Home | History | Annotate | Line # | Download | only in procfs
procfs_linux.c revision 1.76
      1 /*      $NetBSD: procfs_linux.c,v 1.76 2019/09/07 19:08:28 chs Exp $      */
      2 
      3 /*
      4  * Copyright (c) 2001 Wasabi Systems, Inc.
      5  * All rights reserved.
      6  *
      7  * Written by Frank van der Linden for Wasabi Systems, Inc.
      8  *
      9  * Redistribution and use in source and binary forms, with or without
     10  * modification, are permitted provided that the following conditions
     11  * are met:
     12  * 1. Redistributions of source code must retain the above copyright
     13  *    notice, this list of conditions and the following disclaimer.
     14  * 2. Redistributions in binary form must reproduce the above copyright
     15  *    notice, this list of conditions and the following disclaimer in the
     16  *    documentation and/or other materials provided with the distribution.
     17  * 3. All advertising materials mentioning features or use of this software
     18  *    must display the following acknowledgement:
     19  *      This product includes software developed for the NetBSD Project by
     20  *      Wasabi Systems, Inc.
     21  * 4. The name of Wasabi Systems, Inc. may not be used to endorse
     22  *    or promote products derived from this software without specific prior
     23  *    written permission.
     24  *
     25  * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
     26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     27  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     28  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL WASABI SYSTEMS, INC
     29  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     30  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     31  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     32  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     33  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     34  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     35  * POSSIBILITY OF SUCH DAMAGE.
     36  */
     37 
     38 #include <sys/cdefs.h>
     39 __KERNEL_RCSID(0, "$NetBSD: procfs_linux.c,v 1.76 2019/09/07 19:08:28 chs Exp $");
     40 
     41 #include <sys/param.h>
     42 #include <sys/systm.h>
     43 #include <sys/time.h>
     44 #include <sys/cpu.h>
     45 #include <sys/kernel.h>
     46 #include <sys/proc.h>
     47 #include <sys/vnode.h>
     48 #include <sys/exec.h>
     49 #include <sys/resource.h>
     50 #include <sys/resourcevar.h>
     51 #include <sys/signal.h>
     52 #include <sys/signalvar.h>
     53 #include <sys/tty.h>
     54 #include <sys/malloc.h>
     55 #include <sys/mount.h>
     56 #include <sys/conf.h>
     57 #include <sys/sysctl.h>
     58 #include <sys/kauth.h>
     59 #include <sys/filedesc.h>
     60 
     61 #include <miscfs/procfs/procfs.h>
     62 
     63 #include <compat/linux/common/linux_exec.h>
     64 #include <compat/linux32/common/linux32_sysctl.h>
     65 
     66 #include <uvm/uvm_extern.h>
     67 #include <uvm/uvm.h>
     68 
     69 extern struct devsw_conv *devsw_conv;
     70 extern int max_devsw_convs;
     71 
     72 #define PGTOB(p)	((unsigned long)(p) << PAGE_SHIFT)
     73 #define PGTOKB(p)	((unsigned long)(p) << (PAGE_SHIFT - 10))
     74 
     75 #define LBFSZ (8 * 1024)
     76 
     77 static void
     78 get_proc_size_info(struct proc *p, struct vm_map *map, unsigned long *stext,
     79     unsigned long *etext, unsigned long *sstack)
     80 {
     81 	struct vm_map_entry *entry;
     82 
     83 	*stext = 0;
     84 	*etext = 0;
     85 	*sstack = 0;
     86 
     87 	vm_map_lock_read(map);
     88 
     89 	for (entry = map->header.next; entry != &map->header;
     90 	    entry = entry->next) {
     91 		if (UVM_ET_ISSUBMAP(entry))
     92 			continue;
     93 		/* assume text is the first entry */
     94 		if (*stext == *etext) {
     95 			*stext = entry->start;
     96 			*etext = entry->end;
     97 			break;
     98 		}
     99 	}
    100 #if defined(LINUX_USRSTACK32) && defined(USRSTACK32)
    101 	if (strcmp(p->p_emul->e_name, "linux32") == 0 &&
    102 	    LINUX_USRSTACK32 < USRSTACK32)
    103 		*sstack = (unsigned long)LINUX_USRSTACK32;
    104 	else
    105 #endif
    106 #ifdef LINUX_USRSTACK
    107 	if (strcmp(p->p_emul->e_name, "linux") == 0 &&
    108 	    LINUX_USRSTACK < USRSTACK)
    109 		*sstack = (unsigned long)LINUX_USRSTACK;
    110 	else
    111 #endif
    112 #ifdef	USRSTACK32
    113 	if (strstr(p->p_emul->e_name, "32") != NULL)
    114 		*sstack = (unsigned long)USRSTACK32;
    115 	else
    116 #endif
    117 		*sstack = (unsigned long)USRSTACK;
    118 
    119 	/*
    120 	 * jdk 1.6 compares low <= addr && addr < high
    121 	 * if we put addr == high, then the test fails
    122 	 * so eat one page.
    123 	 */
    124 	*sstack -= PAGE_SIZE;
    125 
    126 	vm_map_unlock_read(map);
    127 }
    128 
    129 /*
    130  * Linux compatible /proc/meminfo. Only active when the -o linux
    131  * mountflag is used.
    132  */
    133 int
    134 procfs_domeminfo(struct lwp *curl, struct proc *p,
    135     struct pfsnode *pfs, struct uio *uio)
    136 {
    137 	char *bf;
    138 	int len;
    139 	int error = 0;
    140 
    141 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    142 
    143 	len = snprintf(bf, LBFSZ,
    144 		"        total:    used:    free:  shared: buffers: cached:\n"
    145 		"Mem:  %8lu %8lu %8lu %8lu %8lu %8lu\n"
    146 		"Swap: %8lu %8lu %8lu\n"
    147 		"MemTotal:  %8lu kB\n"
    148 		"MemFree:   %8lu kB\n"
    149 		"MemShared: %8lu kB\n"
    150 		"Buffers:   %8lu kB\n"
    151 		"Cached:    %8lu kB\n"
    152 		"SwapTotal: %8lu kB\n"
    153 		"SwapFree:  %8lu kB\n",
    154 		PGTOB(uvmexp.npages),
    155 		PGTOB(uvmexp.npages - uvmexp.free),
    156 		PGTOB(uvmexp.free),
    157 		0L,
    158 		PGTOB(uvmexp.filepages),
    159 		PGTOB(uvmexp.anonpages + uvmexp.filepages + uvmexp.execpages),
    160 		PGTOB(uvmexp.swpages),
    161 		PGTOB(uvmexp.swpginuse),
    162 		PGTOB(uvmexp.swpages - uvmexp.swpginuse),
    163 		PGTOKB(uvmexp.npages),
    164 		PGTOKB(uvmexp.free),
    165 		0L,
    166 		PGTOKB(uvmexp.filepages),
    167 		PGTOKB(uvmexp.anonpages + uvmexp.filepages + uvmexp.execpages),
    168 		PGTOKB(uvmexp.swpages),
    169 		PGTOKB(uvmexp.swpages - uvmexp.swpginuse));
    170 
    171 	if (len == 0)
    172 		goto out;
    173 
    174 	error = uiomove_frombuf(bf, len, uio);
    175 out:
    176 	free(bf, M_TEMP);
    177 	return error;
    178 }
    179 
    180 /*
    181  * Linux compatible /proc/devices. Only active when the -o linux
    182  * mountflag is used.
    183  */
    184 int
    185 procfs_dodevices(struct lwp *curl, struct proc *p,
    186     struct pfsnode *pfs, struct uio *uio)
    187 {
    188 	char *bf;
    189 	int offset = 0;
    190 	int i, error = ENAMETOOLONG;
    191 
    192 	/* XXX elad - may need filtering. */
    193 
    194 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    195 
    196 	offset += snprintf(&bf[offset], LBFSZ - offset, "Character devices:\n");
    197 	if (offset >= LBFSZ)
    198 		goto out;
    199 
    200 	mutex_enter(&device_lock);
    201 	for (i = 0; i < max_devsw_convs; i++) {
    202 		if ((devsw_conv[i].d_name == NULL) ||
    203 		    (devsw_conv[i].d_cmajor == -1))
    204 			continue;
    205 
    206 		offset += snprintf(&bf[offset], LBFSZ - offset,
    207 		    "%3d %s\n", devsw_conv[i].d_cmajor, devsw_conv[i].d_name);
    208 		if (offset >= LBFSZ) {
    209 			mutex_exit(&device_lock);
    210 			goto out;
    211 		}
    212 	}
    213 
    214 	offset += snprintf(&bf[offset], LBFSZ - offset, "\nBlock devices:\n");
    215 	if (offset >= LBFSZ) {
    216 		mutex_exit(&device_lock);
    217 		goto out;
    218 	}
    219 
    220 	for (i = 0; i < max_devsw_convs; i++) {
    221 		if ((devsw_conv[i].d_name == NULL) ||
    222 		    (devsw_conv[i].d_bmajor == -1))
    223 			continue;
    224 
    225 		offset += snprintf(&bf[offset], LBFSZ - offset,
    226 		    "%3d %s\n", devsw_conv[i].d_bmajor, devsw_conv[i].d_name);
    227 		if (offset >= LBFSZ) {
    228 			mutex_exit(&device_lock);
    229 			goto out;
    230 		}
    231 	}
    232 	mutex_exit(&device_lock);
    233 
    234 	error = uiomove_frombuf(bf, offset, uio);
    235 out:
    236 	free(bf, M_TEMP);
    237 	return error;
    238 }
    239 
    240 /*
    241  * Linux compatible /proc/stat. Only active when the -o linux
    242  * mountflag is used.
    243  */
    244 int
    245 procfs_docpustat(struct lwp *curl, struct proc *p,
    246     struct pfsnode *pfs, struct uio *uio)
    247 {
    248 	char		*bf;
    249 	int	 	 error;
    250 	int	 	 len;
    251 #if defined(MULTIPROCESSOR)
    252         struct cpu_info *ci;
    253         CPU_INFO_ITERATOR cii;
    254 #endif
    255 	int	 	 i;
    256 	uint64_t	nintr;
    257 	uint64_t	nswtch;
    258 
    259 	error = ENAMETOOLONG;
    260 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    261 
    262 	len = snprintf(bf, LBFSZ,
    263 		"cpu %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64 "\n",
    264 		curcpu()->ci_schedstate.spc_cp_time[CP_USER],
    265 		curcpu()->ci_schedstate.spc_cp_time[CP_NICE],
    266 		curcpu()->ci_schedstate.spc_cp_time[CP_SYS] /*+ [CP_INTR]*/,
    267 		curcpu()->ci_schedstate.spc_cp_time[CP_IDLE]);
    268 	if (len == 0)
    269 		goto out;
    270 
    271 #if defined(MULTIPROCESSOR)
    272 #define ALLCPUS	CPU_INFO_FOREACH(cii, ci)
    273 #define CPUNAME	ci
    274 #else
    275 #define ALLCPUS	; i < 1 ;
    276 #define CPUNAME	curcpu()
    277 #endif
    278 
    279 	i = 0;
    280 	nintr = 0;
    281 	nswtch = 0;
    282 	for (ALLCPUS) {
    283 		len += snprintf(&bf[len], LBFSZ - len,
    284 			"cpu%d %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64
    285 			"\n", i,
    286 			CPUNAME->ci_schedstate.spc_cp_time[CP_USER],
    287 			CPUNAME->ci_schedstate.spc_cp_time[CP_NICE],
    288 			CPUNAME->ci_schedstate.spc_cp_time[CP_SYS],
    289 			CPUNAME->ci_schedstate.spc_cp_time[CP_IDLE]);
    290 		if (len >= LBFSZ)
    291 			goto out;
    292 		i += 1;
    293 		nintr += CPUNAME->ci_data.cpu_nintr;
    294 		nswtch += CPUNAME->ci_data.cpu_nswtch;
    295 	}
    296 
    297 	len += snprintf(&bf[len], LBFSZ - len,
    298 			"disk 0 0 0 0\n"
    299 			"page %u %u\n"
    300 			"swap %u %u\n"
    301 			"intr %"PRIu64"\n"
    302 			"ctxt %"PRIu64"\n"
    303 			"btime %"PRId64"\n",
    304 			uvmexp.pageins, uvmexp.pdpageouts,
    305 			uvmexp.pgswapin, uvmexp.pgswapout,
    306 			nintr,
    307 			nswtch,
    308 			boottime.tv_sec);
    309 	if (len >= LBFSZ)
    310 		goto out;
    311 
    312 	error = uiomove_frombuf(bf, len, uio);
    313 out:
    314 	free(bf, M_TEMP);
    315 	return error;
    316 }
    317 
    318 /*
    319  * Linux compatible /proc/loadavg. Only active when the -o linux
    320  * mountflag is used.
    321  */
    322 int
    323 procfs_doloadavg(struct lwp *curl, struct proc *p,
    324     struct pfsnode *pfs, struct uio *uio)
    325 {
    326 	char	*bf;
    327 	int 	 error;
    328 	int 	 len;
    329 
    330 	error = ENAMETOOLONG;
    331 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    332 
    333 	averunnable.fscale = FSCALE;
    334 	len = snprintf(bf, LBFSZ,
    335 	        "%d.%02d %d.%02d %d.%02d %d/%d %d\n",
    336 		(int)(averunnable.ldavg[0] / averunnable.fscale),
    337 		(int)(averunnable.ldavg[0] * 100 / averunnable.fscale % 100),
    338 		(int)(averunnable.ldavg[1] / averunnable.fscale),
    339 		(int)(averunnable.ldavg[1] * 100 / averunnable.fscale % 100),
    340 		(int)(averunnable.ldavg[2] / averunnable.fscale),
    341 		(int)(averunnable.ldavg[2] * 100 / averunnable.fscale % 100),
    342 		1,		/* number of ONPROC processes */
    343 		nprocs,
    344 		30000);		/* last pid */
    345 	if (len == 0)
    346 		goto out;
    347 
    348 	error = uiomove_frombuf(bf, len, uio);
    349 out:
    350 	free(bf, M_TEMP);
    351 	return error;
    352 }
    353 
    354 /*
    355  * Linux compatible /proc/<pid>/statm. Only active when the -o linux
    356  * mountflag is used.
    357  */
    358 int
    359 procfs_do_pid_statm(struct lwp *curl, struct lwp *l,
    360     struct pfsnode *pfs, struct uio *uio)
    361 {
    362 	struct vmspace	*vm;
    363 	struct proc	*p = l->l_proc;
    364 	char		*bf;
    365 	int	 	 error;
    366 	int	 	 len;
    367 	struct kinfo_proc2 ki;
    368 
    369 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    370 
    371 	/* XXX - we use values from vmspace, since dsl says that ru figures
    372 	   are always 0 except for zombies. See kvm_proc.c::kvm_getproc2() */
    373 	if ((error = proc_vmspace_getref(p, &vm)) != 0) {
    374 		goto out;
    375 	}
    376 
    377 	mutex_enter(proc_lock);
    378 	mutex_enter(p->p_lock);
    379 
    380 	/* retrieve RSS size */
    381 	memset(&ki, 0, sizeof(ki));
    382 	fill_kproc2(p, &ki, false, false);
    383 
    384 	mutex_exit(p->p_lock);
    385 	mutex_exit(proc_lock);
    386 
    387 	uvmspace_free(vm);
    388 
    389 	len = snprintf(bf, LBFSZ,
    390 	        "%lu %lu %lu %lu %lu %lu %lu\n",
    391 		(unsigned long)(ki.p_vm_msize),	/* size */
    392 		(unsigned long)(ki.p_vm_rssize),/* resident */
    393 		(unsigned long)(ki.p_uru_ixrss),/* shared */
    394 		(unsigned long)(ki.p_vm_tsize),	/* text */
    395 		(unsigned long) 0,		/* library (unused) */
    396 		(unsigned long)(ki.p_vm_dsize + ki.p_vm_ssize),	/* data+stack */
    397 		(unsigned long) 0);		/* dirty */
    398 
    399 	if (len == 0)
    400 		goto out;
    401 
    402 	error = uiomove_frombuf(bf, len, uio);
    403 out:
    404 	free(bf, M_TEMP);
    405 	return error;
    406 }
    407 
    408 #define UTIME2TICKS(s,u)	(((uint64_t)(s) * 1000000 + (u)) / 10000)
    409 
    410 /*
    411  * Linux compatible /proc/<pid>/stat. Only active when the -o linux
    412  * mountflag is used.
    413  */
    414 int
    415 procfs_do_pid_stat(struct lwp *curl, struct lwp *l,
    416     struct pfsnode *pfs, struct uio *uio)
    417 {
    418 	char *bf;
    419 	struct proc *p = l->l_proc;
    420 	int len;
    421 	struct rusage *cru = &p->p_stats->p_cru;
    422 	unsigned long stext = 0, etext = 0, sstack = 0;
    423 	struct timeval rt;
    424 	struct vmspace	*vm;
    425 	struct kinfo_proc2 ki;
    426 	int error;
    427 
    428 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    429 
    430 	if ((error = proc_vmspace_getref(p, &vm)) != 0) {
    431 		goto out;
    432 	}
    433 
    434 	get_proc_size_info(p, &vm->vm_map, &stext, &etext, &sstack);
    435 
    436 	mutex_enter(proc_lock);
    437 	mutex_enter(p->p_lock);
    438 
    439 	memset(&ki, 0, sizeof(ki));
    440 	fill_kproc2(p, &ki, false, false);
    441 	calcru(p, NULL, NULL, NULL, &rt);
    442 
    443 	len = snprintf(bf, LBFSZ,
    444 	    "%d (%s) %c %d %d %d %u %d "
    445 	    "%u "
    446 	    "%"PRIu64" %lu %"PRIu64" %lu %"PRIu64" %"PRIu64" %"PRIu64" %"PRIu64" "
    447 	    "%d %d %"PRIu64" "
    448 	    "%lld %"PRIu64" %"PRId64" %lu %"PRIu64" "
    449 	    "%lu %lu %lu "
    450 	    "%u %u "
    451 	    "%u %u %u %u "
    452 	    "%"PRIu64" %"PRIu64" %"PRIu64" %d %"PRIu64"\n",
    453 
    454 	    ki.p_pid,						/* 1 pid */
    455 	    ki.p_comm,						/* 2 tcomm */
    456 	    "0RRSTZXR8"[(ki.p_stat > 8) ? 0 : (int)ki.p_stat],	/* 3 state */
    457 	    ki.p_ppid,						/* 4 ppid */
    458 	    ki.p__pgid,						/* 5 pgrp */
    459 	    ki.p_sid,						/* 6 sid */
    460 	    (ki.p_tdev != (uint32_t)NODEV) ? ki.p_tdev : 0,	/* 7 tty_nr */
    461 	    ki.p_tpgid,						/* 8 tty_pgrp */
    462 
    463 	    ki.p_flag,						/* 9 flags */
    464 
    465 	    ki.p_uru_minflt,					/* 10 min_flt */
    466 	    cru->ru_minflt,
    467 	    ki.p_uru_majflt,					/* 12 maj_flt */
    468 	    cru->ru_majflt,
    469 	    UTIME2TICKS(ki.p_uutime_sec, ki.p_uutime_usec),	/* 14 utime */
    470 	    UTIME2TICKS(ki.p_ustime_sec, ki.p_ustime_usec),	/* 15 stime */
    471 	    UTIME2TICKS(cru->ru_utime.tv_sec, cru->ru_utime.tv_usec), /* 16 cutime */
    472 	    UTIME2TICKS(cru->ru_stime.tv_sec, cru->ru_stime.tv_usec), /* 17 cstime */
    473 
    474 	    ki.p_priority,				/* XXX: 18 priority */
    475 	    ki.p_nice - NZERO,				/* 19 nice */
    476 	    ki.p_nlwps,					/* 20 num_threads */
    477 
    478 	    (long long)rt.tv_sec,
    479 	    UTIME2TICKS(ki.p_ustart_sec, ki.p_ustart_usec), /* 22 start_time */
    480 	    ki.p_vm_msize,				/* 23 vsize */
    481 	    PGTOKB(ki.p_vm_rssize),			/* 24 rss */
    482 	    p->p_rlimit[RLIMIT_RSS].rlim_cur,		/* 25 rsslim */
    483 
    484 	    stext,					/* 26 start_code */
    485 	    etext,					/* 27 end_code */
    486 	    sstack,					/* 28 start_stack */
    487 
    488 	    0,						/* XXX: 29 esp */
    489 	    0,						/* XXX: 30 eip */
    490 
    491 	    ki.p_siglist.__bits[0],			/* XXX: 31 pending */
    492 	    0,						/* XXX: 32 blocked */
    493 	    ki.p_sigignore.__bits[0],		/* 33 sigign */
    494 	    ki.p_sigcatch.__bits[0],		/* 34 sigcatch */
    495 
    496 	    ki.p_wchan,					/* 35 wchan */
    497 	    ki.p_uru_nvcsw,
    498 	    ki.p_uru_nivcsw,
    499 	    ki.p_exitsig,				/* 38 exit_signal */
    500 	    ki.p_cpuid);				/* 39 task_cpu */
    501 
    502 	mutex_exit(p->p_lock);
    503 	mutex_exit(proc_lock);
    504 
    505 	uvmspace_free(vm);
    506 
    507 	if (len == 0)
    508 		goto out;
    509 
    510 	error = uiomove_frombuf(bf, len, uio);
    511 out:
    512 	free(bf, M_TEMP);
    513 	return error;
    514 }
    515 
    516 int
    517 procfs_docpuinfo(struct lwp *curl, struct proc *p,
    518     struct pfsnode *pfs, struct uio *uio)
    519 {
    520 	size_t len = LBFSZ;
    521 	char *bf = NULL;
    522 	int error;
    523 
    524 	do {
    525 		if (bf)
    526 			free(bf, M_TEMP);
    527 		bf = malloc(len, M_TEMP, M_WAITOK);
    528 	} while (procfs_getcpuinfstr(bf, &len) < 0);
    529 
    530 	if (len == 0) {
    531 		error = 0;
    532 		goto done;
    533 	}
    534 
    535 	error = uiomove_frombuf(bf, len, uio);
    536 done:
    537 	free(bf, M_TEMP);
    538 	return error;
    539 }
    540 
    541 int
    542 procfs_douptime(struct lwp *curl, struct proc *p,
    543     struct pfsnode *pfs, struct uio *uio)
    544 {
    545 	char *bf;
    546 	int len;
    547 	struct timeval runtime;
    548 	u_int64_t idle;
    549 	int error = 0;
    550 
    551 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    552 
    553 	microuptime(&runtime);
    554 	idle = curcpu()->ci_schedstate.spc_cp_time[CP_IDLE];
    555 	len = snprintf(bf, LBFSZ,
    556 	    "%lld.%02lu %" PRIu64 ".%02" PRIu64 "\n",
    557 	    (long long)runtime.tv_sec, (long)runtime.tv_usec / 10000,
    558 	    idle / hz, (((idle % hz) * 100) / hz) % 100);
    559 
    560 	if (len == 0)
    561 		goto out;
    562 
    563 	error = uiomove_frombuf(bf, len, uio);
    564 out:
    565 	free(bf, M_TEMP);
    566 	return error;
    567 }
    568 
    569 static int
    570 procfs_format_sfs(char **mtab, size_t *mlen, char *buf, size_t blen,
    571     const struct statvfs *sfs, struct lwp *curl, int suser)
    572 {
    573 	const char *fsname;
    574 
    575 	/* Linux uses different names for some filesystems */
    576 	fsname = sfs->f_fstypename;
    577 	if (strcmp(fsname, "procfs") == 0)
    578 		fsname = "proc";
    579 	else if (strcmp(fsname, "ext2fs") == 0)
    580 		fsname = "ext2";
    581 
    582 	blen = snprintf(buf, blen, "%s %s %s %s%s%s%s%s%s 0 0\n",
    583 	    sfs->f_mntfromname, sfs->f_mntonname, fsname,
    584 	    (sfs->f_flag & ST_RDONLY) ? "ro" : "rw",
    585 	    (sfs->f_flag & ST_NOSUID) ? ",nosuid" : "",
    586 	    (sfs->f_flag & ST_NOEXEC) ? ",noexec" : "",
    587 	    (sfs->f_flag & ST_NODEV) ? ",nodev" : "",
    588 	    (sfs->f_flag & ST_SYNCHRONOUS) ? ",sync" : "",
    589 	    (sfs->f_flag & ST_NOATIME) ? ",noatime" : "");
    590 
    591 	*mtab = realloc(*mtab, *mlen + blen, M_TEMP, M_WAITOK);
    592 	memcpy(*mtab + *mlen, buf, blen);
    593 	*mlen += blen;
    594 	return sfs->f_mntonname[0] == '/' && sfs->f_mntonname[1] == '\0';
    595 }
    596 
    597 int
    598 procfs_domounts(struct lwp *curl, struct proc *p,
    599     struct pfsnode *pfs, struct uio *uio)
    600 {
    601 	char *bf, *mtab = NULL;
    602 	size_t mtabsz = 0;
    603 	mount_iterator_t *iter;
    604 	struct mount *mp;
    605 	int error = 0, root = 0;
    606 	struct cwdinfo *cwdi = curl->l_proc->p_cwdi;
    607 
    608 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    609 
    610 	mountlist_iterator_init(&iter);
    611 	while ((mp = mountlist_iterator_next(iter)) != NULL) {
    612 		struct statvfs sfs;
    613 
    614 		if ((error = dostatvfs(mp, &sfs, curl, MNT_WAIT, 0)) == 0)
    615 			root |= procfs_format_sfs(&mtab, &mtabsz, bf, LBFSZ,
    616 			    &sfs, curl, 0);
    617 	}
    618 	mountlist_iterator_destroy(iter);
    619 
    620 	/*
    621 	 * If we are inside a chroot that is not itself a mount point,
    622 	 * fake a root entry.
    623 	 */
    624 	if (!root && cwdi->cwdi_rdir)
    625 		(void)procfs_format_sfs(&mtab, &mtabsz, bf, LBFSZ,
    626 		    &cwdi->cwdi_rdir->v_mount->mnt_stat, curl, 1);
    627 
    628 	free(bf, M_TEMP);
    629 
    630 	if (mtabsz > 0) {
    631 		error = uiomove_frombuf(mtab, mtabsz, uio);
    632 		free(mtab, M_TEMP);
    633 	}
    634 
    635 	return error;
    636 }
    637 
    638 /*
    639  * Linux compatible /proc/version. Only active when the -o linux
    640  * mountflag is used.
    641  */
    642 int
    643 procfs_doversion(struct lwp *curl, struct proc *p,
    644     struct pfsnode *pfs, struct uio *uio)
    645 {
    646 	char *bf;
    647 	char lostype[20], losrelease[20], lversion[80];
    648 	const char *postype, *posrelease, *pversion;
    649 	const char *emulname = curlwp->l_proc->p_emul->e_name;
    650 	int len;
    651 	int error = 0;
    652 	int nm[4];
    653 	size_t buflen;
    654 
    655 	CTASSERT(EMUL_LINUX_KERN_OSTYPE == EMUL_LINUX32_KERN_OSTYPE);
    656 	CTASSERT(EMUL_LINUX_KERN_OSRELEASE == EMUL_LINUX32_KERN_OSRELEASE);
    657 	CTASSERT(EMUL_LINUX_KERN_VERSION == EMUL_LINUX32_KERN_VERSION);
    658 
    659 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    660 
    661 	sysctl_lock(false);
    662 
    663 	if (strncmp(emulname, "linux", 5) == 0) {
    664 		/*
    665 		 * Lookup the emulation ostype, osrelease, and version.
    666 		 * Since compat_linux and compat_linux32 can be built as
    667 		 * modules, we use sysctl to obtain the values instead of
    668 		 * using the symbols directly.
    669 		 */
    670 
    671 		if (strcmp(emulname, "linux32") == 0) {
    672 			nm[0] = CTL_EMUL;
    673 			nm[1] = EMUL_LINUX32;
    674 			nm[2] = EMUL_LINUX32_KERN;
    675 		} else {
    676 			nm[0] = CTL_EMUL;
    677 			nm[1] = EMUL_LINUX;
    678 			nm[2] = EMUL_LINUX_KERN;
    679 		}
    680 
    681 		nm[3] = EMUL_LINUX_KERN_OSTYPE;
    682 		buflen = sizeof(lostype);
    683 		error = sysctl_dispatch(nm, __arraycount(nm),
    684 		    lostype, &buflen,
    685 		    NULL, 0, NULL, NULL, NULL);
    686 		if (error)
    687 			goto out;
    688 
    689 		nm[3] = EMUL_LINUX_KERN_OSRELEASE;
    690 		buflen = sizeof(losrelease);
    691 		error = sysctl_dispatch(nm, __arraycount(nm),
    692 		    losrelease, &buflen,
    693 		    NULL, 0, NULL, NULL, NULL);
    694 		if (error)
    695 			goto out;
    696 
    697 		nm[3] = EMUL_LINUX_KERN_VERSION;
    698 		buflen = sizeof(lversion);
    699 		error = sysctl_dispatch(nm, __arraycount(nm),
    700 		    lversion, &buflen,
    701 		    NULL, 0, NULL, NULL, NULL);
    702 		if (error)
    703 			goto out;
    704 
    705 		postype = lostype;
    706 		posrelease = losrelease;
    707 		pversion = lversion;
    708 	} else {
    709 		postype = ostype;
    710 		posrelease = osrelease;
    711 		strlcpy(lversion, version, sizeof(lversion));
    712 		if (strchr(lversion, '\n'))
    713 			*strchr(lversion, '\n') = '\0';
    714 		pversion = lversion;
    715 	}
    716 
    717 	len = snprintf(bf, LBFSZ,
    718 		"%s version %s (%s@localhost) (gcc version %s) %s\n",
    719 		postype, posrelease, emulname,
    720 #ifdef __VERSION__
    721 		__VERSION__,
    722 #else
    723 		"unknown",
    724 #endif
    725 		pversion);
    726 
    727 	if (len == 0)
    728 		goto out;
    729 
    730 	error = uiomove_frombuf(bf, len, uio);
    731 out:
    732 	free(bf, M_TEMP);
    733 	sysctl_unlock();
    734 	return error;
    735 }
    736