1.. SPDX-License-Identifier: GPL-2.0 2 3==================== 4The /proc Filesystem 5==================== 6 7===================== ======================================= ================ 8/proc/sys Terrehon Bowden <terrehon@pacbell.net>, October 7 1999 9 Bodo Bauer <bb@ricochet.net> 102.4.x update Jorge Nerin <comandante@zaralinux.com> November 14 2000 11move /proc/sys Shen Feng <shen@cn.fujitsu.com> April 1 2009 12fixes/update part 1.1 Stefani Seibold <stefani@seibold.net> June 9 2009 13===================== ======================================= ================ 14 15 16 17.. Table of Contents 18 19 0 Preface 20 0.1 Introduction/Credits 21 0.2 Legal Stuff 22 23 1 Collecting System Information 24 1.1 Process-Specific Subdirectories 25 1.2 Kernel data 26 1.3 IDE devices in /proc/ide 27 1.4 Networking info in /proc/net 28 1.5 SCSI info 29 1.6 Parallel port info in /proc/parport 30 1.7 TTY info in /proc/tty 31 1.8 Miscellaneous kernel statistics in /proc/stat 32 1.9 Ext4 file system parameters 33 34 2 Modifying System Parameters 35 36 3 Per-Process Parameters 37 3.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj - Adjust the oom-killer 38 score 39 3.2 /proc/<pid>/oom_score - Display current oom-killer score 40 3.3 /proc/<pid>/io - Display the IO accounting fields 41 3.4 /proc/<pid>/coredump_filter - Core dump filtering settings 42 3.5 /proc/<pid>/mountinfo - Information about mounts 43 3.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm 44 3.7 /proc/<pid>/task/<tid>/children - Information about task children 45 3.8 /proc/<pid>/fdinfo/<fd> - Information about opened file 46 3.9 /proc/<pid>/map_files - Information about memory mapped files 47 3.10 /proc/<pid>/timerslack_ns - Task timerslack value 48 3.11 /proc/<pid>/patch_state - Livepatch patch operation state 49 3.12 /proc/<pid>/arch_status - Task architecture specific information 50 51 4 Configuring procfs 52 4.1 Mount options 53 54 5 Filesystem behavior 55 56Preface 57======= 58 590.1 Introduction/Credits 60------------------------ 61 62This documentation is part of a soon (or so we hope) to be released book on 63the SuSE Linux distribution. As there is no complete documentation for the 64/proc file system and we've used many freely available sources to write these 65chapters, it seems only fair to give the work back to the Linux community. 66This work is based on the 2.2.* kernel version and the upcoming 2.4.*. I'm 67afraid it's still far from complete, but we hope it will be useful. As far as 68we know, it is the first 'all-in-one' document about the /proc file system. It 69is focused on the Intel x86 hardware, so if you are looking for PPC, ARM, 70SPARC, AXP, etc., features, you probably won't find what you are looking for. 71It also only covers IPv4 networking, not IPv6 nor other protocols - sorry. But 72additions and patches are welcome and will be added to this document if you 73mail them to Bodo. 74 75We'd like to thank Alan Cox, Rik van Riel, and Alexey Kuznetsov and a lot of 76other people for help compiling this documentation. We'd also like to extend a 77special thank you to Andi Kleen for documentation, which we relied on heavily 78to create this document, as well as the additional information he provided. 79Thanks to everybody else who contributed source or docs to the Linux kernel 80and helped create a great piece of software... :) 81 82If you have any comments, corrections or additions, please don't hesitate to 83contact Bodo Bauer at bb@ricochet.net. We'll be happy to add them to this 84document. 85 86The latest version of this document is available online at 87http://tldp.org/LDP/Linux-Filesystem-Hierarchy/html/proc.html 88 89If the above direction does not works for you, you could try the kernel 90mailing list at linux-kernel@vger.kernel.org and/or try to reach me at 91comandante@zaralinux.com. 92 930.2 Legal Stuff 94--------------- 95 96We don't guarantee the correctness of this document, and if you come to us 97complaining about how you screwed up your system because of incorrect 98documentation, we won't feel responsible... 99 100Chapter 1: Collecting System Information 101======================================== 102 103In This Chapter 104--------------- 105* Investigating the properties of the pseudo file system /proc and its 106 ability to provide information on the running Linux system 107* Examining /proc's structure 108* Uncovering various information about the kernel and the processes running 109 on the system 110 111------------------------------------------------------------------------------ 112 113The proc file system acts as an interface to internal data structures in the 114kernel. It can be used to obtain information about the system and to change 115certain kernel parameters at runtime (sysctl). 116 117First, we'll take a look at the read-only parts of /proc. In Chapter 2, we 118show you how you can use /proc/sys to change settings. 119 1201.1 Process-Specific Subdirectories 121----------------------------------- 122 123The directory /proc contains (among other things) one subdirectory for each 124process running on the system, which is named after the process ID (PID). 125 126The link 'self' points to the process reading the file system. Each process 127subdirectory has the entries listed in Table 1-1. 128 129Note that an open file descriptor to /proc/<pid> or to any of its 130contained files or subdirectories does not prevent <pid> being reused 131for some other process in the event that <pid> exits. Operations on 132open /proc/<pid> file descriptors corresponding to dead processes 133never act on any new process that the kernel may, through chance, have 134also assigned the process ID <pid>. Instead, operations on these FDs 135usually fail with ESRCH. 136 137.. table:: Table 1-1: Process specific entries in /proc 138 139 ============= =============================================================== 140 File Content 141 ============= =============================================================== 142 clear_refs Clears page referenced bits shown in smaps output 143 cmdline Command line arguments 144 cpu Current and last cpu in which it was executed (2.4)(smp) 145 cwd Link to the current working directory 146 environ Values of environment variables 147 exe Link to the executable of this process 148 fd Directory, which contains all file descriptors 149 maps Memory maps to executables and library files (2.4) 150 mem Memory held by this process 151 root Link to the root directory of this process 152 stat Process status 153 statm Process memory status information 154 status Process status in human readable form 155 wchan Present with CONFIG_KALLSYMS=y: it shows the kernel function 156 symbol the task is blocked in - or "0" if not blocked. 157 pagemap Page table 158 stack Report full stack trace, enable via CONFIG_STACKTRACE 159 smaps An extension based on maps, showing the memory consumption of 160 each mapping and flags associated with it 161 smaps_rollup Accumulated smaps stats for all mappings of the process. This 162 can be derived from smaps, but is faster and more convenient 163 numa_maps An extension based on maps, showing the memory locality and 164 binding policy as well as mem usage (in pages) of each mapping. 165 ============= =============================================================== 166 167For example, to get the status information of a process, all you have to do is 168read the file /proc/PID/status:: 169 170 >cat /proc/self/status 171 Name: cat 172 State: R (running) 173 Tgid: 5452 174 Pid: 5452 175 PPid: 743 176 TracerPid: 0 (2.4) 177 Uid: 501 501 501 501 178 Gid: 100 100 100 100 179 FDSize: 256 180 Groups: 100 14 16 181 VmPeak: 5004 kB 182 VmSize: 5004 kB 183 VmLck: 0 kB 184 VmHWM: 476 kB 185 VmRSS: 476 kB 186 RssAnon: 352 kB 187 RssFile: 120 kB 188 RssShmem: 4 kB 189 VmData: 156 kB 190 VmStk: 88 kB 191 VmExe: 68 kB 192 VmLib: 1412 kB 193 VmPTE: 20 kb 194 VmSwap: 0 kB 195 HugetlbPages: 0 kB 196 CoreDumping: 0 197 THP_enabled: 1 198 Threads: 1 199 SigQ: 0/28578 200 SigPnd: 0000000000000000 201 ShdPnd: 0000000000000000 202 SigBlk: 0000000000000000 203 SigIgn: 0000000000000000 204 SigCgt: 0000000000000000 205 CapInh: 00000000fffffeff 206 CapPrm: 0000000000000000 207 CapEff: 0000000000000000 208 CapBnd: ffffffffffffffff 209 CapAmb: 0000000000000000 210 NoNewPrivs: 0 211 Seccomp: 0 212 Speculation_Store_Bypass: thread vulnerable 213 SpeculationIndirectBranch: conditional enabled 214 voluntary_ctxt_switches: 0 215 nonvoluntary_ctxt_switches: 1 216 217This shows you nearly the same information you would get if you viewed it with 218the ps command. In fact, ps uses the proc file system to obtain its 219information. But you get a more detailed view of the process by reading the 220file /proc/PID/status. It fields are described in table 1-2. 221 222The statm file contains more detailed information about the process 223memory usage. Its seven fields are explained in Table 1-3. The stat file 224contains detailed information about the process itself. Its fields are 225explained in Table 1-4. 226 227(for SMP CONFIG users) 228 229For making accounting scalable, RSS related information are handled in an 230asynchronous manner and the value may not be very precise. To see a precise 231snapshot of a moment, you can see /proc/<pid>/smaps file and scan page table. 232It's slow but very precise. 233 234.. table:: Table 1-2: Contents of the status files (as of 4.19) 235 236 ========================== =================================================== 237 Field Content 238 ========================== =================================================== 239 Name filename of the executable 240 Umask file mode creation mask 241 State state (R is running, S is sleeping, D is sleeping 242 in an uninterruptible wait, Z is zombie, 243 T is traced or stopped) 244 Tgid thread group ID 245 Ngid NUMA group ID (0 if none) 246 Pid process id 247 PPid process id of the parent process 248 TracerPid PID of process tracing this process (0 if not, or 249 the tracer is outside of the current pid namespace) 250 Uid Real, effective, saved set, and file system UIDs 251 Gid Real, effective, saved set, and file system GIDs 252 FDSize number of file descriptor slots currently allocated 253 Groups supplementary group list 254 NStgid descendant namespace thread group ID hierarchy 255 NSpid descendant namespace process ID hierarchy 256 NSpgid descendant namespace process group ID hierarchy 257 NSsid descendant namespace session ID hierarchy 258 VmPeak peak virtual memory size 259 VmSize total program size 260 VmLck locked memory size 261 VmPin pinned memory size 262 VmHWM peak resident set size ("high water mark") 263 VmRSS size of memory portions. It contains the three 264 following parts 265 (VmRSS = RssAnon + RssFile + RssShmem) 266 RssAnon size of resident anonymous memory 267 RssFile size of resident file mappings 268 RssShmem size of resident shmem memory (includes SysV shm, 269 mapping of tmpfs and shared anonymous mappings) 270 VmData size of private data segments 271 VmStk size of stack segments 272 VmExe size of text segment 273 VmLib size of shared library code 274 VmPTE size of page table entries 275 VmSwap amount of swap used by anonymous private data 276 (shmem swap usage is not included) 277 HugetlbPages size of hugetlb memory portions 278 CoreDumping process's memory is currently being dumped 279 (killing the process may lead to a corrupted core) 280 THP_enabled process is allowed to use THP (returns 0 when 281 PR_SET_THP_DISABLE is set on the process 282 Threads number of threads 283 SigQ number of signals queued/max. number for queue 284 SigPnd bitmap of pending signals for the thread 285 ShdPnd bitmap of shared pending signals for the process 286 SigBlk bitmap of blocked signals 287 SigIgn bitmap of ignored signals 288 SigCgt bitmap of caught signals 289 CapInh bitmap of inheritable capabilities 290 CapPrm bitmap of permitted capabilities 291 CapEff bitmap of effective capabilities 292 CapBnd bitmap of capabilities bounding set 293 CapAmb bitmap of ambient capabilities 294 NoNewPrivs no_new_privs, like prctl(PR_GET_NO_NEW_PRIV, ...) 295 Seccomp seccomp mode, like prctl(PR_GET_SECCOMP, ...) 296 Speculation_Store_Bypass speculative store bypass mitigation status 297 SpeculationIndirectBranch indirect branch speculation mode 298 Cpus_allowed mask of CPUs on which this process may run 299 Cpus_allowed_list Same as previous, but in "list format" 300 Mems_allowed mask of memory nodes allowed to this process 301 Mems_allowed_list Same as previous, but in "list format" 302 voluntary_ctxt_switches number of voluntary context switches 303 nonvoluntary_ctxt_switches number of non voluntary context switches 304 ========================== =================================================== 305 306 307.. table:: Table 1-3: Contents of the statm files (as of 2.6.8-rc3) 308 309 ======== =============================== ============================== 310 Field Content 311 ======== =============================== ============================== 312 size total program size (pages) (same as VmSize in status) 313 resident size of memory portions (pages) (same as VmRSS in status) 314 shared number of pages that are shared (i.e. backed by a file, same 315 as RssFile+RssShmem in status) 316 trs number of pages that are 'code' (not including libs; broken, 317 includes data segment) 318 lrs number of pages of library (always 0 on 2.6) 319 drs number of pages of data/stack (including libs; broken, 320 includes library text) 321 dt number of dirty pages (always 0 on 2.6) 322 ======== =============================== ============================== 323 324 325.. table:: Table 1-4: Contents of the stat files (as of 2.6.30-rc7) 326 327 ============= =============================================================== 328 Field Content 329 ============= =============================================================== 330 pid process id 331 tcomm filename of the executable 332 state state (R is running, S is sleeping, D is sleeping in an 333 uninterruptible wait, Z is zombie, T is traced or stopped) 334 ppid process id of the parent process 335 pgrp pgrp of the process 336 sid session id 337 tty_nr tty the process uses 338 tty_pgrp pgrp of the tty 339 flags task flags 340 min_flt number of minor faults 341 cmin_flt number of minor faults with child's 342 maj_flt number of major faults 343 cmaj_flt number of major faults with child's 344 utime user mode jiffies 345 stime kernel mode jiffies 346 cutime user mode jiffies with child's 347 cstime kernel mode jiffies with child's 348 priority priority level 349 nice nice level 350 num_threads number of threads 351 it_real_value (obsolete, always 0) 352 start_time time the process started after system boot 353 vsize virtual memory size 354 rss resident set memory size 355 rsslim current limit in bytes on the rss 356 start_code address above which program text can run 357 end_code address below which program text can run 358 start_stack address of the start of the main process stack 359 esp current value of ESP 360 eip current value of EIP 361 pending bitmap of pending signals 362 blocked bitmap of blocked signals 363 sigign bitmap of ignored signals 364 sigcatch bitmap of caught signals 365 0 (place holder, used to be the wchan address, 366 use /proc/PID/wchan instead) 367 0 (place holder) 368 0 (place holder) 369 exit_signal signal to send to parent thread on exit 370 task_cpu which CPU the task is scheduled on 371 rt_priority realtime priority 372 policy scheduling policy (man sched_setscheduler) 373 blkio_ticks time spent waiting for block IO 374 gtime guest time of the task in jiffies 375 cgtime guest time of the task children in jiffies 376 start_data address above which program data+bss is placed 377 end_data address below which program data+bss is placed 378 start_brk address above which program heap can be expanded with brk() 379 arg_start address above which program command line is placed 380 arg_end address below which program command line is placed 381 env_start address above which program environment is placed 382 env_end address below which program environment is placed 383 exit_code the thread's exit_code in the form reported by the waitpid 384 system call 385 ============= =============================================================== 386 387The /proc/PID/maps file contains the currently mapped memory regions and 388their access permissions. 389 390The format is:: 391 392 address perms offset dev inode pathname 393 394 08048000-08049000 r-xp 00000000 03:00 8312 /opt/test 395 08049000-0804a000 rw-p 00001000 03:00 8312 /opt/test 396 0804a000-0806b000 rw-p 00000000 00:00 0 [heap] 397 a7cb1000-a7cb2000 ---p 00000000 00:00 0 398 a7cb2000-a7eb2000 rw-p 00000000 00:00 0 399 a7eb2000-a7eb3000 ---p 00000000 00:00 0 400 a7eb3000-a7ed5000 rw-p 00000000 00:00 0 401 a7ed5000-a8008000 r-xp 00000000 03:00 4222 /lib/libc.so.6 402 a8008000-a800a000 r--p 00133000 03:00 4222 /lib/libc.so.6 403 a800a000-a800b000 rw-p 00135000 03:00 4222 /lib/libc.so.6 404 a800b000-a800e000 rw-p 00000000 00:00 0 405 a800e000-a8022000 r-xp 00000000 03:00 14462 /lib/libpthread.so.0 406 a8022000-a8023000 r--p 00013000 03:00 14462 /lib/libpthread.so.0 407 a8023000-a8024000 rw-p 00014000 03:00 14462 /lib/libpthread.so.0 408 a8024000-a8027000 rw-p 00000000 00:00 0 409 a8027000-a8043000 r-xp 00000000 03:00 8317 /lib/ld-linux.so.2 410 a8043000-a8044000 r--p 0001b000 03:00 8317 /lib/ld-linux.so.2 411 a8044000-a8045000 rw-p 0001c000 03:00 8317 /lib/ld-linux.so.2 412 aff35000-aff4a000 rw-p 00000000 00:00 0 [stack] 413 ffffe000-fffff000 r-xp 00000000 00:00 0 [vdso] 414 415where "address" is the address space in the process that it occupies, "perms" 416is a set of permissions:: 417 418 r = read 419 w = write 420 x = execute 421 s = shared 422 p = private (copy on write) 423 424"offset" is the offset into the mapping, "dev" is the device (major:minor), and 425"inode" is the inode on that device. 0 indicates that no inode is associated 426with the memory region, as the case would be with BSS (uninitialized data). 427The "pathname" shows the name associated file for this mapping. If the mapping 428is not associated with a file: 429 430 ============= ==================================== 431 [heap] the heap of the program 432 [stack] the stack of the main process 433 [vdso] the "virtual dynamic shared object", 434 the kernel system call handler 435 [anon:<name>] an anonymous mapping that has been 436 named by userspace 437 ============= ==================================== 438 439 or if empty, the mapping is anonymous. 440 441The /proc/PID/smaps is an extension based on maps, showing the memory 442consumption for each of the process's mappings. For each mapping (aka Virtual 443Memory Area, or VMA) there is a series of lines such as the following:: 444 445 08048000-080bc000 r-xp 00000000 03:02 13130 /bin/bash 446 447 Size: 1084 kB 448 KernelPageSize: 4 kB 449 MMUPageSize: 4 kB 450 Rss: 892 kB 451 Pss: 374 kB 452 Pss_Dirty: 0 kB 453 Shared_Clean: 892 kB 454 Shared_Dirty: 0 kB 455 Private_Clean: 0 kB 456 Private_Dirty: 0 kB 457 Referenced: 892 kB 458 Anonymous: 0 kB 459 LazyFree: 0 kB 460 AnonHugePages: 0 kB 461 ShmemPmdMapped: 0 kB 462 Shared_Hugetlb: 0 kB 463 Private_Hugetlb: 0 kB 464 Swap: 0 kB 465 SwapPss: 0 kB 466 KernelPageSize: 4 kB 467 MMUPageSize: 4 kB 468 Locked: 0 kB 469 THPeligible: 0 470 VmFlags: rd ex mr mw me dw 471 472The first of these lines shows the same information as is displayed for the 473mapping in /proc/PID/maps. Following lines show the size of the mapping 474(size); the size of each page allocated when backing a VMA (KernelPageSize), 475which is usually the same as the size in the page table entries; the page size 476used by the MMU when backing a VMA (in most cases, the same as KernelPageSize); 477the amount of the mapping that is currently resident in RAM (RSS); the 478process' proportional share of this mapping (PSS); and the number of clean and 479dirty shared and private pages in the mapping. 480 481The "proportional set size" (PSS) of a process is the count of pages it has 482in memory, where each page is divided by the number of processes sharing it. 483So if a process has 1000 pages all to itself, and 1000 shared with one other 484process, its PSS will be 1500. "Pss_Dirty" is the portion of PSS which 485consists of dirty pages. ("Pss_Clean" is not included, but it can be 486calculated by subtracting "Pss_Dirty" from "Pss".) 487 488Note that even a page which is part of a MAP_SHARED mapping, but has only 489a single pte mapped, i.e. is currently used by only one process, is accounted 490as private and not as shared. 491 492"Referenced" indicates the amount of memory currently marked as referenced or 493accessed. 494 495"Anonymous" shows the amount of memory that does not belong to any file. Even 496a mapping associated with a file may contain anonymous pages: when MAP_PRIVATE 497and a page is modified, the file page is replaced by a private anonymous copy. 498 499"LazyFree" shows the amount of memory which is marked by madvise(MADV_FREE). 500The memory isn't freed immediately with madvise(). It's freed in memory 501pressure if the memory is clean. Please note that the printed value might 502be lower than the real value due to optimizations used in the current 503implementation. If this is not desirable please file a bug report. 504 505"AnonHugePages" shows the ammount of memory backed by transparent hugepage. 506 507"ShmemPmdMapped" shows the ammount of shared (shmem/tmpfs) memory backed by 508huge pages. 509 510"Shared_Hugetlb" and "Private_Hugetlb" show the ammounts of memory backed by 511hugetlbfs page which is *not* counted in "RSS" or "PSS" field for historical 512reasons. And these are not included in {Shared,Private}_{Clean,Dirty} field. 513 514"Swap" shows how much would-be-anonymous memory is also used, but out on swap. 515 516For shmem mappings, "Swap" includes also the size of the mapped (and not 517replaced by copy-on-write) part of the underlying shmem object out on swap. 518"SwapPss" shows proportional swap share of this mapping. Unlike "Swap", this 519does not take into account swapped out page of underlying shmem objects. 520"Locked" indicates whether the mapping is locked in memory or not. 521 522"THPeligible" indicates whether the mapping is eligible for allocating THP 523pages as well as the THP is PMD mappable or not - 1 if true, 0 otherwise. 524It just shows the current status. 525 526"VmFlags" field deserves a separate description. This member represents the 527kernel flags associated with the particular virtual memory area in two letter 528encoded manner. The codes are the following: 529 530 == ======================================= 531 rd readable 532 wr writeable 533 ex executable 534 sh shared 535 mr may read 536 mw may write 537 me may execute 538 ms may share 539 gd stack segment growns down 540 pf pure PFN range 541 dw disabled write to the mapped file 542 lo pages are locked in memory 543 io memory mapped I/O area 544 sr sequential read advise provided 545 rr random read advise provided 546 dc do not copy area on fork 547 de do not expand area on remapping 548 ac area is accountable 549 nr swap space is not reserved for the area 550 ht area uses huge tlb pages 551 sf synchronous page fault 552 ar architecture specific flag 553 wf wipe on fork 554 dd do not include area into core dump 555 sd soft dirty flag 556 mm mixed map area 557 hg huge page advise flag 558 nh no huge page advise flag 559 mg mergable advise flag 560 bt arm64 BTI guarded page 561 mt arm64 MTE allocation tags are enabled 562 um userfaultfd missing tracking 563 uw userfaultfd wr-protect tracking 564 == ======================================= 565 566Note that there is no guarantee that every flag and associated mnemonic will 567be present in all further kernel releases. Things get changed, the flags may 568be vanished or the reverse -- new added. Interpretation of their meaning 569might change in future as well. So each consumer of these flags has to 570follow each specific kernel version for the exact semantic. 571 572This file is only present if the CONFIG_MMU kernel configuration option is 573enabled. 574 575Note: reading /proc/PID/maps or /proc/PID/smaps is inherently racy (consistent 576output can be achieved only in the single read call). 577 578This typically manifests when doing partial reads of these files while the 579memory map is being modified. Despite the races, we do provide the following 580guarantees: 581 5821) The mapped addresses never go backwards, which implies no two 583 regions will ever overlap. 5842) If there is something at a given vaddr during the entirety of the 585 life of the smaps/maps walk, there will be some output for it. 586 587The /proc/PID/smaps_rollup file includes the same fields as /proc/PID/smaps, 588but their values are the sums of the corresponding values for all mappings of 589the process. Additionally, it contains these fields: 590 591- Pss_Anon 592- Pss_File 593- Pss_Shmem 594 595They represent the proportional shares of anonymous, file, and shmem pages, as 596described for smaps above. These fields are omitted in smaps since each 597mapping identifies the type (anon, file, or shmem) of all pages it contains. 598Thus all information in smaps_rollup can be derived from smaps, but at a 599significantly higher cost. 600 601The /proc/PID/clear_refs is used to reset the PG_Referenced and ACCESSED/YOUNG 602bits on both physical and virtual pages associated with a process, and the 603soft-dirty bit on pte (see Documentation/admin-guide/mm/soft-dirty.rst 604for details). 605To clear the bits for all the pages associated with the process:: 606 607 > echo 1 > /proc/PID/clear_refs 608 609To clear the bits for the anonymous pages associated with the process:: 610 611 > echo 2 > /proc/PID/clear_refs 612 613To clear the bits for the file mapped pages associated with the process:: 614 615 > echo 3 > /proc/PID/clear_refs 616 617To clear the soft-dirty bit:: 618 619 > echo 4 > /proc/PID/clear_refs 620 621To reset the peak resident set size ("high water mark") to the process's 622current value:: 623 624 > echo 5 > /proc/PID/clear_refs 625 626Any other value written to /proc/PID/clear_refs will have no effect. 627 628The /proc/pid/pagemap gives the PFN, which can be used to find the pageflags 629using /proc/kpageflags and number of times a page is mapped using 630/proc/kpagecount. For detailed explanation, see 631Documentation/admin-guide/mm/pagemap.rst. 632 633The /proc/pid/numa_maps is an extension based on maps, showing the memory 634locality and binding policy, as well as the memory usage (in pages) of 635each mapping. The output follows a general format where mapping details get 636summarized separated by blank spaces, one mapping per each file line:: 637 638 address policy mapping details 639 640 00400000 default file=/usr/local/bin/app mapped=1 active=0 N3=1 kernelpagesize_kB=4 641 00600000 default file=/usr/local/bin/app anon=1 dirty=1 N3=1 kernelpagesize_kB=4 642 3206000000 default file=/lib64/ld-2.12.so mapped=26 mapmax=6 N0=24 N3=2 kernelpagesize_kB=4 643 320621f000 default file=/lib64/ld-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4 644 3206220000 default file=/lib64/ld-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4 645 3206221000 default anon=1 dirty=1 N3=1 kernelpagesize_kB=4 646 3206800000 default file=/lib64/libc-2.12.so mapped=59 mapmax=21 active=55 N0=41 N3=18 kernelpagesize_kB=4 647 320698b000 default file=/lib64/libc-2.12.so 648 3206b8a000 default file=/lib64/libc-2.12.so anon=2 dirty=2 N3=2 kernelpagesize_kB=4 649 3206b8e000 default file=/lib64/libc-2.12.so anon=1 dirty=1 N3=1 kernelpagesize_kB=4 650 3206b8f000 default anon=3 dirty=3 active=1 N3=3 kernelpagesize_kB=4 651 7f4dc10a2000 default anon=3 dirty=3 N3=3 kernelpagesize_kB=4 652 7f4dc10b4000 default anon=2 dirty=2 active=1 N3=2 kernelpagesize_kB=4 653 7f4dc1200000 default file=/anon_hugepage\040(deleted) huge anon=1 dirty=1 N3=1 kernelpagesize_kB=2048 654 7fff335f0000 default stack anon=3 dirty=3 N3=3 kernelpagesize_kB=4 655 7fff3369d000 default mapped=1 mapmax=35 active=0 N3=1 kernelpagesize_kB=4 656 657Where: 658 659"address" is the starting address for the mapping; 660 661"policy" reports the NUMA memory policy set for the mapping (see Documentation/admin-guide/mm/numa_memory_policy.rst); 662 663"mapping details" summarizes mapping data such as mapping type, page usage counters, 664node locality page counters (N0 == node0, N1 == node1, ...) and the kernel page 665size, in KB, that is backing the mapping up. 666 6671.2 Kernel data 668--------------- 669 670Similar to the process entries, the kernel data files give information about 671the running kernel. The files used to obtain this information are contained in 672/proc and are listed in Table 1-5. Not all of these will be present in your 673system. It depends on the kernel configuration and the loaded modules, which 674files are there, and which are missing. 675 676.. table:: Table 1-5: Kernel info in /proc 677 678 ============ =============================================================== 679 File Content 680 ============ =============================================================== 681 apm Advanced power management info 682 buddyinfo Kernel memory allocator information (see text) (2.5) 683 bus Directory containing bus specific information 684 cmdline Kernel command line 685 cpuinfo Info about the CPU 686 devices Available devices (block and character) 687 dma Used DMS channels 688 filesystems Supported filesystems 689 driver Various drivers grouped here, currently rtc (2.4) 690 execdomains Execdomains, related to security (2.4) 691 fb Frame Buffer devices (2.4) 692 fs File system parameters, currently nfs/exports (2.4) 693 ide Directory containing info about the IDE subsystem 694 interrupts Interrupt usage 695 iomem Memory map (2.4) 696 ioports I/O port usage 697 irq Masks for irq to cpu affinity (2.4)(smp?) 698 isapnp ISA PnP (Plug&Play) Info (2.4) 699 kcore Kernel core image (can be ELF or A.OUT(deprecated in 2.4)) 700 kmsg Kernel messages 701 ksyms Kernel symbol table 702 loadavg Load average of last 1, 5 & 15 minutes; 703 number of processes currently runnable (running or on ready queue); 704 total number of processes in system; 705 last pid created. 706 All fields are separated by one space except "number of 707 processes currently runnable" and "total number of processes 708 in system", which are separated by a slash ('/'). Example: 709 0.61 0.61 0.55 3/828 22084 710 locks Kernel locks 711 meminfo Memory info 712 misc Miscellaneous 713 modules List of loaded modules 714 mounts Mounted filesystems 715 net Networking info (see text) 716 pagetypeinfo Additional page allocator information (see text) (2.5) 717 partitions Table of partitions known to the system 718 pci Deprecated info of PCI bus (new way -> /proc/bus/pci/, 719 decoupled by lspci (2.4) 720 rtc Real time clock 721 scsi SCSI info (see text) 722 slabinfo Slab pool info 723 softirqs softirq usage 724 stat Overall statistics 725 swaps Swap space utilization 726 sys See chapter 2 727 sysvipc Info of SysVIPC Resources (msg, sem, shm) (2.4) 728 tty Info of tty drivers 729 uptime Wall clock since boot, combined idle time of all cpus 730 version Kernel version 731 video bttv info of video resources (2.4) 732 vmallocinfo Show vmalloced areas 733 ============ =============================================================== 734 735You can, for example, check which interrupts are currently in use and what 736they are used for by looking in the file /proc/interrupts:: 737 738 > cat /proc/interrupts 739 CPU0 740 0: 8728810 XT-PIC timer 741 1: 895 XT-PIC keyboard 742 2: 0 XT-PIC cascade 743 3: 531695 XT-PIC aha152x 744 4: 2014133 XT-PIC serial 745 5: 44401 XT-PIC pcnet_cs 746 8: 2 XT-PIC rtc 747 11: 8 XT-PIC i82365 748 12: 182918 XT-PIC PS/2 Mouse 749 13: 1 XT-PIC fpu 750 14: 1232265 XT-PIC ide0 751 15: 7 XT-PIC ide1 752 NMI: 0 753 754In 2.4.* a couple of lines where added to this file LOC & ERR (this time is the 755output of a SMP machine):: 756 757 > cat /proc/interrupts 758 759 CPU0 CPU1 760 0: 1243498 1214548 IO-APIC-edge timer 761 1: 8949 8958 IO-APIC-edge keyboard 762 2: 0 0 XT-PIC cascade 763 5: 11286 10161 IO-APIC-edge soundblaster 764 8: 1 0 IO-APIC-edge rtc 765 9: 27422 27407 IO-APIC-edge 3c503 766 12: 113645 113873 IO-APIC-edge PS/2 Mouse 767 13: 0 0 XT-PIC fpu 768 14: 22491 24012 IO-APIC-edge ide0 769 15: 2183 2415 IO-APIC-edge ide1 770 17: 30564 30414 IO-APIC-level eth0 771 18: 177 164 IO-APIC-level bttv 772 NMI: 2457961 2457959 773 LOC: 2457882 2457881 774 ERR: 2155 775 776NMI is incremented in this case because every timer interrupt generates a NMI 777(Non Maskable Interrupt) which is used by the NMI Watchdog to detect lockups. 778 779LOC is the local interrupt counter of the internal APIC of every CPU. 780 781ERR is incremented in the case of errors in the IO-APIC bus (the bus that 782connects the CPUs in a SMP system. This means that an error has been detected, 783the IO-APIC automatically retry the transmission, so it should not be a big 784problem, but you should read the SMP-FAQ. 785 786In 2.6.2* /proc/interrupts was expanded again. This time the goal was for 787/proc/interrupts to display every IRQ vector in use by the system, not 788just those considered 'most important'. The new vectors are: 789 790THR 791 interrupt raised when a machine check threshold counter 792 (typically counting ECC corrected errors of memory or cache) exceeds 793 a configurable threshold. Only available on some systems. 794 795TRM 796 a thermal event interrupt occurs when a temperature threshold 797 has been exceeded for the CPU. This interrupt may also be generated 798 when the temperature drops back to normal. 799 800SPU 801 a spurious interrupt is some interrupt that was raised then lowered 802 by some IO device before it could be fully processed by the APIC. Hence 803 the APIC sees the interrupt but does not know what device it came from. 804 For this case the APIC will generate the interrupt with a IRQ vector 805 of 0xff. This might also be generated by chipset bugs. 806 807RES, CAL, TLB 808 rescheduling, call and TLB flush interrupts are 809 sent from one CPU to another per the needs of the OS. Typically, 810 their statistics are used by kernel developers and interested users to 811 determine the occurrence of interrupts of the given type. 812 813The above IRQ vectors are displayed only when relevant. For example, 814the threshold vector does not exist on x86_64 platforms. Others are 815suppressed when the system is a uniprocessor. As of this writing, only 816i386 and x86_64 platforms support the new IRQ vector displays. 817 818Of some interest is the introduction of the /proc/irq directory to 2.4. 819It could be used to set IRQ to CPU affinity. This means that you can "hook" an 820IRQ to only one CPU, or to exclude a CPU of handling IRQs. The contents of the 821irq subdir is one subdir for each IRQ, and two files; default_smp_affinity and 822prof_cpu_mask. 823 824For example:: 825 826 > ls /proc/irq/ 827 0 10 12 14 16 18 2 4 6 8 prof_cpu_mask 828 1 11 13 15 17 19 3 5 7 9 default_smp_affinity 829 > ls /proc/irq/0/ 830 smp_affinity 831 832smp_affinity is a bitmask, in which you can specify which CPUs can handle the 833IRQ. You can set it by doing:: 834 835 > echo 1 > /proc/irq/10/smp_affinity 836 837This means that only the first CPU will handle the IRQ, but you can also echo 8385 which means that only the first and third CPU can handle the IRQ. 839 840The contents of each smp_affinity file is the same by default:: 841 842 > cat /proc/irq/0/smp_affinity 843 ffffffff 844 845There is an alternate interface, smp_affinity_list which allows specifying 846a CPU range instead of a bitmask:: 847 848 > cat /proc/irq/0/smp_affinity_list 849 1024-1031 850 851The default_smp_affinity mask applies to all non-active IRQs, which are the 852IRQs which have not yet been allocated/activated, and hence which lack a 853/proc/irq/[0-9]* directory. 854 855The node file on an SMP system shows the node to which the device using the IRQ 856reports itself as being attached. This hardware locality information does not 857include information about any possible driver locality preference. 858 859prof_cpu_mask specifies which CPUs are to be profiled by the system wide 860profiler. Default value is ffffffff (all CPUs if there are only 32 of them). 861 862The way IRQs are routed is handled by the IO-APIC, and it's Round Robin 863between all the CPUs which are allowed to handle it. As usual the kernel has 864more info than you and does a better job than you, so the defaults are the 865best choice for almost everyone. [Note this applies only to those IO-APIC's 866that support "Round Robin" interrupt distribution.] 867 868There are three more important subdirectories in /proc: net, scsi, and sys. 869The general rule is that the contents, or even the existence of these 870directories, depend on your kernel configuration. If SCSI is not enabled, the 871directory scsi may not exist. The same is true with the net, which is there 872only when networking support is present in the running kernel. 873 874The slabinfo file gives information about memory usage at the slab level. 875Linux uses slab pools for memory management above page level in version 2.2. 876Commonly used objects have their own slab pool (such as network buffers, 877directory cache, and so on). 878 879:: 880 881 > cat /proc/buddyinfo 882 883 Node 0, zone DMA 0 4 5 4 4 3 ... 884 Node 0, zone Normal 1 0 0 1 101 8 ... 885 Node 0, zone HighMem 2 0 0 1 1 0 ... 886 887External fragmentation is a problem under some workloads, and buddyinfo is a 888useful tool for helping diagnose these problems. Buddyinfo will give you a 889clue as to how big an area you can safely allocate, or why a previous 890allocation failed. 891 892Each column represents the number of pages of a certain order which are 893available. In this case, there are 0 chunks of 2^0*PAGE_SIZE available in 894ZONE_DMA, 4 chunks of 2^1*PAGE_SIZE in ZONE_DMA, 101 chunks of 2^4*PAGE_SIZE 895available in ZONE_NORMAL, etc... 896 897More information relevant to external fragmentation can be found in 898pagetypeinfo:: 899 900 > cat /proc/pagetypeinfo 901 Page block order: 9 902 Pages per block: 512 903 904 Free pages count per migrate type at order 0 1 2 3 4 5 6 7 8 9 10 905 Node 0, zone DMA, type Unmovable 0 0 0 1 1 1 1 1 1 1 0 906 Node 0, zone DMA, type Reclaimable 0 0 0 0 0 0 0 0 0 0 0 907 Node 0, zone DMA, type Movable 1 1 2 1 2 1 1 0 1 0 2 908 Node 0, zone DMA, type Reserve 0 0 0 0 0 0 0 0 0 1 0 909 Node 0, zone DMA, type Isolate 0 0 0 0 0 0 0 0 0 0 0 910 Node 0, zone DMA32, type Unmovable 103 54 77 1 1 1 11 8 7 1 9 911 Node 0, zone DMA32, type Reclaimable 0 0 2 1 0 0 0 0 1 0 0 912 Node 0, zone DMA32, type Movable 169 152 113 91 77 54 39 13 6 1 452 913 Node 0, zone DMA32, type Reserve 1 2 2 2 2 0 1 1 1 1 0 914 Node 0, zone DMA32, type Isolate 0 0 0 0 0 0 0 0 0 0 0 915 916 Number of blocks type Unmovable Reclaimable Movable Reserve Isolate 917 Node 0, zone DMA 2 0 5 1 0 918 Node 0, zone DMA32 41 6 967 2 0 919 920Fragmentation avoidance in the kernel works by grouping pages of different 921migrate types into the same contiguous regions of memory called page blocks. 922A page block is typically the size of the default hugepage size, e.g. 2MB on 923X86-64. By keeping pages grouped based on their ability to move, the kernel 924can reclaim pages within a page block to satisfy a high-order allocation. 925 926The pagetypinfo begins with information on the size of a page block. It 927then gives the same type of information as buddyinfo except broken down 928by migrate-type and finishes with details on how many page blocks of each 929type exist. 930 931If min_free_kbytes has been tuned correctly (recommendations made by hugeadm 932from libhugetlbfs https://github.com/libhugetlbfs/libhugetlbfs/), one can 933make an estimate of the likely number of huge pages that can be allocated 934at a given point in time. All the "Movable" blocks should be allocatable 935unless memory has been mlock()'d. Some of the Reclaimable blocks should 936also be allocatable although a lot of filesystem metadata may have to be 937reclaimed to achieve this. 938 939 940meminfo 941~~~~~~~ 942 943Provides information about distribution and utilization of memory. This 944varies by architecture and compile options. Some of the counters reported 945here overlap. The memory reported by the non overlapping counters may not 946add up to the overall memory usage and the difference for some workloads 947can be substantial. In many cases there are other means to find out 948additional memory using subsystem specific interfaces, for instance 949/proc/net/sockstat for TCP memory allocations. 950 951Example output. You may not have all of these fields. 952 953:: 954 955 > cat /proc/meminfo 956 957 MemTotal: 32858820 kB 958 MemFree: 21001236 kB 959 MemAvailable: 27214312 kB 960 Buffers: 581092 kB 961 Cached: 5587612 kB 962 SwapCached: 0 kB 963 Active: 3237152 kB 964 Inactive: 7586256 kB 965 Active(anon): 94064 kB 966 Inactive(anon): 4570616 kB 967 Active(file): 3143088 kB 968 Inactive(file): 3015640 kB 969 Unevictable: 0 kB 970 Mlocked: 0 kB 971 SwapTotal: 0 kB 972 SwapFree: 0 kB 973 Zswap: 1904 kB 974 Zswapped: 7792 kB 975 Dirty: 12 kB 976 Writeback: 0 kB 977 AnonPages: 4654780 kB 978 Mapped: 266244 kB 979 Shmem: 9976 kB 980 KReclaimable: 517708 kB 981 Slab: 660044 kB 982 SReclaimable: 517708 kB 983 SUnreclaim: 142336 kB 984 KernelStack: 11168 kB 985 PageTables: 20540 kB 986 SecPageTables: 0 kB 987 NFS_Unstable: 0 kB 988 Bounce: 0 kB 989 WritebackTmp: 0 kB 990 CommitLimit: 16429408 kB 991 Committed_AS: 7715148 kB 992 VmallocTotal: 34359738367 kB 993 VmallocUsed: 40444 kB 994 VmallocChunk: 0 kB 995 Percpu: 29312 kB 996 HardwareCorrupted: 0 kB 997 AnonHugePages: 4149248 kB 998 ShmemHugePages: 0 kB 999 ShmemPmdMapped: 0 kB 1000 FileHugePages: 0 kB 1001 FilePmdMapped: 0 kB 1002 CmaTotal: 0 kB 1003 CmaFree: 0 kB 1004 HugePages_Total: 0 1005 HugePages_Free: 0 1006 HugePages_Rsvd: 0 1007 HugePages_Surp: 0 1008 Hugepagesize: 2048 kB 1009 Hugetlb: 0 kB 1010 DirectMap4k: 401152 kB 1011 DirectMap2M: 10008576 kB 1012 DirectMap1G: 24117248 kB 1013 1014MemTotal 1015 Total usable RAM (i.e. physical RAM minus a few reserved 1016 bits and the kernel binary code) 1017MemFree 1018 Total free RAM. On highmem systems, the sum of LowFree+HighFree 1019MemAvailable 1020 An estimate of how much memory is available for starting new 1021 applications, without swapping. Calculated from MemFree, 1022 SReclaimable, the size of the file LRU lists, and the low 1023 watermarks in each zone. 1024 The estimate takes into account that the system needs some 1025 page cache to function well, and that not all reclaimable 1026 slab will be reclaimable, due to items being in use. The 1027 impact of those factors will vary from system to system. 1028Buffers 1029 Relatively temporary storage for raw disk blocks 1030 shouldn't get tremendously large (20MB or so) 1031Cached 1032 In-memory cache for files read from the disk (the 1033 pagecache) as well as tmpfs & shmem. 1034 Doesn't include SwapCached. 1035SwapCached 1036 Memory that once was swapped out, is swapped back in but 1037 still also is in the swapfile (if memory is needed it 1038 doesn't need to be swapped out AGAIN because it is already 1039 in the swapfile. This saves I/O) 1040Active 1041 Memory that has been used more recently and usually not 1042 reclaimed unless absolutely necessary. 1043Inactive 1044 Memory which has been less recently used. It is more 1045 eligible to be reclaimed for other purposes 1046Unevictable 1047 Memory allocated for userspace which cannot be reclaimed, such 1048 as mlocked pages, ramfs backing pages, secret memfd pages etc. 1049Mlocked 1050 Memory locked with mlock(). 1051HighTotal, HighFree 1052 Highmem is all memory above ~860MB of physical memory. 1053 Highmem areas are for use by userspace programs, or 1054 for the pagecache. The kernel must use tricks to access 1055 this memory, making it slower to access than lowmem. 1056LowTotal, LowFree 1057 Lowmem is memory which can be used for everything that 1058 highmem can be used for, but it is also available for the 1059 kernel's use for its own data structures. Among many 1060 other things, it is where everything from the Slab is 1061 allocated. Bad things happen when you're out of lowmem. 1062SwapTotal 1063 total amount of swap space available 1064SwapFree 1065 Memory which has been evicted from RAM, and is temporarily 1066 on the disk 1067Zswap 1068 Memory consumed by the zswap backend (compressed size) 1069Zswapped 1070 Amount of anonymous memory stored in zswap (original size) 1071Dirty 1072 Memory which is waiting to get written back to the disk 1073Writeback 1074 Memory which is actively being written back to the disk 1075AnonPages 1076 Non-file backed pages mapped into userspace page tables 1077Mapped 1078 files which have been mmaped, such as libraries 1079Shmem 1080 Total memory used by shared memory (shmem) and tmpfs 1081KReclaimable 1082 Kernel allocations that the kernel will attempt to reclaim 1083 under memory pressure. Includes SReclaimable (below), and other 1084 direct allocations with a shrinker. 1085Slab 1086 in-kernel data structures cache 1087SReclaimable 1088 Part of Slab, that might be reclaimed, such as caches 1089SUnreclaim 1090 Part of Slab, that cannot be reclaimed on memory pressure 1091KernelStack 1092 Memory consumed by the kernel stacks of all tasks 1093PageTables 1094 Memory consumed by userspace page tables 1095SecPageTables 1096 Memory consumed by secondary page tables, this currently 1097 currently includes KVM mmu allocations on x86 and arm64. 1098NFS_Unstable 1099 Always zero. Previous counted pages which had been written to 1100 the server, but has not been committed to stable storage. 1101Bounce 1102 Memory used for block device "bounce buffers" 1103WritebackTmp 1104 Memory used by FUSE for temporary writeback buffers 1105CommitLimit 1106 Based on the overcommit ratio ('vm.overcommit_ratio'), 1107 this is the total amount of memory currently available to 1108 be allocated on the system. This limit is only adhered to 1109 if strict overcommit accounting is enabled (mode 2 in 1110 'vm.overcommit_memory'). 1111 1112 The CommitLimit is calculated with the following formula:: 1113 1114 CommitLimit = ([total RAM pages] - [total huge TLB pages]) * 1115 overcommit_ratio / 100 + [total swap pages] 1116 1117 For example, on a system with 1G of physical RAM and 7G 1118 of swap with a `vm.overcommit_ratio` of 30 it would 1119 yield a CommitLimit of 7.3G. 1120 1121 For more details, see the memory overcommit documentation 1122 in mm/overcommit-accounting. 1123Committed_AS 1124 The amount of memory presently allocated on the system. 1125 The committed memory is a sum of all of the memory which 1126 has been allocated by processes, even if it has not been 1127 "used" by them as of yet. A process which malloc()'s 1G 1128 of memory, but only touches 300M of it will show up as 1129 using 1G. This 1G is memory which has been "committed" to 1130 by the VM and can be used at any time by the allocating 1131 application. With strict overcommit enabled on the system 1132 (mode 2 in 'vm.overcommit_memory'), allocations which would 1133 exceed the CommitLimit (detailed above) will not be permitted. 1134 This is useful if one needs to guarantee that processes will 1135 not fail due to lack of memory once that memory has been 1136 successfully allocated. 1137VmallocTotal 1138 total size of vmalloc virtual address space 1139VmallocUsed 1140 amount of vmalloc area which is used 1141VmallocChunk 1142 largest contiguous block of vmalloc area which is free 1143Percpu 1144 Memory allocated to the percpu allocator used to back percpu 1145 allocations. This stat excludes the cost of metadata. 1146HardwareCorrupted 1147 The amount of RAM/memory in KB, the kernel identifies as 1148 corrupted. 1149AnonHugePages 1150 Non-file backed huge pages mapped into userspace page tables 1151ShmemHugePages 1152 Memory used by shared memory (shmem) and tmpfs allocated 1153 with huge pages 1154ShmemPmdMapped 1155 Shared memory mapped into userspace with huge pages 1156FileHugePages 1157 Memory used for filesystem data (page cache) allocated 1158 with huge pages 1159FilePmdMapped 1160 Page cache mapped into userspace with huge pages 1161CmaTotal 1162 Memory reserved for the Contiguous Memory Allocator (CMA) 1163CmaFree 1164 Free remaining memory in the CMA reserves 1165HugePages_Total, HugePages_Free, HugePages_Rsvd, HugePages_Surp, Hugepagesize, Hugetlb 1166 See Documentation/admin-guide/mm/hugetlbpage.rst. 1167DirectMap4k, DirectMap2M, DirectMap1G 1168 Breakdown of page table sizes used in the kernel's 1169 identity mapping of RAM 1170 1171vmallocinfo 1172~~~~~~~~~~~ 1173 1174Provides information about vmalloced/vmaped areas. One line per area, 1175containing the virtual address range of the area, size in bytes, 1176caller information of the creator, and optional information depending 1177on the kind of area: 1178 1179 ========== =================================================== 1180 pages=nr number of pages 1181 phys=addr if a physical address was specified 1182 ioremap I/O mapping (ioremap() and friends) 1183 vmalloc vmalloc() area 1184 vmap vmap()ed pages 1185 user VM_USERMAP area 1186 vpages buffer for pages pointers was vmalloced (huge area) 1187 N<node>=nr (Only on NUMA kernels) 1188 Number of pages allocated on memory node <node> 1189 ========== =================================================== 1190 1191:: 1192 1193 > cat /proc/vmallocinfo 1194 0xffffc20000000000-0xffffc20000201000 2101248 alloc_large_system_hash+0x204 ... 1195 /0x2c0 pages=512 vmalloc N0=128 N1=128 N2=128 N3=128 1196 0xffffc20000201000-0xffffc20000302000 1052672 alloc_large_system_hash+0x204 ... 1197 /0x2c0 pages=256 vmalloc N0=64 N1=64 N2=64 N3=64 1198 0xffffc20000302000-0xffffc20000304000 8192 acpi_tb_verify_table+0x21/0x4f... 1199 phys=7fee8000 ioremap 1200 0xffffc20000304000-0xffffc20000307000 12288 acpi_tb_verify_table+0x21/0x4f... 1201 phys=7fee7000 ioremap 1202 0xffffc2000031d000-0xffffc2000031f000 8192 init_vdso_vars+0x112/0x210 1203 0xffffc2000031f000-0xffffc2000032b000 49152 cramfs_uncompress_init+0x2e ... 1204 /0x80 pages=11 vmalloc N0=3 N1=3 N2=2 N3=3 1205 0xffffc2000033a000-0xffffc2000033d000 12288 sys_swapon+0x640/0xac0 ... 1206 pages=2 vmalloc N1=2 1207 0xffffc20000347000-0xffffc2000034c000 20480 xt_alloc_table_info+0xfe ... 1208 /0x130 [x_tables] pages=4 vmalloc N0=4 1209 0xffffffffa0000000-0xffffffffa000f000 61440 sys_init_module+0xc27/0x1d00 ... 1210 pages=14 vmalloc N2=14 1211 0xffffffffa000f000-0xffffffffa0014000 20480 sys_init_module+0xc27/0x1d00 ... 1212 pages=4 vmalloc N1=4 1213 0xffffffffa0014000-0xffffffffa0017000 12288 sys_init_module+0xc27/0x1d00 ... 1214 pages=2 vmalloc N1=2 1215 0xffffffffa0017000-0xffffffffa0022000 45056 sys_init_module+0xc27/0x1d00 ... 1216 pages=10 vmalloc N0=10 1217 1218 1219softirqs 1220~~~~~~~~ 1221 1222Provides counts of softirq handlers serviced since boot time, for each CPU. 1223 1224:: 1225 1226 > cat /proc/softirqs 1227 CPU0 CPU1 CPU2 CPU3 1228 HI: 0 0 0 0 1229 TIMER: 27166 27120 27097 27034 1230 NET_TX: 0 0 0 17 1231 NET_RX: 42 0 0 39 1232 BLOCK: 0 0 107 1121 1233 TASKLET: 0 0 0 290 1234 SCHED: 27035 26983 26971 26746 1235 HRTIMER: 0 0 0 0 1236 RCU: 1678 1769 2178 2250 1237 12381.3 Networking info in /proc/net 1239-------------------------------- 1240 1241The subdirectory /proc/net follows the usual pattern. Table 1-8 shows the 1242additional values you get for IP version 6 if you configure the kernel to 1243support this. Table 1-9 lists the files and their meaning. 1244 1245 1246.. table:: Table 1-8: IPv6 info in /proc/net 1247 1248 ========== ===================================================== 1249 File Content 1250 ========== ===================================================== 1251 udp6 UDP sockets (IPv6) 1252 tcp6 TCP sockets (IPv6) 1253 raw6 Raw device statistics (IPv6) 1254 igmp6 IP multicast addresses, which this host joined (IPv6) 1255 if_inet6 List of IPv6 interface addresses 1256 ipv6_route Kernel routing table for IPv6 1257 rt6_stats Global IPv6 routing tables statistics 1258 sockstat6 Socket statistics (IPv6) 1259 snmp6 Snmp data (IPv6) 1260 ========== ===================================================== 1261 1262.. table:: Table 1-9: Network info in /proc/net 1263 1264 ============= ================================================================ 1265 File Content 1266 ============= ================================================================ 1267 arp Kernel ARP table 1268 dev network devices with statistics 1269 dev_mcast the Layer2 multicast groups a device is listening too 1270 (interface index, label, number of references, number of bound 1271 addresses). 1272 dev_stat network device status 1273 ip_fwchains Firewall chain linkage 1274 ip_fwnames Firewall chain names 1275 ip_masq Directory containing the masquerading tables 1276 ip_masquerade Major masquerading table 1277 netstat Network statistics 1278 raw raw device statistics 1279 route Kernel routing table 1280 rpc Directory containing rpc info 1281 rt_cache Routing cache 1282 snmp SNMP data 1283 sockstat Socket statistics 1284 tcp TCP sockets 1285 udp UDP sockets 1286 unix UNIX domain sockets 1287 wireless Wireless interface data (Wavelan etc) 1288 igmp IP multicast addresses, which this host joined 1289 psched Global packet scheduler parameters. 1290 netlink List of PF_NETLINK sockets 1291 ip_mr_vifs List of multicast virtual interfaces 1292 ip_mr_cache List of multicast routing cache 1293 ============= ================================================================ 1294 1295You can use this information to see which network devices are available in 1296your system and how much traffic was routed over those devices:: 1297 1298 > cat /proc/net/dev 1299 Inter-|Receive |[... 1300 face |bytes packets errs drop fifo frame compressed multicast|[... 1301 lo: 908188 5596 0 0 0 0 0 0 [... 1302 ppp0:15475140 20721 410 0 0 410 0 0 [... 1303 eth0: 614530 7085 0 0 0 0 0 1 [... 1304 1305 ...] Transmit 1306 ...] bytes packets errs drop fifo colls carrier compressed 1307 ...] 908188 5596 0 0 0 0 0 0 1308 ...] 1375103 17405 0 0 0 0 0 0 1309 ...] 1703981 5535 0 0 0 3 0 0 1310 1311In addition, each Channel Bond interface has its own directory. For 1312example, the bond0 device will have a directory called /proc/net/bond0/. 1313It will contain information that is specific to that bond, such as the 1314current slaves of the bond, the link status of the slaves, and how 1315many times the slaves link has failed. 1316 13171.4 SCSI info 1318------------- 1319 1320If you have a SCSI host adapter in your system, you'll find a subdirectory 1321named after the driver for this adapter in /proc/scsi. You'll also see a list 1322of all recognized SCSI devices in /proc/scsi:: 1323 1324 >cat /proc/scsi/scsi 1325 Attached devices: 1326 Host: scsi0 Channel: 00 Id: 00 Lun: 00 1327 Vendor: IBM Model: DGHS09U Rev: 03E0 1328 Type: Direct-Access ANSI SCSI revision: 03 1329 Host: scsi0 Channel: 00 Id: 06 Lun: 00 1330 Vendor: PIONEER Model: CD-ROM DR-U06S Rev: 1.04 1331 Type: CD-ROM ANSI SCSI revision: 02 1332 1333 1334The directory named after the driver has one file for each adapter found in 1335the system. These files contain information about the controller, including 1336the used IRQ and the IO address range. The amount of information shown is 1337dependent on the adapter you use. The example shows the output for an Adaptec 1338AHA-2940 SCSI adapter:: 1339 1340 > cat /proc/scsi/aic7xxx/0 1341 1342 Adaptec AIC7xxx driver version: 5.1.19/3.2.4 1343 Compile Options: 1344 TCQ Enabled By Default : Disabled 1345 AIC7XXX_PROC_STATS : Disabled 1346 AIC7XXX_RESET_DELAY : 5 1347 Adapter Configuration: 1348 SCSI Adapter: Adaptec AHA-294X Ultra SCSI host adapter 1349 Ultra Wide Controller 1350 PCI MMAPed I/O Base: 0xeb001000 1351 Adapter SEEPROM Config: SEEPROM found and used. 1352 Adaptec SCSI BIOS: Enabled 1353 IRQ: 10 1354 SCBs: Active 0, Max Active 2, 1355 Allocated 15, HW 16, Page 255 1356 Interrupts: 160328 1357 BIOS Control Word: 0x18b6 1358 Adapter Control Word: 0x005b 1359 Extended Translation: Enabled 1360 Disconnect Enable Flags: 0xffff 1361 Ultra Enable Flags: 0x0001 1362 Tag Queue Enable Flags: 0x0000 1363 Ordered Queue Tag Flags: 0x0000 1364 Default Tag Queue Depth: 8 1365 Tagged Queue By Device array for aic7xxx host instance 0: 1366 {255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255} 1367 Actual queue depth per device for aic7xxx host instance 0: 1368 {1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1} 1369 Statistics: 1370 (scsi0:0:0:0) 1371 Device using Wide/Sync transfers at 40.0 MByte/sec, offset 8 1372 Transinfo settings: current(12/8/1/0), goal(12/8/1/0), user(12/15/1/0) 1373 Total transfers 160151 (74577 reads and 85574 writes) 1374 (scsi0:0:6:0) 1375 Device using Narrow/Sync transfers at 5.0 MByte/sec, offset 15 1376 Transinfo settings: current(50/15/0/0), goal(50/15/0/0), user(50/15/0/0) 1377 Total transfers 0 (0 reads and 0 writes) 1378 1379 13801.5 Parallel port info in /proc/parport 1381--------------------------------------- 1382 1383The directory /proc/parport contains information about the parallel ports of 1384your system. It has one subdirectory for each port, named after the port 1385number (0,1,2,...). 1386 1387These directories contain the four files shown in Table 1-10. 1388 1389 1390.. table:: Table 1-10: Files in /proc/parport 1391 1392 ========= ==================================================================== 1393 File Content 1394 ========= ==================================================================== 1395 autoprobe Any IEEE-1284 device ID information that has been acquired. 1396 devices list of the device drivers using that port. A + will appear by the 1397 name of the device currently using the port (it might not appear 1398 against any). 1399 hardware Parallel port's base address, IRQ line and DMA channel. 1400 irq IRQ that parport is using for that port. This is in a separate 1401 file to allow you to alter it by writing a new value in (IRQ 1402 number or none). 1403 ========= ==================================================================== 1404 14051.6 TTY info in /proc/tty 1406------------------------- 1407 1408Information about the available and actually used tty's can be found in the 1409directory /proc/tty. You'll find entries for drivers and line disciplines in 1410this directory, as shown in Table 1-11. 1411 1412 1413.. table:: Table 1-11: Files in /proc/tty 1414 1415 ============= ============================================== 1416 File Content 1417 ============= ============================================== 1418 drivers list of drivers and their usage 1419 ldiscs registered line disciplines 1420 driver/serial usage statistic and status of single tty lines 1421 ============= ============================================== 1422 1423To see which tty's are currently in use, you can simply look into the file 1424/proc/tty/drivers:: 1425 1426 > cat /proc/tty/drivers 1427 pty_slave /dev/pts 136 0-255 pty:slave 1428 pty_master /dev/ptm 128 0-255 pty:master 1429 pty_slave /dev/ttyp 3 0-255 pty:slave 1430 pty_master /dev/pty 2 0-255 pty:master 1431 serial /dev/cua 5 64-67 serial:callout 1432 serial /dev/ttyS 4 64-67 serial 1433 /dev/tty0 /dev/tty0 4 0 system:vtmaster 1434 /dev/ptmx /dev/ptmx 5 2 system 1435 /dev/console /dev/console 5 1 system:console 1436 /dev/tty /dev/tty 5 0 system:/dev/tty 1437 unknown /dev/tty 4 1-63 console 1438 1439 14401.7 Miscellaneous kernel statistics in /proc/stat 1441------------------------------------------------- 1442 1443Various pieces of information about kernel activity are available in the 1444/proc/stat file. All of the numbers reported in this file are aggregates 1445since the system first booted. For a quick look, simply cat the file:: 1446 1447 > cat /proc/stat 1448 cpu 2255 34 2290 22625563 6290 127 456 0 0 0 1449 cpu0 1132 34 1441 11311718 3675 127 438 0 0 0 1450 cpu1 1123 0 849 11313845 2614 0 18 0 0 0 1451 intr 114930548 113199788 3 0 5 263 0 4 [... lots more numbers ...] 1452 ctxt 1990473 1453 btime 1062191376 1454 processes 2915 1455 procs_running 1 1456 procs_blocked 0 1457 softirq 183433 0 21755 12 39 1137 231 21459 2263 1458 1459The very first "cpu" line aggregates the numbers in all of the other "cpuN" 1460lines. These numbers identify the amount of time the CPU has spent performing 1461different kinds of work. Time units are in USER_HZ (typically hundredths of a 1462second). The meanings of the columns are as follows, from left to right: 1463 1464- user: normal processes executing in user mode 1465- nice: niced processes executing in user mode 1466- system: processes executing in kernel mode 1467- idle: twiddling thumbs 1468- iowait: In a word, iowait stands for waiting for I/O to complete. But there 1469 are several problems: 1470 1471 1. CPU will not wait for I/O to complete, iowait is the time that a task is 1472 waiting for I/O to complete. When CPU goes into idle state for 1473 outstanding task I/O, another task will be scheduled on this CPU. 1474 2. In a multi-core CPU, the task waiting for I/O to complete is not running 1475 on any CPU, so the iowait of each CPU is difficult to calculate. 1476 3. The value of iowait field in /proc/stat will decrease in certain 1477 conditions. 1478 1479 So, the iowait is not reliable by reading from /proc/stat. 1480- irq: servicing interrupts 1481- softirq: servicing softirqs 1482- steal: involuntary wait 1483- guest: running a normal guest 1484- guest_nice: running a niced guest 1485 1486The "intr" line gives counts of interrupts serviced since boot time, for each 1487of the possible system interrupts. The first column is the total of all 1488interrupts serviced including unnumbered architecture specific interrupts; 1489each subsequent column is the total for that particular numbered interrupt. 1490Unnumbered interrupts are not shown, only summed into the total. 1491 1492The "ctxt" line gives the total number of context switches across all CPUs. 1493 1494The "btime" line gives the time at which the system booted, in seconds since 1495the Unix epoch. 1496 1497The "processes" line gives the number of processes and threads created, which 1498includes (but is not limited to) those created by calls to the fork() and 1499clone() system calls. 1500 1501The "procs_running" line gives the total number of threads that are 1502running or ready to run (i.e., the total number of runnable threads). 1503 1504The "procs_blocked" line gives the number of processes currently blocked, 1505waiting for I/O to complete. 1506 1507The "softirq" line gives counts of softirqs serviced since boot time, for each 1508of the possible system softirqs. The first column is the total of all 1509softirqs serviced; each subsequent column is the total for that particular 1510softirq. 1511 1512 15131.8 Ext4 file system parameters 1514------------------------------- 1515 1516Information about mounted ext4 file systems can be found in 1517/proc/fs/ext4. Each mounted filesystem will have a directory in 1518/proc/fs/ext4 based on its device name (i.e., /proc/fs/ext4/hdc or 1519/proc/fs/ext4/dm-0). The files in each per-device directory are shown 1520in Table 1-12, below. 1521 1522.. table:: Table 1-12: Files in /proc/fs/ext4/<devname> 1523 1524 ============== ========================================================== 1525 File Content 1526 mb_groups details of multiblock allocator buddy cache of free blocks 1527 ============== ========================================================== 1528 15291.9 /proc/consoles 1530------------------- 1531Shows registered system console lines. 1532 1533To see which character device lines are currently used for the system console 1534/dev/console, you may simply look into the file /proc/consoles:: 1535 1536 > cat /proc/consoles 1537 tty0 -WU (ECp) 4:7 1538 ttyS0 -W- (Ep) 4:64 1539 1540The columns are: 1541 1542+--------------------+-------------------------------------------------------+ 1543| device | name of the device | 1544+====================+=======================================================+ 1545| operations | * R = can do read operations | 1546| | * W = can do write operations | 1547| | * U = can do unblank | 1548+--------------------+-------------------------------------------------------+ 1549| flags | * E = it is enabled | 1550| | * C = it is preferred console | 1551| | * B = it is primary boot console | 1552| | * p = it is used for printk buffer | 1553| | * b = it is not a TTY but a Braille device | 1554| | * a = it is safe to use when cpu is offline | 1555+--------------------+-------------------------------------------------------+ 1556| major:minor | major and minor number of the device separated by a | 1557| | colon | 1558+--------------------+-------------------------------------------------------+ 1559 1560Summary 1561------- 1562 1563The /proc file system serves information about the running system. It not only 1564allows access to process data but also allows you to request the kernel status 1565by reading files in the hierarchy. 1566 1567The directory structure of /proc reflects the types of information and makes 1568it easy, if not obvious, where to look for specific data. 1569 1570Chapter 2: Modifying System Parameters 1571====================================== 1572 1573In This Chapter 1574--------------- 1575 1576* Modifying kernel parameters by writing into files found in /proc/sys 1577* Exploring the files which modify certain parameters 1578* Review of the /proc/sys file tree 1579 1580------------------------------------------------------------------------------ 1581 1582A very interesting part of /proc is the directory /proc/sys. This is not only 1583a source of information, it also allows you to change parameters within the 1584kernel. Be very careful when attempting this. You can optimize your system, 1585but you can also cause it to crash. Never alter kernel parameters on a 1586production system. Set up a development machine and test to make sure that 1587everything works the way you want it to. You may have no alternative but to 1588reboot the machine once an error has been made. 1589 1590To change a value, simply echo the new value into the file. 1591You need to be root to do this. You can create your own boot script 1592to perform this every time your system boots. 1593 1594The files in /proc/sys can be used to fine tune and monitor miscellaneous and 1595general things in the operation of the Linux kernel. Since some of the files 1596can inadvertently disrupt your system, it is advisable to read both 1597documentation and source before actually making adjustments. In any case, be 1598very careful when writing to any of these files. The entries in /proc may 1599change slightly between the 2.1.* and the 2.2 kernel, so if there is any doubt 1600review the kernel documentation in the directory /usr/src/linux/Documentation. 1601This chapter is heavily based on the documentation included in the pre 2.2 1602kernels, and became part of it in version 2.2.1 of the Linux kernel. 1603 1604Please see: Documentation/admin-guide/sysctl/ directory for descriptions of these 1605entries. 1606 1607Summary 1608------- 1609 1610Certain aspects of kernel behavior can be modified at runtime, without the 1611need to recompile the kernel, or even to reboot the system. The files in the 1612/proc/sys tree can not only be read, but also modified. You can use the echo 1613command to write value into these files, thereby changing the default settings 1614of the kernel. 1615 1616 1617Chapter 3: Per-process Parameters 1618================================= 1619 16203.1 /proc/<pid>/oom_adj & /proc/<pid>/oom_score_adj- Adjust the oom-killer score 1621-------------------------------------------------------------------------------- 1622 1623These files can be used to adjust the badness heuristic used to select which 1624process gets killed in out of memory (oom) conditions. 1625 1626The badness heuristic assigns a value to each candidate task ranging from 0 1627(never kill) to 1000 (always kill) to determine which process is targeted. The 1628units are roughly a proportion along that range of allowed memory the process 1629may allocate from based on an estimation of its current memory and swap use. 1630For example, if a task is using all allowed memory, its badness score will be 16311000. If it is using half of its allowed memory, its score will be 500. 1632 1633The amount of "allowed" memory depends on the context in which the oom killer 1634was called. If it is due to the memory assigned to the allocating task's cpuset 1635being exhausted, the allowed memory represents the set of mems assigned to that 1636cpuset. If it is due to a mempolicy's node(s) being exhausted, the allowed 1637memory represents the set of mempolicy nodes. If it is due to a memory 1638limit (or swap limit) being reached, the allowed memory is that configured 1639limit. Finally, if it is due to the entire system being out of memory, the 1640allowed memory represents all allocatable resources. 1641 1642The value of /proc/<pid>/oom_score_adj is added to the badness score before it 1643is used to determine which task to kill. Acceptable values range from -1000 1644(OOM_SCORE_ADJ_MIN) to +1000 (OOM_SCORE_ADJ_MAX). This allows userspace to 1645polarize the preference for oom killing either by always preferring a certain 1646task or completely disabling it. The lowest possible value, -1000, is 1647equivalent to disabling oom killing entirely for that task since it will always 1648report a badness score of 0. 1649 1650Consequently, it is very simple for userspace to define the amount of memory to 1651consider for each task. Setting a /proc/<pid>/oom_score_adj value of +500, for 1652example, is roughly equivalent to allowing the remainder of tasks sharing the 1653same system, cpuset, mempolicy, or memory controller resources to use at least 165450% more memory. A value of -500, on the other hand, would be roughly 1655equivalent to discounting 50% of the task's allowed memory from being considered 1656as scoring against the task. 1657 1658For backwards compatibility with previous kernels, /proc/<pid>/oom_adj may also 1659be used to tune the badness score. Its acceptable values range from -16 1660(OOM_ADJUST_MIN) to +15 (OOM_ADJUST_MAX) and a special value of -17 1661(OOM_DISABLE) to disable oom killing entirely for that task. Its value is 1662scaled linearly with /proc/<pid>/oom_score_adj. 1663 1664The value of /proc/<pid>/oom_score_adj may be reduced no lower than the last 1665value set by a CAP_SYS_RESOURCE process. To reduce the value any lower 1666requires CAP_SYS_RESOURCE. 1667 1668 16693.2 /proc/<pid>/oom_score - Display current oom-killer score 1670------------------------------------------------------------- 1671 1672This file can be used to check the current score used by the oom-killer for 1673any given <pid>. Use it together with /proc/<pid>/oom_score_adj to tune which 1674process should be killed in an out-of-memory situation. 1675 1676Please note that the exported value includes oom_score_adj so it is 1677effectively in range [0,2000]. 1678 1679 16803.3 /proc/<pid>/io - Display the IO accounting fields 1681------------------------------------------------------- 1682 1683This file contains IO statistics for each running process. 1684 1685Example 1686~~~~~~~ 1687 1688:: 1689 1690 test:/tmp # dd if=/dev/zero of=/tmp/test.dat & 1691 [1] 3828 1692 1693 test:/tmp # cat /proc/3828/io 1694 rchar: 323934931 1695 wchar: 323929600 1696 syscr: 632687 1697 syscw: 632675 1698 read_bytes: 0 1699 write_bytes: 323932160 1700 cancelled_write_bytes: 0 1701 1702 1703Description 1704~~~~~~~~~~~ 1705 1706rchar 1707^^^^^ 1708 1709I/O counter: chars read 1710The number of bytes which this task has caused to be read from storage. This 1711is simply the sum of bytes which this process passed to read() and pread(). 1712It includes things like tty IO and it is unaffected by whether or not actual 1713physical disk IO was required (the read might have been satisfied from 1714pagecache). 1715 1716 1717wchar 1718^^^^^ 1719 1720I/O counter: chars written 1721The number of bytes which this task has caused, or shall cause to be written 1722to disk. Similar caveats apply here as with rchar. 1723 1724 1725syscr 1726^^^^^ 1727 1728I/O counter: read syscalls 1729Attempt to count the number of read I/O operations, i.e. syscalls like read() 1730and pread(). 1731 1732 1733syscw 1734^^^^^ 1735 1736I/O counter: write syscalls 1737Attempt to count the number of write I/O operations, i.e. syscalls like 1738write() and pwrite(). 1739 1740 1741read_bytes 1742^^^^^^^^^^ 1743 1744I/O counter: bytes read 1745Attempt to count the number of bytes which this process really did cause to 1746be fetched from the storage layer. Done at the submit_bio() level, so it is 1747accurate for block-backed filesystems. <please add status regarding NFS and 1748CIFS at a later time> 1749 1750 1751write_bytes 1752^^^^^^^^^^^ 1753 1754I/O counter: bytes written 1755Attempt to count the number of bytes which this process caused to be sent to 1756the storage layer. This is done at page-dirtying time. 1757 1758 1759cancelled_write_bytes 1760^^^^^^^^^^^^^^^^^^^^^ 1761 1762The big inaccuracy here is truncate. If a process writes 1MB to a file and 1763then deletes the file, it will in fact perform no writeout. But it will have 1764been accounted as having caused 1MB of write. 1765In other words: The number of bytes which this process caused to not happen, 1766by truncating pagecache. A task can cause "negative" IO too. If this task 1767truncates some dirty pagecache, some IO which another task has been accounted 1768for (in its write_bytes) will not be happening. We _could_ just subtract that 1769from the truncating task's write_bytes, but there is information loss in doing 1770that. 1771 1772 1773.. Note:: 1774 1775 At its current implementation state, this is a bit racy on 32-bit machines: 1776 if process A reads process B's /proc/pid/io while process B is updating one 1777 of those 64-bit counters, process A could see an intermediate result. 1778 1779 1780More information about this can be found within the taskstats documentation in 1781Documentation/accounting. 1782 17833.4 /proc/<pid>/coredump_filter - Core dump filtering settings 1784--------------------------------------------------------------- 1785When a process is dumped, all anonymous memory is written to a core file as 1786long as the size of the core file isn't limited. But sometimes we don't want 1787to dump some memory segments, for example, huge shared memory or DAX. 1788Conversely, sometimes we want to save file-backed memory segments into a core 1789file, not only the individual files. 1790 1791/proc/<pid>/coredump_filter allows you to customize which memory segments 1792will be dumped when the <pid> process is dumped. coredump_filter is a bitmask 1793of memory types. If a bit of the bitmask is set, memory segments of the 1794corresponding memory type are dumped, otherwise they are not dumped. 1795 1796The following 9 memory types are supported: 1797 1798 - (bit 0) anonymous private memory 1799 - (bit 1) anonymous shared memory 1800 - (bit 2) file-backed private memory 1801 - (bit 3) file-backed shared memory 1802 - (bit 4) ELF header pages in file-backed private memory areas (it is 1803 effective only if the bit 2 is cleared) 1804 - (bit 5) hugetlb private memory 1805 - (bit 6) hugetlb shared memory 1806 - (bit 7) DAX private memory 1807 - (bit 8) DAX shared memory 1808 1809 Note that MMIO pages such as frame buffer are never dumped and vDSO pages 1810 are always dumped regardless of the bitmask status. 1811 1812 Note that bits 0-4 don't affect hugetlb or DAX memory. hugetlb memory is 1813 only affected by bit 5-6, and DAX is only affected by bits 7-8. 1814 1815The default value of coredump_filter is 0x33; this means all anonymous memory 1816segments, ELF header pages and hugetlb private memory are dumped. 1817 1818If you don't want to dump all shared memory segments attached to pid 1234, 1819write 0x31 to the process's proc file:: 1820 1821 $ echo 0x31 > /proc/1234/coredump_filter 1822 1823When a new process is created, the process inherits the bitmask status from its 1824parent. It is useful to set up coredump_filter before the program runs. 1825For example:: 1826 1827 $ echo 0x7 > /proc/self/coredump_filter 1828 $ ./some_program 1829 18303.5 /proc/<pid>/mountinfo - Information about mounts 1831-------------------------------------------------------- 1832 1833This file contains lines of the form:: 1834 1835 36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue 1836 (1)(2)(3) (4) (5) (6) (n…m) (m+1)(m+2) (m+3) (m+4) 1837 1838 (1) mount ID: unique identifier of the mount (may be reused after umount) 1839 (2) parent ID: ID of parent (or of self for the top of the mount tree) 1840 (3) major:minor: value of st_dev for files on filesystem 1841 (4) root: root of the mount within the filesystem 1842 (5) mount point: mount point relative to the process's root 1843 (6) mount options: per mount options 1844 (n…m) optional fields: zero or more fields of the form "tag[:value]" 1845 (m+1) separator: marks the end of the optional fields 1846 (m+2) filesystem type: name of filesystem of the form "type[.subtype]" 1847 (m+3) mount source: filesystem specific information or "none" 1848 (m+4) super options: per super block options 1849 1850Parsers should ignore all unrecognised optional fields. Currently the 1851possible optional fields are: 1852 1853================ ============================================================== 1854shared:X mount is shared in peer group X 1855master:X mount is slave to peer group X 1856propagate_from:X mount is slave and receives propagation from peer group X [#]_ 1857unbindable mount is unbindable 1858================ ============================================================== 1859 1860.. [#] X is the closest dominant peer group under the process's root. If 1861 X is the immediate master of the mount, or if there's no dominant peer 1862 group under the same root, then only the "master:X" field is present 1863 and not the "propagate_from:X" field. 1864 1865For more information on mount propagation see: 1866 1867 Documentation/filesystems/sharedsubtree.rst 1868 1869 18703.6 /proc/<pid>/comm & /proc/<pid>/task/<tid>/comm 1871-------------------------------------------------------- 1872These files provide a method to access a task's comm value. It also allows for 1873a task to set its own or one of its thread siblings comm value. The comm value 1874is limited in size compared to the cmdline value, so writing anything longer 1875then the kernel's TASK_COMM_LEN (currently 16 chars) will result in a truncated 1876comm value. 1877 1878 18793.7 /proc/<pid>/task/<tid>/children - Information about task children 1880------------------------------------------------------------------------- 1881This file provides a fast way to retrieve first level children pids 1882of a task pointed by <pid>/<tid> pair. The format is a space separated 1883stream of pids. 1884 1885Note the "first level" here -- if a child has its own children they will 1886not be listed here; one needs to read /proc/<children-pid>/task/<tid>/children 1887to obtain the descendants. 1888 1889Since this interface is intended to be fast and cheap it doesn't 1890guarantee to provide precise results and some children might be 1891skipped, especially if they've exited right after we printed their 1892pids, so one needs to either stop or freeze processes being inspected 1893if precise results are needed. 1894 1895 18963.8 /proc/<pid>/fdinfo/<fd> - Information about opened file 1897--------------------------------------------------------------- 1898This file provides information associated with an opened file. The regular 1899files have at least four fields -- 'pos', 'flags', 'mnt_id' and 'ino'. 1900The 'pos' represents the current offset of the opened file in decimal 1901form [see lseek(2) for details], 'flags' denotes the octal O_xxx mask the 1902file has been created with [see open(2) for details] and 'mnt_id' represents 1903mount ID of the file system containing the opened file [see 3.5 1904/proc/<pid>/mountinfo for details]. 'ino' represents the inode number of 1905the file. 1906 1907A typical output is:: 1908 1909 pos: 0 1910 flags: 0100002 1911 mnt_id: 19 1912 ino: 63107 1913 1914All locks associated with a file descriptor are shown in its fdinfo too:: 1915 1916 lock: 1: FLOCK ADVISORY WRITE 359 00:13:11691 0 EOF 1917 1918The files such as eventfd, fsnotify, signalfd, epoll among the regular pos/flags 1919pair provide additional information particular to the objects they represent. 1920 1921Eventfd files 1922~~~~~~~~~~~~~ 1923 1924:: 1925 1926 pos: 0 1927 flags: 04002 1928 mnt_id: 9 1929 ino: 63107 1930 eventfd-count: 5a 1931 1932where 'eventfd-count' is hex value of a counter. 1933 1934Signalfd files 1935~~~~~~~~~~~~~~ 1936 1937:: 1938 1939 pos: 0 1940 flags: 04002 1941 mnt_id: 9 1942 ino: 63107 1943 sigmask: 0000000000000200 1944 1945where 'sigmask' is hex value of the signal mask associated 1946with a file. 1947 1948Epoll files 1949~~~~~~~~~~~ 1950 1951:: 1952 1953 pos: 0 1954 flags: 02 1955 mnt_id: 9 1956 ino: 63107 1957 tfd: 5 events: 1d data: ffffffffffffffff pos:0 ino:61af sdev:7 1958 1959where 'tfd' is a target file descriptor number in decimal form, 1960'events' is events mask being watched and the 'data' is data 1961associated with a target [see epoll(7) for more details]. 1962 1963The 'pos' is current offset of the target file in decimal form 1964[see lseek(2)], 'ino' and 'sdev' are inode and device numbers 1965where target file resides, all in hex format. 1966 1967Fsnotify files 1968~~~~~~~~~~~~~~ 1969For inotify files the format is the following:: 1970 1971 pos: 0 1972 flags: 02000000 1973 mnt_id: 9 1974 ino: 63107 1975 inotify wd:3 ino:9e7e sdev:800013 mask:800afce ignored_mask:0 fhandle-bytes:8 fhandle-type:1 f_handle:7e9e0000640d1b6d 1976 1977where 'wd' is a watch descriptor in decimal form, i.e. a target file 1978descriptor number, 'ino' and 'sdev' are inode and device where the 1979target file resides and the 'mask' is the mask of events, all in hex 1980form [see inotify(7) for more details]. 1981 1982If the kernel was built with exportfs support, the path to the target 1983file is encoded as a file handle. The file handle is provided by three 1984fields 'fhandle-bytes', 'fhandle-type' and 'f_handle', all in hex 1985format. 1986 1987If the kernel is built without exportfs support the file handle won't be 1988printed out. 1989 1990If there is no inotify mark attached yet the 'inotify' line will be omitted. 1991 1992For fanotify files the format is:: 1993 1994 pos: 0 1995 flags: 02 1996 mnt_id: 9 1997 ino: 63107 1998 fanotify flags:10 event-flags:0 1999 fanotify mnt_id:12 mflags:40 mask:38 ignored_mask:40000003 2000 fanotify ino:4f969 sdev:800013 mflags:0 mask:3b ignored_mask:40000000 fhandle-bytes:8 fhandle-type:1 f_handle:69f90400c275b5b4 2001 2002where fanotify 'flags' and 'event-flags' are values used in fanotify_init 2003call, 'mnt_id' is the mount point identifier, 'mflags' is the value of 2004flags associated with mark which are tracked separately from events 2005mask. 'ino' and 'sdev' are target inode and device, 'mask' is the events 2006mask and 'ignored_mask' is the mask of events which are to be ignored. 2007All are in hex format. Incorporation of 'mflags', 'mask' and 'ignored_mask' 2008provide information about flags and mask used in fanotify_mark 2009call [see fsnotify manpage for details]. 2010 2011While the first three lines are mandatory and always printed, the rest is 2012optional and may be omitted if no marks created yet. 2013 2014Timerfd files 2015~~~~~~~~~~~~~ 2016 2017:: 2018 2019 pos: 0 2020 flags: 02 2021 mnt_id: 9 2022 ino: 63107 2023 clockid: 0 2024 ticks: 0 2025 settime flags: 01 2026 it_value: (0, 49406829) 2027 it_interval: (1, 0) 2028 2029where 'clockid' is the clock type and 'ticks' is the number of the timer expirations 2030that have occurred [see timerfd_create(2) for details]. 'settime flags' are 2031flags in octal form been used to setup the timer [see timerfd_settime(2) for 2032details]. 'it_value' is remaining time until the timer expiration. 2033'it_interval' is the interval for the timer. Note the timer might be set up 2034with TIMER_ABSTIME option which will be shown in 'settime flags', but 'it_value' 2035still exhibits timer's remaining time. 2036 2037DMA Buffer files 2038~~~~~~~~~~~~~~~~ 2039 2040:: 2041 2042 pos: 0 2043 flags: 04002 2044 mnt_id: 9 2045 ino: 63107 2046 size: 32768 2047 count: 2 2048 exp_name: system-heap 2049 2050where 'size' is the size of the DMA buffer in bytes. 'count' is the file count of 2051the DMA buffer file. 'exp_name' is the name of the DMA buffer exporter. 2052 20533.9 /proc/<pid>/map_files - Information about memory mapped files 2054--------------------------------------------------------------------- 2055This directory contains symbolic links which represent memory mapped files 2056the process is maintaining. Example output:: 2057 2058 | lr-------- 1 root root 64 Jan 27 11:24 333c600000-333c620000 -> /usr/lib64/ld-2.18.so 2059 | lr-------- 1 root root 64 Jan 27 11:24 333c81f000-333c820000 -> /usr/lib64/ld-2.18.so 2060 | lr-------- 1 root root 64 Jan 27 11:24 333c820000-333c821000 -> /usr/lib64/ld-2.18.so 2061 | ... 2062 | lr-------- 1 root root 64 Jan 27 11:24 35d0421000-35d0422000 -> /usr/lib64/libselinux.so.1 2063 | lr-------- 1 root root 64 Jan 27 11:24 400000-41a000 -> /usr/bin/ls 2064 2065The name of a link represents the virtual memory bounds of a mapping, i.e. 2066vm_area_struct::vm_start-vm_area_struct::vm_end. 2067 2068The main purpose of the map_files is to retrieve a set of memory mapped 2069files in a fast way instead of parsing /proc/<pid>/maps or 2070/proc/<pid>/smaps, both of which contain many more records. At the same 2071time one can open(2) mappings from the listings of two processes and 2072comparing their inode numbers to figure out which anonymous memory areas 2073are actually shared. 2074 20753.10 /proc/<pid>/timerslack_ns - Task timerslack value 2076--------------------------------------------------------- 2077This file provides the value of the task's timerslack value in nanoseconds. 2078This value specifies an amount of time that normal timers may be deferred 2079in order to coalesce timers and avoid unnecessary wakeups. 2080 2081This allows a task's interactivity vs power consumption tradeoff to be 2082adjusted. 2083 2084Writing 0 to the file will set the task's timerslack to the default value. 2085 2086Valid values are from 0 - ULLONG_MAX 2087 2088An application setting the value must have PTRACE_MODE_ATTACH_FSCREDS level 2089permissions on the task specified to change its timerslack_ns value. 2090 20913.11 /proc/<pid>/patch_state - Livepatch patch operation state 2092----------------------------------------------------------------- 2093When CONFIG_LIVEPATCH is enabled, this file displays the value of the 2094patch state for the task. 2095 2096A value of '-1' indicates that no patch is in transition. 2097 2098A value of '0' indicates that a patch is in transition and the task is 2099unpatched. If the patch is being enabled, then the task hasn't been 2100patched yet. If the patch is being disabled, then the task has already 2101been unpatched. 2102 2103A value of '1' indicates that a patch is in transition and the task is 2104patched. If the patch is being enabled, then the task has already been 2105patched. If the patch is being disabled, then the task hasn't been 2106unpatched yet. 2107 21083.12 /proc/<pid>/arch_status - task architecture specific status 2109------------------------------------------------------------------- 2110When CONFIG_PROC_PID_ARCH_STATUS is enabled, this file displays the 2111architecture specific status of the task. 2112 2113Example 2114~~~~~~~ 2115 2116:: 2117 2118 $ cat /proc/6753/arch_status 2119 AVX512_elapsed_ms: 8 2120 2121Description 2122~~~~~~~~~~~ 2123 2124x86 specific entries 2125~~~~~~~~~~~~~~~~~~~~~ 2126 2127AVX512_elapsed_ms 2128^^^^^^^^^^^^^^^^^^ 2129 2130 If AVX512 is supported on the machine, this entry shows the milliseconds 2131 elapsed since the last time AVX512 usage was recorded. The recording 2132 happens on a best effort basis when a task is scheduled out. This means 2133 that the value depends on two factors: 2134 2135 1) The time which the task spent on the CPU without being scheduled 2136 out. With CPU isolation and a single runnable task this can take 2137 several seconds. 2138 2139 2) The time since the task was scheduled out last. Depending on the 2140 reason for being scheduled out (time slice exhausted, syscall ...) 2141 this can be arbitrary long time. 2142 2143 As a consequence the value cannot be considered precise and authoritative 2144 information. The application which uses this information has to be aware 2145 of the overall scenario on the system in order to determine whether a 2146 task is a real AVX512 user or not. Precise information can be obtained 2147 with performance counters. 2148 2149 A special value of '-1' indicates that no AVX512 usage was recorded, thus 2150 the task is unlikely an AVX512 user, but depends on the workload and the 2151 scheduling scenario, it also could be a false negative mentioned above. 2152 2153Chapter 4: Configuring procfs 2154============================= 2155 21564.1 Mount options 2157--------------------- 2158 2159The following mount options are supported: 2160 2161 ========= ======================================================== 2162 hidepid= Set /proc/<pid>/ access mode. 2163 gid= Set the group authorized to learn processes information. 2164 subset= Show only the specified subset of procfs. 2165 ========= ======================================================== 2166 2167hidepid=off or hidepid=0 means classic mode - everybody may access all 2168/proc/<pid>/ directories (default). 2169 2170hidepid=noaccess or hidepid=1 means users may not access any /proc/<pid>/ 2171directories but their own. Sensitive files like cmdline, sched*, status are now 2172protected against other users. This makes it impossible to learn whether any 2173user runs specific program (given the program doesn't reveal itself by its 2174behaviour). As an additional bonus, as /proc/<pid>/cmdline is unaccessible for 2175other users, poorly written programs passing sensitive information via program 2176arguments are now protected against local eavesdroppers. 2177 2178hidepid=invisible or hidepid=2 means hidepid=1 plus all /proc/<pid>/ will be 2179fully invisible to other users. It doesn't mean that it hides a fact whether a 2180process with a specific pid value exists (it can be learned by other means, e.g. 2181by "kill -0 $PID"), but it hides process' uid and gid, which may be learned by 2182stat()'ing /proc/<pid>/ otherwise. It greatly complicates an intruder's task of 2183gathering information about running processes, whether some daemon runs with 2184elevated privileges, whether other user runs some sensitive program, whether 2185other users run any program at all, etc. 2186 2187hidepid=ptraceable or hidepid=4 means that procfs should only contain 2188/proc/<pid>/ directories that the caller can ptrace. 2189 2190gid= defines a group authorized to learn processes information otherwise 2191prohibited by hidepid=. If you use some daemon like identd which needs to learn 2192information about processes information, just add identd to this group. 2193 2194subset=pid hides all top level files and directories in the procfs that 2195are not related to tasks. 2196 2197Chapter 5: Filesystem behavior 2198============================== 2199 2200Originally, before the advent of pid namepsace, procfs was a global file 2201system. It means that there was only one procfs instance in the system. 2202 2203When pid namespace was added, a separate procfs instance was mounted in 2204each pid namespace. So, procfs mount options are global among all 2205mountpoints within the same namespace:: 2206 2207 # grep ^proc /proc/mounts 2208 proc /proc proc rw,relatime,hidepid=2 0 0 2209 2210 # strace -e mount mount -o hidepid=1 -t proc proc /tmp/proc 2211 mount("proc", "/tmp/proc", "proc", 0, "hidepid=1") = 0 2212 +++ exited with 0 +++ 2213 2214 # grep ^proc /proc/mounts 2215 proc /proc proc rw,relatime,hidepid=2 0 0 2216 proc /tmp/proc proc rw,relatime,hidepid=2 0 0 2217 2218and only after remounting procfs mount options will change at all 2219mountpoints:: 2220 2221 # mount -o remount,hidepid=1 -t proc proc /tmp/proc 2222 2223 # grep ^proc /proc/mounts 2224 proc /proc proc rw,relatime,hidepid=1 0 0 2225 proc /tmp/proc proc rw,relatime,hidepid=1 0 0 2226 2227This behavior is different from the behavior of other filesystems. 2228 2229The new procfs behavior is more like other filesystems. Each procfs mount 2230creates a new procfs instance. Mount options affect own procfs instance. 2231It means that it became possible to have several procfs instances 2232displaying tasks with different filtering options in one pid namespace:: 2233 2234 # mount -o hidepid=invisible -t proc proc /proc 2235 # mount -o hidepid=noaccess -t proc proc /tmp/proc 2236 # grep ^proc /proc/mounts 2237 proc /proc proc rw,relatime,hidepid=invisible 0 0 2238 proc /tmp/proc proc rw,relatime,hidepid=noaccess 0 0 2239