1 /* $NetBSD: procfs_linux.c,v 1.39 2007/05/26 16:21:04 agc 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 1.39 2007/05/26 16:21:04 agc 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 char *bf; 230 int error; 231 int len; 232 #if defined(MULTIPROCESSOR) 233 struct cpu_info *ci; 234 CPU_INFO_ITERATOR cii; 235 #endif 236 int i; 237 238 error = ENAMETOOLONG; 239 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 240 241 len = snprintf(bf, LBFSZ, 242 "cpu %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64 "\n", 243 curcpu()->ci_schedstate.spc_cp_time[CP_USER], 244 curcpu()->ci_schedstate.spc_cp_time[CP_NICE], 245 curcpu()->ci_schedstate.spc_cp_time[CP_SYS] /*+ [CP_INTR]*/, 246 curcpu()->ci_schedstate.spc_cp_time[CP_IDLE]); 247 if (len == 0) 248 goto out; 249 250 #if defined(MULTIPROCESSOR) 251 #define ALLCPUS CPU_INFO_FOREACH(cii, ci) 252 #define CPUNAME ci 253 #else 254 #define ALLCPUS ; i < 1 ; 255 #define CPUNAME curcpu() 256 #endif 257 258 i = 0; 259 for (ALLCPUS) { 260 len += snprintf(&bf[len], LBFSZ - len, 261 "cpu%d %" PRIu64 " %" PRIu64 " %" PRIu64 " %" PRIu64 262 "\n", i, 263 CPUNAME->ci_schedstate.spc_cp_time[CP_USER], 264 CPUNAME->ci_schedstate.spc_cp_time[CP_NICE], 265 CPUNAME->ci_schedstate.spc_cp_time[CP_SYS], 266 CPUNAME->ci_schedstate.spc_cp_time[CP_IDLE]); 267 if (len >= LBFSZ) 268 goto out; 269 i += 1; 270 } 271 272 timersub(&curcpu()->ci_schedstate.spc_runtime, &boottime, &runtime); 273 len += snprintf(&bf[len], LBFSZ - len, 274 "disk 0 0 0 0\n" 275 "page %u %u\n" 276 "swap %u %u\n" 277 "intr %u\n" 278 "ctxt %u\n" 279 "btime %lld\n", 280 uvmexp.pageins, uvmexp.pdpageouts, 281 uvmexp.pgswapin, uvmexp.pgswapout, 282 uvmexp.intrs, 283 uvmexp.swtch, 284 (long long)boottime.tv_sec); 285 if (len >= LBFSZ) 286 goto out; 287 288 error = uiomove_frombuf(bf, len, uio); 289 out: 290 free(bf, M_TEMP); 291 return error; 292 } 293 294 /* 295 * Linux compatible /proc/loadavg. Only active when the -o linux 296 * mountflag is used. 297 */ 298 int 299 procfs_doloadavg(struct lwp *curl, struct proc *p, 300 struct pfsnode *pfs, struct uio *uio) 301 { 302 char *bf; 303 int error; 304 int len; 305 306 error = ENAMETOOLONG; 307 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 308 309 averunnable.fscale = FSCALE; 310 len = snprintf(bf, LBFSZ, 311 "%d.%02d %d.%02d %d.%02d %d/%d %d\n", 312 (int)(averunnable.ldavg[0] / averunnable.fscale), 313 (int)(averunnable.ldavg[0] * 100 / averunnable.fscale % 100), 314 (int)(averunnable.ldavg[1] / averunnable.fscale), 315 (int)(averunnable.ldavg[1] * 100 / averunnable.fscale % 100), 316 (int)(averunnable.ldavg[2] / averunnable.fscale), 317 (int)(averunnable.ldavg[2] * 100 / averunnable.fscale % 100), 318 1, /* number of ONPROC processes */ 319 nprocs, 320 30000); /* last pid */ 321 if (len == 0) 322 goto out; 323 324 error = uiomove_frombuf(bf, len, uio); 325 out: 326 free(bf, M_TEMP); 327 return error; 328 } 329 330 /* 331 * Linux compatible /proc/<pid>/statm. Only active when the -o linux 332 * mountflag is used. 333 */ 334 int 335 procfs_do_pid_statm(struct lwp *curl, struct lwp *l, 336 struct pfsnode *pfs, struct uio *uio) 337 { 338 struct vmspace *vm; 339 struct proc *p = l->l_proc; 340 struct rusage *ru = &p->p_stats->p_ru; 341 char *bf; 342 int error; 343 int len; 344 345 error = ENAMETOOLONG; 346 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 347 348 /* XXX - we use values from vmspace, since dsl says that ru figures 349 are always 0 except for zombies. See kvm_proc.c::kvm_getproc2() */ 350 if ((error = proc_vmspace_getref(p, &vm)) != 0) { 351 goto out; 352 } 353 354 len = snprintf(bf, LBFSZ, 355 "%lu %lu %lu %lu %lu %lu %lu\n", 356 (unsigned long)(vm->vm_tsize + vm->vm_dsize + vm->vm_ssize), /* size */ 357 (unsigned long)(vm->vm_rssize), /* resident */ 358 (unsigned long)(ru->ru_ixrss), /* shared */ 359 (unsigned long)(vm->vm_tsize), /* text size in pages */ 360 (unsigned long)(vm->vm_dsize), /* data size in pages */ 361 (unsigned long)(vm->vm_ssize), /* stack size in pages */ 362 (unsigned long) 0); 363 364 if (len == 0) 365 goto out; 366 367 error = uiomove_frombuf(bf, len, uio); 368 out: 369 free(bf, M_TEMP); 370 return error; 371 } 372 373 #define USEC_2_TICKS(x) ((x) / 10000) 374 375 /* 376 * Linux compatible /proc/<pid>/stat. Only active when the -o linux 377 * mountflag is used. 378 */ 379 int 380 procfs_do_pid_stat(struct lwp *curl, struct lwp *l, 381 struct pfsnode *pfs, struct uio *uio) 382 { 383 char *bf; 384 struct proc *p = l->l_proc; 385 int len; 386 struct tty *tty = p->p_session->s_ttyp; 387 struct rusage *ru = &p->p_stats->p_ru; 388 struct rusage *cru = &p->p_stats->p_cru; 389 unsigned long stext = 0, etext = 0, sstack = 0; 390 struct timeval rt; 391 struct vmspace *vm; 392 int error = 0; 393 394 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 395 396 if ((error = proc_vmspace_getref(p, &vm)) != 0) { 397 goto out; 398 } 399 400 get_proc_size_info(l, &stext, &etext, &sstack); 401 402 mutex_enter(&proclist_lock); 403 mutex_enter(&p->p_mutex); 404 mutex_enter(&p->p_smutex); 405 406 calcru(p, NULL, NULL, NULL, &rt); 407 408 len = snprintf(bf, LBFSZ, 409 "%d (%s) %c %d %d %d %d %d " 410 "%u " 411 "%lu %lu %lu %lu %lu %lu %lu %lu " 412 "%d %d %d " 413 "%lu %lu %lu %lu %" PRIu64 " " 414 "%lu %lu %lu " 415 "%u %u " 416 "%u %u %u %u " 417 "%lu %lu %lu %d %d\n", 418 419 p->p_pid, 420 p->p_comm, 421 "0IR3SZD"[(p->p_stat > 6) ? 0 : (int)p->p_stat], 422 (p->p_pptr != NULL) ? p->p_pptr->p_pid : 0, 423 424 p->p_pgid, 425 p->p_session->s_sid, 426 tty ? tty->t_dev : 0, 427 (tty && tty->t_pgrp) ? tty->t_pgrp->pg_id : 0, 428 429 p->p_flag, 430 431 ru->ru_minflt, 432 cru->ru_minflt, 433 ru->ru_majflt, 434 cru->ru_majflt, 435 USEC_2_TICKS(ru->ru_utime.tv_usec), 436 USEC_2_TICKS(ru->ru_stime.tv_usec), 437 USEC_2_TICKS(cru->ru_utime.tv_usec), 438 USEC_2_TICKS(cru->ru_stime.tv_usec), 439 440 l->l_priority, /* XXX: priority */ 441 p->p_nice - 20, 442 0, 443 444 rt.tv_sec, 445 p->p_stats->p_start.tv_sec, 446 (unsigned long)(vm->vm_tsize + vm->vm_dsize + vm->vm_ssize), /* size */ 447 (unsigned long)(vm->vm_rssize), /* resident */ 448 p->p_rlimit[RLIMIT_RSS].rlim_cur, 449 450 stext, /* start code */ 451 etext, /* end code */ 452 sstack, /* mm start stack */ 453 0, /* XXX: pc */ 454 0, /* XXX: sp */ 455 p->p_sigpend.sp_set.__bits[0], /* XXX: pending */ 456 0, /* XXX: held */ 457 p->p_sigctx.ps_sigignore.__bits[0], /* ignored */ 458 p->p_sigctx.ps_sigcatch.__bits[0], /* caught */ 459 460 (unsigned long)(intptr_t)l->l_wchan, 461 ru->ru_nvcsw, 462 ru->ru_nivcsw, 463 p->p_exitsig, 464 0); /* XXX: processor */ 465 466 mutex_exit(&p->p_smutex); 467 mutex_exit(&p->p_mutex); 468 mutex_exit(&proclist_lock); 469 470 if (len == 0) 471 goto out; 472 473 error = uiomove_frombuf(bf, len, uio); 474 out: 475 free(bf, M_TEMP); 476 return error; 477 } 478 479 int 480 procfs_docpuinfo(struct lwp *curl, struct proc *p, 481 struct pfsnode *pfs, struct uio *uio) 482 { 483 int len = LBFSZ; 484 char *bf = malloc(len, M_TEMP, M_WAITOK); 485 int error; 486 487 if (procfs_getcpuinfstr(bf, &len) < 0) { 488 error = ENOSPC; 489 goto done; 490 } 491 492 if (len == 0) { 493 error = 0; 494 goto done; 495 } 496 497 error = uiomove_frombuf(bf, len, uio); 498 done: 499 free(bf, M_TEMP); 500 return error; 501 } 502 503 int 504 procfs_douptime(struct lwp *curl, struct proc *p, 505 struct pfsnode *pfs, struct uio *uio) 506 { 507 char *bf; 508 int len; 509 struct timeval runtime; 510 u_int64_t idle; 511 int error = 0; 512 513 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 514 515 timersub(&curcpu()->ci_schedstate.spc_runtime, &boottime, &runtime); 516 idle = curcpu()->ci_schedstate.spc_cp_time[CP_IDLE]; 517 len = snprintf(bf, LBFSZ, 518 "%lu.%02lu %" PRIu64 ".%02" PRIu64 "\n", 519 runtime.tv_sec, runtime.tv_usec / 10000, 520 idle / hz, (((idle % hz) * 100) / hz) % 100); 521 522 if (len == 0) 523 goto out; 524 525 error = uiomove_frombuf(bf, len, uio); 526 out: 527 free(bf, M_TEMP); 528 return error; 529 } 530 531 int 532 procfs_domounts(struct lwp *curl, struct proc *p, 533 struct pfsnode *pfs, struct uio *uio) 534 { 535 char *bf, *mtab = NULL; 536 const char *fsname; 537 size_t len, mtabsz = 0; 538 struct mount *mp, *nmp; 539 struct statvfs *sfs; 540 int error = 0; 541 542 /* XXX elad - may need filtering. */ 543 544 bf = malloc(LBFSZ, M_TEMP, M_WAITOK); 545 simple_lock(&mountlist_slock); 546 for (mp = CIRCLEQ_FIRST(&mountlist); mp != (void *)&mountlist; 547 mp = nmp) { 548 if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock)) { 549 nmp = CIRCLEQ_NEXT(mp, mnt_list); 550 continue; 551 } 552 553 sfs = &mp->mnt_stat; 554 555 /* Linux uses different names for some filesystems */ 556 fsname = sfs->f_fstypename; 557 if (strcmp(fsname, "procfs") == 0) 558 fsname = "proc"; 559 else if (strcmp(fsname, "ext2fs") == 0) 560 fsname = "ext2"; 561 562 len = snprintf(bf, LBFSZ, "%s %s %s %s%s%s%s%s%s 0 0\n", 563 sfs->f_mntfromname, 564 sfs->f_mntonname, 565 fsname, 566 (mp->mnt_flag & MNT_RDONLY) ? "ro" : "rw", 567 (mp->mnt_flag & MNT_NOSUID) ? ",nosuid" : "", 568 (mp->mnt_flag & MNT_NOEXEC) ? ",noexec" : "", 569 (mp->mnt_flag & MNT_NODEV) ? ",nodev" : "", 570 (mp->mnt_flag & MNT_SYNCHRONOUS) ? ",sync" : "", 571 (mp->mnt_flag & MNT_NOATIME) ? ",noatime" : "" 572 ); 573 574 mtab = realloc(mtab, mtabsz + len, M_TEMP, M_WAITOK); 575 memcpy(mtab + mtabsz, bf, len); 576 mtabsz += len; 577 578 simple_lock(&mountlist_slock); 579 nmp = CIRCLEQ_NEXT(mp, mnt_list); 580 vfs_unbusy(mp); 581 } 582 simple_unlock(&mountlist_slock); 583 free(bf, M_TEMP); 584 585 if (mtabsz > 0) { 586 error = uiomove_frombuf(mtab, mtabsz, uio); 587 free(mtab, M_TEMP); 588 } 589 590 return error; 591 } 592