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 <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 static const char *path_slash_sys = "/sys";
 407 static const char *fstype_sysfs = "sysfs";
 408
 409 static int
 410 _mtab_helper(const struct pfs_node *pn, const struct statfs *sp,
 411     const char **mntfrom, const char **mntto, const char **fstype)
 412 {
 413         /* determine device name */
 414         *mntfrom = sp->f_mntfromname;
 415
 416         /* determine mount point */
 417         *mntto = sp->f_mntonname;
 418
 419         /* determine fs type */
 420         *fstype = sp->f_fstypename;
 421         if (strcmp(*fstype, pn->pn_info->pi_name) == 0)
 422                 *mntfrom = *fstype = "proc";
 423         else if (strcmp(*fstype, "procfs") == 0)
 424                 return (ECANCELED);
 425
 426         if (strcmp(*fstype, "autofs") == 0) {
 427                 /*
 428                  * FreeBSD uses eg "map -hosts", whereas Linux
 429                  * expects just "-hosts".
 430                  */
 431                 if (strncmp(*mntfrom, "map ", 4) == 0)
 432                         *mntfrom += 4;
 433         }
 434
 435         if (strcmp(*fstype, "linsysfs") == 0) {
 436                 *mntfrom = path_slash_sys;
 437                 *fstype = fstype_sysfs;
 438         } else {
 439                 /* For Linux msdosfs is called vfat */
 440                 if (strcmp(*fstype, "msdosfs") == 0)
 441                         *fstype = "vfat";
 442         }
 443         return (0);
 444 }
 445
 446 static void
 447 _sbuf_mntoptions_helper(struct sbuf *sb, uint64_t f_flags)
 448 {
 449         sbuf_cat(sb, (f_flags & MNT_RDONLY) ? "ro" : "rw");
 450 #define ADD_OPTION(opt, name) \
 451         if (f_flags & (opt)) sbuf_cat(sb, "," name);
 452         ADD_OPTION(MNT_SYNCHRONOUS,     "sync");
 453         ADD_OPTION(MNT_NOEXEC,          "noexec");
 454         ADD_OPTION(MNT_NOSUID,          "nosuid");
 455         ADD_OPTION(MNT_UNION,           "union");
 456         ADD_OPTION(MNT_ASYNC,           "async");
 457         ADD_OPTION(MNT_SUIDDIR,         "suiddir");
 458         ADD_OPTION(MNT_NOSYMFOLLOW,     "nosymfollow");
 459         ADD_OPTION(MNT_NOATIME,         "noatime");
 460 #undef ADD_OPTION
 461 }
 462
 463 /*
 464  * Filler function for proc/mtab and proc/<pid>/mounts.
 465  *
 466  * /proc/mtab doesn't exist in Linux' procfs, but is included here so
 467  * users can symlink /compat/linux/etc/mtab to /proc/mtab
 468  */
 469 static int
 470 linprocfs_domtab(PFS_FILL_ARGS)
 471 {
 472         struct nameidata nd;
 473         const char *lep, *mntto, *mntfrom, *fstype;
 474         char *dlep, *flep;
 475         size_t lep_len;
 476         int error;
 477         struct statfs *buf, *sp;
 478         size_t count;
 479
 480         /* resolve symlinks etc. in the emulation tree prefix */
 481         /*
 482          * Ideally, this would use the current chroot rather than some
 483          * hardcoded path.
 484          */
 485         NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, linux_emul_path, td);
 486         flep = NULL;
 487         error = namei(&nd);
 488         lep = linux_emul_path;
 489         if (error == 0) {
 490                 if (vn_fullpath(nd.ni_vp, &dlep, &flep) == 0)
 491                         lep = dlep;
 492                 vrele(nd.ni_vp);
 493         }
 494         lep_len = strlen(lep);
 495
 496         buf = NULL;
 497         error = kern_getfsstat(td, &buf, SIZE_T_MAX, &count,
 498             UIO_SYSSPACE, MNT_WAIT);
 499         if (error != 0) {
 500                 free(buf, M_TEMP);
 501                 free(flep, M_TEMP);
 502                 return (error);
 503         }
 504
 505         for (sp = buf; count > 0; sp++, count--) {
 506                 error = _mtab_helper(pn, sp, &mntfrom, &mntto, &fstype);
 507                 if (error != 0) {
 508                         MPASS(error == ECANCELED);
 509                         continue;
 510                 }
 511
 512                 /* determine mount point */
 513                 if (strncmp(mntto, lep, lep_len) == 0 && mntto[lep_len] == '/')
 514                         mntto += lep_len;
 515
 516                 sbuf_printf(sb, "%s %s %s ", mntfrom, mntto, fstype);
 517                 _sbuf_mntoptions_helper(sb, sp->f_flags);
 518                 /* a real Linux mtab will also show NFS options */
 519                 sbuf_printf(sb, " 0 0\n");
 520         }
 521
 522         free(buf, M_TEMP);
 523         free(flep, M_TEMP);
 524         return (error);
 525 }
 526
 527 static int
 528 linprocfs_doprocmountinfo(PFS_FILL_ARGS)
 529 {
 530         struct nameidata nd;
 531         const char *mntfrom, *mntto, *fstype;
 532         const char *lep;
 533         char *dlep, *flep;
 534         struct statfs *buf, *sp;
 535         size_t count, lep_len;
 536         int error;
 537
 538         /*
 539          * Ideally, this would use the current chroot rather than some
 540          * hardcoded path.
 541          */
 542         NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, linux_emul_path, td);
 543         flep = NULL;
 544         error = namei(&nd);
 545         lep = linux_emul_path;
 546         if (error == 0) {
 547                 if (vn_fullpath(nd.ni_vp, &dlep, &flep) == 0)
 548                         lep = dlep;
 549                 vrele(nd.ni_vp);
 550         }
 551         lep_len = strlen(lep);
 552
 553         buf = NULL;
 554         error = kern_getfsstat(td, &buf, SIZE_T_MAX, &count,
 555             UIO_SYSSPACE, MNT_WAIT);
 556         if (error != 0)
 557                 goto out;
 558
 559         for (sp = buf; count > 0; sp++, count--) {
 560                 error = _mtab_helper(pn, sp, &mntfrom, &mntto, &fstype);
 561                 if (error != 0) {
 562                         MPASS(error == ECANCELED);
 563                         continue;
 564                 }
 565
 566                 if (strncmp(mntto, lep, lep_len) == 0 && mntto[lep_len] == '/')
 567                         mntto += lep_len;
 568 #if 0
 569                 /*
 570                  * If the prefix is a chroot, and this mountpoint is not under
 571                  * the prefix, we should skip it.  Leave it for now for
 572                  * consistency with procmtab above.
 573                  */
 574                 else
 575                         continue;
 576 #endif
 577
 578                 /*
 579                  * (1) mount id
 580                  *
 581                  * (2) parent mount id -- we don't have this cheaply, so
 582                  * provide a dummy value
 583                  *
 584                  * (3) major:minor -- ditto
 585                  *
 586                  * (4) root filesystem mount -- probably a namespaces thing
 587                  *
 588                  * (5) mountto path
 589                  */
 590                 sbuf_printf(sb, "%u 0 0:0 / %s ",
 591                     sp->f_fsid.val[0] ^ sp->f_fsid.val[1], mntto);
 592                 /* (6) mount options */
 593                 _sbuf_mntoptions_helper(sb, sp->f_flags);
 594                 /*
 595                  * (7) zero or more optional fields -- again, namespace related
 596                  *
 597                  * (8) End of variable length fields separator ("-")
 598                  *
 599                  * (9) fstype
 600                  *
 601                  * (10) mount from
 602                  *
 603                  * (11) "superblock" options -- like (6), but different
 604                  * semantics in Linux
 605                  */
 606                 sbuf_printf(sb, " - %s %s %s\n", fstype, mntfrom,
 607                     (sp->f_flags & MNT_RDONLY) ? "ro" : "rw");
 608         }
 609
 610         error = 0;
 611 out:
 612         free(buf, M_TEMP);
 613         free(flep, M_TEMP);
 614         return (error);
 615 }
 616
 617 /*
 618  * Filler function for proc/partitions
 619  */
 620 static int
 621 linprocfs_dopartitions(PFS_FILL_ARGS)
 622 {
 623         struct g_class *cp;
 624         struct g_geom *gp;
 625         struct g_provider *pp;
 626         int major, minor;
 627
 628         g_topology_lock();
 629         sbuf_printf(sb, "major minor  #blocks  name rio rmerge rsect "
 630             "ruse wio wmerge wsect wuse running use aveq\n");
 631
 632         LIST_FOREACH(cp, &g_classes, class) {
 633                 if (strcmp(cp->name, "DISK") == 0 ||
 634                     strcmp(cp->name, "PART") == 0)
 635                         LIST_FOREACH(gp, &cp->geom, geom) {
 636                                 LIST_FOREACH(pp, &gp->provider, provider) {
 637                                         if (linux_driver_get_major_minor(
 638                                             pp->name, &major, &minor) != 0) {
 639                                                 major = 0;
 640                                                 minor = 0;
 641                                         }
 642                                         sbuf_printf(sb, "%d %d %lld %s "
 643                                             "%d %d %d %d %d "
 644                                              "%d %d %d %d %d %d\n",
 645                                              major, minor,
 646                                              (long long)pp->mediasize, pp->name,
 647                                              0, 0, 0, 0, 0,
 648                                              0, 0, 0, 0, 0, 0);
 649                                 }
 650                         }
 651         }
 652         g_topology_unlock();
 653
 654         return (0);
 655 }
 656
 657 /*
 658  * Filler function for proc/stat
 659  *
 660  * Output depends on kernel version:
 661  *
 662  * v2.5.40 <=
 663  *   user nice system idle
 664  * v2.5.41
 665  *   user nice system idle iowait
 666  * v2.6.11
 667  *   user nice system idle iowait irq softirq steal
 668  * v2.6.24
 669  *   user nice system idle iowait irq softirq steal guest
 670  * v2.6.33 >=
 671  *   user nice system idle iowait irq softirq steal guest guest_nice
 672  */
 673 static int
 674 linprocfs_dostat(PFS_FILL_ARGS)
 675 {
 676         struct pcpu *pcpu;
 677         long cp_time[CPUSTATES];
 678         long *cp;
 679         struct timeval boottime;
 680         int i;
 681         char *zero_pad;
 682         bool has_intr = true;
 683
 684         if (linux_kernver(td) >= LINUX_KERNVER(2,6,33)) {
 685                 zero_pad = " 0 0 0 0\n";
 686         } else if (linux_kernver(td) >= LINUX_KERNVER(2,6,24)) {
 687                 zero_pad = " 0 0 0\n";
 688         } else if (linux_kernver(td) >= LINUX_KERNVER(2,6,11)) {
 689                 zero_pad = " 0 0\n";
 690         } else if (linux_kernver(td) >= LINUX_KERNVER(2,5,41)) {
 691                 has_intr = false;
 692                 zero_pad = " 0\n";
 693         } else {
 694                 has_intr = false;
 695                 zero_pad = "\n";
 696         }
 697
 698         read_cpu_time(cp_time);
 699         getboottime(&boottime);
 700         /* Parameters common to all versions */
 701         sbuf_printf(sb, "cpu %lu %lu %lu %lu",
 702             T2J(cp_time[CP_USER]),
 703             T2J(cp_time[CP_NICE]),
 704             T2J(cp_time[CP_SYS]),
 705             T2J(cp_time[CP_IDLE]));
 706
 707         /* Print interrupt stats if available */
 708         if (has_intr) {
 709                 sbuf_printf(sb, " 0 %lu", T2J(cp_time[CP_INTR]));
 710         }
 711
 712         /* Pad out remaining fields depending on version */
 713         sbuf_printf(sb, "%s", zero_pad);
 714
 715         CPU_FOREACH(i) {
 716                 pcpu = pcpu_find(i);
 717                 cp = pcpu->pc_cp_time;
 718                 sbuf_printf(sb, "cpu%d %lu %lu %lu %lu", i,
 719                     T2J(cp[CP_USER]),
 720                     T2J(cp[CP_NICE]),
 721                     T2J(cp[CP_SYS]),
 722                     T2J(cp[CP_IDLE]));
 723
 724                 if (has_intr) {
 725                         sbuf_printf(sb, " 0 %lu", T2J(cp[CP_INTR]));
 726                 }
 727
 728                 sbuf_printf(sb, "%s", zero_pad);
 729         }
 730         sbuf_printf(sb,
 731             "disk 0 0 0 0\n"
 732             "page %ju %ju\n"
 733             "swap %ju %ju\n"
 734             "intr %ju\n"
 735             "ctxt %ju\n"
 736             "btime %lld\n",
 737             (uintmax_t)VM_CNT_FETCH(v_vnodepgsin),
 738             (uintmax_t)VM_CNT_FETCH(v_vnodepgsout),
 739             (uintmax_t)VM_CNT_FETCH(v_swappgsin),
 740             (uintmax_t)VM_CNT_FETCH(v_swappgsout),
 741             (uintmax_t)VM_CNT_FETCH(v_intr),
 742             (uintmax_t)VM_CNT_FETCH(v_swtch),
 743             (long long)boottime.tv_sec);
 744         return (0);
 745 }
 746
 747 static int
 748 linprocfs_doswaps(PFS_FILL_ARGS)
 749 {
 750         struct xswdev xsw;
 751         uintmax_t total, used;
 752         int n;
 753         char devname[SPECNAMELEN + 1];
 754
 755         sbuf_printf(sb, "Filename\t\t\t\tType\t\tSize\tUsed\tPriority\n");
 756         for (n = 0; ; n++) {
 757                 if (swap_dev_info(n, &xsw, devname, sizeof(devname)) != 0)
 758                         break;
 759                 total = (uintmax_t)xsw.xsw_nblks * PAGE_SIZE / 1024;
 760                 used  = (uintmax_t)xsw.xsw_used * PAGE_SIZE / 1024;
 761
 762                 /*
 763                  * The space and not tab after the device name is on
 764                  * purpose.  Linux does so.
 765                  */
 766                 sbuf_printf(sb, "/dev/%-34s unknown\t\t%jd\t%jd\t-1\n",
 767                     devname, total, used);
 768         }
 769         return (0);
 770 }
 771
 772 /*
 773  * Filler function for proc/uptime
 774  */
 775 static int
 776 linprocfs_douptime(PFS_FILL_ARGS)
 777 {
 778         long cp_time[CPUSTATES];
 779         struct timeval tv;
 780
 781         getmicrouptime(&tv);
 782         read_cpu_time(cp_time);
 783         sbuf_printf(sb, "%lld.%02ld %ld.%02lu\n",
 784             (long long)tv.tv_sec, tv.tv_usec / 10000,
 785             T2S(cp_time[CP_IDLE] / mp_ncpus),
 786             T2CS(cp_time[CP_IDLE] / mp_ncpus) % 100);
 787         return (0);
 788 }
 789
 790 /*
 791  * Get OS build date
 792  */
 793 static void
 794 linprocfs_osbuild(struct thread *td, struct sbuf *sb)
 795 {
 796 #if 0
 797         char osbuild[256];
 798         char *cp1, *cp2;
 799
 800         strncpy(osbuild, version, 256);
 801         osbuild[255] = '\0';
 802         cp1 = strstr(osbuild, "\n");
 803         cp2 = strstr(osbuild, ":");
 804         if (cp1 && cp2) {
 805                 *cp1 = *cp2 = '\0';
 806                 cp1 = strstr(osbuild, "#");
 807         } else
 808                 cp1 = NULL;
 809         if (cp1)
 810                 sbuf_printf(sb, "%s%s", cp1, cp2 + 1);
 811         else
 812 #endif
 813                 sbuf_cat(sb, "#4 Sun Dec 18 04:30:00 CET 1977");
 814 }
 815
 816 /*
 817  * Get OS builder
 818  */
 819 static void
 820 linprocfs_osbuilder(struct thread *td, struct sbuf *sb)
 821 {
 822 #if 0
 823         char builder[256];
 824         char *cp;
 825
 826         cp = strstr(version, "\n    ");
 827         if (cp) {
 828                 strncpy(builder, cp + 5, 256);
 829                 builder[255] = '\0';
 830                 cp = strstr(builder, ":");
 831                 if (cp)
 832                         *cp = '\0';
 833         }
 834         if (cp)
 835                 sbuf_cat(sb, builder);
 836         else
 837 #endif
 838                 sbuf_cat(sb, "des@freebsd.org");
 839 }
 840
 841 /*
 842  * Filler function for proc/version
 843  */
 844 static int
 845 linprocfs_doversion(PFS_FILL_ARGS)
 846 {
 847         char osname[LINUX_MAX_UTSNAME];
 848         char osrelease[LINUX_MAX_UTSNAME];
 849
 850         linux_get_osname(td, osname);
 851         linux_get_osrelease(td, osrelease);
 852         sbuf_printf(sb, "%s version %s (", osname, osrelease);
 853         linprocfs_osbuilder(td, sb);
 854         sbuf_cat(sb, ") (gcc version " __VERSION__ ") ");
 855         linprocfs_osbuild(td, sb);
 856         sbuf_cat(sb, "\n");
 857
 858         return (0);
 859 }
 860
 861 /*
 862  * Filler function for proc/loadavg
 863  */
 864 static int
 865 linprocfs_doloadavg(PFS_FILL_ARGS)
 866 {
 867
 868         sbuf_printf(sb,
 869             "%d.%02d %d.%02d %d.%02d %d/%d %d\n",
 870             (int)(averunnable.ldavg[0] / averunnable.fscale),
 871             (int)(averunnable.ldavg[0] * 100 / averunnable.fscale % 100),
 872             (int)(averunnable.ldavg[1] / averunnable.fscale),
 873             (int)(averunnable.ldavg[1] * 100 / averunnable.fscale % 100),
 874             (int)(averunnable.ldavg[2] / averunnable.fscale),
 875             (int)(averunnable.ldavg[2] * 100 / averunnable.fscale % 100),
 876             1,                          /* number of running tasks */
 877             nprocs,                     /* number of tasks */
 878             lastpid                     /* the last pid */
 879         );
 880         return (0);
 881 }
 882
 883 static int
 884 linprocfs_get_tty_nr(struct proc *p)
 885 {
 886         struct session *sp;
 887         const char *ttyname;
 888         int error, major, minor, nr;
 889
 890         PROC_LOCK_ASSERT(p, MA_OWNED);
 891         sx_assert(&proctree_lock, SX_LOCKED);
 892
 893         if ((p->p_flag & P_CONTROLT) == 0)
 894                 return (-1);
 895
 896         sp = p->p_pgrp->pg_session;
 897         if (sp == NULL)
 898                 return (-1);
 899
 900         ttyname = devtoname(sp->s_ttyp->t_dev);
 901         error = linux_driver_get_major_minor(ttyname, &major, &minor);
 902         if (error != 0)
 903                 return (-1);
 904
 905         nr = makedev(major, minor);
 906         return (nr);
 907 }
 908
 909 /*
 910  * Filler function for proc/pid/stat
 911  */
 912 static int
 913 linprocfs_doprocstat(PFS_FILL_ARGS)
 914 {
 915         struct kinfo_proc kp;
 916         struct timeval boottime;
 917         char state;
 918         static int ratelimit = 0;
 919         int tty_nr;
 920         vm_offset_t startcode, startdata;
 921
 922         getboottime(&boottime);
 923         sx_slock(&proctree_lock);
 924         PROC_LOCK(p);
 925         fill_kinfo_proc(p, &kp);
 926         tty_nr = linprocfs_get_tty_nr(p);
 927         sx_sunlock(&proctree_lock);
 928         if (p->p_vmspace) {
 929            startcode = (vm_offset_t)p->p_vmspace->vm_taddr;
 930            startdata = (vm_offset_t)p->p_vmspace->vm_daddr;
 931         } else {
 932            startcode = 0;
 933            startdata = 0;
 934         }
 935         sbuf_printf(sb, "%d", p->p_pid);
 936 #define PS_ADD(name, fmt, arg) sbuf_printf(sb, " " fmt, arg)
 937         PS_ADD("comm",          "(%s)", p->p_comm);
 938         if (kp.ki_stat > sizeof(linux_state)) {
 939                 state = 'R';
 940
 941                 if (ratelimit == 0) {
 942                         printf("linprocfs: don't know how to handle unknown FreeBSD state %d/%zd, mapping to R\n",
 943                             kp.ki_stat, sizeof(linux_state));
 944                         ++ratelimit;
 945                 }
 946         } else
 947                 state = linux_state[kp.ki_stat - 1];
 948         PS_ADD("state",         "%c",   state);
 949         PS_ADD("ppid",          "%d",   p->p_pptr ? p->p_pptr->p_pid : 0);
 950         PS_ADD("pgrp",          "%d",   p->p_pgid);
 951         PS_ADD("session",       "%d",   p->p_session->s_sid);
 952         PROC_UNLOCK(p);
 953         PS_ADD("tty",           "%d",   tty_nr);
 954         PS_ADD("tpgid",         "%d",   kp.ki_tpgid);
 955         PS_ADD("flags",         "%u",   0); /* XXX */
 956         PS_ADD("minflt",        "%lu",  kp.ki_rusage.ru_minflt);
 957         PS_ADD("cminflt",       "%lu",  kp.ki_rusage_ch.ru_minflt);
 958         PS_ADD("majflt",        "%lu",  kp.ki_rusage.ru_majflt);
 959         PS_ADD("cmajflt",       "%lu",  kp.ki_rusage_ch.ru_majflt);
 960         PS_ADD("utime",         "%ld",  TV2J(&kp.ki_rusage.ru_utime));
 961         PS_ADD("stime",         "%ld",  TV2J(&kp.ki_rusage.ru_stime));
 962         PS_ADD("cutime",        "%ld",  TV2J(&kp.ki_rusage_ch.ru_utime));
 963         PS_ADD("cstime",        "%ld",  TV2J(&kp.ki_rusage_ch.ru_stime));
 964         PS_ADD("priority",      "%d",   kp.ki_pri.pri_user);
 965         PS_ADD("nice",          "%d",   kp.ki_nice); /* 19 (nicest) to -19 */
 966         PS_ADD("0",             "%d",   0); /* removed field */
 967         PS_ADD("itrealvalue",   "%d",   0); /* XXX */
 968         PS_ADD("starttime",     "%lu",  TV2J(&kp.ki_start) - TV2J(&boottime));
 969         PS_ADD("vsize",         "%ju",  P2K((uintmax_t)kp.ki_size));
 970         PS_ADD("rss",           "%ju",  (uintmax_t)kp.ki_rssize);
 971         PS_ADD("rlim",          "%lu",  kp.ki_rusage.ru_maxrss);
 972         PS_ADD("startcode",     "%ju",  (uintmax_t)startcode);
 973         PS_ADD("endcode",       "%ju",  (uintmax_t)startdata);
 974         PS_ADD("startstack",    "%u",   0); /* XXX */
 975         PS_ADD("kstkesp",       "%u",   0); /* XXX */
 976         PS_ADD("kstkeip",       "%u",   0); /* XXX */
 977         PS_ADD("signal",        "%u",   0); /* XXX */
 978         PS_ADD("blocked",       "%u",   0); /* XXX */
 979         PS_ADD("sigignore",     "%u",   0); /* XXX */
 980         PS_ADD("sigcatch",      "%u",   0); /* XXX */
 981         PS_ADD("wchan",         "%u",   0); /* XXX */
 982         PS_ADD("nswap",         "%lu",  kp.ki_rusage.ru_nswap);
 983         PS_ADD("cnswap",        "%lu",  kp.ki_rusage_ch.ru_nswap);
 984         PS_ADD("exitsignal",    "%d",   0); /* XXX */
 985         PS_ADD("processor",     "%u",   kp.ki_lastcpu);
 986         PS_ADD("rt_priority",   "%u",   0); /* XXX */ /* >= 2.5.19 */
 987         PS_ADD("policy",        "%u",   kp.ki_pri.pri_class); /* >= 2.5.19 */
 988 #undef PS_ADD
 989         sbuf_putc(sb, '\n');
 990
 991         return (0);
 992 }
 993
 994 /*
 995  * Filler function for proc/pid/statm
 996  */
 997 static int
 998 linprocfs_doprocstatm(PFS_FILL_ARGS)
 999 {
1000         struct kinfo_proc kp;
1001         segsz_t lsize;
1002
1003         sx_slock(&proctree_lock);
1004         PROC_LOCK(p);
1005         fill_kinfo_proc(p, &kp);
1006         PROC_UNLOCK(p);
1007         sx_sunlock(&proctree_lock);
1008
1009         /*
1010          * See comments in linprocfs_doprocstatus() regarding the
1011          * computation of lsize.
1012          */
1013         /* size resident share trs drs lrs dt */
1014         sbuf_printf(sb, "%ju ", B2P((uintmax_t)kp.ki_size));
1015         sbuf_printf(sb, "%ju ", (uintmax_t)kp.ki_rssize);
1016         sbuf_printf(sb, "%ju ", (uintmax_t)0); /* XXX */
1017         sbuf_printf(sb, "%ju ", (uintmax_t)kp.ki_tsize);
1018         sbuf_printf(sb, "%ju ", (uintmax_t)(kp.ki_dsize + kp.ki_ssize));
1019         lsize = B2P(kp.ki_size) - kp.ki_dsize -
1020             kp.ki_ssize - kp.ki_tsize - 1;
1021         sbuf_printf(sb, "%ju ", (uintmax_t)lsize);
1022         sbuf_printf(sb, "%ju\n", (uintmax_t)0); /* XXX */
1023
1024         return (0);
1025 }
1026
1027 /*
1028  * Filler function for proc/pid/status
1029  */
1030 static int
1031 linprocfs_doprocstatus(PFS_FILL_ARGS)
1032 {
1033         struct kinfo_proc kp;
1034         char *state;
1035         segsz_t lsize;
1036         struct thread *td2;
1037         struct sigacts *ps;
1038         l_sigset_t siglist, sigignore, sigcatch;
1039         int i;
1040
1041         sx_slock(&proctree_lock);
1042         PROC_LOCK(p);
1043         td2 = FIRST_THREAD_IN_PROC(p); /* XXXKSE pretend only one thread */
1044
1045         if (P_SHOULDSTOP(p)) {
1046                 state = "T (stopped)";
1047         } else {
1048                 switch(p->p_state) {
1049                 case PRS_NEW:
1050                         state = "I (idle)";
1051                         break;
1052                 case PRS_NORMAL:
1053                         if (p->p_flag & P_WEXIT) {
1054                                 state = "X (exiting)";
1055                                 break;
1056                         }
1057                         switch(TD_GET_STATE(td2)) {
1058                         case TDS_INHIBITED:
1059                                 state = "S (sleeping)";
1060                                 break;
1061                         case TDS_RUNQ:
1062                         case TDS_RUNNING:
1063                                 state = "R (running)";
1064                                 break;
1065                         default:
1066                                 state = "? (unknown)";
1067                                 break;
1068                         }
1069                         break;
1070                 case PRS_ZOMBIE:
1071                         state = "Z (zombie)";
1072                         break;
1073                 default:
1074                         state = "? (unknown)";
1075                         break;
1076                 }
1077         }
1078
1079         fill_kinfo_proc(p, &kp);
1080         sx_sunlock(&proctree_lock);
1081
1082         sbuf_printf(sb, "Name:\t%s\n",          p->p_comm); /* XXX escape */
1083         sbuf_printf(sb, "State:\t%s\n",         state);
1084
1085         /*
1086          * Credentials
1087          */
1088         sbuf_printf(sb, "Tgid:\t%d\n",          p->p_pid);
1089         sbuf_printf(sb, "Pid:\t%d\n",           p->p_pid);
1090         sbuf_printf(sb, "PPid:\t%d\n",          kp.ki_ppid );
1091         sbuf_printf(sb, "TracerPid:\t%d\n",     kp.ki_tracer );
1092         sbuf_printf(sb, "Uid:\t%d %d %d %d\n",  p->p_ucred->cr_ruid,
1093                                                 p->p_ucred->cr_uid,
1094                                                 p->p_ucred->cr_svuid,
1095                                                 /* FreeBSD doesn't have fsuid */
1096                                                 p->p_ucred->cr_uid);
1097         sbuf_printf(sb, "Gid:\t%d %d %d %d\n",  p->p_ucred->cr_rgid,
1098                                                 p->p_ucred->cr_gid,
1099                                                 p->p_ucred->cr_svgid,
1100                                                 /* FreeBSD doesn't have fsgid */
1101                                                 p->p_ucred->cr_gid);
1102         sbuf_cat(sb, "Groups:\t");
1103         for (i = 0; i < p->p_ucred->cr_ngroups; i++)
1104                 sbuf_printf(sb, "%d ",          p->p_ucred->cr_groups[i]);
1105         PROC_UNLOCK(p);
1106         sbuf_putc(sb, '\n');
1107
1108         /*
1109          * Memory
1110          *
1111          * While our approximation of VmLib may not be accurate (I
1112          * don't know of a simple way to verify it, and I'm not sure
1113          * it has much meaning anyway), I believe it's good enough.
1114          *
1115          * The same code that could (I think) accurately compute VmLib
1116          * could also compute VmLck, but I don't really care enough to
1117          * implement it. Submissions are welcome.
1118          */
1119         sbuf_printf(sb, "VmSize:\t%8ju kB\n",   B2K((uintmax_t)kp.ki_size));
1120         sbuf_printf(sb, "VmLck:\t%8u kB\n",     P2K(0)); /* XXX */
1121         sbuf_printf(sb, "VmRSS:\t%8ju kB\n",    P2K((uintmax_t)kp.ki_rssize));
1122         sbuf_printf(sb, "VmData:\t%8ju kB\n",   P2K((uintmax_t)kp.ki_dsize));
1123         sbuf_printf(sb, "VmStk:\t%8ju kB\n",    P2K((uintmax_t)kp.ki_ssize));
1124         sbuf_printf(sb, "VmExe:\t%8ju kB\n",    P2K((uintmax_t)kp.ki_tsize));
1125         lsize = B2P(kp.ki_size) - kp.ki_dsize -
1126             kp.ki_ssize - kp.ki_tsize - 1;
1127         sbuf_printf(sb, "VmLib:\t%8ju kB\n",    P2K((uintmax_t)lsize));
1128
1129         /*
1130          * Signal masks
1131          */
1132         PROC_LOCK(p);
1133         bsd_to_linux_sigset(&p->p_siglist, &siglist);
1134         ps = p->p_sigacts;
1135         mtx_lock(&ps->ps_mtx);
1136         bsd_to_linux_sigset(&ps->ps_sigignore, &sigignore);
1137         bsd_to_linux_sigset(&ps->ps_sigcatch, &sigcatch);
1138         mtx_unlock(&ps->ps_mtx);
1139         PROC_UNLOCK(p);
1140
1141         sbuf_printf(sb, "SigPnd:\t%016jx\n",    siglist.__mask);
1142         /*
1143          * XXX. SigBlk - target thread's signal mask, td_sigmask.
1144          * To implement SigBlk pseudofs should support proc/tid dir entries.
1145          */
1146         sbuf_printf(sb, "SigBlk:\t%016x\n",     0);
1147         sbuf_printf(sb, "SigIgn:\t%016jx\n",    sigignore.__mask);
1148         sbuf_printf(sb, "SigCgt:\t%016jx\n",    sigcatch.__mask);
1149
1150         /*
1151          * Linux also prints the capability masks, but we don't have
1152          * capabilities yet, and when we do get them they're likely to
1153          * be meaningless to Linux programs, so we lie. XXX
1154          */
1155         sbuf_printf(sb, "CapInh:\t%016x\n",     0);
1156         sbuf_printf(sb, "CapPrm:\t%016x\n",     0);
1157         sbuf_printf(sb, "CapEff:\t%016x\n",     0);
1158
1159         return (0);
1160 }
1161
1162 /*
1163  * Filler function for proc/pid/cwd
1164  */
1165 static int
1166 linprocfs_doproccwd(PFS_FILL_ARGS)
1167 {
1168         struct pwd *pwd;
1169         char *fullpath = "unknown";
1170         char *freepath = NULL;
1171
1172         pwd = pwd_hold(td);
1173         vn_fullpath(pwd->pwd_cdir, &fullpath, &freepath);
1174         sbuf_printf(sb, "%s", fullpath);
1175         if (freepath)
1176                 free(freepath, M_TEMP);
1177         pwd_drop(pwd);
1178         return (0);
1179 }
1180
1181 /*
1182  * Filler function for proc/pid/root
1183  */
1184 static int
1185 linprocfs_doprocroot(PFS_FILL_ARGS)
1186 {
1187         struct pwd *pwd;
1188         struct vnode *vp;
1189         char *fullpath = "unknown";
1190         char *freepath = NULL;
1191
1192         pwd = pwd_hold(td);
1193         vp = jailed(p->p_ucred) ? pwd->pwd_jdir : pwd->pwd_rdir;
1194         vn_fullpath(vp, &fullpath, &freepath);
1195         sbuf_printf(sb, "%s", fullpath);
1196         if (freepath)
1197                 free(freepath, M_TEMP);
1198         pwd_drop(pwd);
1199         return (0);
1200 }
1201
1202 /*
1203  * Filler function for proc/pid/cmdline
1204  */
1205 static int
1206 linprocfs_doproccmdline(PFS_FILL_ARGS)
1207 {
1208         int ret;
1209
1210         PROC_LOCK(p);
1211         if ((ret = p_cansee(td, p)) != 0) {
1212                 PROC_UNLOCK(p);
1213                 return (ret);
1214         }
1215
1216         /*
1217          * Mimic linux behavior and pass only processes with usermode
1218          * address space as valid.  Return zero silently otherwize.
1219          */
1220         if (p->p_vmspace == &vmspace0) {
1221                 PROC_UNLOCK(p);
1222                 return (0);
1223         }
1224         if (p->p_args != NULL) {
1225                 sbuf_bcpy(sb, p->p_args->ar_args, p->p_args->ar_length);
1226                 PROC_UNLOCK(p);
1227                 return (0);
1228         }
1229
1230         if ((p->p_flag & P_SYSTEM) != 0) {
1231                 PROC_UNLOCK(p);
1232                 return (0);
1233         }
1234
1235         PROC_UNLOCK(p);
1236
1237         ret = proc_getargv(td, p, sb);
1238         return (ret);
1239 }
1240
1241 /*
1242  * Filler function for proc/pid/environ
1243  */
1244 static int
1245 linprocfs_doprocenviron(PFS_FILL_ARGS)
1246 {
1247
1248         /*
1249          * Mimic linux behavior and pass only processes with usermode
1250          * address space as valid.  Return zero silently otherwize.
1251          */
1252         if (p->p_vmspace == &vmspace0)
1253                 return (0);
1254
1255         return (proc_getenvv(td, p, sb));
1256 }
1257
1258 static char l32_map_str[] = "%08lx-%08lx %s%s%s%s %08lx %02x:%02x %lu%s%s\n";
1259 static char l64_map_str[] = "%016lx-%016lx %s%s%s%s %08lx %02x:%02x %lu%s%s\n";
1260 static char vdso_str[] = "      [vdso]";
1261 static char stack_str[] = "      [stack]";
1262
1263 /*
1264  * Filler function for proc/pid/maps
1265  */
1266 static int
1267 linprocfs_doprocmaps(PFS_FILL_ARGS)
1268 {
1269         struct vmspace *vm;
1270         vm_map_t map;
1271         vm_map_entry_t entry, tmp_entry;
1272         vm_object_t obj, tobj, lobj;
1273         vm_offset_t e_start, e_end;
1274         vm_ooffset_t off;
1275         vm_prot_t e_prot;
1276         unsigned int last_timestamp;
1277         char *name = "", *freename = NULL;
1278         const char *l_map_str;
1279         ino_t ino;
1280         int ref_count, shadow_count, flags;
1281         int error;
1282         struct vnode *vp;
1283         struct vattr vat;
1284         bool private;
1285
1286         PROC_LOCK(p);
1287         error = p_candebug(td, p);
1288         PROC_UNLOCK(p);
1289         if (error)
1290                 return (error);
1291
1292         if (uio->uio_rw != UIO_READ)
1293                 return (EOPNOTSUPP);
1294
1295         error = 0;
1296         vm = vmspace_acquire_ref(p);
1297         if (vm == NULL)
1298                 return (ESRCH);
1299
1300         if (SV_CURPROC_FLAG(SV_LP64))
1301                 l_map_str = l64_map_str;
1302         else
1303                 l_map_str = l32_map_str;
1304         map = &vm->vm_map;
1305         vm_map_lock_read(map);
1306         VM_MAP_ENTRY_FOREACH(entry, map) {
1307                 name = "";
1308                 freename = NULL;
1309                 if (entry->eflags & MAP_ENTRY_IS_SUB_MAP)
1310                         continue;
1311                 e_prot = entry->protection;
1312                 e_start = entry->start;
1313                 e_end = entry->end;
1314                 obj = entry->object.vm_object;
1315                 off = entry->offset;
1316                 for (lobj = tobj = obj; tobj != NULL;
1317                     lobj = tobj, tobj = tobj->backing_object) {
1318                         VM_OBJECT_RLOCK(tobj);
1319                         off += lobj->backing_object_offset;
1320                         if (lobj != obj)
1321                                 VM_OBJECT_RUNLOCK(lobj);
1322                 }
1323                 private = (entry->eflags & MAP_ENTRY_COW) != 0 || obj == NULL ||
1324                     (obj->flags & OBJ_ANON) != 0;
1325                 last_timestamp = map->timestamp;
1326                 vm_map_unlock_read(map);
1327                 ino = 0;
1328                 if (lobj) {
1329                         vp = vm_object_vnode(lobj);
1330                         if (vp != NULL)
1331                                 vref(vp);
1332                         if (lobj != obj)
1333                                 VM_OBJECT_RUNLOCK(lobj);
1334                         flags = obj->flags;
1335                         ref_count = obj->ref_count;
1336                         shadow_count = obj->shadow_count;
1337                         VM_OBJECT_RUNLOCK(obj);
1338                         if (vp != NULL) {
1339                                 vn_fullpath(vp, &name, &freename);
1340                                 vn_lock(vp, LK_SHARED | LK_RETRY);
1341                                 VOP_GETATTR(vp, &vat, td->td_ucred);
1342                                 ino = vat.va_fileid;
1343                                 vput(vp);
1344                         } else if (SV_PROC_ABI(p) == SV_ABI_LINUX) {
1345                                 /*
1346                                  * sv_shared_page_base pointed out to the
1347                                  * FreeBSD sharedpage, PAGE_SIZE is a size
1348                                  * of it. The vDSO page is above.
1349                                  */
1350                                 if (e_start == p->p_sysent->sv_shared_page_base +
1351                                     PAGE_SIZE)
1352                                         name = vdso_str;
1353                                 if (e_end == p->p_sysent->sv_usrstack)
1354                                         name = stack_str;
1355                         }
1356                 } else {
1357                         flags = 0;
1358                         ref_count = 0;
1359                         shadow_count = 0;
1360                 }
1361
1362                 /*
1363                  * format:
1364                  *  start, end, access, offset, major, minor, inode, name.
1365                  */
1366                 error = sbuf_printf(sb, l_map_str,
1367                     (u_long)e_start, (u_long)e_end,
1368                     (e_prot & VM_PROT_READ)?"r":"-",
1369                     (e_prot & VM_PROT_WRITE)?"w":"-",
1370                     (e_prot & VM_PROT_EXECUTE)?"x":"-",
1371                     private ? "p" : "s",
1372                     (u_long)off,
1373                     0,
1374                     0,
1375                     (u_long)ino,
1376                     *name ? "     " : " ",
1377                     name
1378                     );
1379                 if (freename)
1380                         free(freename, M_TEMP);
1381                 vm_map_lock_read(map);
1382                 if (error == -1) {
1383                         error = 0;
1384                         break;
1385                 }
1386                 if (last_timestamp != map->timestamp) {
1387                         /*
1388                          * Look again for the entry because the map was
1389                          * modified while it was unlocked.  Specifically,
1390                          * the entry may have been clipped, merged, or deleted.
1391                          */
1392                         vm_map_lookup_entry(map, e_end - 1, &tmp_entry);
1393                         entry = tmp_entry;
1394                 }
1395         }
1396         vm_map_unlock_read(map);
1397         vmspace_free(vm);
1398
1399         return (error);
1400 }
1401
1402 /*
1403  * Filler function for proc/pid/mem
1404  */
1405 static int
1406 linprocfs_doprocmem(PFS_FILL_ARGS)
1407 {
1408         ssize_t resid;
1409         int error;
1410
1411         resid = uio->uio_resid;
1412         error = procfs_doprocmem(PFS_FILL_ARGNAMES);
1413
1414         if (uio->uio_rw == UIO_READ && resid != uio->uio_resid)
1415                 return (0);
1416
1417         if (error == EFAULT)
1418                 error = EIO;
1419
1420         return (error);
1421 }
1422
1423 /*
1424  * Criteria for interface name translation
1425  */
1426 #define IFP_IS_ETH(ifp) (ifp->if_type == IFT_ETHER)
1427
1428 static int
1429 linux_ifname(struct ifnet *ifp, char *buffer, size_t buflen)
1430 {
1431         struct ifnet *ifscan;
1432         int ethno;
1433
1434         IFNET_RLOCK_ASSERT();
1435
1436         /* Short-circuit non ethernet interfaces */
1437         if (!IFP_IS_ETH(ifp))
1438                 return (strlcpy(buffer, ifp->if_xname, buflen));
1439
1440         /* Determine the (relative) unit number for ethernet interfaces */
1441         ethno = 0;
1442         CK_STAILQ_FOREACH(ifscan, &V_ifnet, if_link) {
1443                 if (ifscan == ifp)
1444                         return (snprintf(buffer, buflen, "eth%d", ethno));
1445                 if (IFP_IS_ETH(ifscan))
1446                         ethno++;
1447         }
1448
1449         return (0);
1450 }
1451
1452 /*
1453  * Filler function for proc/net/dev
1454  */
1455 static int
1456 linprocfs_donetdev(PFS_FILL_ARGS)
1457 {
1458         char ifname[16]; /* XXX LINUX_IFNAMSIZ */
1459         struct ifnet *ifp;
1460
1461         sbuf_printf(sb, "%6s|%58s|%s\n"
1462             "%6s|%58s|%58s\n",
1463             "Inter-", "   Receive", "  Transmit",
1464             " face",
1465             "bytes    packets errs drop fifo frame compressed multicast",
1466             "bytes    packets errs drop fifo colls carrier compressed");
1467
1468         CURVNET_SET(TD_TO_VNET(curthread));
1469         IFNET_RLOCK();
1470         CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
1471                 linux_ifname(ifp, ifname, sizeof ifname);
1472                 sbuf_printf(sb, "%6.6s: ", ifname);
1473                 sbuf_printf(sb, "%7ju %7ju %4ju %4ju %4lu %5lu %10lu %9ju ",
1474                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IBYTES),
1475                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IPACKETS),
1476                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IERRORS),
1477                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IQDROPS),
1478                                                         /* rx_missed_errors */
1479                     0UL,                                /* rx_fifo_errors */
1480                     0UL,                                /* rx_length_errors +
1481                                                          * rx_over_errors +
1482                                                          * rx_crc_errors +
1483                                                          * rx_frame_errors */
1484                     0UL,                                /* rx_compressed */
1485                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_IMCASTS));
1486                                                         /* XXX-BZ rx only? */
1487                 sbuf_printf(sb, "%8ju %7ju %4ju %4ju %4lu %5ju %7lu %10lu\n",
1488                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_OBYTES),
1489                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_OPACKETS),
1490                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_OERRORS),
1491                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_OQDROPS),
1492                     0UL,                                /* tx_fifo_errors */
1493                     (uintmax_t )ifp->if_get_counter(ifp, IFCOUNTER_COLLISIONS),
1494                     0UL,                                /* tx_carrier_errors +
1495                                                          * tx_aborted_errors +
1496                                                          * tx_window_errors +
1497                                                          * tx_heartbeat_errors*/
1498                     0UL);                               /* tx_compressed */
1499         }
1500         IFNET_RUNLOCK();
1501         CURVNET_RESTORE();
1502
1503         return (0);
1504 }
1505
1506 /*
1507  * Filler function for proc/sys/kernel/osrelease
1508  */
1509 static int
1510 linprocfs_doosrelease(PFS_FILL_ARGS)
1511 {
1512         char osrelease[LINUX_MAX_UTSNAME];
1513
1514         linux_get_osrelease(td, osrelease);
1515         sbuf_printf(sb, "%s\n", osrelease);
1516
1517         return (0);
1518 }
1519
1520 /*
1521  * Filler function for proc/sys/kernel/ostype
1522  */
1523 static int
1524 linprocfs_doostype(PFS_FILL_ARGS)
1525 {
1526         char osname[LINUX_MAX_UTSNAME];
1527
1528         linux_get_osname(td, osname);
1529         sbuf_printf(sb, "%s\n", osname);
1530
1531         return (0);
1532 }
1533
1534 /*
1535  * Filler function for proc/sys/kernel/version
1536  */
1537 static int
1538 linprocfs_doosbuild(PFS_FILL_ARGS)
1539 {
1540
1541         linprocfs_osbuild(td, sb);
1542         sbuf_cat(sb, "\n");
1543         return (0);
1544 }
1545
1546 /*
1547  * Filler function for proc/sys/kernel/msgmax
1548  */
1549 static int
1550 linprocfs_domsgmax(PFS_FILL_ARGS)
1551 {
1552
1553         sbuf_printf(sb, "%d\n", msginfo.msgmax);
1554         return (0);
1555 }
1556
1557 /*
1558  * Filler function for proc/sys/kernel/msgmni
1559  */
1560 static int
1561 linprocfs_domsgmni(PFS_FILL_ARGS)
1562 {
1563
1564         sbuf_printf(sb, "%d\n", msginfo.msgmni);
1565         return (0);
1566 }
1567
1568 /*
1569  * Filler function for proc/sys/kernel/msgmnb
1570  */
1571 static int
1572 linprocfs_domsgmnb(PFS_FILL_ARGS)
1573 {
1574
1575         sbuf_printf(sb, "%d\n", msginfo.msgmnb);
1576         return (0);
1577 }
1578
1579 /*
1580  * Filler function for proc/sys/kernel/ngroups_max
1581  *
1582  * Note that in Linux it defaults to 65536, not 1023.
1583  */
1584 static int
1585 linprocfs_dongroups_max(PFS_FILL_ARGS)
1586 {
1587
1588         sbuf_printf(sb, "%d\n", ngroups_max);
1589         return (0);
1590 }
1591
1592 /*
1593  * Filler function for proc/sys/kernel/pid_max
1594  */
1595 static int
1596 linprocfs_dopid_max(PFS_FILL_ARGS)
1597 {
1598
1599         sbuf_printf(sb, "%i\n", PID_MAX);
1600         return (0);
1601 }
1602
1603 /*
1604  * Filler function for proc/sys/kernel/sem
1605  */
1606 static int
1607 linprocfs_dosem(PFS_FILL_ARGS)
1608 {
1609
1610         sbuf_printf(sb, "%d %d %d %d\n", seminfo.semmsl, seminfo.semmns,
1611             seminfo.semopm, seminfo.semmni);
1612         return (0);
1613 }
1614
1615 /*
1616  * Filler function for proc/sys/kernel/shmall
1617  */
1618 static int
1619 linprocfs_doshmall(PFS_FILL_ARGS)
1620 {
1621
1622         sbuf_printf(sb, "%lu\n", shminfo.shmall);
1623         return (0);
1624 }
1625
1626 /*
1627  * Filler function for proc/sys/kernel/shmmax
1628  */
1629 static int
1630 linprocfs_doshmmax(PFS_FILL_ARGS)
1631 {
1632
1633         sbuf_printf(sb, "%lu\n", shminfo.shmmax);
1634         return (0);
1635 }
1636
1637 /*
1638  * Filler function for proc/sys/kernel/shmmni
1639  */
1640 static int
1641 linprocfs_doshmmni(PFS_FILL_ARGS)
1642 {
1643
1644         sbuf_printf(sb, "%lu\n", shminfo.shmmni);
1645         return (0);
1646 }
1647
1648 /*
1649  * Filler function for proc/sys/kernel/tainted
1650  */
1651 static int
1652 linprocfs_dotainted(PFS_FILL_ARGS)
1653 {
1654
1655         sbuf_printf(sb, "0\n");
1656         return (0);
1657 }
1658
1659 /*
1660  * Filler function for proc/sys/vm/min_free_kbytes
1661  *
1662  * This mirrors the approach in illumos to return zero for reads. Effectively,
1663  * it says, no memory is kept in reserve for "atomic allocations". This class
1664  * of allocation can be used at times when a thread cannot be suspended.
1665  */
1666 static int
1667 linprocfs_dominfree(PFS_FILL_ARGS)
1668 {
1669
1670         sbuf_printf(sb, "%d\n", 0);
1671         return (0);
1672 }
1673
1674 /*
1675  * Filler function for proc/scsi/device_info
1676  */
1677 static int
1678 linprocfs_doscsidevinfo(PFS_FILL_ARGS)
1679 {
1680
1681         return (0);
1682 }
1683
1684 /*
1685  * Filler function for proc/scsi/scsi
1686  */
1687 static int
1688 linprocfs_doscsiscsi(PFS_FILL_ARGS)
1689 {
1690
1691         return (0);
1692 }
1693
1694 /*
1695  * Filler function for proc/devices
1696  */
1697 static int
1698 linprocfs_dodevices(PFS_FILL_ARGS)
1699 {
1700         char *char_devices;
1701         sbuf_printf(sb, "Character devices:\n");
1702
1703         char_devices = linux_get_char_devices();
1704         sbuf_printf(sb, "%s", char_devices);
1705         linux_free_get_char_devices(char_devices);
1706
1707         sbuf_printf(sb, "\nBlock devices:\n");
1708
1709         return (0);
1710 }
1711
1712 /*
1713  * Filler function for proc/cmdline
1714  */
1715 static int
1716 linprocfs_docmdline(PFS_FILL_ARGS)
1717 {
1718
1719         sbuf_printf(sb, "BOOT_IMAGE=%s", kernelname);
1720         sbuf_printf(sb, " ro root=302\n");
1721         return (0);
1722 }
1723
1724 /*
1725  * Filler function for proc/filesystems
1726  */
1727 static int
1728 linprocfs_dofilesystems(PFS_FILL_ARGS)
1729 {
1730         struct vfsconf *vfsp;
1731
1732         vfsconf_slock();
1733         TAILQ_FOREACH(vfsp, &vfsconf, vfc_list) {
1734                 if (vfsp->vfc_flags & VFCF_SYNTHETIC)
1735                         sbuf_printf(sb, "nodev");
1736                 sbuf_printf(sb, "\t%s\n", vfsp->vfc_name);
1737         }
1738         vfsconf_sunlock();
1739         return(0);
1740 }
1741
1742 /*
1743  * Filler function for proc/modules
1744  */
1745 static int
1746 linprocfs_domodules(PFS_FILL_ARGS)
1747 {
1748 #if 0
1749         struct linker_file *lf;
1750
1751         TAILQ_FOREACH(lf, &linker_files, link) {
1752                 sbuf_printf(sb, "%-20s%8lu%4d\n", lf->filename,
1753                     (unsigned long)lf->size, lf->refs);
1754         }
1755 #endif
1756         return (0);
1757 }
1758
1759 /*
1760  * Filler function for proc/pid/fd
1761  */
1762 static int
1763 linprocfs_dofdescfs(PFS_FILL_ARGS)
1764 {
1765
1766         if (p == curproc)
1767                 sbuf_printf(sb, "/dev/fd");
1768         else
1769                 sbuf_printf(sb, "unknown");
1770         return (0);
1771 }
1772
1773 /*
1774  * Filler function for proc/pid/limits
1775  */
1776 static const struct linux_rlimit_ident {
1777         const char      *desc;
1778         const char      *unit;
1779         unsigned int    rlim_id;
1780 } linux_rlimits_ident[] = {
1781         { "Max cpu time",       "seconds",      RLIMIT_CPU },
1782         { "Max file size",      "bytes",        RLIMIT_FSIZE },
1783         { "Max data size",      "bytes",        RLIMIT_DATA },
1784         { "Max stack size",     "bytes",        RLIMIT_STACK },
1785         { "Max core file size",  "bytes",       RLIMIT_CORE },
1786         { "Max resident set",   "bytes",        RLIMIT_RSS },
1787         { "Max processes",      "processes",    RLIMIT_NPROC },
1788         { "Max open files",     "files",        RLIMIT_NOFILE },
1789         { "Max locked memory",  "bytes",        RLIMIT_MEMLOCK },
1790         { "Max address space",  "bytes",        RLIMIT_AS },
1791         { "Max file locks",     "locks",        LINUX_RLIMIT_LOCKS },
1792         { "Max pending signals", "signals",     LINUX_RLIMIT_SIGPENDING },
1793         { "Max msgqueue size",  "bytes",        LINUX_RLIMIT_MSGQUEUE },
1794         { "Max nice priority",          "",     LINUX_RLIMIT_NICE },
1795         { "Max realtime priority",      "",     LINUX_RLIMIT_RTPRIO },
1796         { "Max realtime timeout",       "us",   LINUX_RLIMIT_RTTIME },
1797         { 0, 0, 0 }
1798 };
1799
1800 static int
1801 linprocfs_doproclimits(PFS_FILL_ARGS)
1802 {
1803         const struct linux_rlimit_ident *li;
1804         struct plimit *limp;
1805         struct rlimit rl;
1806         ssize_t size;
1807         int res, error;
1808
1809         error = 0;
1810
1811         PROC_LOCK(p);
1812         limp = lim_hold(p->p_limit);
1813         PROC_UNLOCK(p);
1814         size = sizeof(res);
1815         sbuf_printf(sb, "%-26s%-21s%-21s%-21s\n", "Limit", "Soft Limit",
1816                         "Hard Limit", "Units");
1817         for (li = linux_rlimits_ident; li->desc != NULL; ++li) {
1818                 switch (li->rlim_id)
1819                 {
1820                 case LINUX_RLIMIT_LOCKS:
1821                         /* FALLTHROUGH */
1822                 case LINUX_RLIMIT_RTTIME:
1823                         rl.rlim_cur = RLIM_INFINITY;
1824                         break;
1825                 case LINUX_RLIMIT_SIGPENDING:
1826                         error = kernel_sysctlbyname(td,
1827                             "kern.sigqueue.max_pending_per_proc",
1828                             &res, &size, 0, 0, 0, 0);
1829                         if (error != 0)
1830                                 goto out;
1831                         rl.rlim_cur = res;
1832                         rl.rlim_max = res;
1833                         break;
1834                 case LINUX_RLIMIT_MSGQUEUE:
1835                         error = kernel_sysctlbyname(td,
1836                             "kern.ipc.msgmnb", &res, &size, 0, 0, 0, 0);
1837                         if (error != 0)
1838                                 goto out;
1839                         rl.rlim_cur = res;
1840                         rl.rlim_max = res;
1841                         break;
1842                 case LINUX_RLIMIT_NICE:
1843                         /* FALLTHROUGH */
1844                 case LINUX_RLIMIT_RTPRIO:
1845                         rl.rlim_cur = 0;
1846                         rl.rlim_max = 0;
1847                         break;
1848                 default:
1849                         rl = limp->pl_rlimit[li->rlim_id];
1850                         break;
1851                 }
1852                 if (rl.rlim_cur == RLIM_INFINITY)
1853                         sbuf_printf(sb, "%-26s%-21s%-21s%-10s\n",
1854                             li->desc, "unlimited", "unlimited", li->unit);
1855                 else
1856                         sbuf_printf(sb, "%-26s%-21llu%-21llu%-10s\n",
1857                             li->desc, (unsigned long long)rl.rlim_cur,
1858                             (unsigned long long)rl.rlim_max, li->unit);
1859         }
1860 out:
1861         lim_free(limp);
1862         return (error);
1863 }
1864
1865 /*
1866  * The point of the following two functions is to work around
1867  * an assertion in Chromium; see kern/240991 for details.
1868  */
1869 static int
1870 linprocfs_dotaskattr(PFS_ATTR_ARGS)
1871 {
1872
1873         vap->va_nlink = 3;
1874         return (0);
1875 }
1876
1877 /*
1878  * Filler function for proc/<pid>/task/.dummy
1879  */
1880 static int
1881 linprocfs_dotaskdummy(PFS_FILL_ARGS)
1882 {
1883
1884         return (0);
1885 }
1886
1887 /*
1888  * Filler function for proc/sys/kernel/random/uuid
1889  */
1890 static int
1891 linprocfs_douuid(PFS_FILL_ARGS)
1892 {
1893         struct uuid uuid;
1894
1895         kern_uuidgen(&uuid, 1);
1896         sbuf_printf_uuid(sb, &uuid);
1897         sbuf_printf(sb, "\n");
1898         return(0);
1899 }
1900
1901 /*
1902  * Filler function for proc/pid/auxv
1903  */
1904 static int
1905 linprocfs_doauxv(PFS_FILL_ARGS)
1906 {
1907         struct sbuf *asb;
1908         off_t buflen, resid;
1909         int error;
1910
1911         /*
1912          * Mimic linux behavior and pass only processes with usermode
1913          * address space as valid. Return zero silently otherwise.
1914          */
1915         if (p->p_vmspace == &vmspace0)
1916                 return (0);
1917
1918         if (uio->uio_resid == 0)
1919                 return (0);
1920         if (uio->uio_offset < 0 || uio->uio_resid < 0)
1921                 return (EINVAL);
1922
1923         asb = sbuf_new_auto();
1924         if (asb == NULL)
1925                 return (ENOMEM);
1926         error = proc_getauxv(td, p, asb);
1927         if (error == 0)
1928                 error = sbuf_finish(asb);
1929
1930         resid = sbuf_len(asb) - uio->uio_offset;
1931         if (resid > uio->uio_resid)
1932                 buflen = uio->uio_resid;
1933         else
1934                 buflen = resid;
1935         if (buflen > IOSIZE_MAX)
1936                 return (EINVAL);
1937         if (buflen > maxphys)
1938                 buflen = maxphys;
1939         if (resid <= 0)
1940                 return (0);
1941
1942         if (error == 0)
1943                 error = uiomove(sbuf_data(asb) + uio->uio_offset, buflen, uio);
1944         sbuf_delete(asb);
1945         return (error);
1946 }
1947
1948 /*
1949  * Constructor
1950  */
1951 static int
1952 linprocfs_init(PFS_INIT_ARGS)
1953 {
1954         struct pfs_node *root;
1955         struct pfs_node *dir;
1956         struct pfs_node *sys;
1957
1958         root = pi->pi_root;
1959
1960         /* /proc/... */
1961         pfs_create_file(root, "cmdline", &linprocfs_docmdline,
1962             NULL, NULL, NULL, PFS_RD);
1963         pfs_create_file(root, "cpuinfo", &linprocfs_docpuinfo,
1964             NULL, NULL, NULL, PFS_RD);
1965         pfs_create_file(root, "devices", &linprocfs_dodevices,
1966             NULL, NULL, NULL, PFS_RD);
1967         pfs_create_file(root, "filesystems", &linprocfs_dofilesystems,
1968             NULL, NULL, NULL, PFS_RD);
1969         pfs_create_file(root, "loadavg", &linprocfs_doloadavg,
1970             NULL, NULL, NULL, PFS_RD);
1971         pfs_create_file(root, "meminfo", &linprocfs_domeminfo,
1972             NULL, NULL, NULL, PFS_RD);
1973         pfs_create_file(root, "modules", &linprocfs_domodules,
1974             NULL, NULL, NULL, PFS_RD);
1975         pfs_create_file(root, "mounts", &linprocfs_domtab,
1976             NULL, NULL, NULL, PFS_RD);
1977         pfs_create_file(root, "mtab", &linprocfs_domtab,
1978             NULL, NULL, NULL, PFS_RD);
1979         pfs_create_file(root, "partitions", &linprocfs_dopartitions,
1980             NULL, NULL, NULL, PFS_RD);
1981         pfs_create_link(root, "self", &procfs_docurproc,
1982             NULL, NULL, NULL, 0);
1983         pfs_create_file(root, "stat", &linprocfs_dostat,
1984             NULL, NULL, NULL, PFS_RD);
1985         pfs_create_file(root, "swaps", &linprocfs_doswaps,
1986             NULL, NULL, NULL, PFS_RD);
1987         pfs_create_file(root, "uptime", &linprocfs_douptime,
1988             NULL, NULL, NULL, PFS_RD);
1989         pfs_create_file(root, "version", &linprocfs_doversion,
1990             NULL, NULL, NULL, PFS_RD);
1991
1992         /* /proc/bus/... */
1993         dir = pfs_create_dir(root, "bus", NULL, NULL, NULL, 0);
1994         dir = pfs_create_dir(dir, "pci", NULL, NULL, NULL, 0);
1995         dir = pfs_create_dir(dir, "devices", NULL, NULL, NULL, 0);
1996
1997         /* /proc/net/... */
1998         dir = pfs_create_dir(root, "net", NULL, NULL, NULL, 0);
1999         pfs_create_file(dir, "dev", &linprocfs_donetdev,
2000             NULL, NULL, NULL, PFS_RD);
2001
2002         /* /proc/<pid>/... */
2003         dir = pfs_create_dir(root, "pid", NULL, NULL, NULL, PFS_PROCDEP);
2004         pfs_create_file(dir, "cmdline", &linprocfs_doproccmdline,
2005             NULL, NULL, NULL, PFS_RD);
2006         pfs_create_link(dir, "cwd", &linprocfs_doproccwd,
2007             NULL, NULL, NULL, 0);
2008         pfs_create_file(dir, "environ", &linprocfs_doprocenviron,
2009             NULL, &procfs_candebug, NULL, PFS_RD);
2010         pfs_create_link(dir, "exe", &procfs_doprocfile,
2011             NULL, &procfs_notsystem, NULL, 0);
2012         pfs_create_file(dir, "maps", &linprocfs_doprocmaps,
2013             NULL, NULL, NULL, PFS_RD | PFS_AUTODRAIN);
2014         pfs_create_file(dir, "mem", &linprocfs_doprocmem,
2015             procfs_attr_rw, &procfs_candebug, NULL, PFS_RDWR | PFS_RAW);
2016         pfs_create_file(dir, "mountinfo", &linprocfs_doprocmountinfo,
2017             NULL, NULL, NULL, PFS_RD);
2018         pfs_create_file(dir, "mounts", &linprocfs_domtab,
2019             NULL, NULL, NULL, PFS_RD);
2020         pfs_create_link(dir, "root", &linprocfs_doprocroot,
2021             NULL, NULL, NULL, 0);
2022         pfs_create_file(dir, "stat", &linprocfs_doprocstat,
2023             NULL, NULL, NULL, PFS_RD);
2024         pfs_create_file(dir, "statm", &linprocfs_doprocstatm,
2025             NULL, NULL, NULL, PFS_RD);
2026         pfs_create_file(dir, "status", &linprocfs_doprocstatus,
2027             NULL, NULL, NULL, PFS_RD);
2028         pfs_create_link(dir, "fd", &linprocfs_dofdescfs,
2029             NULL, NULL, NULL, 0);
2030         pfs_create_file(dir, "auxv", &linprocfs_doauxv,
2031             NULL, &procfs_candebug, NULL, PFS_RD|PFS_RAWRD);
2032         pfs_create_file(dir, "limits", &linprocfs_doproclimits,
2033             NULL, NULL, NULL, PFS_RD);
2034
2035         /* /proc/<pid>/task/... */
2036         dir = pfs_create_dir(dir, "task", linprocfs_dotaskattr, NULL, NULL, 0);
2037         pfs_create_file(dir, ".dummy", &linprocfs_dotaskdummy,
2038             NULL, NULL, NULL, PFS_RD);
2039
2040         /* /proc/scsi/... */
2041         dir = pfs_create_dir(root, "scsi", NULL, NULL, NULL, 0);
2042         pfs_create_file(dir, "device_info", &linprocfs_doscsidevinfo,
2043             NULL, NULL, NULL, PFS_RD);
2044         pfs_create_file(dir, "scsi", &linprocfs_doscsiscsi,
2045             NULL, NULL, NULL, PFS_RD);
2046
2047         /* /proc/sys/... */
2048         sys = pfs_create_dir(root, "sys", NULL, NULL, NULL, 0);
2049
2050         /* /proc/sys/kernel/... */
2051         dir = pfs_create_dir(sys, "kernel", NULL, NULL, NULL, 0);
2052         pfs_create_file(dir, "osrelease", &linprocfs_doosrelease,
2053             NULL, NULL, NULL, PFS_RD);
2054         pfs_create_file(dir, "ostype", &linprocfs_doostype,
2055             NULL, NULL, NULL, PFS_RD);
2056         pfs_create_file(dir, "version", &linprocfs_doosbuild,
2057             NULL, NULL, NULL, PFS_RD);
2058         pfs_create_file(dir, "msgmax", &linprocfs_domsgmax,
2059             NULL, NULL, NULL, PFS_RD);
2060         pfs_create_file(dir, "msgmni", &linprocfs_domsgmni,
2061             NULL, NULL, NULL, PFS_RD);
2062         pfs_create_file(dir, "msgmnb", &linprocfs_domsgmnb,
2063             NULL, NULL, NULL, PFS_RD);
2064         pfs_create_file(dir, "ngroups_max", &linprocfs_dongroups_max,
2065             NULL, NULL, NULL, PFS_RD);
2066         pfs_create_file(dir, "pid_max", &linprocfs_dopid_max,
2067             NULL, NULL, NULL, PFS_RD);
2068         pfs_create_file(dir, "sem", &linprocfs_dosem,
2069             NULL, NULL, NULL, PFS_RD);
2070         pfs_create_file(dir, "shmall", &linprocfs_doshmall,
2071             NULL, NULL, NULL, PFS_RD);
2072         pfs_create_file(dir, "shmmax", &linprocfs_doshmmax,
2073             NULL, NULL, NULL, PFS_RD);
2074         pfs_create_file(dir, "shmmni", &linprocfs_doshmmni,
2075             NULL, NULL, NULL, PFS_RD);
2076         pfs_create_file(dir, "tainted", &linprocfs_dotainted,
2077             NULL, NULL, NULL, PFS_RD);
2078
2079         /* /proc/sys/kernel/random/... */
2080         dir = pfs_create_dir(dir, "random", NULL, NULL, NULL, 0);
2081         pfs_create_file(dir, "uuid", &linprocfs_douuid,
2082             NULL, NULL, NULL, PFS_RD);
2083
2084         /* /proc/sys/vm/.... */
2085         dir = pfs_create_dir(sys, "vm", NULL, NULL, NULL, 0);
2086         pfs_create_file(dir, "min_free_kbytes", &linprocfs_dominfree,
2087             NULL, NULL, NULL, PFS_RD);
2088
2089         return (0);
2090 }
2091
2092 /*
2093  * Destructor
2094  */
2095 static int
2096 linprocfs_uninit(PFS_INIT_ARGS)
2097 {
2098
2099         /* nothing to do, pseudofs will GC */
2100         return (0);
2101 }
2102
2103 PSEUDOFS(linprocfs, 1, VFCF_JAIL);
2104 #if defined(__aarch64__) || defined(__amd64__)
2105 MODULE_DEPEND(linprocfs, linux_common, 1, 1, 1);
2106 #else
2107 MODULE_DEPEND(linprocfs, linux, 1, 1, 1);
2108 #endif
2109 MODULE_DEPEND(linprocfs, procfs, 1, 1, 1);
2110 MODULE_DEPEND(linprocfs, sysvmsg, 1, 1, 1);
2111 MODULE_DEPEND(linprocfs, sysvsem, 1, 1, 1);
2112 MODULE_DEPEND(linprocfs, sysvshm, 1, 1, 1);