%d",   kp.ki_tpgid);
1011         PS_ADD("flags",         "%u",   0); /* XXX */
1012         PS_ADD("minflt",        "%lu",  kp.ki_rusage.ru_minflt);
1013         PS_ADD("cminflt",       "%lu",  kp.ki_rusage_ch.ru_minflt);
1014         PS_ADD("majflt",        "%lu",  kp.ki_rusage.ru_majflt);
1015         PS_ADD("cmajflt",       "%lu",  kp.ki_rusage_ch.ru_majflt);
1016         PS_ADD("utime",         "%ld",  TV2J(&kp.ki_rusage.ru_utime));
1017         PS_ADD("stime",         "%ld",  TV2J(&kp.ki_rusage.ru_stime));
1018         PS_ADD("cutime",        "%ld",  TV2J(&kp.ki_rusage_ch.ru_utime));
1019         PS_ADD("cstime",        "%ld",  TV2J(&kp.ki_rusage_ch.ru_stime));
1020         PS_ADD("priority",      "%d",   kp.ki_pri.pri_user);
1021         PS_ADD("nice",          "%d",   kp.ki_nice); /* 19 (nicest) to -19 */
1022         PS_ADD("0",             "%d",   0); /* removed field */
1023         PS_ADD("itrealvalue",   "%d",   0); /* XXX */
1024         PS_ADD("starttime",     "%lu",  TV2J(&kp.ki_start) - TV2J(&boottime));
1025         PS_ADD("vsize",         "%ju",  P2K((uintmax_t)kp.ki_size));
1026         PS_ADD("rss",           "%ju",  (uintmax_t)kp.ki_rssize);
1027         PS_ADD("rlim",          "%lu",  kp.ki_rusage.ru_maxrss);
1028         PS_ADD("startcode",     "%ju",  (uintmax_t)startcode);
1029         PS_ADD("endcode",       "%ju",  (uintmax_t)startdata);
1030         PS_ADD("startstack",    "%u",   0); /* XXX */
1031         PS_ADD("kstkesp",       "%u",   0); /* XXX */
1032         PS_ADD("kstkeip",       "%u",   0); /* XXX */
1033         PS_ADD("signal",        "%u",   0); /* XXX */
1034         PS_ADD("blocked",       "%u",   0); /* XXX */
1035         PS_ADD("sigignore",     "%u",   0); /* XXX */
1036         PS_ADD("sigcatch",      "%u",   0); /* XXX */
1037         PS_ADD("wchan",         "%u",   0); /* XXX */
1038         PS_ADD("nswap",         "%lu",  kp.ki_rusage.ru_nswap);
1039         PS_ADD("cnswap",        "%lu",  kp.ki_rusage_ch.ru_nswap);
1040         PS_ADD("exitsignal",    "%d",   0); /* XXX */
1041         PS_ADD("processor",     "%u",   kp.ki_lastcpu);
1042         PS_ADD("rt_priority",   "%u",   0); /* XXX */ /* >= 2.5.19 */
1043         PS_ADD("policy",        "%u",   kp.ki_pri.pri_class); /* >= 2.5.19 */
1044 #undef PS_ADD
1045         sbuf_putc(sb, '\n');
1046
1047         return (0);
1048 }
1049
1050 /*
1051  * Filler function for proc/pid/statm
1052  */
1053 static int
1054 linprocfs_doprocstatm(PFS_FILL_ARGS)
1055 {
1056         struct kinfo_proc kp;
1057         segsz_t lsize;
1058
1059         sx_slock(&proctree_lock);
1060         PROC_LOCK(p);
1061         fill_kinfo_proc(p, &kp);
1062         PROC_UNLOCK(p);
1063         sx_sunlock(&proctree_lock);
1064
1065         /*
1066          * See comments in linprocfs_doprocstatus() regarding the
1067          * computation of lsize.
1068          */
1069         /* size resident share trs drs lrs dt */
1070         sbuf_printf(sb, "%ju ", B2P((uintmax_t)kp.ki_size));
1071         sbuf_printf(sb, "%ju ", (uintmax_t)kp.ki_rssize);
1072         sbuf_printf(sb, "%ju ", (uintmax_t)0); /* XXX */
1073         sbuf_printf(sb, "%ju ", (uintmax_t)kp.ki_tsize);
1074         sbuf_printf(sb, "%ju ", (uintmax_t)(kp.ki_dsize + kp.ki_ssize));
1075         lsize = B2P(kp.ki_size) - kp.ki_dsize -
1076             kp.ki_ssize - kp.ki_tsize - 1;
1077         sbuf_printf(sb, "%ju ", (uintmax_t)lsize);
1078         sbuf_printf(sb, "%ju\n", (uintmax_t)0); /* XXX */
1079
1080         return (0);
1081 }
1082
1083 /*
1084  * Filler function for proc/pid/status
1085  */
1086 static int
1087 linprocfs_doprocstatus(PFS_FILL_ARGS)
1088 {
1089         struct kinfo_proc kp;
1090         char *state;
1091         segsz_t lsize;
1092         struct thread *td2;
1093         struct sigacts *ps;
1094         l_sigset_t siglist, sigignore, sigcatch;
1095         int i;
1096
1097         sx_slock(&proctree_lock);
1098         PROC_LOCK(p);
1099         td2 = FIRST_THREAD_IN_PROC(p);
1100
1101         if (P_SHOULDSTOP(p)) {
1102                 state = "T (stopped)";
1103         } else {
1104                 switch(p->p_state) {
1105                 case PRS_NEW:
1106                         state = "I (idle)";
1107                         break;
1108                 case PRS_NORMAL:
1109                         if (p->p_flag & P_WEXIT) {
1110                                 state = "X (exiting)";
1111                                 break;
1112                         }
1113                         switch(TD_GET_STATE(td2)) {
1114                         case TDS_INHIBITED:
1115                                 state = "S (sleeping)";
1116                                 break;
1117                         case TDS_RUNQ:
1118                         case TDS_RUNNING:
1119                                 state = "R (running)";
1120                                 break;
1121                         default:
1122                                 state = "? (unknown)";
1123                                 break;
1124                         }
1125                         break;
1126                 case PRS_ZOMBIE:
1127                         state = "Z (zombie)";
1128                         break;
1129                 default:
1130                         state = "? (unknown)";
1131                         break;
1132                 }
1133         }
1134
1135         fill_kinfo_proc(p, &kp);
1136         sx_sunlock(&proctree_lock);
1137
1138         sbuf_printf(sb, "Name:\t%s\n",          p->p_comm); /* XXX escape */
1139         sbuf_printf(sb, "State:\t%s\n",         state);
1140
1141         /*
1142          * Credentials
1143          */
1144         sbuf_printf(sb, "Tgid:\t%d\n",          p->p_pid);
1145         sbuf_printf(sb, "Pid:\t%d\n",           p->p_pid);
1146         sbuf_printf(sb, "PPid:\t%d\n",          kp.ki_ppid );
1147         sbuf_printf(sb, "TracerPid:\t%d\n",     kp.ki_tracer );
1148         sbuf_printf(sb, "Uid:\t%d\t%d\t%d\t%d\n", p->p_ucred->cr_ruid,
1149                                                 p->p_ucred->cr_uid,
1150                                                 p->p_ucred->cr_svuid,
1151                                                 /* FreeBSD doesn't have fsuid */
1152                                                 p->p_ucred->cr_uid);
1153         sbuf_printf(sb, "Gid:\t%d\t%d\t%d\t%d\n", p->p_ucred->cr_rgid,
1154                                                 p->p_ucred->cr_gid,
1155                                                 p->p_ucred->cr_svgid,
1156                                                 /* FreeBSD doesn't have fsgid */
1157                                                 p->p_ucred->cr_gid);
1158         sbuf_cat(sb, "Groups:\t");
1159         for (i = 0; i < p->p_ucred->cr_ngroups; i++)
1160                 sbuf_printf(sb, "%d ",          p->p_ucred->cr_groups[i]);
1161         PROC_UNLOCK(p);
1162         sbuf_putc(sb, '\n');
1163
1164         /*
1165          * Memory
1166          *
1167          * While our approximation of VmLib may not be accurate (I
1168          * don't know of a simple way to verify it, and I'm not sure
1169          * it has much meaning anyway), I believe it's good enough.
1170          *
1171          * The same code that could (I think) accurately compute VmLib
1172          * could also compute VmLck, but I don't really care enough to
1173          * implement it. Submissions are welcome.
1174          */
1175         sbuf_printf(sb, "VmSize:\t%8ju kB\n",   B2K((uintmax_t)kp.ki_size));
1176         sbuf_printf(sb, "VmLck:\t%8u kB\n",     P2K(0)); /* XXX */
1177         sbuf_printf(sb, "VmRSS:\t%8ju kB\n",    P2K((uintmax_t)kp.ki_rssize));
1178         sbuf_printf(sb, "VmData:\t%8ju kB\n",   P2K((uintmax_t)kp.ki_dsize));
1179         sbuf_printf(sb, "VmStk:\t%8ju kB\n",    P2K((uintmax_t)kp.ki_ssize));
1180         sbuf_printf(sb, "VmExe:\t%8ju kB\n",    P2K((uintmax_t)kp.ki_tsize));
1181         lsize = B2P(kp.ki_size) - kp.ki_dsize -
1182             kp.ki_ssize - kp.ki_tsize - 1;
1183         sbuf_printf(sb, "VmLib:\t%8ju kB\n",    P2K((uintmax_t)lsize));
1184
1185         /*
1186          * Signal masks
1187          */
1188         PROC_LOCK(p);
1189         bsd_to_linux_sigset(&p->p_siglist, &siglist);
1190         ps = p->p_sigacts;
1191         mtx_lock(&ps->ps_mtx);
1192         bsd_to_linux_sigset(&ps->ps_sigignore, &sigignore);
1193         bsd_to_linux_sigset(&ps->ps_sigcatch, &sigcatch);
1194         mtx_unlock(&ps->ps_mtx);
1195         PROC_UNLOCK(p);
1196
1197         sbuf_printf(sb, "SigPnd:\t%016jx\n",    siglist.__mask);
1198         /*
1199          * XXX. SigBlk - target thread's signal mask, td_sigmask.
1200          * To implement SigBlk pseudofs should support proc/tid dir entries.
1201          */
1202         sbuf_printf(sb, "SigBlk:\t%016x\n",     0);
1203         sbuf_printf(sb, "SigIgn:\t%016jx\n",    sigignore.__mask);
1204         sbuf_printf(sb, "SigCgt:\t%016jx\n",    sigcatch.__mask);
1205
1206         /*
1207          * Linux also prints the capability masks, but we don't have
1208          * capabilities yet, and when we do get them they're likely to
1209          * be meaningless to Linux programs, so we lie. XXX
1210          */
1211         sbuf_printf(sb, "CapInh:\t%016x\n",     0);
1212         sbuf_printf(sb, "CapPrm:\t%016x\n",     0);
1213         sbuf_printf(sb, "CapEff:\t%016x\n",     0);
1214
1215         return (0);
1216 }
1217
1218 /*
1219  * Filler function for proc/pid/cwd
1220  */
1221 static int
1222 linprocfs_doproccwd(PFS_FILL_ARGS)
1223 {
1224         struct pwd *pwd;
1225         char *fullpath = "unknown";
1226         char *freepath = NULL;
1227
1228         pwd = pwd_hold_proc(p);
1229         vn_fullpath(pwd->pwd_cdir, &fullpath, &freepath);
1230         sbuf_printf(sb, "%s", fullpath);
1231         if (freepath)
1232                 free(freepath, M_TEMP);
1233         pwd_drop(pwd);
1234         return (0);
1235 }
1236
1237 /*
1238  * Filler function for proc/pid/root
1239  */
1240 static int
1241 linprocfs_doprocroot(PFS_FILL_ARGS)
1242 {
1243         struct pwd *pwd;
1244         struct vnode *vp;
1245         char *fullpath = "unknown";
1246         char *freepath = NULL;
1247
1248         pwd = pwd_hold_proc(p);
1249         vp = jailed(p->p_ucred) ? pwd->pwd_jdir : pwd->pwd_rdir;
1250         vn_fullpath(vp, &fullpath, &freepath);
1251         sbuf_printf(sb, "%s", fullpath);
1252         if (freepath)
1253                 free(freepath, M_TEMP);
1254         pwd_drop(pwd);
1255         return (0);
1256 }
1257
1258 /*
1259  * Filler function for proc/pid/cmdline
1260  */
1261 static int
1262 linprocfs_doproccmdline(PFS_FILL_ARGS)
1263 {
1264         int ret;
1265
1266         PROC_LOCK(p);
1267         if ((ret = p_cansee(td, p)) != 0) {
1268                 PROC_UNLOCK(p);
1269                 return (ret);
1270         }
1271
1272         /*
1273          * Mimic linux behavior and pass only processes with usermode
1274          * address space as valid.  Return zero silently otherwize.
1275          */
1276         if (p->p_vmspace == &vmspace0) {
1277                 PROC_UNLOCK(p);
1278                 return (0);
1279         }
1280         if (p->p_args != NULL) {
1281                 sbuf_bcpy(sb, p->p_args->ar_args, p->p_args->ar_length);
1282                 PROC_UNLOCK(p);
1283                 return (0);
1284         }
1285
1286         if ((p->p_flag & P_SYSTEM) != 0) {
1287                 PROC_UNLOCK(p);
1288                 return (0);
1289         }
1290
1291         PROC_UNLOCK(p);
1292
1293         ret = proc_getargv(td, p, sb);
1294         return (ret);
1295 }
1296
1297 /*
1298  * Filler function for proc/pid/environ
1299  */
1300 static int
1301 linprocfs_doprocenviron(PFS_FILL_ARGS)
1302 {
1303
1304         /*
1305          * Mimic linux behavior and pass only processes with usermode
1306          * address space as valid.  Return zero silently otherwize.
1307          */
1308         if (p->p_vmspace == &vmspace0)
1309                 return (0);
1310
1311         return (proc_getenvv(td, p, sb));
1312 }
1313
1314 static char l32_map_str[] = "%08lx-%08lx %s%s%s%s %08lx %02x:%02x %lu%s%s\n";
1315 static char l64_map_str[] = "%016lx-%016lx %s%s%s%s %08lx %02x:%02x %lu%s%s\n";
1316 static char vdso_str[] = "      [vdso]";
1317 static char stack_str[] = "      [stack]";
1318
1319 /*
1320  * Filler function for proc/pid/maps
1321  */
1322 static int
1323 linprocfs_doprocmaps(PFS_FILL_ARGS)
1324 {
1325         struct vmspace *vm;
1326         vm_map_t map;
1327         vm_map_entry_t entry, tmp_entry;
1328         vm_object_t obj, tobj, lobj;
1329         vm_offset_t e_start, e_end;
1330         vm_ooffset_t off;
1331         vm_prot_t e_prot;
1332         unsigned int last_timestamp;
1333         char *name = "", *freename = NULL;
1334         const char *l_map_str;
1335         ino_t ino;
1336         int error;
1337         struct vnode *vp;
1338         struct vattr vat;
1339         bool private;
1340
1341         PROC_LOCK(p);
1342         error = p_candebug(td, p);
1343         PROC_UNLOCK(p);
1344         if (error)
1345                 return (error);
1346
1347         if (uio->uio_rw != UIO_READ)
1348                 return (EOPNOTSUPP);
1349
1350         error = 0;
1351         vm = vmspace_acquire_ref(p);
1352         if (vm == NULL)
1353                 return (ESRCH);
1354
1355         if (SV_CURPROC_FLAG(SV_LP64))
1356                 l_map_str = l64_map_str;
1357         else
1358                 l_map_str = l32_map_str;
1359         map = &vm->vm_map;
1360         vm_map_lock_read(map);
1361         VM_MAP_ENTRY_FOREACH(entry, map) {
1362                 name = "";
1363                 freename = NULL;
1364                 /*
1365                  * Skip printing of the guard page of the stack region, as
1366                  * it confuses glibc pthread_getattr_np() method, where both
1367                  * the base address and size of the stack of the initial thread
1368                  * are calculated.
1369                  */
1370                 if ((entry->eflags & (MAP_ENTRY_IS_SUB_MAP | MAP_ENTRY_GUARD)) != 0)
1371                         continue;
1372                 e_prot = entry->protection;
1373                 e_start = entry->start;
1374                 e_end = entry->end;
1375                 obj = entry->object.vm_object;
1376                 off = entry->offset;
1377                 for (lobj = tobj = obj; tobj != NULL;
1378                     lobj = tobj, tobj = tobj->backing_object) {
1379                         VM_OBJECT_RLOCK(tobj);
1380                         off += lobj->backing_object_offset;
1381                         if (lobj != obj)
1382                                 VM_OBJECT_RUNLOCK(lobj);
1383                 }
1384                 private = (entry->eflags & MAP_ENTRY_COW) != 0 || obj == NULL ||
1385                     (obj->flags & OBJ_ANON) != 0;
1386                 last_timestamp = map->timestamp;
1387                 vm_map_unlock_read(map);
1388                 ino = 0;
1389                 if (lobj) {
1390                         vp = vm_object_vnode(lobj);
1391                         if (vp != NULL)
1392                                 vref(vp);
1393                         if (lobj != obj)
1394                                 VM_OBJECT_RUNLOCK(lobj);
1395                         VM_OBJECT_RUNLOCK(obj);
1396                         if (vp != NULL) {
1397                                 vn_fullpath(vp, &name, &freename);
1398                                 vn_lock(vp, LK_SHARED | LK_RETRY);
1399                                 VOP_GETATTR(vp, &vat, td->td_ucred);
1400                                 ino = vat.va_fileid;
1401                                 vput(vp);
1402                         } else if (SV_PROC_ABI(p) == SV_ABI_LINUX) {
1403                                 /*
1404                                  * sv_shared_page_base pointed out to the
1405                                  * FreeBSD sharedpage, PAGE_SIZE is a size
1406                                  * of it. The vDSO page is above.
1407                                  */
1408                                 if (e_start == p->p_sysent->sv_shared_page_base +
1409                                     PAGE_SIZE)
1410                                         name = vdso_str;
1411                                 if (e_end == p->p_sysent->sv_usrstack)
1412                                         name = stack_str;
1413                         }
1414                 }
1415
1416                 /*
1417                  * format:
1418                  *  start, end, access, offset, major, minor, inode, name.
1419                  */
1420                 error = sbuf_printf(sb, l_map_str,
1421                     (u_long)e_start, (u_long)e_end,
1422                     (e_prot & VM_PROT_READ)?"r":"-",
1423                     (e_prot & VM_PROT_WRITE)?"w":"-",
1424                     (e_prot & VM_PROT_EXECUTE)?"x":"-",
1425                     private ? "p" : "s",
1426                     (u_long)off,
1427                     0,
1428                     0,
1429                     (u_long)ino,
1430                     *name ? "     " : " ",
1431                     name
1432                     );
1433                 if (freename)
1434                         free(freename, M_TEMP);
1435                 vm_map_lock_read(map);
1436                 if (error == -1) {
1437                         error = 0;
1438                         break;
1439                 }
1440                 if (last_timestamp != map->timestamp) {
1441                         /*
1442                          * Look again for the entry because the map was
1443                          * modified while it was unlocked.  Specifically,
1444                          * the entry may have been clipped, merged, or deleted.
1445                          */
1446                         vm_map_lookup_entry(map, e_end - 1, &tmp_entry);
1447                         entry = tmp_entry;
1448                 }
1449         }
1450         vm_map_unlock_read(map);
1451         vmspace_free(vm);
1452
1453         return (error);
1454 }
1455
1456 /*
1457  * Filler function for proc/pid/mem
1458  */
1459 static int
1460 linprocfs_doprocmem(PFS_FILL_ARGS)
1461 {
1462         ssize_t resid;
1463         int error;
1464
1465         resid = uio->uio_resid;
1466         error = procfs_doprocmem(PFS_FILL_ARGNAMES);
1467
1468         if (uio->uio_rw == UIO_READ && resid != uio->uio_resid)
1469                 return (0);
1470
1471         if (error == EFAULT)
1472                 error = EIO;
1473
1474         return (error);
1475 }
1476
1477 static int
1478 linux_ifname(struct ifnet *ifp, char *buffer, size_t buflen)
1479 {
1480         struct ifnet *ifscan;
1481         int ethno;
1482
1483         IFNET_RLOCK_ASSERT();
1484
1485         /* Short-circuit non ethernet interfaces */
1486         if (linux_use_real_ifname(ifp))
1487                 return (strlcpy(buffer, ifp->if_xname, buflen));
1488
1489         /* Determine the (relative) unit number for ethernet interfaces */
1490         ethno = 0;
1491         CK_STAILQ_FOREACH(ifscan, &V_ifnet, if_link) {
1492                 if (ifscan == ifp)
1493                         return (snprintf(buffer, buflen, "eth%d", ethno));
1494                 if (!linux_use_real_ifname(ifscan))
1495                         ethno++;
1496         }
1497
1498         return (0);
1499 }
1500
1501 /*
1502  * Filler function for proc/net/dev
1503  */
1504 static int
1505 linprocfs_donetdev(PFS_FILL_ARGS)
1506 {
1507         char ifname[16]; /* XXX LINUX_IFNAMSIZ */
1508         struct ifnet *ifp;
1509
1510         sbuf_printf(sb, "%6s|%58s|%s\n"
1511             "%6s|%58s|%58s\n",
1512             "Inter-", "   Receive", "  Transmit",
1513             " face",
1514             "bytes    packets errs drop fifo frame compressed multicast",
1515             "bytes    packets errs drop fifo colls carrier compressed");
1516
1517         CURVNET_SET(TD_TO_VNET(curthread));
1518         IFNET_RLOCK();
1519         CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
1520                 linux_ifname(ifp, ifname, sizeof ifname);
1521                 sbuf_printf(sb, "%6.6s: ", ifname);
1522                 sbuf_printf(sb, "%7ju %7ju %4ju %4ju %4lu %5lu %10lu %9ju ",
1523                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IBYTES),
1524                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IPACKETS),
1525                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IERRORS),
1526                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IQDROPS),
1527                                                         /* rx_missed_errors */
1528                     0UL,                                /* rx_fifo_errors */
1529                     0UL,                                /* rx_length_errors +
1530                                                          * rx_over_errors +
1531                                                          * rx_crc_errors +
1532                                                          * rx_frame_errors */
1533                     0UL,                                /* rx_compressed */
1534                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IMCASTS));
1535                                                         /* XXX-BZ rx only? */
1536                 sbuf_printf(sb, "%8ju %7ju %4ju %4ju %4lu %5ju %7lu %10lu\n",
1537                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_OBYTES),
1538                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_OPACKETS),
1539                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_OERRORS),
1540                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_OQDROPS),
1541                     0UL,                                /* tx_fifo_errors */
1542                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_COLLISIONS),
1543                     0UL,                                /* tx_carrier_errors +
1544                                                          * tx_aborted_errors +
1545                                                          * tx_window_errors +
1546                                                          * tx_heartbeat_errors*/
1547                     0UL);                               /* tx_compressed */
1548         }
1549         IFNET_RUNLOCK();
1550         CURVNET_RESTORE();
1551
1552         return (0);
1553 }
1554
1555 struct walkarg {
1556         struct sbuf *sb;
1557 };
1558
1559 static int
1560 linux_route_print(struct rtentry *rt, void *vw)
1561 {
1562 #ifdef INET
1563         struct walkarg *w = vw;
1564         struct route_nhop_data rnd;
1565         struct in_addr dst, mask;
1566         struct nhop_object *nh;
1567         char ifname[16];
1568         uint32_t scopeid = 0;
1569         uint32_t gw = 0;
1570         uint32_t linux_flags = 0;
1571
1572         rt_get_inet_prefix_pmask(rt, &dst, &mask, &scopeid);
1573
1574         rt_get_rnd(rt, &rnd);
1575
1576         /* select only first route in case of multipath */
1577         nh = nhop_select_func(rnd.rnd_nhop, 0);
1578
1579         linux_ifname(nh->nh_ifp, ifname, sizeof(ifname));
1580
1581         gw = (nh->nh_flags & NHF_GATEWAY)
1582                 ? nh->gw4_sa.sin_addr.s_addr : 0;
1583
1584         linux_flags = RTF_UP |
1585                 (nhop_get_rtflags(nh) & (RTF_GATEWAY | RTF_HOST));
1586
1587         sbuf_printf(w->sb,
1588                 "%s\t"
1589                 "%08X\t%08X\t%04X\t"
1590                 "%d\t%u\t%d\t"
1591                 "%08X\t%d\t%u\t%u",
1592                 ifname,
1593                 dst.s_addr, gw, linux_flags,
1594                 0, 0, rnd.rnd_weight,
1595                 mask.s_addr, nh->nh_mtu, 0, 0);
1596
1597         sbuf_printf(w->sb, "\n\n");
1598 #endif
1599         return (0);
1600 }
1601
1602 /*
1603  * Filler function for proc/net/route
1604  */
1605 static int
1606 linprocfs_donetroute(PFS_FILL_ARGS)
1607 {
1608         struct walkarg w = {
1609                 .sb = sb
1610         };
1611         uint32_t fibnum = curthread->td_proc->p_fibnum;
1612
1613         sbuf_printf(w.sb, "%-127s\n", "Iface\tDestination\tGateway "
1614                "\tFlags\tRefCnt\tUse\tMetric\tMask\t\tMTU"
1615                "\tWindow\tIRTT");
1616
1617         CURVNET_SET(TD_TO_VNET(curthread));
1618         IFNET_RLOCK();
1619         rib_walk(fibnum, AF_INET, false, linux_route_print, &w);
1620         IFNET_RUNLOCK();
1621         CURVNET_RESTORE();
1622
1623         return (0);
1624 }
1625
1626 /*
1627  * Filler function for proc/sys/kernel/osrelease
1628  */
1629 static int
1630 linprocfs_doosrelease(PFS_FILL_ARGS)
1631 {
1632         char osrelease[LINUX_MAX_UTSNAME];
1633
1634         linux_get_osrelease(td, osrelease);
1635         sbuf_printf(sb, "%s\n", osrelease);
1636
1637         return (0);
1638 }
1639
1640 /*
1641  * Filler function for proc/sys/kernel/ostype
1642  */
1643 static int
1644 linprocfs_doostype(PFS_FILL_ARGS)
1645 {
1646         char osname[LINUX_MAX_UTSNAME];
1647
1648         linux_get_osname(td, osname);
1649         sbuf_printf(sb, "%s\n", osname);
1650
1651         return (0);
1652 }
1653
1654 /*
1655  * Filler function for proc/sys/kernel/version
1656  */
1657 static int
1658 linprocfs_doosbuild(PFS_FILL_ARGS)
1659 {
1660
1661         linprocfs_osbuild(td, sb);
1662         sbuf_cat(sb, "\n");
1663         return (0);
1664 }
1665
1666 /*
1667  * Filler function for proc/sys/kernel/msgmax
1668  */
1669 static int
1670 linprocfs_domsgmax(PFS_FILL_ARGS)
1671 {
1672
1673         sbuf_printf(sb, "%d\n", msginfo.msgmax);
1674         return (0);
1675 }
1676
1677 /*
1678  * Filler function for proc/sys/kernel/msgmni
1679  */
1680 static int
1681 linprocfs_domsgmni(PFS_FILL_ARGS)
1682 {
1683
1684         sbuf_printf(sb, "%d\n", msginfo.msgmni);
1685         return (0);
1686 }
1687
1688 /*
1689  * Filler function for proc/sys/kernel/msgmnb
1690  */
1691 static int
1692 linprocfs_domsgmnb(PFS_FILL_ARGS)
1693 {
1694
1695         sbuf_printf(sb, "%d\n", msginfo.msgmnb);
1696         return (0);
1697 }
1698
1699 /*
1700  * Filler function for proc/sys/kernel/ngroups_max
1701  *
1702  * Note that in Linux it defaults to 65536, not 1023.
1703  */
1704 static int
1705 linprocfs_dongroups_max(PFS_FILL_ARGS)
1706 {
1707
1708         sbuf_printf(sb, "%d\n", ngroups_max);
1709         return (0);
1710 }
1711
1712 /*
1713  * Filler function for proc/sys/kernel/pid_max
1714  */
1715 static int
1716 linprocfs_dopid_max(PFS_FILL_ARGS)
1717 {
1718
1719         sbuf_printf(sb, "%i\n", PID_MAX);
1720         return (0);
1721 }
1722
1723 /*
1724  * Filler function for proc/sys/kernel/sem
1725  */
1726 static int
1727 linprocfs_dosem(PFS_FILL_ARGS)
1728 {
1729
1730         sbuf_printf(sb, "%d %d %d %d\n", seminfo.semmsl, seminfo.semmns,
1731             seminfo.semopm, seminfo.semmni);
1732         return (0);
1733 }
1734
1735 /*
1736  * Filler function for proc/sys/kernel/shmall
1737  */
1738 static int
1739 linprocfs_doshmall(PFS_FILL_ARGS)
1740 {
1741
1742         sbuf_printf(sb, "%lu\n", shminfo.shmall);
1743         return (0);
1744 }
1745
1746 /*
1747  * Filler function for proc/sys/kernel/shmmax
1748  */
1749 static int
1750 linprocfs_doshmmax(PFS_FILL_ARGS)
1751 {
1752
1753         sbuf_printf(sb, "%lu\n", shminfo.shmmax);
1754         return (0);
1755 }
1756
1757 /*
1758  * Filler function for proc/sys/kernel/shmmni
1759  */
1760 static int
1761 linprocfs_doshmmni(PFS_FILL_ARGS)
1762 {
1763
1764         sbuf_printf(sb, "%lu\n", shminfo.shmmni);
1765         return (0);
1766 }
1767
1768 /*
1769  * Filler function for proc/sys/kernel/tainted
1770  */
1771 static int
1772 linprocfs_dotainted(PFS_FILL_ARGS)
1773 {
1774
1775         sbuf_printf(sb, "0\n");
1776         return (0);
1777 }
1778
1779 /*
1780  * Filler function for proc/sys/vm/min_free_kbytes
1781  *
1782  * This mirrors the approach in illumos to return zero for reads. Effectively,
1783  * it says, no memory is kept in reserve for "atomic allocations". This class
1784  * of allocation can be used at times when a thread cannot be suspended.
1785  */
1786 static int
1787 linprocfs_dominfree(PFS_FILL_ARGS)
1788 {
1789
1790         sbuf_printf(sb, "%d\n", 0);
1791         return (0);
1792 }
1793
1794 /*
1795  * Filler function for proc/scsi/device_info
1796  */
1797 static int
1798 linprocfs_doscsidevinfo(PFS_FILL_ARGS)
1799 {
1800
1801         return (0);
1802 }
1803
1804 /*
1805  * Filler function for proc/scsi/scsi
1806  */
1807 static int
1808 linprocfs_doscsiscsi(PFS_FILL_ARGS)
1809 {
1810
1811         return (0);
1812 }
1813
1814 /*
1815  * Filler function for proc/devices
1816  */
1817 static int
1818 linprocfs_dodevices(PFS_FILL_ARGS)
1819 {
1820         char *char_devices;
1821         sbuf_printf(sb, "Character devices:\n");
1822
1823         char_devices = linux_get_char_devices();
1824         sbuf_printf(sb, "%s", char_devices);
1825         linux_free_get_char_devices(char_devices);
1826
1827         sbuf_printf(sb, "\nBlock devices:\n");
1828
1829         return (0);
1830 }
1831
1832 /*
1833  * Filler function for proc/cmdline
1834  */
1835 static int
1836 linprocfs_docmdline(PFS_FILL_ARGS)
1837 {
1838
1839         sbuf_printf(sb, "BOOT_IMAGE=%s", kernelname);
1840         sbuf_printf(sb, " ro root=302\n");
1841         return (0);
1842 }
1843
1844 /*
1845  * Filler function for proc/filesystems
1846  */
1847 static int
1848 linprocfs_dofilesystems(PFS_FILL_ARGS)
1849 {
1850         struct vfsconf *vfsp;
1851
1852         vfsconf_slock();
1853         TAILQ_FOREACH(vfsp, &vfsconf, vfc_list) {
1854                 if (vfsp->vfc_flags & VFCF_SYNTHETIC)
1855                         sbuf_printf(sb, "nodev");
1856                 sbuf_printf(sb, "\t%s\n", vfsp->vfc_name);
1857         }
1858         vfsconf_sunlock();
1859         return(0);
1860 }
1861
1862 /*
1863  * Filler function for proc/modules
1864  */
1865 static int
1866 linprocfs_domodules(PFS_FILL_ARGS)
1867 {
1868 #if 0
1869         struct linker_file *lf;
1870
1871         TAILQ_FOREACH(lf, &linker_files, link) {
1872                 sbuf_printf(sb, "%-20s%8lu%4d\n", lf->filename,
1873                     (unsigned long)lf->size, lf->refs);
1874         }
1875 #endif
1876         return (0);
1877 }
1878
1879 /*
1880  * Filler function for proc/pid/fd
1881  */
1882 static int
1883 linprocfs_dofdescfs(PFS_FILL_ARGS)
1884 {
1885
1886         if (p == curproc)
1887                 sbuf_printf(sb, "/dev/fd");
1888         else
1889                 sbuf_printf(sb, "unknown");
1890         return (0);
1891 }
1892
1893 /*
1894  * Filler function for proc/pid/limits
1895  */
1896 static const struct linux_rlimit_ident {
1897         const char      *desc;
1898         const char      *unit;
1899         unsigned int    rlim_id;
1900 } linux_rlimits_ident[] = {
1901         { "Max cpu time",       "seconds",      RLIMIT_CPU },
1902         { "Max file size",      "bytes",        RLIMIT_FSIZE },
1903         { "Max data size",      "bytes",        RLIMIT_DATA },
1904         { "Max stack size",     "bytes",        RLIMIT_STACK },
1905         { "Max core file size",  "bytes",       RLIMIT_CORE },
1906         { "Max resident set",   "bytes",        RLIMIT_RSS },
1907         { "Max processes",      "processes",    RLIMIT_NPROC },
1908         { "Max open files",     "files",        RLIMIT_NOFILE },
1909         { "Max locked memory",  "bytes",        RLIMIT_MEMLOCK },
1910         { "Max address space",  "bytes",        RLIMIT_AS },
1911         { "Max file locks",     "locks",        LINUX_RLIMIT_LOCKS },
1912         { "Max pending signals", "signals",     LINUX_RLIMIT_SIGPENDING },
1913         { "Max msgqueue size",  "bytes",        LINUX_RLIMIT_MSGQUEUE },
1914         { "Max nice priority",          "",     LINUX_RLIMIT_NICE },
1915         { "Max realtime priority",      "",     LINUX_RLIMIT_RTPRIO },
1916         { "Max realtime timeout",       "us",   LINUX_RLIMIT_RTTIME },
1917         { 0, 0, 0 }
1918 };
1919
1920 static int
1921 linprocfs_doproclimits(PFS_FILL_ARGS)
1922 {
1923         const struct linux_rlimit_ident *li;
1924         struct plimit *limp;
1925         struct rlimit rl;
1926         ssize_t size;
1927         int res, error;
1928
1929         error = 0;
1930
1931         PROC_LOCK(p);
1932         limp = lim_hold(p->p_limit);
1933         PROC_UNLOCK(p);
1934         size = sizeof(res);
1935         sbuf_printf(sb, "%-26s%-21s%-21s%-21s\n", "Limit", "Soft Limit",
1936                         "Hard Limit", "Units");
1937         for (li = linux_rlimits_ident; li->desc != NULL; ++li) {
1938                 switch (li->rlim_id)
1939                 {
1940                 case LINUX_RLIMIT_LOCKS:
1941                         /* FALLTHROUGH */
1942                 case LINUX_RLIMIT_RTTIME:
1943                         rl.rlim_cur = RLIM_INFINITY;
1944                         break;
1945                 case LINUX_RLIMIT_SIGPENDING:
1946                         error = kernel_sysctlbyname(td,
1947                             "kern.sigqueue.max_pending_per_proc",
1948                             &res, &size, 0, 0, 0, 0);
1949                         if (error != 0)
1950                                 goto out;
1951                         rl.rlim_cur = res;
1952                         rl.rlim_max = res;
1953                         break;
1954                 case LINUX_RLIMIT_MSGQUEUE:
1955                         error = kernel_sysctlbyname(td,
1956                             "kern.ipc.msgmnb", &res, &size, 0, 0, 0, 0);
1957                         if (error != 0)
1958                                 goto out;
1959                         rl.rlim_cur = res;
1960                         rl.rlim_max = res;
1961                         break;
1962                 case LINUX_RLIMIT_NICE:
1963                         /* FALLTHROUGH */
1964                 case LINUX_RLIMIT_RTPRIO:
1965                         rl.rlim_cur = 0;
1966                         rl.rlim_max = 0;
1967                         break;
1968                 default:
1969                         rl = limp->pl_rlimit[li->rlim_id];
1970                         break;
1971                 }
1972                 if (rl.rlim_cur == RLIM_INFINITY)
1973                         sbuf_printf(sb, "%-26s%-21s%-21s%-10s\n",
1974                             li->desc, "unlimited", "unlimited", li->unit);
1975                 else
1976                         sbuf_printf(sb, "%-26s%-21llu%-21llu%-10s\n",
1977                             li->desc, (unsigned long long)rl.rlim_cur,
1978                             (unsigned long long)rl.rlim_max, li->unit);
1979         }
1980 out:
1981         lim_free(limp);
1982         return (error);
1983 }
1984
1985 /*
1986  * The point of the following two functions is to work around
1987  * an assertion in Chromium; see kern/240991 for details.
1988  */
1989 static int
1990 linprocfs_dotaskattr(PFS_ATTR_ARGS)
1991 {
1992
1993         vap->va_nlink = 3;
1994         return (0);
1995 }
1996
1997 /*
1998  * Filler function for proc/<pid>/task/.dummy
1999  */
2000 static int
2001 linprocfs_dotaskdummy(PFS_FILL_ARGS)
2002 {
2003
2004         return (0);
2005 }
2006
2007 /*
2008  * Filler function for proc/sys/kernel/random/uuid
2009  */
2010 static int
2011 linprocfs_douuid(PFS_FILL_ARGS)
2012 {
2013         struct uuid uuid;
2014
2015         kern_uuidgen(&uuid, 1);
2016         sbuf_printf_uuid(sb, &uuid);
2017         sbuf_printf(sb, "\n");
2018         return(0);
2019 }
2020
2021 /*
2022  * Filler function for proc/sys/kernel/random/boot_id
2023  */
2024 static int
2025 linprocfs_doboot_id(PFS_FILL_ARGS)
2026 {
2027        static bool firstboot = 1;
2028        static struct uuid uuid;
2029
2030        if (firstboot) {
2031                kern_uuidgen(&uuid, 1);
2032                firstboot = 0;
2033        }
2034        sbuf_printf_uuid(sb, &uuid);
2035        sbuf_printf(sb, "\n");
2036        return(0);
2037 }
2038
2039 /*
2040  * Filler function for proc/pid/auxv
2041  */
2042 static int
2043 linprocfs_doauxv(PFS_FILL_ARGS)
2044 {
2045         struct sbuf *asb;
2046         off_t buflen, resid;
2047         int error;
2048
2049         /*
2050          * Mimic linux behavior and pass only processes with usermode
2051          * address space as valid. Return zero silently otherwise.
2052          */
2053         if (p->p_vmspace == &vmspace0)
2054                 return (0);
2055
2056         if (uio->uio_resid == 0)
2057                 return (0);
2058         if (uio->uio_offset < 0 || uio->uio_resid < 0)
2059                 return (EINVAL);
2060
2061         asb = sbuf_new_auto();
2062         if (asb == NULL)
2063                 return (ENOMEM);
2064         error = proc_getauxv(td, p, asb);
2065         if (error == 0)
2066                 error = sbuf_finish(asb);
2067
2068         resid = sbuf_len(asb) - uio->uio_offset;
2069         if (resid > uio->uio_resid)
2070                 buflen = uio->uio_resid;
2071         else
2072                 buflen = resid;
2073         if (buflen > IOSIZE_MAX)
2074                 return (EINVAL);
2075         if (buflen > maxphys)
2076                 buflen = maxphys;
2077         if (resid <= 0)
2078                 return (0);
2079
2080         if (error == 0)
2081                 error = uiomove(sbuf_data(asb) + uio->uio_offset, buflen, uio);
2082         sbuf_delete(asb);
2083         return (error);
2084 }
2085
2086 /*
2087  * Filler function for proc/self/oom_score_adj
2088  */
2089 static int
2090 linprocfs_do_oom_score_adj(PFS_FILL_ARGS)
2091 {
2092         struct linux_pemuldata *pem;
2093         long oom;
2094
2095         pem = pem_find(p);
2096         if (pem == NULL || uio == NULL)
2097                 return (EOPNOTSUPP);
2098         if (uio->uio_rw == UIO_READ) {
2099                 sbuf_printf(sb, "%d\n", pem->oom_score_adj);
2100         } else {
2101                 sbuf_trim(sb);
2102                 sbuf_finish(sb);
2103                 oom = strtol(sbuf_data(sb), NULL, 10);
2104                 if (oom < LINUX_OOM_SCORE_ADJ_MIN ||
2105                     oom > LINUX_OOM_SCORE_ADJ_MAX)
2106                         return (EINVAL);
2107                 pem->oom_score_adj = oom;
2108         }
2109         return (0);
2110 }
2111
2112 /*
2113  * Filler function for proc/sys/vm/max_map_count
2114  *
2115  * Maximum number of active map areas, on Linux this limits the number
2116  * of vmaps per mm struct. We don't limit mappings, return a suitable
2117  * large value.
2118  */
2119 static int
2120 linprocfs_domax_map_cnt(PFS_FILL_ARGS)
2121 {
2122
2123         sbuf_printf(sb, "%d\n", INT32_MAX);
2124         return (0);
2125 }
2126
2127 /*
2128  * Constructor
2129  */
2130 static int
2131 linprocfs_init(PFS_INIT_ARGS)
2132 {
2133         struct pfs_node *root;
2134         struct pfs_node *dir;
2135         struct pfs_node *sys;
2136
2137         root = pi->pi_root;
2138
2139         /* /proc/... */
2140         pfs_create_file(root, "cmdline", &linprocfs_docmdline,
2141             NULL, NULL, NULL, PFS_RD);
2142         pfs_create_file(root, "cpuinfo", &linprocfs_docpuinfo,
2143             NULL, NULL, NULL, PFS_RD);
2144         pfs_create_file(root, "devices", &linprocfs_dodevices,
2145             NULL, NULL, NULL, PFS_RD);
2146         pfs_create_file(root, "filesystems", &linprocfs_dofilesystems,
2147             NULL, NULL, NULL, PFS_RD);
2148         pfs_create_file(root, "loadavg", &linprocfs_doloadavg,
2149             NULL, NULL, NULL, PFS_RD);
2150         pfs_create_file(root, "meminfo", &linprocfs_domeminfo,
2151             NULL, NULL, NULL, PFS_RD);
2152         pfs_create_file(root, "modules", &linprocfs_domodules,
2153             NULL, NULL, NULL, PFS_RD);
2154         pfs_create_file(root, "mounts", &linprocfs_domtab,
2155             NULL, NULL, NULL, PFS_RD);
2156         pfs_create_file(root, "mtab", &linprocfs_domtab,
2157             NULL, NULL, NULL, PFS_RD);
2158         pfs_create_file(root, "partitions", &linprocfs_dopartitions,
2159             NULL, NULL, NULL, PFS_RD);
2160         pfs_create_link(root, "self", &procfs_docurproc,
2161             NULL, NULL, NULL, 0);
2162         pfs_create_file(root, "stat", &linprocfs_dostat,
2163             NULL, NULL, NULL, PFS_RD);
2164         pfs_create_file(root, "swaps", &linprocfs_doswaps,
2165             NULL, NULL, NULL, PFS_RD);
2166         pfs_create_file(root, "uptime", &linprocfs_douptime,
2167             NULL, NULL, NULL, PFS_RD);
2168         pfs_create_file(root, "version", &linprocfs_doversion,
2169             NULL, NULL, NULL, PFS_RD);
2170
2171         /* /proc/bus/... */
2172         dir = pfs_create_dir(root, "bus", NULL, NULL, NULL, 0);
2173         dir = pfs_create_dir(dir, "pci", NULL, NULL, NULL, 0);
2174         dir = pfs_create_dir(dir, "devices", NULL, NULL, NULL, 0);
2175
2176         /* /proc/net/... */
2177         dir = pfs_create_dir(root, "net", NULL, NULL, NULL, 0);
2178         pfs_create_file(dir, "dev", &linprocfs_donetdev,
2179             NULL, NULL, NULL, PFS_RD);
2180         pfs_create_file(dir, "route", &linprocfs_donetroute,
2181             NULL, NULL, NULL, PFS_RD);
2182
2183         /* /proc/<pid>/... */
2184         dir = pfs_create_dir(root, "pid", NULL, NULL, NULL, PFS_PROCDEP);
2185         pfs_create_file(dir, "cmdline", &linprocfs_doproccmdline,
2186             NULL, NULL, NULL, PFS_RD);
2187         pfs_create_link(dir, "cwd", &linprocfs_doproccwd,
2188             NULL, NULL, NULL, 0);
2189         pfs_create_file(dir, "environ", &linprocfs_doprocenviron,
2190             NULL, &procfs_candebug, NULL, PFS_RD);
2191         pfs_create_link(dir, "exe", &procfs_doprocfile,
2192             NULL, &procfs_notsystem, NULL, 0);
2193         pfs_create_file(dir, "maps", &linprocfs_doprocmaps,
2194             NULL, NULL, NULL, PFS_RD | PFS_AUTODRAIN);
2195         pfs_create_file(dir, "mem", &linprocfs_doprocmem,
2196             procfs_attr_rw, &procfs_candebug, NULL, PFS_RDWR | PFS_RAW);
2197         pfs_create_file(dir, "mountinfo", &linprocfs_doprocmountinfo,
2198             NULL, NULL, NULL, PFS_RD);
2199         pfs_create_file(dir, "mounts", &linprocfs_domtab,
2200             NULL, NULL, NULL, PFS_RD);
2201         pfs_create_link(dir, "root", &linprocfs_doprocroot,
2202             NULL, NULL, NULL, 0);
2203         pfs_create_file(dir, "stat", &linprocfs_doprocstat,
2204             NULL, NULL, NULL, PFS_RD);
2205         pfs_create_file(dir, "statm", &linprocfs_doprocstatm,
2206             NULL, NULL, NULL, PFS_RD);
2207         pfs_create_file(dir, "status", &linprocfs_doprocstatus,
2208             NULL, NULL, NULL, PFS_RD);
2209         pfs_create_link(dir, "fd", &linprocfs_dofdescfs,
2210             NULL, NULL, NULL, 0);
2211         pfs_create_file(dir, "auxv", &linprocfs_doauxv,
2212             NULL, &procfs_candebug, NULL, PFS_RD|PFS_RAWRD);
2213         pfs_create_file(dir, "limits", &linprocfs_doproclimits,
2214             NULL, NULL, NULL, PFS_RD);
2215         pfs_create_file(dir, "oom_score_adj", &linprocfs_do_oom_score_adj,
2216             procfs_attr_rw, &procfs_candebug, NULL, PFS_RDWR);
2217
2218         /* /proc/<pid>/task/... */
2219         dir = pfs_create_dir(dir, "task", linprocfs_dotaskattr, NULL, NULL, 0);
2220         pfs_create_file(dir, ".dummy", &linprocfs_dotaskdummy,
2221             NULL, NULL, NULL, PFS_RD);
2222
2223         /* /proc/scsi/... */
2224         dir = pfs_create_dir(root, "scsi", NULL, NULL, NULL, 0);
2225         pfs_create_file(dir, "device_info", &linprocfs_doscsidevinfo,
2226             NULL, NULL, NULL, PFS_RD);
2227         pfs_create_file(dir, "scsi", &linprocfs_doscsiscsi,
2228             NULL, NULL, NULL, PFS_RD);
2229
2230         /* /proc/sys/... */
2231         sys = pfs_create_dir(root, "sys", NULL, NULL, NULL, 0);
2232
2233         /* /proc/sys/kernel/... */
2234         dir = pfs_create_dir(sys, "kernel", NULL, NULL, NULL, 0);
2235         pfs_create_file(dir, "osrelease", &linprocfs_doosrelease,
2236             NULL, NULL, NULL, PFS_RD);
2237         pfs_create_file(dir, "ostype", &linprocfs_doostype,
2238             NULL, NULL, NULL, PFS_RD);
2239         pfs_create_file(dir, "version", &linprocfs_doosbuild,
2240             NULL, NULL, NULL, PFS_RD);
2241         pfs_create_file(dir, "msgmax", &linprocfs_domsgmax,
2242             NULL, NULL, NULL, PFS_RD);
2243         pfs_create_file(dir, "msgmni", &linprocfs_domsgmni,
2244             NULL, NULL, NULL, PFS_RD);
2245         pfs_create_file(dir, "msgmnb", &linprocfs_domsgmnb,
2246             NULL, NULL, NULL, PFS_RD);
2247         pfs_create_file(dir, "ngroups_max", &linprocfs_dongroups_max,
2248             NULL, NULL, NULL, PFS_RD);
2249         pfs_create_file(dir, "pid_max", &linprocfs_dopid_max,
2250             NULL, NULL, NULL, PFS_RD);
2251         pfs_create_file(dir, "sem", &linprocfs_dosem,
2252             NULL, NULL, NULL, PFS_RD);
2253         pfs_create_file(dir, "shmall", &linprocfs_doshmall,
2254             NULL, NULL, NULL, PFS_RD);
2255         pfs_create_file(dir, "shmmax", &linprocfs_doshmmax,
2256             NULL, NULL, NULL, PFS_RD);
2257         pfs_create_file(dir, "shmmni", &linprocfs_doshmmni,
2258             NULL, NULL, NULL, PFS_RD);
2259         pfs_create_file(dir, "tainted", &linprocfs_dotainted,
2260             NULL, NULL, NULL, PFS_RD);
2261
2262         /* /proc/sys/kernel/random/... */
2263         dir = pfs_create_dir(dir, "random", NULL, NULL, NULL, 0);
2264         pfs_create_file(dir, "uuid", &linprocfs_douuid,
2265             NULL, NULL, NULL, PFS_RD);
2266         pfs_create_file(dir, "boot_id", &linprocfs_doboot_id,
2267             NULL, NULL, NULL, PFS_RD);
2268
2269         /* /proc/sys/vm/.... */
2270         dir = pfs_create_dir(sys, "vm", NULL, NULL, NULL, 0);
2271         pfs_create_file(dir, "min_free_kbytes", &linprocfs_dominfree,
2272             NULL, NULL, NULL, PFS_RD);
2273         pfs_create_file(dir, "max_map_count", &linprocfs_domax_map_cnt,
2274             NULL, NULL, NULL, PFS_RD);
2275
2276         return (0);
2277 }
2278
2279 /*
2280  * Destructor
2281  */
2282 static int
2283 linprocfs_uninit(PFS_INIT_ARGS)
2284 {
2285
2286         /* nothing to do, pseudofs will GC */
2287         return (0);
2288 }
2289
2290 PSEUDOFS(linprocfs, 1, VFCF_JAIL);
2291 #if defined(__aarch64__) || defined(__amd64__)
2292 MODULE_DEPEND(linprocfs, linux_common, 1, 1, 1);
2293 #else
2294 MODULE_DEPEND(linprocfs, linux, 1, 1, 1);
2295 #endif
2296 MODULE_DEPEND(linprocfs, procfs, 1, 1, 1);
2297 MODULE_DEPEND(linprocfs, sysvmsg, 1, 1, 1);
2298 MODULE_DEPEND(linprocfs, sysvsem, 1, 1, 1);
2299 MODULE_DEPEND(linprocfs, sysvshm, 1, 1, 1);