1 /* $NetBSD: procfs_linux.c,v 1.36 2007/05/24 05:33:08 dogcow 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 1.36 2007/05/24 05:33:08 dogcow Exp $"); 40 41 #include <sys/param.h> 42 #include <sys/systm.h> 43 #include <sys/time.h> 44 #include <sys/kernel.h> 45 #include <sys/proc.h> 46 #include <sys/vnode.h> 47 #include <sys/exec.h> 48 #include <sys/resource.h> 49 #include <sys/resourcevar.h> 50 #include <sys/signal.h> 51 #include <sys/signalvar.h> 52 #include <sys/tty.h> 53 #include <sys/malloc.h> 54 #include <sys/mount.h> 55 #include <sys/conf.h> 56 57 #include <miscfs/procfs/procfs.h> 58 #include <compat/linux/common/linux_exec.h> 59 60 #include <uvm/uvm_extern.h> 61 #include <uvm/uvm.h> 62 63 extern struct devsw_conv *devsw_conv; 64 extern int max_devsw_convs; 65 66 #define PGTOB(p) ((unsigned long)(p) << PAGE_SHIFT) 67 #define PGTOKB(p) ((unsigned long)(p) << (PAGE_SHIFT - 10)) 68 69 #define LBFSZ (8 * 1024) 70 71 static void 72 get_proc_size_info(struct lwp *l, unsigned long *stext, unsigned long *etext, unsigned long *sstack) 73 { 74 struct proc *p = l->l_proc; 75 struct vmspace *vm; 76 struct vm_map *map; 77 struct vm_map_entry *entry; 78 79 *stext = 0; 80 *etext = 0; 81 *sstack = 0; 82 83 proc_vmspace_getref(p, &vm); 84 map = &vm->vm_map; 85 vm_map_lock_read(map); 86 87 for (entry = map->header.next; entry != &map->header; 88 entry = entry->next) { 89 if (UVM_ET_ISSUBMAP(entry)) 90 continue; 91 /* assume text is the first entry */ 92 if (*stext == *etext) { 93 *stext = entry->start; 94 *etext = entry->end; 95 break; 96 } 97 } 98 #ifdef LINUX_USRSTACK 99 if (strcmp(p->p_emul->e_name, "linux") == 0 && 100 LINUX_USRSTACK < USRSTACK) 101 *sstack = (unsigned long) LINUX_USRSTACK; 102 else 103 #endif 104 *sstack = (unsigned long) USRSTACK; 105 106 /* 107 * jdk 1.6 compares low <= addr && addr < high 108 * if we put addr == high, then the test fails 109 * so eat one page. 110 */ 111 *sstack -= PAGE_SIZE; 112 113 vm_map_unlock_read(map); 114 uvmspace_free(vm); 115 } 116 117 /* 118 * Linux compatible /proc/meminfo. Only active when the -o linux 119 * mountflag is used. 120 */ 121 int 122 procfs_domeminfo(struct lwp *curl, struct proc *p, 123 struct pfsnode *pfs, struct uio *uio) 124 { 125 char *bf; 126 int len; 127 int error = 0; 128 129 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 130 131 len = snprintf(bf, LBFSZ, 132 " total: used: free: shared: buffers: cached:\n" 133 "Mem: %8lu %8lu %8lu %8lu %8lu %8lu\n" 134 "Swap: %8lu %8lu %8lu\n" 135 "MemTotal: %8lu kB\n" 136 "MemFree: %8lu kB\n" 137 "MemShared: %8lu kB\n" 138 "Buffers: %8lu kB\n" 139 "Cached: %8lu kB\n" 140 "SwapTotal: %8lu kB\n" 141 "SwapFree: %8lu kB\n", 142 PGTOB(uvmexp.npages), 143 PGTOB(uvmexp.npages - uvmexp.free), 144 PGTOB(uvmexp.free), 145 0L, 146 PGTOB(uvmexp.filepages), 147 PGTOB(uvmexp.anonpages + uvmexp.filepages + uvmexp.execpages), 148 PGTOB(uvmexp.swpages), 149 PGTOB(uvmexp.swpginuse), 150 PGTOB(uvmexp.swpages - uvmexp.swpginuse), 151 PGTOKB(uvmexp.npages), 152 PGTOKB(uvmexp.free), 153 0L, 154 PGTOKB(uvmexp.filepages), 155 PGTOKB(uvmexp.anonpages + uvmexp.filepages + uvmexp.execpages), 156 PGTOKB(uvmexp.swpages), 157 PGTOKB(uvmexp.swpages - uvmexp.swpginuse)); 158 159 if (len == 0) 160 goto out; 161 162 error = uiomove_frombuf(bf, len, uio); 163 out: 164 free(bf, M_TEMP); 165 return error; 166 } 167 168 /* 169 * Linux compatible /proc/devices. Only active when the -o linux 170 * mountflag is used. 171 */ 172 int 173 procfs_dodevices(struct lwp *curl, struct proc *p, 174 struct pfsnode *pfs, struct uio *uio) 175 { 176 char *bf; 177 int offset = 0; 178 int i, error = ENAMETOOLONG; 179 180 /* XXX elad - may need filtering. */ 181 182 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 183 184 offset += snprintf(&bf[offset], LBFSZ - offset, "Character devices:\n"); 185 if (offset >= LBFSZ) 186 goto out; 187 188 for (i = 0; i < max_devsw_convs; i++) { 189 if ((devsw_conv[i].d_name == NULL) || 190 (devsw_conv[i].d_cmajor == -1)) 191 continue; 192 193 offset += snprintf(&bf[offset], LBFSZ - offset, 194 "%3d %s\n", devsw_conv[i].d_cmajor, devsw_conv[i].d_name); 195 if (offset >= LBFSZ) 196 goto out; 197 } 198 199 offset += snprintf(&bf[offset], LBFSZ - offset, "\nBlock devices:\n"); 200 if (offset >= LBFSZ) 201 goto out; 202 203 for (i = 0; i < max_devsw_convs; i++) { 204 if ((devsw_conv[i].d_name == NULL) || 205 (devsw_conv[i].d_bmajor == -1)) 206 continue; 207 208 offset += snprintf(&bf[offset], LBFSZ - offset, 209 "%3d %s\n", devsw_conv[i].d_bmajor, devsw_conv[i].d_name); 210 if (offset >= LBFSZ) 211 goto out; 212 } 213 214 error = uiomove_frombuf(bf, offset, uio); 215 out: 216 free(bf, M_TEMP); 217 return error; 218 } 219 220 /* 221 * Linux compatible /proc/stat. Only active when the -o linux 222 * mountflag is used. 223 */ 224 int 225 procfs_docpustat(struct lwp *curl, struct proc *p, 226 struct pfsnode *pfs, struct uio *uio) 227 { 228 struct timeval runtime; 229 struct cpu_info *ci; 230 CPU_INFO_ITERATOR cii; 231 char *bf; 232 int error; 233 int len; 234 int i; 235 236 error = ENAMETOOLONG; 237 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 238 239 len = snprintf(bf, LBFSZ, 240 "cpu %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64 "\n", 241 curcpu()->ci_schedstate.spc_cp_time[CP_USER], 242 curcpu()->ci_schedstate.spc_cp_time[CP_NICE], 243 curcpu()->ci_schedstate.spc_cp_time[CP_SYS] /*+ [CP_INTR]*/, 244 curcpu()->ci_schedstate.spc_cp_time[CP_IDLE]); 245 if (len == 0) 246 goto out; 247 248 i = 0; 249 for (CPU_INFO_FOREACH(cii, ci)) { 250 len += snprintf(&bf[len], LBFSZ - len, 251 "cpu%d %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64 252 "\n", i, 253 ci->ci_schedstate.spc_cp_time[CP_USER], 254 ci->ci_schedstate.spc_cp_time[CP_NICE], 255 ci->ci_schedstate.spc_cp_time[CP_SYS], 256 ci->ci_schedstate.spc_cp_time[CP_IDLE]); 257 if (len >= LBFSZ) 258 goto out; 259 i += 1; 260 } 261 262 timersub(&curcpu()->ci_schedstate.spc_runtime, &boottime, &runtime); 263 len += snprintf(&bf[len], LBFSZ - len, 264 "disk 0 0 0 0\n" 265 "page %u %u\n" 266 "swap %u %u\n" 267 "intr %u\n" 268 "ctxt %u\n" 269 "btime %lld\n", 270 uvmexp.pageins, uvmexp.pdpageouts, 271 uvmexp.pgswapin, uvmexp.pgswapout, 272 uvmexp.intrs, 273 uvmexp.swtch, 274 (long long)boottime.tv_sec); 275 if (len >= LBFSZ) 276 goto out; 277 278 error = uiomove_frombuf(bf, len, uio); 279 out: 280 free(bf, M_TEMP); 281 return error; 282 } 283 284 /* 285 * Linux compatible /proc/loadavg. Only active when the -o linux 286 * mountflag is used. 287 */ 288 int 289 procfs_doloadavg(struct lwp *curl, struct proc *p, 290 struct pfsnode *pfs, struct uio *uio) 291 { 292 char *bf; 293 int error; 294 int len; 295 296 error = ENAMETOOLONG; 297 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 298 299 averunnable.fscale = FSCALE; 300 len = snprintf(bf, LBFSZ, 301 "%d.%02d %d.%02d %d.%02d %d/%d %d\n", 302 (int)(averunnable.ldavg[0] / averunnable.fscale), 303 (int)(averunnable.ldavg[0] * 100 / averunnable.fscale % 100), 304 (int)(averunnable.ldavg[1] / averunnable.fscale), 305 (int)(averunnable.ldavg[1] * 100 / averunnable.fscale % 100), 306 (int)(averunnable.ldavg[2] / averunnable.fscale), 307 (int)(averunnable.ldavg[2] * 100 / averunnable.fscale % 100), 308 1, /* number of ONPROC processes */ 309 nprocs, 310 30000); /* last pid */ 311 if (len == 0) 312 goto out; 313 314 error = uiomove_frombuf(bf, len, uio); 315 out: 316 free(bf, M_TEMP); 317 return error; 318 } 319 320 /* 321 * Linux compatible /proc/<pid>/statm. Only active when the -o linux 322 * mountflag is used. 323 */ 324 int 325 procfs_do_pid_statm(struct lwp *curl, struct lwp *l, 326 struct pfsnode *pfs, struct uio *uio) 327 { 328 unsigned long stext = 0, etext = 0, sstack = 0; 329 struct proc *p = l->l_proc; 330 struct rusage *ru = &p->p_stats->p_ru; 331 char *bf; 332 int error; 333 int len; 334 335 error = ENAMETOOLONG; 336 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 337 338 get_proc_size_info(l, &stext, &etext, &sstack); 339 340 len = snprintf(bf, LBFSZ, 341 "%lu %lu %lu %lu %lu %lu %lu\n", 342 (ru->ru_ixrss + ru->ru_idrss + ru->ru_isrss), /* size */ 343 ru->ru_maxrss, /* resident */ 344 ru->ru_ixrss, /* shared */ 345 stext, 346 etext, 347 sstack, 348 (unsigned long) 0); 349 if (len == 0) 350 goto out; 351 352 error = uiomove_frombuf(bf, len, uio); 353 out: 354 free(bf, M_TEMP); 355 return error; 356 } 357 358 /* 359 * Linux compatible /proc/<pid>/stat. Only active when the -o linux 360 * mountflag is used. 361 */ 362 int 363 procfs_do_pid_stat(struct lwp *curl, struct lwp *l, 364 struct pfsnode *pfs, struct uio *uio) 365 { 366 char *bf; 367 struct proc *p = l->l_proc; 368 int len; 369 struct tty *tty = p->p_session->s_ttyp; 370 struct rusage *ru = &p->p_stats->p_ru; 371 struct rusage *cru = &p->p_stats->p_cru; 372 unsigned long stext = 0, etext = 0, sstack = 0; 373 struct timeval rt; 374 int error = 0; 375 376 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 377 378 get_proc_size_info(l, &stext, &etext, &sstack); 379 380 mutex_enter(&proclist_lock); 381 mutex_enter(&p->p_mutex); 382 mutex_enter(&p->p_smutex); 383 384 calcru(p, NULL, NULL, NULL, &rt); 385 386 len = snprintf(bf, LBFSZ, 387 "%d (%s) %c %d %d %d %d %d " 388 "%u " 389 "%lu %lu %lu %lu %lu %lu %lu %lu " 390 "%d %d %d " 391 "%lu %lu %lu %lu %" PRIu64 " " 392 "%lu %lu %lu " 393 "%u %u " 394 "%u %u %u %u " 395 "%lu %lu %lu %d %d\n", 396 397 p->p_pid, 398 p->p_comm, 399 "0IR3SZD"[(p->p_stat > 6) ? 0 : (int)p->p_stat], 400 (p->p_pptr != NULL) ? p->p_pptr->p_pid : 0, 401 402 p->p_pgid, 403 p->p_session->s_sid, 404 tty ? tty->t_dev : 0, 405 (tty && tty->t_pgrp) ? tty->t_pgrp->pg_id : 0, 406 407 p->p_flag, 408 409 ru->ru_minflt, 410 cru->ru_minflt, 411 ru->ru_majflt, 412 cru->ru_majflt, 413 ru->ru_utime.tv_sec, 414 ru->ru_stime.tv_sec, 415 cru->ru_utime.tv_sec, 416 cru->ru_stime.tv_sec, 417 418 p->p_nice, /* XXX: priority */ 419 p->p_nice, 420 0, 421 422 rt.tv_sec, 423 p->p_stats->p_start.tv_sec, 424 ru->ru_ixrss + ru->ru_idrss + ru->ru_isrss, 425 ru->ru_maxrss, 426 p->p_rlimit[RLIMIT_RSS].rlim_cur, 427 428 stext, /* start code */ 429 etext, /* end code */ 430 sstack, /* mm start stack */ 431 0, /* XXX: pc */ 432 0, /* XXX: sp */ 433 p->p_sigpend.sp_set.__bits[0], /* XXX: pending */ 434 0, /* XXX: held */ 435 p->p_sigctx.ps_sigignore.__bits[0], /* ignored */ 436 p->p_sigctx.ps_sigcatch.__bits[0], /* caught */ 437 438 (unsigned long)(intptr_t)l->l_wchan, 439 ru->ru_nvcsw, 440 ru->ru_nivcsw, 441 p->p_exitsig, 442 0); /* XXX: processor */ 443 444 mutex_exit(&p->p_smutex); 445 mutex_exit(&p->p_mutex); 446 mutex_exit(&proclist_lock); 447 448 if (len == 0) 449 goto out; 450 451 error = uiomove_frombuf(bf, len, uio); 452 out: 453 free(bf, M_TEMP); 454 return error; 455 } 456 457 int 458 procfs_docpuinfo(struct lwp *curl, struct proc *p, 459 struct pfsnode *pfs, struct uio *uio) 460 { 461 int len = LBFSZ; 462 char *bf = malloc(len, M_TEMP, M_WAITOK); 463 int error; 464 465 if (procfs_getcpuinfstr(bf, &len) < 0) { 466 error = ENOSPC; 467 goto done; 468 } 469 470 if (len == 0) { 471 error = 0; 472 goto done; 473 } 474 475 error = uiomove_frombuf(bf, len, uio); 476 done: 477 free(bf, M_TEMP); 478 return error; 479 } 480 481 int 482 procfs_douptime(struct lwp *curl, struct proc *p, 483 struct pfsnode *pfs, struct uio *uio) 484 { 485 char *bf; 486 int len; 487 struct timeval runtime; 488 u_int64_t idle; 489 int error = 0; 490 491 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 492 493 timersub(&curcpu()->ci_schedstate.spc_runtime, &boottime, &runtime); 494 idle = curcpu()->ci_schedstate.spc_cp_time[CP_IDLE]; 495 len = snprintf(bf, LBFSZ, 496 "%lu.%02lu %" PRIu64 ".%02" PRIu64 "\n", 497 runtime.tv_sec, runtime.tv_usec / 10000, 498 idle / hz, (((idle % hz) * 100) / hz) % 100); 499 500 if (len == 0) 501 goto out; 502 503 error = uiomove_frombuf(bf, len, uio); 504 out: 505 free(bf, M_TEMP); 506 return error; 507 } 508 509 int 510 procfs_domounts(struct lwp *curl, struct proc *p, 511 struct pfsnode *pfs, struct uio *uio) 512 { 513 char *bf, *mtab = NULL; 514 const char *fsname; 515 size_t len, mtabsz = 0; 516 struct mount *mp, *nmp; 517 struct statvfs *sfs; 518 int error = 0; 519 520 /* XXX elad - may need filtering. */ 521 522 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 523 simple_lock(&mountlist_slock); 524 for (mp = CIRCLEQ_FIRST(&mountlist); mp != (void *)&mountlist; 525 mp = nmp) { 526 if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock)) { 527 nmp = CIRCLEQ_NEXT(mp, mnt_list); 528 continue; 529 } 530 531 sfs = &mp->mnt_stat; 532 533 /* Linux uses different names for some filesystems */ 534 fsname = sfs->f_fstypename; 535 if (strcmp(fsname, "procfs") == 0) 536 fsname = "proc"; 537 else if (strcmp(fsname, "ext2fs") == 0) 538 fsname = "ext2"; 539 540 len = snprintf(bf, LBFSZ, "%s %s %s %s%s%s%s%s%s 0 0\n", 541 sfs->f_mntfromname, 542 sfs->f_mntonname, 543 fsname, 544 (mp->mnt_flag & MNT_RDONLY) ? "ro" : "rw", 545 (mp->mnt_flag & MNT_NOSUID) ? ",nosuid" : "", 546 (mp->mnt_flag & MNT_NOEXEC) ? ",noexec" : "", 547 (mp->mnt_flag & MNT_NODEV) ? ",nodev" : "", 548 (mp->mnt_flag & MNT_SYNCHRONOUS) ? ",sync" : "", 549 (mp->mnt_flag & MNT_NOATIME) ? ",noatime" : "" 550 ); 551 552 mtab = realloc(mtab, mtabsz + len, M_TEMP, M_WAITOK); 553 memcpy(mtab + mtabsz, bf, len); 554 mtabsz += len; 555 556 simple_lock(&mountlist_slock); 557 nmp = CIRCLEQ_NEXT(mp, mnt_list); 558 vfs_unbusy(mp); 559 } 560 simple_unlock(&mountlist_slock); 561 free(bf, M_TEMP); 562 563 if (mtabsz > 0) { 564 error = uiomove_frombuf(mtab, mtabsz, uio); 565 free(mtab, M_TEMP); 566 } 567 568 return error; 569 } 570