Home | History | Annotate | Line # | Download | only in procfs
procfs_linux.c revision 1.72.2.1
      1 /*      $NetBSD: procfs_linux.c,v 1.72.2.1 2017/04/26 02:53:28 pgoyette 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 1.72.2.1 2017/04/26 02:53:28 pgoyette 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 lwp *l, unsigned long *stext, unsigned long *etext, unsigned long *sstack)
     79 {
     80 	struct proc *p = l->l_proc;
     81 	struct vmspace *vm;
     82 	struct vm_map *map;
     83 	struct vm_map_entry *entry;
     84 
     85 	*stext = 0;
     86 	*etext = 0;
     87 	*sstack = 0;
     88 
     89 	proc_vmspace_getref(p, &vm);
     90 	map = &vm->vm_map;
     91 	vm_map_lock_read(map);
     92 
     93 	for (entry = map->header.next; entry != &map->header;
     94 	    entry = entry->next) {
     95 		if (UVM_ET_ISSUBMAP(entry))
     96 			continue;
     97 		/* assume text is the first entry */
     98 		if (*stext == *etext) {
     99 			*stext = entry->start;
    100 			*etext = entry->end;
    101 			break;
    102 		}
    103 	}
    104 #if defined(LINUX_USRSTACK32) && defined(USRSTACK32)
    105 	if (strcmp(p->p_emul->e_name, "linux32") == 0 &&
    106 	    LINUX_USRSTACK32 < USRSTACK32)
    107 		*sstack = (unsigned long)LINUX_USRSTACK32;
    108 	else
    109 #endif
    110 #ifdef LINUX_USRSTACK
    111 	if (strcmp(p->p_emul->e_name, "linux") == 0 &&
    112 	    LINUX_USRSTACK < USRSTACK)
    113 		*sstack = (unsigned long)LINUX_USRSTACK;
    114 	else
    115 #endif
    116 #ifdef	USRSTACK32
    117 	if (strstr(p->p_emul->e_name, "32") != NULL)
    118 		*sstack = (unsigned long)USRSTACK32;
    119 	else
    120 #endif
    121 		*sstack = (unsigned long)USRSTACK;
    122 
    123 	/*
    124 	 * jdk 1.6 compares low <= addr && addr < high
    125 	 * if we put addr == high, then the test fails
    126 	 * so eat one page.
    127 	 */
    128 	*sstack -= PAGE_SIZE;
    129 
    130 	vm_map_unlock_read(map);
    131 	uvmspace_free(vm);
    132 }
    133 
    134 /*
    135  * Linux compatible /proc/meminfo. Only active when the -o linux
    136  * mountflag is used.
    137  */
    138 int
    139 procfs_domeminfo(struct lwp *curl, struct proc *p,
    140     struct pfsnode *pfs, struct uio *uio)
    141 {
    142 	char *bf;
    143 	int len;
    144 	int error = 0;
    145 
    146 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    147 
    148 	len = snprintf(bf, LBFSZ,
    149 		"        total:    used:    free:  shared: buffers: cached:\n"
    150 		"Mem:  %8lu %8lu %8lu %8lu %8lu %8lu\n"
    151 		"Swap: %8lu %8lu %8lu\n"
    152 		"MemTotal:  %8lu kB\n"
    153 		"MemFree:   %8lu kB\n"
    154 		"MemShared: %8lu kB\n"
    155 		"Buffers:   %8lu kB\n"
    156 		"Cached:    %8lu kB\n"
    157 		"SwapTotal: %8lu kB\n"
    158 		"SwapFree:  %8lu kB\n",
    159 		PGTOB(uvmexp.npages),
    160 		PGTOB(uvmexp.npages - uvmexp.free),
    161 		PGTOB(uvmexp.free),
    162 		0L,
    163 		PGTOB(uvmexp.filepages),
    164 		PGTOB(uvmexp.anonpages + uvmexp.filepages + uvmexp.execpages),
    165 		PGTOB(uvmexp.swpages),
    166 		PGTOB(uvmexp.swpginuse),
    167 		PGTOB(uvmexp.swpages - uvmexp.swpginuse),
    168 		PGTOKB(uvmexp.npages),
    169 		PGTOKB(uvmexp.free),
    170 		0L,
    171 		PGTOKB(uvmexp.filepages),
    172 		PGTOKB(uvmexp.anonpages + uvmexp.filepages + uvmexp.execpages),
    173 		PGTOKB(uvmexp.swpages),
    174 		PGTOKB(uvmexp.swpages - uvmexp.swpginuse));
    175 
    176 	if (len == 0)
    177 		goto out;
    178 
    179 	error = uiomove_frombuf(bf, len, uio);
    180 out:
    181 	free(bf, M_TEMP);
    182 	return error;
    183 }
    184 
    185 /*
    186  * Linux compatible /proc/devices. Only active when the -o linux
    187  * mountflag is used.
    188  */
    189 int
    190 procfs_dodevices(struct lwp *curl, struct proc *p,
    191     struct pfsnode *pfs, struct uio *uio)
    192 {
    193 	char *bf;
    194 	int offset = 0;
    195 	int i, error = ENAMETOOLONG;
    196 
    197 	/* XXX elad - may need filtering. */
    198 
    199 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    200 
    201 	offset += snprintf(&bf[offset], LBFSZ - offset, "Character devices:\n");
    202 	if (offset >= LBFSZ)
    203 		goto out;
    204 
    205 	mutex_enter(&device_lock);
    206 	for (i = 0; i < max_devsw_convs; i++) {
    207 		if ((devsw_conv[i].d_name == NULL) ||
    208 		    (devsw_conv[i].d_cmajor == -1))
    209 			continue;
    210 
    211 		offset += snprintf(&bf[offset], LBFSZ - offset,
    212 		    "%3d %s\n", devsw_conv[i].d_cmajor, devsw_conv[i].d_name);
    213 		if (offset >= LBFSZ) {
    214 			mutex_exit(&device_lock);
    215 			goto out;
    216 		}
    217 	}
    218 
    219 	offset += snprintf(&bf[offset], LBFSZ - offset, "\nBlock devices:\n");
    220 	if (offset >= LBFSZ) {
    221 		mutex_exit(&device_lock);
    222 		goto out;
    223 	}
    224 
    225 	for (i = 0; i < max_devsw_convs; i++) {
    226 		if ((devsw_conv[i].d_name == NULL) ||
    227 		    (devsw_conv[i].d_bmajor == -1))
    228 			continue;
    229 
    230 		offset += snprintf(&bf[offset], LBFSZ - offset,
    231 		    "%3d %s\n", devsw_conv[i].d_bmajor, devsw_conv[i].d_name);
    232 		if (offset >= LBFSZ) {
    233 			mutex_exit(&device_lock);
    234 			goto out;
    235 		}
    236 	}
    237 	mutex_exit(&device_lock);
    238 
    239 	error = uiomove_frombuf(bf, offset, uio);
    240 out:
    241 	free(bf, M_TEMP);
    242 	return error;
    243 }
    244 
    245 /*
    246  * Linux compatible /proc/stat. Only active when the -o linux
    247  * mountflag is used.
    248  */
    249 int
    250 procfs_docpustat(struct lwp *curl, struct proc *p,
    251     struct pfsnode *pfs, struct uio *uio)
    252 {
    253 	char		*bf;
    254 	int	 	 error;
    255 	int	 	 len;
    256 #if defined(MULTIPROCESSOR)
    257         struct cpu_info *ci;
    258         CPU_INFO_ITERATOR cii;
    259 #endif
    260 	int	 	 i;
    261 	uint64_t	nintr;
    262 	uint64_t	nswtch;
    263 
    264 	error = ENAMETOOLONG;
    265 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    266 
    267 	len = snprintf(bf, LBFSZ,
    268 		"cpu %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64 "\n",
    269 		curcpu()->ci_schedstate.spc_cp_time[CP_USER],
    270 		curcpu()->ci_schedstate.spc_cp_time[CP_NICE],
    271 		curcpu()->ci_schedstate.spc_cp_time[CP_SYS] /*+ [CP_INTR]*/,
    272 		curcpu()->ci_schedstate.spc_cp_time[CP_IDLE]);
    273 	if (len == 0)
    274 		goto out;
    275 
    276 #if defined(MULTIPROCESSOR)
    277 #define ALLCPUS	CPU_INFO_FOREACH(cii, ci)
    278 #define CPUNAME	ci
    279 #else
    280 #define ALLCPUS	; i < 1 ;
    281 #define CPUNAME	curcpu()
    282 #endif
    283 
    284 	i = 0;
    285 	nintr = 0;
    286 	nswtch = 0;
    287 	for (ALLCPUS) {
    288 		len += snprintf(&bf[len], LBFSZ - len,
    289 			"cpu%d %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64
    290 			"\n", i,
    291 			CPUNAME->ci_schedstate.spc_cp_time[CP_USER],
    292 			CPUNAME->ci_schedstate.spc_cp_time[CP_NICE],
    293 			CPUNAME->ci_schedstate.spc_cp_time[CP_SYS],
    294 			CPUNAME->ci_schedstate.spc_cp_time[CP_IDLE]);
    295 		if (len >= LBFSZ)
    296 			goto out;
    297 		i += 1;
    298 		nintr += CPUNAME->ci_data.cpu_nintr;
    299 		nswtch += CPUNAME->ci_data.cpu_nswtch;
    300 	}
    301 
    302 	len += snprintf(&bf[len], LBFSZ - len,
    303 			"disk 0 0 0 0\n"
    304 			"page %u %u\n"
    305 			"swap %u %u\n"
    306 			"intr %"PRIu64"\n"
    307 			"ctxt %"PRIu64"\n"
    308 			"btime %"PRId64"\n",
    309 			uvmexp.pageins, uvmexp.pdpageouts,
    310 			uvmexp.pgswapin, uvmexp.pgswapout,
    311 			nintr,
    312 			nswtch,
    313 			boottime.tv_sec);
    314 	if (len >= LBFSZ)
    315 		goto out;
    316 
    317 	error = uiomove_frombuf(bf, len, uio);
    318 out:
    319 	free(bf, M_TEMP);
    320 	return error;
    321 }
    322 
    323 /*
    324  * Linux compatible /proc/loadavg. Only active when the -o linux
    325  * mountflag is used.
    326  */
    327 int
    328 procfs_doloadavg(struct lwp *curl, struct proc *p,
    329     struct pfsnode *pfs, struct uio *uio)
    330 {
    331 	char	*bf;
    332 	int 	 error;
    333 	int 	 len;
    334 
    335 	error = ENAMETOOLONG;
    336 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    337 
    338 	averunnable.fscale = FSCALE;
    339 	len = snprintf(bf, LBFSZ,
    340 	        "%d.%02d %d.%02d %d.%02d %d/%d %d\n",
    341 		(int)(averunnable.ldavg[0] / averunnable.fscale),
    342 		(int)(averunnable.ldavg[0] * 100 / averunnable.fscale % 100),
    343 		(int)(averunnable.ldavg[1] / averunnable.fscale),
    344 		(int)(averunnable.ldavg[1] * 100 / averunnable.fscale % 100),
    345 		(int)(averunnable.ldavg[2] / averunnable.fscale),
    346 		(int)(averunnable.ldavg[2] * 100 / averunnable.fscale % 100),
    347 		1,		/* number of ONPROC processes */
    348 		nprocs,
    349 		30000);		/* last pid */
    350 	if (len == 0)
    351 		goto out;
    352 
    353 	error = uiomove_frombuf(bf, len, uio);
    354 out:
    355 	free(bf, M_TEMP);
    356 	return error;
    357 }
    358 
    359 /*
    360  * Linux compatible /proc/<pid>/statm. Only active when the -o linux
    361  * mountflag is used.
    362  */
    363 int
    364 procfs_do_pid_statm(struct lwp *curl, struct lwp *l,
    365     struct pfsnode *pfs, struct uio *uio)
    366 {
    367 	struct vmspace	*vm;
    368 	struct proc	*p = l->l_proc;
    369 	char		*bf;
    370 	int	 	 error;
    371 	int	 	 len;
    372 	struct kinfo_proc2 ki;
    373 
    374 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    375 
    376 	/* XXX - we use values from vmspace, since dsl says that ru figures
    377 	   are always 0 except for zombies. See kvm_proc.c::kvm_getproc2() */
    378 	if ((error = proc_vmspace_getref(p, &vm)) != 0) {
    379 		goto out;
    380 	}
    381 
    382 	mutex_enter(proc_lock);
    383 	mutex_enter(p->p_lock);
    384 
    385 	/* retrieve RSS size */
    386 	fill_kproc2(p, &ki, false);
    387 
    388 	mutex_exit(p->p_lock);
    389 	mutex_exit(proc_lock);
    390 
    391 	uvmspace_free(vm);
    392 
    393 	len = snprintf(bf, LBFSZ,
    394 	        "%lu %lu %lu %lu %lu %lu %lu\n",
    395 		(unsigned long)(ki.p_vm_msize),	/* size */
    396 		(unsigned long)(ki.p_vm_rssize),/* resident */
    397 		(unsigned long)(ki.p_uru_ixrss),/* shared */
    398 		(unsigned long)(ki.p_vm_tsize),	/* text */
    399 		(unsigned long) 0,		/* library (unused) */
    400 		(unsigned long)(ki.p_vm_dsize + ki.p_vm_ssize),	/* data+stack */
    401 		(unsigned long) 0);		/* dirty */
    402 
    403 	if (len == 0)
    404 		goto out;
    405 
    406 	error = uiomove_frombuf(bf, len, uio);
    407 out:
    408 	free(bf, M_TEMP);
    409 	return error;
    410 }
    411 
    412 #define UTIME2TICKS(s,u)	(((uint64_t)(s) * 1000000 + (u)) / 10000)
    413 
    414 /*
    415  * Linux compatible /proc/<pid>/stat. Only active when the -o linux
    416  * mountflag is used.
    417  */
    418 int
    419 procfs_do_pid_stat(struct lwp *curl, struct lwp *l,
    420     struct pfsnode *pfs, struct uio *uio)
    421 {
    422 	char *bf;
    423 	struct proc *p = l->l_proc;
    424 	int len;
    425 	struct rusage *cru = &p->p_stats->p_cru;
    426 	unsigned long stext = 0, etext = 0, sstack = 0;
    427 	struct timeval rt;
    428 	struct vmspace	*vm;
    429 	struct kinfo_proc2 ki;
    430 	int error;
    431 
    432 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    433 
    434 	if ((error = proc_vmspace_getref(p, &vm)) != 0) {
    435 		goto out;
    436 	}
    437 
    438 	get_proc_size_info(l, &stext, &etext, &sstack);
    439 
    440 	mutex_enter(proc_lock);
    441 	mutex_enter(p->p_lock);
    442 
    443 	fill_kproc2(p, &ki, false);
    444 	calcru(p, NULL, NULL, NULL, &rt);
    445 
    446 	len = snprintf(bf, LBFSZ,
    447 	    "%d (%s) %c %d %d %d %u %d "
    448 	    "%u "
    449 	    "%"PRIu64" %lu %"PRIu64" %lu %"PRIu64" %"PRIu64" %"PRIu64" %"PRIu64" "
    450 	    "%d %d %"PRIu64" "
    451 	    "%lld %"PRIu64" %"PRId64" %lu %"PRIu64" "
    452 	    "%lu %lu %lu "
    453 	    "%u %u "
    454 	    "%u %u %u %u "
    455 	    "%"PRIu64" %"PRIu64" %"PRIu64" %d %"PRIu64"\n",
    456 
    457 	    ki.p_pid,						/* 1 pid */
    458 	    ki.p_comm,						/* 2 tcomm */
    459 	    "0RRSTZXR8"[(ki.p_stat > 8) ? 0 : (int)ki.p_stat],	/* 3 state */
    460 	    ki.p_ppid,						/* 4 ppid */
    461 	    ki.p__pgid,						/* 5 pgrp */
    462 	    ki.p_sid,						/* 6 sid */
    463 	    (ki.p_tdev != (uint32_t)NODEV) ? ki.p_tdev : 0,	/* 7 tty_nr */
    464 	    ki.p_tpgid,						/* 8 tty_pgrp */
    465 
    466 	    ki.p_flag,						/* 9 flags */
    467 
    468 	    ki.p_uru_minflt,					/* 10 min_flt */
    469 	    cru->ru_minflt,
    470 	    ki.p_uru_majflt,					/* 12 maj_flt */
    471 	    cru->ru_majflt,
    472 	    UTIME2TICKS(ki.p_uutime_sec, ki.p_uutime_usec),	/* 14 utime */
    473 	    UTIME2TICKS(ki.p_ustime_sec, ki.p_ustime_usec),	/* 15 stime */
    474 	    UTIME2TICKS(cru->ru_utime.tv_sec, cru->ru_utime.tv_usec), /* 16 cutime */
    475 	    UTIME2TICKS(cru->ru_stime.tv_sec, cru->ru_stime.tv_usec), /* 17 cstime */
    476 
    477 	    ki.p_priority,				/* XXX: 18 priority */
    478 	    ki.p_nice - NZERO,				/* 19 nice */
    479 	    ki.p_nlwps,					/* 20 num_threads */
    480 
    481 	    (long long)rt.tv_sec,
    482 	    UTIME2TICKS(ki.p_ustart_sec, ki.p_ustart_usec), /* 22 start_time */
    483 	    ki.p_vm_msize,				/* 23 vsize */
    484 	    PGTOKB(ki.p_vm_rssize),			/* 24 rss */
    485 	    p->p_rlimit[RLIMIT_RSS].rlim_cur,		/* 25 rsslim */
    486 
    487 	    stext,					/* 26 start_code */
    488 	    etext,					/* 27 end_code */
    489 	    sstack,					/* 28 start_stack */
    490 
    491 	    0,						/* XXX: 29 esp */
    492 	    0,						/* XXX: 30 eip */
    493 
    494 	    ki.p_siglist.__bits[0],			/* XXX: 31 pending */
    495 	    0,						/* XXX: 32 blocked */
    496 	    ki.p_sigignore.__bits[0],		/* 33 sigign */
    497 	    ki.p_sigcatch.__bits[0],		/* 34 sigcatch */
    498 
    499 	    ki.p_wchan,					/* 35 wchan */
    500 	    ki.p_uru_nvcsw,
    501 	    ki.p_uru_nivcsw,
    502 	    ki.p_exitsig,				/* 38 exit_signal */
    503 	    ki.p_cpuid);				/* 39 task_cpu */
    504 
    505 	mutex_exit(p->p_lock);
    506 	mutex_exit(proc_lock);
    507 
    508 	uvmspace_free(vm);
    509 
    510 	if (len == 0)
    511 		goto out;
    512 
    513 	error = uiomove_frombuf(bf, len, uio);
    514 out:
    515 	free(bf, M_TEMP);
    516 	return error;
    517 }
    518 
    519 int
    520 procfs_docpuinfo(struct lwp *curl, struct proc *p,
    521     struct pfsnode *pfs, struct uio *uio)
    522 {
    523 	size_t len = LBFSZ;
    524 	char *bf = NULL;
    525 	int error;
    526 
    527 	do {
    528 		if (bf)
    529 			free(bf, M_TEMP);
    530 		bf = malloc(len, M_TEMP, M_WAITOK);
    531 	} while (procfs_getcpuinfstr(bf, &len) < 0);
    532 
    533 	if (len == 0) {
    534 		error = 0;
    535 		goto done;
    536 	}
    537 
    538 	error = uiomove_frombuf(bf, len, uio);
    539 done:
    540 	free(bf, M_TEMP);
    541 	return error;
    542 }
    543 
    544 int
    545 procfs_douptime(struct lwp *curl, struct proc *p,
    546     struct pfsnode *pfs, struct uio *uio)
    547 {
    548 	char *bf;
    549 	int len;
    550 	struct timeval runtime;
    551 	u_int64_t idle;
    552 	int error = 0;
    553 
    554 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    555 
    556 	microuptime(&runtime);
    557 	idle = curcpu()->ci_schedstate.spc_cp_time[CP_IDLE];
    558 	len = snprintf(bf, LBFSZ,
    559 	    "%lld.%02lu %" PRIu64 ".%02" PRIu64 "\n",
    560 	    (long long)runtime.tv_sec, (long)runtime.tv_usec / 10000,
    561 	    idle / hz, (((idle % hz) * 100) / hz) % 100);
    562 
    563 	if (len == 0)
    564 		goto out;
    565 
    566 	error = uiomove_frombuf(bf, len, uio);
    567 out:
    568 	free(bf, M_TEMP);
    569 	return error;
    570 }
    571 
    572 static int
    573 procfs_format_sfs(char **mtab, size_t *mlen, char *buf, size_t blen,
    574     const struct statvfs *sfs, struct lwp *curl, int suser)
    575 {
    576 	const char *fsname;
    577 
    578 	/* Linux uses different names for some filesystems */
    579 	fsname = sfs->f_fstypename;
    580 	if (strcmp(fsname, "procfs") == 0)
    581 		fsname = "proc";
    582 	else if (strcmp(fsname, "ext2fs") == 0)
    583 		fsname = "ext2";
    584 
    585 	blen = snprintf(buf, blen, "%s %s %s %s%s%s%s%s%s 0 0\n",
    586 	    sfs->f_mntfromname, sfs->f_mntonname, fsname,
    587 	    (sfs->f_flag & ST_RDONLY) ? "ro" : "rw",
    588 	    (sfs->f_flag & ST_NOSUID) ? ",nosuid" : "",
    589 	    (sfs->f_flag & ST_NOEXEC) ? ",noexec" : "",
    590 	    (sfs->f_flag & ST_NODEV) ? ",nodev" : "",
    591 	    (sfs->f_flag & ST_SYNCHRONOUS) ? ",sync" : "",
    592 	    (sfs->f_flag & ST_NOATIME) ? ",noatime" : "");
    593 
    594 	*mtab = realloc(*mtab, *mlen + blen, M_TEMP, M_WAITOK);
    595 	memcpy(*mtab + *mlen, buf, blen);
    596 	*mlen += blen;
    597 	return sfs->f_mntonname[0] == '/' && sfs->f_mntonname[1] == '\0';
    598 }
    599 
    600 int
    601 procfs_domounts(struct lwp *curl, struct proc *p,
    602     struct pfsnode *pfs, struct uio *uio)
    603 {
    604 	char *bf, *mtab = NULL;
    605 	size_t mtabsz = 0;
    606 	mount_iterator_t *iter;
    607 	struct mount *mp;
    608 	int error = 0, root = 0;
    609 	struct cwdinfo *cwdi = curl->l_proc->p_cwdi;
    610 
    611 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    612 
    613 	mountlist_iterator_init(&iter);
    614 	while ((mp = mountlist_iterator_next(iter)) != NULL) {
    615 		struct statvfs sfs;
    616 
    617 		if ((error = dostatvfs(mp, &sfs, curl, MNT_WAIT, 0)) == 0)
    618 			root |= procfs_format_sfs(&mtab, &mtabsz, bf, LBFSZ,
    619 			    &sfs, curl, 0);
    620 	}
    621 	mountlist_iterator_destroy(iter);
    622 
    623 	/*
    624 	 * If we are inside a chroot that is not itself a mount point,
    625 	 * fake a root entry.
    626 	 */
    627 	if (!root && cwdi->cwdi_rdir)
    628 		(void)procfs_format_sfs(&mtab, &mtabsz, bf, LBFSZ,
    629 		    &cwdi->cwdi_rdir->v_mount->mnt_stat, curl, 1);
    630 
    631 	free(bf, M_TEMP);
    632 
    633 	if (mtabsz > 0) {
    634 		error = uiomove_frombuf(mtab, mtabsz, uio);
    635 		free(mtab, M_TEMP);
    636 	}
    637 
    638 	return error;
    639 }
    640 
    641 /*
    642  * Linux compatible /proc/version. Only active when the -o linux
    643  * mountflag is used.
    644  */
    645 int
    646 procfs_doversion(struct lwp *curl, struct proc *p,
    647     struct pfsnode *pfs, struct uio *uio)
    648 {
    649 	char *bf;
    650 	char lostype[20], losrelease[20], lversion[80];
    651 	const char *postype, *posrelease, *pversion;
    652 	const char *emulname = curlwp->l_proc->p_emul->e_name;
    653 	int len;
    654 	int error = 0;
    655 	int nm[4];
    656 	size_t buflen;
    657 
    658 	CTASSERT(EMUL_LINUX_KERN_OSTYPE == EMUL_LINUX32_KERN_OSTYPE);
    659 	CTASSERT(EMUL_LINUX_KERN_OSRELEASE == EMUL_LINUX32_KERN_OSRELEASE);
    660 	CTASSERT(EMUL_LINUX_KERN_VERSION == EMUL_LINUX32_KERN_VERSION);
    661 
    662 	bf = malloc(LBFSZ, M_TEMP, M_WAITOK);
    663 
    664 	sysctl_lock(false);
    665 
    666 	if (strncmp(emulname, "linux", 5) == 0) {
    667 		/*
    668 		 * Lookup the emulation ostype, osrelease, and version.
    669 		 * Since compat_linux and compat_linux32 can be built as
    670 		 * modules, we use sysctl to obtain the values instead of
    671 		 * using the symbols directly.
    672 		 */
    673 
    674 		if (strcmp(emulname, "linux32") == 0) {
    675 			nm[0] = CTL_EMUL;
    676 			nm[1] = EMUL_LINUX32;
    677 			nm[2] = EMUL_LINUX32_KERN;
    678 		} else {
    679 			nm[0] = CTL_EMUL;
    680 			nm[1] = EMUL_LINUX;
    681 			nm[2] = EMUL_LINUX_KERN;
    682 		}
    683 
    684 		nm[3] = EMUL_LINUX_KERN_OSTYPE;
    685 		buflen = sizeof(lostype);
    686 		error = sysctl_dispatch(nm, __arraycount(nm),
    687 		    lostype, &buflen,
    688 		    NULL, 0, NULL, NULL, NULL);
    689 		if (error)
    690 			goto out;
    691 
    692 		nm[3] = EMUL_LINUX_KERN_OSRELEASE;
    693 		buflen = sizeof(losrelease);
    694 		error = sysctl_dispatch(nm, __arraycount(nm),
    695 		    losrelease, &buflen,
    696 		    NULL, 0, NULL, NULL, NULL);
    697 		if (error)
    698 			goto out;
    699 
    700 		nm[3] = EMUL_LINUX_KERN_VERSION;
    701 		buflen = sizeof(lversion);
    702 		error = sysctl_dispatch(nm, __arraycount(nm),
    703 		    lversion, &buflen,
    704 		    NULL, 0, NULL, NULL, NULL);
    705 		if (error)
    706 			goto out;
    707 
    708 		postype = lostype;
    709 		posrelease = losrelease;
    710 		pversion = lversion;
    711 	} else {
    712 		postype = ostype;
    713 		posrelease = osrelease;
    714 		strlcpy(lversion, version, sizeof(lversion));
    715 		if (strchr(lversion, '\n'))
    716 			*strchr(lversion, '\n') = '\0';
    717 		pversion = lversion;
    718 	}
    719 
    720 	len = snprintf(bf, LBFSZ,
    721 		"%s version %s (%s@localhost) (gcc version %s) %s\n",
    722 		postype, posrelease, emulname,
    723 #ifdef __VERSION__
    724 		__VERSION__,
    725 #else
    726 		"unknown",
    727 #endif
    728 		pversion);
    729 
    730 	if (len == 0)
    731 		goto out;
    732 
    733 	error = uiomove_frombuf(bf, len, uio);
    734 out:
    735 	free(bf, M_TEMP);
    736 	sysctl_unlock();
    737 	return error;
    738 }
    739