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